FRI internal design vol 2

FRI VOLUME 5: FRACTIONATION DESIGN HANDBOOK

ORDERING AND SUPPLY OF PACKING

Issued:

10/30/1985

2.01.1

Revised:

ORDERING AND SUPPLY OF PACKING

ORDERING AND SUPPLY OF PACKING ................................................. ........................... 1 1.

Volume of Packing to be Purchased – Random Packing ................................................... 2

2.

Volume of Packing to be Purchased – Strucutred Packing ................................................ 2

3.

Method of Ordering ................................................... ................................................... ...... 2

Page 1 of 2

Issued: 10/30/1985 Revised:

1

ORDERING AND SUPPLY OF PACK ING

2.01.1

Volume of Packing to be Purchased – Random Packing

The process design will determine the packed bed height and diameter; the volume to be filled will therefore be known. Numerous factors have to be considered, considered, however, when translating translating that volume into a volume of packing to be received from the manufacturer. From a manufacturer's point of view, packing quantity is measured by the number of elements or pieces per unit volume of supply container. Machine counting generally ensures that the number of elements supplied per container remains constant, although the volume of those elements can change as a result of a number of factors, such as metal thickness variation, machine die conditions, machine adjustment, etc. The relationship between the volume of elements so produced and the amount required to fill a given column space is likewise likewise subject to numerous influences. influences. Many of these are virtually virtually impossible impossible to quantify to meet the wide variety of situations encountered encountered in commercial installations. installations. For instance, factors which affect this situation vary from the manufacturers' tests carried out to relate the number of elements to some standard volume, to the numerous factors (described in Section 3.1) which affect the number of elements finally placed into the column. In practice, therefore, the manufacturer's experience should be sought to enable a suitable volume of packing to be supplied which will adequately fill the space required. The manufacturer should also be informed of the intended method of filling the column since this may influence the quantity of packing required. If a purchaser wishes to allow an additional safety margin, for whatever reason, this should be clearly stated on the purchase order.

2

Volume of Packing to be Purchased – Strucutred Packing

With structured packing, the elements or grids are specifically manufactured to fit the geometric dimensions of the the bed concerned. Since there is negligible negligible compression of of those elements, adjustment adjustment factors as required for random packings are not required. The peripheral elements are cut to suit the column curvature, and therefore a knowledge of column out of roundness is desirable. desirable. If this is not available, available, tolerances will be assumed by by the manufacturer. Bed height will be an integer function of the number of elements or grids supplied, and will vary slightly between manufacturers. Possible manufacturing tolerances should be borne in mind where space is critical.

3

Method of Ordering

The recommended method of ordering random packing is to state the column diameter and the finally required bed depth. The filled volume only, without any supporting supporting geometric data, is not enough. In the absence of data to the contrary, a vendor will normally assume that metal packing will be dry packed and ceramic packing will be wet packed. Where doubt exists, a specific reference to the proposed method of installation of the packing should be made and the matter discussed with the vendor.

Page 2 of 2


INSTALLATION OF PACKING

Issued:

04/15/1985

2.01.2

Revised:

INSTALLATION OF PACKING

INSTALLATION OF PACKING ........................................................................... .................. 1 1.

Initial Receipt/Storage ................................................ ................................................ ........ 2

2.

Column Preparation ................................................... ........................................................ .......................................................... 2

3.

Packing Installation ........................................................ ........................................................................................................... ................................................... . 3 3.1

Random Packing (dry packed) ................................................ .................................. 4

3.2

Random Packing (wet packed) ................................................ .................................. 5

3.3

Structured Packing (including grid packing) ............................................... .............. 5

Page 1 of 5


1

INSTALLA TION OF PACKING

2.01.2

Initial Receipt/Storage

Packing is transported in bags, cartons, crates, etc. depending upon the particular manufacturer's policy. Large orders may be containerized, possibly with other column internals. An inventory check should be made as soon as possible after delivery and a record made of the number of bags or boxes supplied. Operational problems have been known to occur because of lack of control during filling. Note that receipt of the correct (i.e. as per purchase contract) amount of packing is not in itself a guarantee of an adequate volume to fill the column packed beds since different packing and filling procedures can lead to different packed bed densities. Modern packings, whether manufactured from ceramics, thin sheet metal or plastics, are vulnerable to impact loads. Disorganized storage can lead to scattering of packing elements over an appreciable area and consequently they can become permanently damaged. Fragmented storage can result in inventory loss. Storage facilities will be required for the period between delivery and installation. Preferably this storage will be covered and away from the immediate plant area. Note that carbon steel packing can rust in damp storage areas. Since the presence of iron oxide is normally deleterious to column operation, dry storage should be provided in such cases.

2

Column Preparation

Before installation of the internals, care should be taken to ensure that the vessel is clean and that pipes leading into the vessel are free of dirt, slag, and other debris which could plug distributors and coat, or plug, the packing. Facilities for hoisting the bags or boxes of packing to the working elevation, and from thence to the column, should be provided by the client or contractor, by agreement. The supply of scaffolding, lifting equipment and internal walking boards will normally be the responsibility of the contractor or subcontractor carrying out the work. Only that quantity of packing required for the bed being filled should be taken to the working area. In that way overfilling (which could occur if the bed was inadvertently compressed during filling), or underfilling (if, for example, previously filled beds had used unexpectedly greater amounts of packing) can be avoided. Packing support rings should be level and free of weld spatter. In revamp situations redundant tray support rings might need to be cut back; the necessity of so doing should be discussed with the supplier of the packing. Where the rings are cut back a projection of about 0.5 inch (10 mm) is frequently regarded as being acceptable. Downcomer bolting bars should also be cut back. Packing bed limiter lugs, where installed, should be checked to ensure the correct depth of packing can be installed. After installation of the bed support plate a continuous guide line should be drawn to indicate the depth of packing required unless this is clearly indicated by the bed limiter lugs or other support means. This guide line should be measured from the top side of the packing support plate. Packing should be installed through a suitable manhole or other means of access to the packed bed. Where appropriate, nozzles, including those with vortex breakers, should be shrouded to ensure that dropped packing elements and other debris do not enter the pipework.

Page 2 of 5


3


2.01.2

Packing Installation

Installation procedures will vary according to the packing type and the preference of different manufacturers. All will have the same objective: to ensure that the installation allows the design performance to be achieved. Note that with some structured packings it may be a condition of purchase that the manufacturer or his agents will also install the packing and associated equipment. Where packing is installed by the manufacturer, appropriate care and attention to detail will almost certainly be given. This cannot always be guaranteed where the work is contracted out. In such cases constant supervision by suitably knowledgeable technical staff is strongly recommended. An incorrect installation may otherwise go undetected and could be the cause of future poor performance. Care should be taken to minimize the ingress of dirt, debris and corrosive atmospheres during installation and before start up. It should be noted that, particularly with large columns (having a diameter greater than about 10-15 ft or 34.5 m), there can be several installation personnel inside the column spreading out packing elements. Whilst correct procedures should minimize the extent to which the elements are compressed, that possibility can never be ignored. If more compression occurs than was allowed for by either the supplier or the client, a greater volume of packing will be installed - assuming its availability - to meet the predetermined level. The pressure gradient, and efficiency, will be higher than anticipated. More importantly, however, is the probability that less packing will remain for subsequent beds in the column with the consequent possibility of delaying start up whilst additional packing is obtained, or of accepting a reduced bed depth. Random packing can be either dry packed or wet packed. With dry packing elements are placed directly into the column. With wet packing the column is filled with water to an appropriate level such that the water allows a more controlled deposition of elements. Wet packing has been largely superseded except when installing ceramic packing, in which case it is necessary to prevent attrition and breakage. Polypropylene and other plastic packings are of course, dry packed, because they float on water. The more controlled element deposition by wet packing results in a less tightly packed bed. The orientation of the elements consequent upon this controlled procedure is a function of packing type and shape, and if in doubt, the reader is strongly advised to consult the appropriate packing manufacturer before making a final decision. Note also that process considerations, or the type of packing material, which ultimately require a dry environment may specifically exclude the possibility of wet packing procedures. Whatever packing is used, wet packing is expected to give a less dense bed. For example, approximately 10% less elements have been reported in one experimental case using metal Pall rings (12). The reader should be aware that such a bed will be subject to normal settlement during operation, and that any benefits initially resulting from less dense packing may not necessarily continue to be obtained. Attention is drawn to the effect of low ambient temperature on the strength of polypropylene type packings, and the possibility of shattering when filling under such conditions. Where such an operation is scheduled for low ambient temperature conditions, advice should be sought from the packing manufacturer. Note that in all cases packing should be installed after the vessel is erected.

Page 3 of 5



2.01.2

The following procedures are recommended and should be considered in conjunction with any specific instructions supplied by the packing manufacturer. 3.1

Random Packing (dry packed)

1.

Where possible start from the lowest packed bed and work upwards. Where this is not practicable, or where beds at different levels are being filled simultaneously, adequate protection must be provided for those working below.

2.

Ensure the upper limit of the bed is clearly recognizable. This will usually be indicated by the bed limiter lugs, where used, welded on to the shell. Where this is not the case - for example when a hold down plate is used - a chalk mark should be drawn as indicated earlier.

3.

The first few feet of packing should be installed by lowering the box or container into the column and emptying from a height of not more than about 3 ft (900 mm). This is to prevent damage to the packing elements and consequent restriction of the apertures in the support plate. When a layer of about 2-3 ft (600-900 mm) has been built up, pouring from a higher level is acceptable providing that level does not exceed about 10 ft (3 M). Care should be taken to ensure that the elements are spread over the whole cross- sectional area of the column and that a reasonably level surface is maintained.

The use of a chute and filling sock is recommended in order to break the fall and to allow the elements to be efficiently distributed over the column cross-sectional area. During filling a reasonably even distribution of packing elements over the column cross-section should be maintained to avoid untypical element orientation and the need for unnecessary handling. 4.

Walking directly on the packing should be avoided since it can cause localized compression and damage. Where the column diameter is such that entry into the column is necessary, walking boards must be available and the supervising engineer should specifically ensure that they are used. The boards should be rigid and used in such a way that the packing elements are not damaged. The utmost care is needed to ensure that those walking boards are subsequently removed.

5.

The use of a pneumatic filing device is not recommended due to the possibility of packing element deformation.

6.

When the required volume of packing has been installed check that the surface is level and that the desired depth has been reached.

Where bed limiter lugs are installed ensure that the bed surface is such as to be level with the lugs. Additional packing should be added if it is necessary to obtain that level. Note that in severe cases this could result in an overall shortage of elements, whereupon additional packing may have to be acquired. This matter should be discussed with the packing supplier. Although packings can be expected to settle in operation, it is recommended that any gap that may be left between the surface of the packing and the bed limiter does not exceed approximately one packing diameter. 7.

At a suitable time during installation, and in any case upon completion, a count should be made to ensure that all boxes or bags have been removed from the column.

Page 4 of 5


3.2


2.01.2 2.01. 2

Random Packing (wet packed)

The above procedure applies equally for wet packing except that the various intermediate leveling procedures are not possible. Even distribution must therefore rely on the ability to discharge the packing uniformly over the cross-sectional area in the first instance and for this reason the maximum possible depth of water is recommended. recommended. A depth of at least least 4 ft (1200 mm) of of water above the packing should be maintained where possible. Final leveling can be accomplished in the manner described above after the water has been drained. 3.3

Structured Packing (including grid packing)

1.

Where possible start from the lowest packed bed and work upwards. Where this is not practicable, or where beds at different levels are being filled simultaneously, adequate protection must be provided provided for those working working below.

2.

If possible the individual layers of packing should be assembled on the ground adjacent to the column in order to ensure ensure that all sections are available. available. Individual elements should should be lifted to the working working manhole and and lowered into place. place. Above the first layer walking boards should be used although in some cases it may be permissible to stand directly on the packing when a certain number of elements have been installed.

3.

Different layers are normally installed such that they are orientated at a definite angle to the layer immediately below.

It is absolutely essential that this this procedure is adhered to throughout the the filling operation. This point should be specifically checked by the the supervising engineer. 4.

Peripheral elements should be in contact with the entire circumference of the column so that any small gaps occur randomly randomly within the bed. Such gaps may be filled with strips strips of packing.

Page 5 of 5


DISCHARGING OF PACKING

Issued:

01/30/1986

2.01.3

Revised:

Packing may need to be removed from the column for a number of reasons, such as vessel inspection, replacement of damaged packing, or to upgrade the packing installed. The ease of removal will depend largely on the type of packing used and the extent to which the packing elements have interlocked together, or have become bound together by coking or other material deposition. The mechanical state of the packing packing elements (for example, presence of sharp edges edges due to corrosion) may also influence the ease and safety by which the elements can be removed. The mode of removal will depend on on the state of the packing and on the volume volume to be removed. Small volumes are generally economically removed manually; suction devices are frequently considered for large volumes. Suction equipment can satisfactorily operate only when the elements are reasonably movable and where excessive coking or interlocking interlocking has not not occurred. Where the packing has sharp sharp edges, plastic flexible flexible hoses should not be used since these tend to become shredded. shredded. Metal pipes (e.g. aluminum) with suitable flexible connections should should be used. Packing removed by suction equipment equipment is liable to suffer damage, particularly if the elements are of the modern thin walled type. Segregation into damaged and non damaged elements is obviously a local decision and will will depend on the volumes being handled. In general it is not advisable to assume that all the packing can be re-loaded (where this is required), but provision of an appropriate quantity of new packing is recommended. Removal of structured packing elements should be discussed with the manufacturers.

Page 1 of 1


LIQUID DISTRIBUTION FOR PACKING

Issued:

10/25/2011

Revised:

11/11/2013

Section 2.02.1

LIQUID DISTRIBUTION FOR PACKING

LIQUID DISTRIBUTION FOR PACKING ..............................................................................................................1 1.0 Introduction ................ .................. ................. .................. .................. ................. .................... .................. ........3 2.0 Distribution Importance................. .................. .................. .................. .................. .................. .................. .......3 2.1 Packing Efficiency................ .................. .................. .................. .................. .................... .................. ..........3 2.2 Redistribution (Other Considerations)................ ................... .................. ................... ................. .................3 3.0 Distributor Design Process ................ ................... .................. ................... .................. ................. .................. ..4 3.1 Types of Distributors ................. .................. .................. .................. .................. .................. .................. .......4 3.1.1 Orifice Pan and Plate (Deck) Distributors ................ .................. ................... .................. ................. .....4 3.1.2 Trough Distributors .................. ................... .................. ................... .................. .................. .................7 3.1.3 Ladder Pipe Distributors.................. ................... .................. .................. ................... ................. .........11 11 3.1.4 Spray Distributors................. .................. .................. .................. .................. .................. ................. ....13 3.1.5 Specialty Distributors ................ .................. ................... .................. ................... ................. ...............16 16 3.2 Distribution Quality ................ ................. .................. ................. .................. ................... .................. .........17 3.2.1 Drip Point Density ................. .................. .................. .................. .................. ................. ................. ....17 3.2.2 Pattern orientation .................. ................... .................. ................... .................. ................. .................19 . 3.2.3 Edge distribution or zone distribution ................. .................. .................. ................... ................... ......19 19 3.2.4 Defining Adequate Distribution Qualities. ................. .................. .................. ................... ..................20 ..................20 3.2.5 Feed Conditions ................. .................. .................. .................. .................. .................. .................. ......23 3.2.6 Feed Variations.................. .................. .................. .................. .................. .................. .................. ......24 3.2.7 Liquid Quantity ................ ................. .................. .................. ................. ................... ................. .........24 3.2.8 Pressure and Pressure Drop Considerations .................. .................. .................. ................... ...............25 25 3.2.9 Special Distribution Considerations for Different Types of Packing .................. ................... .............25 .............25 3.3 Distributor Metering Devices ................ .................. .................. .................. .................. .................... .........26 26 3.3.1 Orifice Types .................. .................. .................. .................. .................. ................... ................. .........26 3.3.2 Circular Orifices (Holes) ................. .................. .................. ................... .................. .................. .........27 3.3.3 Equations for Circular Orifice Flow ................. .................. ................... .................. ................... .........29 29 3.3.4 Rectangular Slots .................. .................. .................. .................. .................. .................. .................. ...32 3.3.5 Triangular Notches ................. .................. ................... .................. ................... ................. .................33 . 3.3.6 Other Gravity-Flow Metering Devices .................. .................. .................. ................... ................. ......34 34 3.3.7 Pressure-Driven Metering Devices................ ................... .................. ................... .................. ............34 34 3.4 Auxiliary Distributor Features ................ .................. .................. .................. .................... ................. .........35 35 3.4.1 Flow Guides.................. .................. .................. .................. .................. ................. .................. ............35 3.4.2 Half-Tubes and Deflector Baffles.................. ................... .................. ................... .................. ............36 36 3.4.3 Flow Baffles and Splitters ................. ................... .................. .................. ................... .................. ......37 3.4.4 Directed tubes ................. .................. .................. .................. .................. .................. .................. .........40 3.4.5 Extended Tubes ................. .................. .................. .................. .................. .................. ................. .......40 3.4.6 Raised Tubes ................ .................. .................. .................. .................. .................. ................. ............40 3.5 Design Features to Handle Solids................... .................. .................. ................... ................... .................41 . 3.5.1 Plan for Solids .................. .................. .................. .................. .................. ................ .................. .........41 3.5.2 Nature and Type of Solids ................ .................. ................. .................. .................. ................... .........42 42 3.5.3 Orifice Size ................ .................. .................. .................. .................. .................. ................. ...............42 Page 1 of 67

Issued: 10/25/2011 Revised:11/11/2013

Liquid Distribution for Packings

Section 2.02.1

3.5.4 Feed Filtering.......................................................................................................................................42 3.5.5 Elevated Pour Points............................................................................................................................43 3.5.6 Notches ................................................................................................................................................ 44 3.5.7 Screens Over Holes, Slots or Predistributors.......................................................................................44 3.6 Foaming ...................................................................................................................................................... 45 3.6.1 Low Velocity Inlets .............................................................................................................................45 3.6.2 Increase Distributor Height .................................................................................................................45 3.6.3 Feed Pipe Submergence....................... ................................................................................................45 3.6.4 Types of Foam and Characteristics .....................................................................................................46 3.6.5 Anti-foam (With and Without) ...........................................................................................................46 3.7 Gas Risers (Chimneys) ...............................................................................................................................47 3.7.1 Separation of Vapor (Gas) and Liquid ................................................................................................47 3.7.2 Riser Hats ............................................................................................................................................47 3.7.3 Gas Riser Area .....................................................................................................................................48 3.7.4 Gas Riser Height.................................................................................................................................. 48 3.7.5 Gas Riser Pressure Drop and Velocity ................................................................................................49 3.7.6 Liquid Gradient between Risers ..........................................................................................................49 3.8 Other Design Considerations......................................................................................................................50 3.8.1 Height Between Packing and Distributor ............................................................................................50 3.8.2 Height Above the Distributor ..............................................................................................................51 3.8.3 Interferences from Bed Limiter ...........................................................................................................51 3.8.4 Predistribution Systems .......................................................................................................................51 3.8.5 Hole Size Considerations............................. ........................................................................................52 3.8.6 Overflow Protection ............................................................................................................................53 3.8.7 Trough Design .....................................................................................................................................53 3.8.8 Distributor Support ..............................................................................................................................54 3.8.9 Distributor Gasketing ..........................................................................................................................55 3.8.10 Distributors in Small-Scale Columns ..................................................................................................55 4.0 Distributor Testing and Installation ................................................................................................................55 4.1 General Overview of Water Testing Liquid Distributors ...........................................................................56 4.2 Desirable Test Stand Facilities ...................................................................................................................57 4.3 Orifice Distributor Test Criteria .................................................................................................................57 4.4 Spray Test Stand .........................................................................................................................................63 4.5 Testing Spray Distributors..........................................................................................................................63 4.6 Field Testing Distributors ...........................................................................................................................63 4.7 Leveling a Distributor.................................................................................................................................63 4.8 Other Distributor Installation Issues ...........................................................................................................64 References ................................................................................................................................................................ 66

Page 2 of 67

Issued: 10/25/2011 Revised:11/11/2013

1


Section 2.02.1

Introduction

Distribute: 1) To divide and give out in portions; 2) To spread or diffuse over an area. (Ref. The American Heritage Dictionary) Liquid distribution transfers the liquid between the inlet nozzle and a packed bed, or guides the liquid flow from one packed bed to another. The object is to evenly spread the liquid in a way that will optimize wetting the surface of the packing.

2

Distribution Importance

FRI, academia and industry have spent much time testing packings for capacity and efficiency. Packing efficiency varies with the type and quality of distribution. FRI and other independent testing have shown that both the number and location of distribution points is important. Various packings have different liquid spreading characteristics which affect the relative impact of distribution quality. Zuiderweg’s development of the Zone Stage Model has enhanced our understanding of distribution issues. (See TR-92) Some of our best knowledge comes from mistakes and experience. One of the most flawed assumptions that many process engineers make is that the supplier always adequately designs and optimizes the distributor to the same criteria that the process designer expects. An optimized distributor comes with proper attention to detail during all phases of design, testing and installation. The first and most important rule is to ASSUME NOTHING (21). Common sense, observation, and communication round out the top three rules. The best trouble-free distributor designs are produced through a partnership of vendor/designer and customer/inspector. And remember, you get what you inspect not what you expect!! 2.1

Packing Efficiency

Poor liquid distribution is by far the most common cause of unexpected poor separation efficiency in packed columns. Liquid distribution usually determines the difference between good and bad performance of tower packings. The potential impact is hard to overstate. In some cases, a high quality liquid distributor can more than double the theoretical stages from the same bed of packing equipped with a poor distributor. Liquid distributors for packed towers can be deceptively simple which makes their design prone to mistakes. Design errors can come from improper specifications (of the process requirements) and/or an inexperienced distributor designer. 2.2

Redistribution (Other Considerations)

Feeds and draws between beds as well as multiple bed zones require collection and redistribution of the liquid. A redistributor can be a simple pan orifice distributor between beds or a sophisticated distributor unit which collects internal reflux and mixes it with a feed or pumparound. To maximize performance, the liquid between beds should ideally be collected, mixed and redistributed. Between beds a collector tray may be used to mix and redirect liquid into a predistributor or distributor. Wall wipers may be used to collect liquid from the walls and direct it into the troughs or predistributor. Collector trays require more shell height than wall wipers, but also give better re-mixing of the liquid. Wall wipers are less expensive, but may require gasketing and may be more difficult to install. Page 3 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Column sections that require many theoretical stages may need to be split into multiple beds for either mechanical or efficiency reasons. Redistribution design is important to maintain the efficiency of each section of a column. In a 4in. (102mm) diameter column, a single point gives you 11.5 pts/ft² (124 pts/m 2). This may be adequate, if the tower is packed with an appropriate size packing, (see Small Scale Tower Section 3.07 subsection 2.1). In that case, and for smaller diameters, one can (arguably) skip the liquid distributor as long as the feed pipe is placed in the center of the column. However, if the tower is packed with very small (e.g. 6 or 10mm (0.24 to 0.39in.) packing, a single point discharge may be inadequate. Experience has shown that in some cases columns as small as 1 in. (25mm) in diameter can perform poorly with a single point distribution.

3

Distributor Design Process

Applying the proper distributor for any application requires understanding the process, plant operating philosophy, and the needs of the specific packing or type of packing. The following section addresses the questions and considerations necessary for specifying and checking distributor designs. 3.1

Types of Distributors

There are many types of distributors out in industry. The scope of this section will be to discuss four general types. The types are pan or plate, trough, pipe, and spray. A distributor can become a combination of a collector and a distributor, and then we call it a re-distributor. A re-distributor will have hats or covers that keeps liquid from the bed above from going through the open vapor chimneys. If the covers are only over the vapor chimneys then it is termed a partial collection redistributor. If there are covers over the open liquid area in the distributor as well then it is termed a total collection re-distributor. For more information on Gas Risers see section 3.7. Distributors also fall into two categories for liquid head on the distributor that drives the distribution and flow. The most common type is the “gravity head” distributor. Here liquid flows into an “open” head box or trough where there is a head of liquid established. This liquid head then produces the driving force for the liquid to flow through the distributor where it is distributed onto the packing. The second type is a pressurized piping system where the feed piping is directly connected to the distributor. This type of feed system is not as common and is more susceptible to maldistribution when used over a large operating range due to pressure drop and velocity variations inherent in closed pipes. See section 3.8 for more detail. 3.1.1

Orifice Pan and Plate (Deck) Distributors

These distributors can be equipped with either orifices in the bottom of the pan or plate or with orifices on the side of distribution tubes. Chimneys rising in a regular pattern between the orifices are typically placed for vapor flow and vapor distribution. The distributors may be designed with a single or multiple point inlets. A multipoint inlet may look like a pipe header designed to put liquid into different sections or locations in the distributor. This type distributor is used primarily with random dumped packing but can be used for structured packed towers.

Page 4 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Pan distributors are typically used in smaller diameter columns (approximately 36 in. (914mm) or less). Pan distributors can be manufactured in one or more pieces. The pans can be designed to be leveled on or from support tabs. Most pan type distributors have the holes placed in the floor of the pan. Some distributors just use a drip tube with no holes/slots and let the liquid overflow into the top of the tube.

For low flow with small holes or for fouling service the holes should be raised above the floor by placing them into the sides of distribution tubes. Pan distributors should be designed so that there is a uniform distribution of vapor flowing to the bed above. This is usually accomplished by risers carefully placed in the deck plus an allowance for flow around the outside of the pan. Pan distributors with holes in the bottom can be susceptible to fouling as well as out of levelness which can cause maldistribution. Minimum head requirements help with out of levelness (section 3.8). Fouling can be partially minimized or mitigated by larger holes sizes. When distribution tubes are used they improve the distributor fouling resistance as well as could allow for a larger operating range of the distributor (section 3.4) The biggest challenge with pan distributors is to ensure that the liquid is distributed uniformly across the entire cross section of the tower. Pans must be small enough in diameter to allow for installation and vapor flow past the pan. This can cause adequate liquid flow distribution near the wall of the vessel to be an issue. Figure 3.1.1 is an example of a pan re-distributor. Overflow notch and flow tubes

Distribution holes in sidewall

Riser caps for liquid collection

Distributor tubes with two levels of holes. Lug for leveling screw Tubes on triangular pitch

Figure 3.1.1 Pan Re-Distributor Courtesy of ExxonMobil

Page 5 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Orifice plate or deck distributors rest on a perimeter support ring and are dependent on the ring for levelness. The plate is sectioned to fit through the vessel manway and can be gasketed or welded. The holes are normally in the deck with vapor chimneys placed in a regular pattern between the orifices. For small towers the distributor can be placed between the shell flanges. A plate or deck distributor is not recommended for low liquid rate applications below 1 gpm/ft 2 (2.44m3/hr/m2). Leakage at the support ring and at each joint can be a significant problem at low flow rates if plate distributors are not seal welded or gasketed properly.

Orifice plate distributors having large liquid flow rates (25 gpm/ft 2 (61m3/hr/m2) or more), have been successfully used without welding or gasketing where leakage is a small percentage of the total liquid flow rate and therefore is not a major concern. Flow guides are sometimes installed on the periphery of the deck to guide the liquid under the support ring to eliminate large areas of un-wetted packing. A minimum elevation between plate deck and top of the packed bed below is recommended to be between 4 to 6 in. (102 to 152 mm), however, interference of support beams, bed limiter, etc may require more space. This type of distributor can be used in most columns 24in. (610 mm) diameter and larger.

Figure 3.1.1a Schematic of an orifice plate re-distributor with liquid collection hats Courtesy of Koch Glitsch

Figure 3.1.1b Orifice Plate re-distributor for a small diameter column Courtesy of The Dow Chemical Company

Page 6 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Gas riser cover for liquid collection Orifice deck Orifice holes for distribution

Figure 3.1.1c Orifice Plate re-distributor for a large diameter column Courtesy of The Dow Chemical Company

Figure 3.1.1d Plate re-distributor with guide pipes Courtesy of Raschig GMBH

3.1.2

Trough Distributors

The category of trough distributors has the most variety of distributor types. A series of individual troughs are typically evenly spaced across the cross section of the column. The troughs may be directly interconnected or totally separate, fed by single or multiple sets of pipes, or by a predistribution trough(s) or parting box (es). The troughs are leveled individually and/or collectively from wall supports and beams. The trough width depends on vapor and liquid flow requirements resulting in vapor open area from about 15 to 60% (section 3.7). Metering devices may be orifices, V-notches, rectangular slots, or tubes with various guides and flow splitters (section 3.4). Trough distributors can be designed to handle a wide range of flows from as low as 0.1 or as high as 60+ gpm/ft 2 (0.24 or 147 m 3/hr/m2). Trough distributors having large troughs can be applied to designs in which the liquid flows are large and the vapor flows are small. They can be standard distributors or redistributors, and are used with random dumped or structured packing. Page 7 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Trough distributors generally come with one of two means of distributing liquid to the individual troughs. The first is an integral feed trough system, and the second uses a separate feed box or parting box that is not directly connected to the distributor troughs. The integral feed trough is characterized by one or more main feed channels that are part of the trough system itself. The main channel is connected to the distributor troughs directly, and requires some kind of gasketing. The liquid in all the channels in the distributor are hydraulically connected. This type of distributor is used most often with large troughs where there is a large liquid flow (> 6 gpm/ft 2 (15 m3/hr/m2)). One advantage of this type is that the total distributor height is shorter so less space is required between beds of packing. This type of distributor can also be designed to be a re-distributor with partial or total liquid collection when used in combination with Ushaped hats and a wall wiper (section 3.7). This design is generally more costly than the parting box design, and is most often used in towers with diameters of 16 ft (4.9m) or less. An example of this type of distributor system can be seen in figure 3.1.2. For low liquid rates (< 2gpm/ft 2 (4.9 m3/hr/m2)), integral trough distributors should be avoided because the percentage of total liquid that might leak is high. An integral trough distributor may also be more difficult to level at installation. The trough distributor utilizing a separate feed box, usually referred to as a parting box, is characterized by one or more boxes located above the troughs in the distributor. This design requires a separate liquid collection system and more space between beds of packing. The liquid in all the channels in the distributor are not hydraulically connected and do not communicate with each other. This design is more cost effective than the integral feed channel design since it is a simpler design and requires less welding. The design is generally used in larger diameter towers (> 6 ft (1.8m) diameter). An example of this type of distributor system can be seen in figure 3.1.2b. This design is also used in towers requiring non-metallic materials such as FRP or plastic where fabrication can be an issue. A separate parting box(es) can be placed above the integral parting box trough type distributor to allow better overall distribution at higher liquid loads to account for uneven feed splits.

Wall wiper band for collection of liquid from tower wall

Liquid distribution trough

Integral liquid feed trough Figure 3.1.2 Wide trough re-distributor with integral trough and wall wiper with collection hats removed Courtesy of The Dow Chemical Company

Page 8 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Narrow trough distributors are preferred for applications having moderate liquid flows (typically about 2 to 15 gpm/ft 2 (4.9 to 37 m 3/hr/m2)) and a high quality distribution is required. When properly designed these distributors can have a very high distribution quality. This distributor can be used with either random dumped packing or structured packing. A narrow trough distributor is typically 3in. (76mm) or less in width.

Figure 3.1.2a Narrow trough distributor with feed troughs removed for clarity Courtesy of Exxon Mobil

Mixing box

Feed troughs or parting box for distributing liquid to each distributor trough

Distributor troughs

Figure 3.1.2b Narrow trough distributor with feed troughs and mixing box Courtesy of Sulzer Chemtech

Some trough distributors use V-notches or rectangular notches in the side of the troughs for flow distribution. This type is typically used in dirty services, as V-notches or rectangular notches are less prone to plugging. They are often used in absorbers and strippers where there is less sensitivity to maldistribution, and also where non-metallic materials such as plastic or FRP (fiberglass reinforced plastic) may be used. The units can handle large flow rates and a reasonable turndown.

Page 9 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

V-notch distributors have a very large operating range, however they are not recommended due to the susceptibility of out of levelness. A rectangular notch distributor is preferred. V-notch distributor flow is proportional to the 5/2 power of the head, while a rectangular notch distributor flow is proportional to the 3/2 power of the head. Occasionally a V-notch and rectangular notch will be used together to form a “Y-notch” distributor (See figure 3.1.2c). This type of trough distributor has a low quality distribution. These distributors are typically used with random dumped packing.

Figure 3.1.2c Trough distributor with Y-notches and parting box Courtesy of The Dow Chemical Company

Figure 3.1.2d Trough distributor with feed trough and guide channels Courtesy of Raschig GMBH

Page 10 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Figure 3.1.2e Trough distributor with V-notches and parting box Courtesy of Raschig GMBH

3.1.3

Ladder Pipe Distributors

Ladder Pipe Distributors are made of a series of pipes with holes drilled into the bottom of the pipes or may have down pipes for distribution. They are sometimes used in towers where low residence time is required, such as where polymerization can occur. They can also be used in cases where two liquid phases are present where higher velocities in the pipes create turbulence to keep the two liquid phases better mixed. They have been used in towers under motion such as in offshore applications. Each downpipe or hole represents one point of distribution. The combination of the distributor header and the individual distribution pipes must be designed so that the liquid is divided into equal flows. Metering orifices (sometimes referred to as restricting orifices) in each down pipe may be used to assure an even split of the liquid. Ladder distributors should not be used if solids are present due to small holes in the bottom of the pipes where fouling can occur. Ladder distributors should not be used in corrosive or erosive service unless the proper materials of construction are used. When gravity head ladder distributors are used they will require a surge pot of sufficient height to provide the head required to satisfy the maximum and minimum design flow rates through the distribution pipes or holes. Pressurized ladder distributors do not require a surge pot and can operate at a higher range of flows rates and pressure drops. However they are more susceptible to maldistribution when used over a large operating range due to the variation of pressure drop and velocities inherent in the pipes of this design.

Page 11 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Figure 3.1.3 Pipe type distributor using a gravity head box Courtesy of The Dow Chemical company

Figure 3.1.3a Plastic pipe-ladder distributor with a gravity head box Courtesy of The Dow Chemical company

Figure 3.1.3b Ladder pipe type distributor Courtesy of Raschig GMBH

Page 12 of 67

Issued: 10/25/2011 Revised:11/11/2013

3.1.4


Section 2.02.1

Spray Distributors

Spray distributors are constructed of a header pipe and laterals that supply liquid to a pattern of spray nozzles. The nozzles have a specific angle, typically with a solid cone spray pattern (liquid is sprayed throughout the inside of the cone). This does not mean that the liquid spray is uniform throughout. Full cone nozzles are required to optimize the coverage on the packing. The combination of number of nozzles, spray angle, and height above the packed bed will determine the pattern of liquid distribution coverage at the top of the packing. Distance to the bed below is set by the spray pattern required and the angle of the spray cone. Since the pattern is one of overlapping circles or square patterns, the distribution may include: dry locations, liquid sprayed on the wall, single or multiple coverage at the overlaps. Spray distributors are typically located 18 (457 mm) (2) to 36in. (914 mm) above a packed bed. The distribution of liquid across the spray pattern will vary with the design and quality of the nozzle, the pressure drop across the nozzle, and the elevation of the spray nozzle above the bed surface. Spray distributors may be specified as single coverage or higher average spray coverage (denoted as total spray coverage divided by column crosssectional area). The coverage level is set by how much wall flow is allowed, likelihood of nozzle pluggage, and whether the distributor is to be used for heat transfer service or separation service. Some spray distributors are designed for average spray coverage equal to a value of two or more. Since the overall distribution quality is typically quite poor (figure 3.1.4), spray distributors are normally limited to heat transfer zones and wash zones in crude vacuum towers. Spray distributors can also be useful in scrubbers where materials of construction are typically non-metallic and the spray has some scrubbing benefit as well as distribution. FRI Progress Report for May-June 1985 reports that single, wide-angle spray nozzle distributors were studied with a 12 ft (3.7 m) bed of 1 in. (25 mm) Pall Rings to determine the effects of nozzle size, nozzle height above packing, vapor loading and column pressure on the separation efficiency. The tests were conducted with the cyclohexane/normal-heptane system at 1 atmosphere, 24 psia (1.65 bar-a) and 30 psia (2.1 bar-a) in the FRI high-pressure 4 ft (1.2 m) diameter column. Three nozzle sizes were used: 1, 1.5, and 2 inches (25, 38, and 51 mm). The height of the nozzle over the packing was varied between 8 and 20 inches (76 and 508 mm) and three loadings were used at each condition. The separation efficiencies with the single spray distributors were much lower than with the FRI drip tube pan distributor. The cone shaped sprays appeared to lose their symmetry more at higher operating pressures, and the separation efficiency is impaired. The larger nozzles consistently resulted in higher separation efficiency at comparable operating conditions. The best efficiency was obtained when considerable amounts of spray wetted the column wall above the bed. Spray distributors have a relatively poor distribution quality, yet may be the best selection for heat transfer where the spray/vapor contact adds to the heat transfer. The quality of distribution and efficiency of spray distributors is dependent on proper design. See the FRI Progress Report for July-August 1985 for test results of a 7-nozzle distributor at various heights above 1in. Pall Rings.

Page 13 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

SPRAY PATTERN ON TOP OF PACKING FROM A 7-SPRAY DISTRIBUTOR KEY TO NUMBERED AREAS: 0 = NO LIQUID 1 = LIQUID FROM 1 SPRAY 2 = LIQUID FROM 2 SPRAYS OVERLAPPING 3 = LIQUID FROM 3 SPRAYS OVERLAPPING

Note that the density profile within circles and the quality of the profile is different with different nozzles. Figure 3.1.4 Spray pattern on packing

"Heat Transfer Performance of Large Structured Packing" by Tony Cai and J. Kunesh concluded the following: The heat transfer coefficient of the tested large size structured packing could not be determined since very little or no additional heat transfer took place in the packed bed with a totally condensing system or non-condensing gas injection. Spray nozzles are an effective direct heat transfer device. In a direct contact heat transfer application where liquid is heated by condensing a dew point vapor, only a simple bank of spray nozzles and a short vapor-liquid contacting space appears to be needed. The spray distributor has more open vapor passage area than any other distributor. This is most important in petroleum refining vacuum column design. The pressure drop across spray distributor nozzles makes them less sensitive to levelness. They require about 20 psi (1.38 bar) additional pump head compared to gravity distributors. However, depending on the nozzle pressure drop and vapor velocity conditions, spray distributors can generate a significant amount of entrainment (up to 25% or more).

Page 14 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Figure 3.1.4a Spray distributor header Courtesy of Sulzer Chemtech

Spray nozzles have a reasonable turndown or turn-up that is a function of the pressure drop through the nozzles. The limitation depends on the nozzle application. Higher pressure drop generates more fine particles of liquid. This could be a good thing for heat transfer or scrubbing. High nozzle pressure drop will result in entrainment of the fine particles in the vapor. Low pressure drop spray nozzles produce a coarser spray with less chance for entrainment. At very low pressure drop (typically < 3 psi (0.21 bar)) the sprays droop. Fouling or plugging can be a concern with spray nozzles. The number of sprays per square foot or square meter of the bed’s cross sectional area is much smaller than the number of drip points in a gravity distributor. This permits larger spray nozzle openings compared to gravity distributors, which reduces their plugging tendency. However, the torturous path inside the spray nozzle, which generates the spray cone, provides a multitude of low-velocity spots where solids can deposit and accumulate. There are “self-cleaning” or “pig tail” nozzles commercially available in which an external spiral that looks like a pig’s tail, replaces the internal tortuous path. These “pig tail” nozzles successfully eliminate the internal tortuous path that promotes solid deposition, but they require smaller hole diameters (typically about half of those in regular nozzles) to produce the same required pressure drop which counters their “self-cleaning” benefit. Filters, strainers, and / or screens should be used to minimize the fouling. The device should be located outside the tower. Consider multiple units for dirty service so cleaning can be made without shutting the system down. One also has to notice that if one spray nozzle fails, a significant cross sectional area of the column will not be wetted properly. It is recommended that all spray nozzle systems be flow tested inside and /or outside the towers with water (if possible) to check the coverage. Testing inside the tower is a final check of the flow distribution and mechanical integrity. A visual observation is usually sufficient, but for better understanding of the quality of the distribution, area samples can be collected to determine the flow variation. Also the visualization gives an idea of the particle size range to be expected. Different type nozzles will produce different droplet size distributions. Spray nozzle vendors may be able to provide drop size distribution data for spray nozzles and may have facilities to flow test spray nozzles with water to check coverage and drop size distributions. It is important to note that fluids with lower surface tensions and densities than water will produce different drop size distributions. Page 15 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

For low liquid flow systems, like wash sections in refinery vacuum towers, it is best to first perform a flow test outside the tower on a new system. This test is invaluable to check spray evenness, turndown, and coverage. A second test should be done inside the tower. Leakage from the spray header, plugging with construction debris, and installation damage will not show up in the outside test. The inside test should be repeated every turnaround as nozzles plug, shake themselves loose, corrode, and erode during service. For high flow systems, like pump-arounds, an outside test may not be practical due to the large quantity of water required. The inside test can still be performed by circulating water through the pump-around if practical. One way of testing nozzles outside the tower is by building a rig that can test one to three nozzles at a time. 3.1.5

Specialty Distributors

There are some distributors available that do not fall directly into the categories of distributors previously discussed. These distributors are for special applications. They may be for very low flow, very high flow, two liquid phase flow applications, or variations of distributors designed to handle fouling conditions. For this type application it is recommended that the engineer contact the distributor suppliers to work out the best design for their application. An example of some special distributors is seen below. See section 3.4 for more information.

Figure 3.1.5 MTS-109 Compartmental Distributor Courtesy of Sulzer Chemtech

Page 16 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Distribution fingers for distributing liquid from side holes in distributor

Bottom of distributor trough

Figure 3.1.5a Distributor with distribution fingers Courtesy of the Dow Chemical Company

3.2

Distribution Quality

Defining adequate distribution is dependent on the purpose of the packing system, the alternatives available for that service, and the economic cost/benefit analysis. One driving force is cost. Project budgets are often limited; therefore vendors may propose less expensive alternatives to increase the probability of winning the bid. Another driving force is to maximize the efficiency of the system. There are other drivers as well, for example fouling resistance.

  

Improved distribution has been shown by FRI tests to improve the “efficiency” of packing up to a point. FRI testing has also shown that for random packing pour (or drip) point density; there is a maximum beyond which no improvement is achieved. Especially in large towers, area distribution inequalities are more important than point-to point inequalities (6).

Proper specifications, correct design, quality of manufacture, hydraulic testing, and precise installation are all required to obtain good distribution quality. There are several different quality considerations, each of which has its own importance. 3.2.1

Drip Point Density

Drip or pour points per square foot (or square meter) define the density of the pattern of separate streams entering the top of the packing. The pattern density requirements are related to the size and type of packing. FRI testing has shown that random packings have a natural internal distribution pattern (1). The optimum number of drip points increases as the random packing size decreases. Drip point densities above this natural internal distribution pattern make no significant improvement in the packing performance. For Pall Rings, FRI (1) showed no improvement in performance at densities exceeding those shown in the Table below. Many member companies use drip point densities that are greater than those shown for Pall Rings in particular for more modern high performance packings. Industry practice beyond FRI test experience can be reviewed in the open literature such as Henry Kister's books (2,3) and various Frank Rukovena and F. D. Moore articles (4,5,6). Page 17 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Table 3.2.1 FRI Minimum Pour Point Densities for Pall Rings Pour Points per Sq. Foot Pour Points per Sq. Meter 5/8 Inch Pall Rings 9.3 100 1 Inch Pall Rings 5.6 60 1.5 Inch Pall Rings 3.7 40 2 Inch Pall Rings 3.3 35 3.5 Inch Pall Rings 2.8 30 9,10

For structured packing FRI’s tests with the Adjustable Liquid Distributor (ALD) showed deterioration of Mellapak 250Y efficiency when the number of drip points was reduced from 9.6 to 4.8 and 3.3 points per sqft (103 to 52 and 36 points per sqm). As a reasonable practice for structured packing use at least as many drip points as are recommended for a random packing of the same specific surface area. Only limited research has been done by FRI on this topic as summarized in the table below. This table may be of limited use because, for example, it does not address the issue of the arrangement of the pour points relative to the orientation of the structured packing layers. Table 3.2.2 FRI tests on the effect of pour point density on HETP for structured packing Packing Area of Reference Distributor Tested Pour Points 2 2 Packing points/ft (points/m ) 2 3 2 3 Ft /ft (m /m ) Montz B1TR 108 67.1 (220) TDP 9.8 , 4.9 (105,53) 200 (1990) Mellapak 78.9 (260 PR 5/1992 ALD 9.6 , 4.8 (103,52) 250Y

Mellapak 250Y

TDP ALD

78.9 (260)

PR 7/1992

ALD

9.6, 3.3, 2.4 (103,35,26)

Tubed Drip Pan distributor Adjustable Liquid Distributor

Since structured packing has multiple sheets that need to be wetted, trough distributors have been developed with various spreader baffles which lay down more of a line of liquid instead of a set of discreet points. When oriented at an angle to the sheets, all of the packing sheets get wetted by this design. It is very important to ensure that the orientation of the line of liquid not be in parallel with the line of the structured packing sheets. Reference 14 indicates that the optimum angle of the baffle line is between 60 and 90 degrees to the line of the structured packing sheets. This is particularly useful for low liquid rate distributors which cannot meet the higher pour point density due to small hole size restrictions. This spreader baffle type of distributor is not intended for use over a random packed bed. Other multiplier designs provide ways of increasing the effective pour point density for low rate systems. Industry practice has shown line and multiplier designs to be used quite frequently but these have not been tested at FRI to date.

Page 18 of 67

Issued: 10/25/2011 Revised:11/11/2013

3.2.2


Section 2.02.1

Pattern orientation

There are three main issues:



 

Pitch. Square or triangular pitch can be used for the orientation of the drip points. Triangular pitch gives the most uniform pattern across the bulk of the distribution area. However, square (or rectangular) pitch often times is the logical choice with troughs or chimneys. Square pitch is typically easier to engineer when dealing with troughs and chimneys. Towers work acceptably with both pitch types, however there appears to be no published data that show how the performance of the two pitch layouts compare. Drift. Changes in pattern with liquid flow rates. Certain types of distributors produce a variable location for the liquid impact on the top of the bed. Two examples are Notched-Trough and Spray distributors. Stability. Under some conditions the individual streams from a distributor may “wander” by clinging to the underside of the distributor or be pushed off vertical by vapor flow. Guide tubes that extend below the distributor help enhance the distribution stability.

Further discussion on drift and guide tubes is found in sections 3.4.1 and 3.8.1. 3.2.3

Edge distribution or zone distribution

It is important that pour point distribution be as uniform as possible over the cross section of the tower. For example, 10% of the pour points should be located in the outer 10% of the tower cross-sectional area. The cross sectional area should be able to be divided into several equal areas and the pour point count should be the same in each resulting area. This allocation of pour points should be uniform regardless of whether the cross sectional area is divided into equal "pie pieces" or equal annular areas (11).

Figure 3.2.3.1 Quadrant partitioning to check pour point counts

Page 19 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Figure 3.2.3.2 Concentric partitioning to check pour point counts

Even when a distributor has an appropriate drip point density, the quality of distribution can be poor. Unless care is taken, manufacturing and installation considerations may (7,8,9,10) yield a lack of irrigation around the perimeter of the column. FRI studies have shown that an under-irrigated periphery can seriously reduce the efficiency of a packed bed. Auxiliary pour points may need to be added to provide uniform distribution over the entire tower cross section. These auxiliary pour points may not "fit" the bulk drip point pitch. However, it is more important to provide uniformity of distribution than constancy of pitch (11). 3.2.4

Defining Adequate Distribution Qualities. 3.2.4.1

Coefficient of Variation.

Coefficient of Variation (Cv) is a dimensionless statistical analysis of the point-to-point rate (and/or area) comparison based on random selection and measurement of drip point rates or groups of drip points. The Cv measures a combination of quality of hole punching or drilling, effectiveness of predistribution boxes or pipes, liquid momentum effects, the effect of liquid head differences caused by high or low rates, velocities in troughs, and liquid heads above single or multiple holes. Please note that all the above sections establish design criteria and they are used to guide the engineer towards a distributor that OUGHT to work. The Cv criterion is a testing criterion – it tells you how well the distributor actually PERFORMS.

The coefficient of variation is a normalized measure of dispersion of a probability distribution. It is defined as the ratio of the standard deviation the mean μ:

c



 

Page 20 of 67

σ

to

Issued: 10/25/2011 Revised:11/11/2013


That is, the standard deviation of (X-μ)2.

Section 2.02.1

(sigma) is the square root of the average value

σ

In the case where X takes random values from a finite data set , with each value having the same probability, the standard deviation is





( x1   ) 2  ( x2   ) 2    ( x N   ) 2 N

,

or, using summation notation,





1

N

 ( x  N i

 )

2

,

i 1

(5,11,12,13)

Numerous people have studied the Cv and found it to be quite useful and effective for distributor flow tests. The section on flow testing (Section 2.02.4) will expand on the topic of how many points, or areas, are to be tested to determine the Cv. 3.2.4.2

Distribution Quality Index

The distributor quality index (D QI) is expressed as follows

4

:

DQI = 0.40 (100-A) + 0.60B -0.33 (C-7.5)

Where A = the percent of the cross sectional tower area not covered by drip point circles, % B = 100 * (the least point circle area in 1/12 tower area) / (tower area/12) Or, B = 100 * (tower area/12) / (the most point circle area in 1/12 tower area) Note: Use whichever B value is less. C = 100 * (area of overlap of point circles) / (tower area) See Figure 3.2.4.2 for example of A, B and C areas.

If the liquid hydraulic design will provide no more than 10% random (micro or small scale) liquid flow variation between drip points throughout, each point circle can be assumed to be of equal area to simplify the distribution quality index evaluation procedure. High performance distributors and redistributors are those which have a Page 21 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

distribution index of 90% or more. Many commercial standard distributors range from 10 to 70%. For new designs of traditional distributors, the distribution quality index should not be less than 75%. Note that in determining the index, the point where the liquid falls onto the packing should be used as the center of the circle and not necessarily where the liquid leaves the distributor. Sample calculations can be found in references 15 and 16.

Figure 3.2.4.2 Areas considered in liquid distribution quality rating A – Cross sectional tower area not covered by drip point circles B – Drip point circle area in 1/12 tower area C – Overlap area of point circles

3.2.4.3

Wetting Index

The distribution quality index (DQI) does not apply to structured packed towers because the use of circles to represent spreading of the liquid cannot work with standing sheets of structured packing. In addition, the DQI cannot be applied to in-line distributor vanes such as those frequently applied to structured packing beds. An alternative is the Wetting Index (WI) which is defined as the ratio of the wetted area to the tower cross section in a reference plane well below the top of the packed bed (14). To calculate the WI, the lateral spreading of liquid in the first structured packing element (layer) needs to be considered. In addition, the "twist angle" of the distributor with respect to the top packing element must be known. Liquid load can also affect the WI for in-line distributors because, at low loads, the wetted line may not be continuously wetted. The WI method shows that in-line distributors have the best contact with the top packing element when the distributor twist angle is between 60° and 90°. Spiegel’s work (14) explicitly recognizes what we’ve implicitly assumed but Page 22 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

technically ignored: that is, the top of a packed bed is really the final distribution element. Although his paper doesn’t say it, the distributor test stand should really include this layer of packing to determine if the liquid emerging from it is evenly distributed. This may never happen due to added cost of the flow test. As will be described in the following sections, the optimum distribution for any specific design is the summation of the effects of Feed Conditions, Feed Variations, Column Operating Conditions, Type of Packing, Mechanical and Space Limitations. 3.2.5

Feed Conditions

Certain column feeds have characteristics that require special design considerations. Solids must be considered in many or most distributor designs.  Many column shells and much piping are constructed of carbon steel. Corrosion produces iron scale which is collected by liquid distributors resulting in pluggage and poor distribution.  Solids may be a part of the process such as entrained catalyst from an upstream operation.  Solids may be generated as byproducts to the process through polymerization or other reactions.  Salts or precipitate may be deposited due to equilibrium conditions.  Coking deposits may occur due to high temperature and residence time.  Biological organisms may grow in the system Techniques used to handle solids include: larger and fewer holes (at the expense of hole density), V-notched and rectangular slots, raised holes, drip tubes, external filters or screens, or internal screens, trough covers, settling zones (for small quantities of solids), minimizing settling zones (for an abundance of solids) and changing operating conditions. There have been some devices used in industry that protect a hole from fouling, but do not eliminate the fouling material (see subsections of section 3.5). Foaming is a problem which may be obvious based on past experience, or subtle due to contaminants affecting a normally non-foaming system. (See section 3.6.) Flashing occurs when a portion of the feed to a column vaporizes upstream or at the point of discharge in the column. This can occur due to a pressure loss, such as through an orifice or control valve. It may also occur when two different temperature streams are mixed upstream of the tower. Flashing as little as 0.1 weight% of the total feed can substantially increase the volumetric flow rate of the feed stream and change the flow characteristics in the distributor itself.

This can be a significant problem if the distribution system is not designed to handle the flashing. Flashing is often encountered in multiple column systems which are at different pressures. Heat integration can also result in feed preheat and a significant quantity of vapor in a mixed phase feed which may best be separated by a pre-distribution system. Care should be taken when a feed is close to the bubble point. Flashing Feeds are beyond the Scope of this Practice. Two Liquid Phases may be present or occur with streams which contain entrained water or are saturated with water. Separation systems for extractions or various chemical systems may contain two phases by entrainment or equilibrium changes. Two liquid

Page 23 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

phase systems should generally be avoided when considering packing for a tower. However, if packing is to be used, then pressurized distributors are recommended for two liquid phase feeds. Corrosive liquids require special metallurgy considerations. Any corrosion to metering orifices in the distributor or pre-distributor will affect the capacity and distribution quality. If zero corrosion metallurgy is not possible, the distributor design must allow for corrosion effects, or consider the use of non-metallic materials. High velocity liquids may exacerbate erosive/corrosive effects. 3.2.6

Feed Variations

Compositional changes coupled with rate variations must be considered for the design capacity. The process engineer must provide the vendor with the full range of flow rates the distributor needs to handle during startup, normal operation, at peak rates, and at turndown. Failure to provide even one of these sets of rates can result in a design that does not meet expectations when it is put into operation. Mid-column feeds and pump-around returns will usually be joined by an internal reflux stream. Differences between the streams in temperature and composition will impact the distributor design. The more dissimilar the liquid streams are, the more important that the streams are properly mixed before making the final distribution. It can never be stated enough times that distributors will always limit the operating range of packing and that available head will always limit the operating range of distributors. 3.2.7

Liquid Quantity 2

3

2

Liquid Rates can be roughly classified as Low (<2 gpm/ft [4.9 m /hr/m ]), Moderate (22 3 2 2 3 2 20 gpm/ft [4.9-49 m /hr/m ]), and High (>20 gpm/ft [49 m /hr/m ]). Low rates have the difficult task of metering or dividing the liquid into very small equal streams. Since there is a practical limit on the minimum size of metering holes, a number of proprietary distributors have been developed for such low rates High rate distributors have the benefit of larger holes or slots, but must control liquid gradient and momentum effects created by high volumes and velocities. These distributors are much more difficult to design than ones for low liquid rates. It is possible to design distributors for flows less 2 3 2 than 0.5 gpm/ft (1.22 m /hr/m ), but they must have very specialized features. There is a practical limit on minimum hole size which is roughly 1.0 to 1.5 mm (0.039 in. to 0.059 in.) diameter. Column Diameter limits the use of some distributors. For example, pan distributors are more likely to be used in small columns than large columns. Structural considerations and vendor conventions will limit the size of some styles of distributors. Turndown, Turn-up and Normal total flow through each distributor must be defined. It’s important to specify a realistic turndown requirement and not an arbitrary value. The larger the turndown requirement the more complex and costly the distributor becomes. Standard distributors (with a single level of drip orifices) are easily designed for a 2:1 High to Low Turndown. For higher turndown ratios special designs will be required. Typically, distributor operating flexibility is limited by the height available above the packed bed in existing columns and vessel manway size may limit trough and parting Page 24 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

box height. The need for drain holes needs to be balanced against the need for good distribution quality. Conditions at startup can be drastically different than the minimum operating rates and must be addressed when specifying turndown conditions. 3.2.8

Pressure and Pressure Drop Considerations Vacuum columns are especially sensitive to pressure drop. Over the years, many of the vacuum column internals have progressed from trays to r andom packing to structured and grid packing to minimize pressure drop. Low L/V ratio fractionation and wash zones are particularly important. Therefore all vacuum column internals have to be designed for minimum pressure drop including the distributors. Pressure drop through a distributor is a function of open area, so narrow trough, and spray distributors (where efficiency is not critical) are preferred designs. Packed vacuum columns are often used where polymerization is a problem and fouling is a concern. Moderate Pressure columns (Atmospheric to about 50 psig [3.4 barg]) can also be sensitive to pressure drop. Revamps from trays to packing may be driven by savings that lower pressure drop, or increases in product yields (as in FCC and Coker Main Columns) provide or to provide higher suction pressures for higher capacity in overhead gas recovery compressors. Pressure columns above 100 psig [6.89 barg] often have high liquid rates leading to large trough area and that leads to low open area for vapor flow. The distributor design must consider the effects of pressure drop and vapor static head on the hydraulic head requirements. For higher system pressures with high density vapors and low surface tension liquids, re-entrainment of the distributed liquid can be a problem which overloads the distributor. Liquid flow guides directing the liquid as close to the packing as possible reduces entrainment.

3.2.9

Special Distribution Considerations for Different Types of Packing Random Packing does not care about the orientation of the distributor as long as the distribution points are equal in flow rate and evenly spread across the cross sectional area. However, bed limiters, or anti-migration screens, that don’t interfere with the liquid distribution must be used to keep the random packing pieces from moving into the distributor during operating or upset conditions. High-quality trough distributors are often not necessary on a bed of random packing because the natural distribution of the packing would degenerate the distribution quality developed by the distributor. Only when a small size random packing is used, a 1 inch size or less, should a high quality distributor be considered, see Table 3.2.1. Structured Packing is formed from a series of corrugated parallel sheets (metal, plastic or ceramic) or wire screen (such as BX or CY). Good distribution should have distribution points that are equal and evenly spread across the cross sectional area. Some designs spread the distribution from individual points into a sheet or line of liquid which can wet several of the packing sheets at a time. The lines should be oriented to wet as many of the corrugated sheets as possible. Woven Wire Gauze Packing has both a high open area and a very high surface area. The gauze will help spread the liquid as the liquid proceeds down the bed, however it will not correct large mal-distributions. There is also a high penalty to pay in lost efficiency due to poor distribution with these packings with their low HETP's.

Page 25 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Grid Packing is often used for wash, scrubbing or heat transfer zones which can operate at the extremes of liquid distribution. Wash zones usually have very low liquid loads and often are found in fouling services where small distribution holes or small orifice spray nozzles may become a fouling problem. The effects of entrainment must also be considered in wash and fractionation zones with low liquid loads. Heat transfer and some scrubbing zones have very high liquid rates which are often distributed with sprays or trough type distributors. 3.3

Distributor Metering Devices 3.3.1

Orifice Types

Conceptually, liquid distributors function by splitting the total flow of liquid that reaches them into smaller streams, and distributing those evenly across the surface of a bed of packing. By far, the most common means of achieving that split is gravity flow of liquid through orifices. This section describes the various types of orifices that are commonly used, their advantages and disadvantages, and recommended methods for evaluating their performance. The discussion of orifice types in this section is relevant both for flow through the orifices in the distributor pan or troughs, and also for any orifices used to pre-distribute liquid, such as from a parting box into other troughs. Imbalances in flow in the predistribution can lead to irregularities in the overall distribution pattern, no matter how high the quality of the distribution pattern below. Before diving too far down into the details, it may be useful to compare the form of the equations for gravity flow through the three major types of orifices encountered in liquid distributors: circular orifices (also called holes), rectangular notches (also called slots, or narrow slots), and triangular notches (also called V-notches). The equations for flow rate through an orifice are of the form:

q circular orifice

 Ao C o 2 g ho

qnarrow slot  C L ho q v  notch



C ho

2.5

1 .5

2g

2g

tan 

Where the terms represent the following, in a consistent set of units: q Volumetric flow rate ho Liquid height above the orifice g Acceleration due to gravity C, Co Orifice flow coefficients L Slot width Ao Orifice area Characteristic angle φ

Page 26 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

The specifics of the terms are discussed in some detail in later subsections. For now, let’s focus only on the liquid height above the orifice, h o. Note that the amount of flow through an orifice is always a function of the liquid height, but that the exponent to which that flow is raised differs greatly, depending on the orifice type. Flow through circular orifices is proportional to the 0.5 power. Flow through rectangular notches is proportional to the 1.5 power, and flow through triangular notches is proportional to the 2.5 power of liquid height. This sensitivity has implications on the choice of an orifice type for a given application. Circular orifices are less sensitive to slight variations in levelness of a distributor, since the difference in level has a smaller impact on flow rate. Thus circular orifices tend to give the most uniform distribution quality. FRI Progress Reports from late 1982 through 1983 show convincingly that this uniformity has direct consequences on the efficiency attained from the packed bed below the distributor. However, this same lack of sensitivity to level also means that, for a given flow rate, the height of a distributor using circular orifices must be greater than one using rectangular or V-notches. 3.3.2

Circular Orifices (Holes) Range of Good Operation: Another important term in the equation for flow through a circular orifice in Section 3.3.1 is the “orifice coefficient”, Co. It is important to note that is a coefficient, and not a constant. This reflects the reality that flow through an orifice is affected by things other than just the liquid height and the force of gravity.

Tests at FRI found that the coefficient becomes highly variable when the liquid head is too low. There is always flow instability in the low liquid head region, below about 25mm (1in) of liquid. This instability can persist in some circumstances even at heads up to 75mm (3in) of liquid. As a working guideline, it is good practice to insure that there is at least 40mm (1.6in) of liquid above orifices at any rates where you might reasonably expect to operate. If you need good efficiency from your packing at turndown conditions, then you should have at least that much liquid head at turndown rates. The only practical restriction on upper-end flow rates is the height of the distributor pan or troughs, which is often on the order of 300mm (12in) in high-efficiency designs. This height must be able to accommodate the liquid head above the orifice at maximum rates, any additional head associated with pressure drop of vapor going past the distributor, any splashing that might occur from introduction of feed, and appropriate freeboard (see section 3.6.2). For this sort of distributor height, the constraint of needing to have enough liquid head at turndown puts the typical turndown range for a distributor with circular orifices in the range of 2:1 to 2.5:1, if only a single tier of orifices is used. Bottom Orifices: The simplest application places the orifices in the bottom of pans or troughs. This approach has several advantages: it is relatively easy to fabricate, it is inherently self-draining, and the liquid exiting the orifices has a purely downward component of velocity, such that they are likely to land on the packed surface directly below them, rather than drifting horizontally.

Page 27 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

There are two significant disadvantages to bottom orifices. First, because they are on the bottom, they are susceptible to being plugged from any solids that might be present in the process. Second, because there is only one level of orifices (the bottom), it is not well suited to handle turndown ratios above about 2.5:1. Larger ratios result in very deep liquid levels at peak rates. Elevated Orifices: Orifices may be elevated above the bottom floor of distributors, most commonly either placing them on raised tubes, or on the sidewalls of troughs.

This design approach allows for small amounts of solids to settle out without plugging the orifices, as long as the solids are heavy enough to do so. Suspended solids can still cause plugging. Note that with elevated holes, some sort of drain hole or holes on the bottom of the distributor are still required, so care must be taken to insure that the size and location of the drain holes is such that they do not significantly interfere with the distribution pattern. The use of elevated orifices introduces an additional level of complexity into the fabrication process. If the holes are not all at the same elevation, it can result in substantial liquid maldistribution, particularly at turndown conditions. This is normally not too difficult to achieve with trough type distributors, since the holes can be punched before the troughs are fully assembled. For raised tubes, however, a great deal of attention to detail in manufacturing is needed to insure all the holes will be at the same elevation. Tubes or other baffles are commonly used to direct liquid flow from the elevated orifices to the packing. It is important to insure that the tube diameters are large enough that flow through them will be self-venting, with a Froude number less than 0.3. Failure to do this can result in unanticipated stack-up of liquid in the distributor. Normally this is less of a concern when holes are elevated by placing them in the side of distributor troughs, since the use of vented flow guides is common in that configuration. This is described in more detail in Section 2.4 of this Design Practice. Multiple Orifice Tiers: Raising the orifices optionally allows for one or more additional rows, or “tiers”, of holes to be installed. Careful choice of the size and position of the orifices can extend turndown range well beyond the typical 2.5:1 of a single tier, without requiring an excessively tall distributor.

The flexibility offered by multiple tiers comes with a downside. There is a range of flows, corresponding from when liquid reaches the level of the upper tier until it is at sufficiently above them, over which liquid distribution quality suffers.

Flow through the lower tier is still uniform, but flow through the upper tier is erratic. The relative flow through the two sets determines just how bad distribution quality is during the transition phase. If the amount of flow through the lower orifices is large relative to that through the upper orifices at the point of transition, the guideline for minimum liquid head above the upper orifices at anticipated operating rates is sometimes relaxed to 25mm (1in).

Page 28 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

When specifying flows for design of a liquid distributor that will have multiple tiers of orifices, it is important to specify not just maximum and minimum rates, but also typical operating rates. Liquid level must not be in the transition region at typical operating rates. This consideration requires much closer attention to the orifice coefficient calculation for multiple-tier distributors. For a single tier, it might make little difference if you designed for a normal level at 150mm (6in), and because of a slight error in the coefficient ended up at 125mm (5in) instead. But if that error put you in the transition region between orifice tiers, packing efficiency could be poorer at your normal operating rates. An example to illustrate this: Consider a 42in. (1.07m) diameter Scrubber with a normal operating liquid flow of 3 3 100,000 lb/hr (45,454kg/hr) at a density of 55lb/ft (881 kg/m ). Process conditions require that it must be able to operate with good efficiency at rates as low as 50,000 lb/hr (22,727kg/hr). Occasionally it must also operate with a peak liquid rate of 200,000 lb/hr (90,909kg/hr).

If we assume a distributor with 61 orifices, an orifice coefficient of 0.70, and a minimum liquid head of 1.5in. (38mm) above the orifice at low rates, then by the equation in section 3.3.3, each orifice would be 0.618in. (15.7mm) diameter. With only a single tier of orifices, at peak rates the liquid height above the orifice would be 24in. (610mm) – unreasonably high in most situations. Now suppose we kept that 0.618in. (15.7mm) orifice diameter for our lower tier, to maintain adequate head at turndown conditions, but added a second tier of orifices above it. There are many possible combinations of orifice diameter and orifice elevation that might work. For example, adding in a tier of 0.50in. (12.7mm) diameter orifices, at an elevation 3in. (76mm) above the lower tier, would reduce the total liquid depth to 10in. (254mm) above the lowest tier. At first glance this might look reasonable. However, at normal operating rates, the liquid level would be only 0.67in. (17.0mm) above the upper tier of orifices – inadequate to insure that the flow going through them (about 22% of the total in this case) would be uniformly distributed. A better solution might be to use a somewhat larger diameter second tier of orifices, and locate it somewhat higher above the first tier. An orifice diameter of 0.618in. (15.7) located 6.25in. (159mm) above the first tier would result in all the flow going through the lower tier at normal operating rates, and still give a reasonable liquid level of 9.5in. (241mm) at peak rates. In summary: the optimum balance between orifice diameter and elevation for a given application will depend on the specifics of the operating conditions for the application. 3.3.3

Equations for Circular Orifice Flow

The traditional equation for flow through circular orifices in the bottom of a plate, based on Torricelli, is: 0.5

V = C (2gh)

C = orifice coefficient, typically 0.5 - 0.8 Page 29 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

g = gravity acceleration h = height above the hole (centerline) V = Flow velocity Many rules of thumb and equations for orifice coefficients can be found in the literature, seldom in exact agreement with one another. There are proposals that a buoyancy term may need to be included in the above equation for high pressure applications. V = C (2gh)

(ρL – ρG) / ρL

where ρL and ρG are the liquid and gas densities respectively.

FRI, in conjunction with OSU, have studied typical relationships for orifice flow through distributor holes. For details on factors which can influence the orifice coefficient, see Section 3.3.3 below, or consult TR-115 “Vertical Orifice Flow in the Fully Turbulent Regime”, TR-118A “Side Orifice Flow”, and TR-122 “Downward Orifice Flow with Transverse Flow Velocity”. Summaries of the equations, and on their range of applicability, are presented below. Over the range of conditions studied in deriving them, these equations are recommended in preference to available open literature methods. Orifice Coefficient with Downward Discharge (TR-115):

When the Liquid Head (h L) is in inches, the correlation is: 0.18

C = [0.807 - 0.0414 ln(h L)] [0.93 + 0.0118h L

+ 0.102e

-0.129 (RDT - 3.0)^2.0

]

When the Liquid Head (h L) is in millimeters, the correlation is: 0.18

C = [0.941 - 0.0414 ln(h L)] [0.93 + 0.0066h L

+ 0.102e

-0.129 (RDT - 3.0)^2.0

]

Orifice Coefficient with Side Discharge (TR-118A): When the dimensions are in inches and h L is the liquid height, D o is the side orifice diameter, and To is the orifice plate thickness.

C

0.75 0.035 ln h L

1 0.027

0.5

0.078

D o

0.083

T o

2

0.083

IF: Do <0.5

C

0.75 0.035 ln h L

1.027 0.078

T o

0.083

2

0.083

IF: Do ≥ 0.5

For SI units dimensions are in millimeters:

C

0.863 0.035 ln h L

1 0.027

Page 30 of 67

12.7

D o

0.078

T o

2.108 2.108

2

Issued: 10/25/2011 Revised:11/11/2013

Liquid Distribution for Packings Packings

Section Section 2.02 2.02.1 .1

IF: Do <12.7

C

0.863 0.035 ln h L

1.027 0.078

T o

2.108

2

2.108

IF: Do ≥ 12.7 Orifice Coefficient with Side Discharge with Transverse Flow (TR-122):

The flow of liquid past past an orifice can affect flow through through the orifice. The effect of the transverse flow velocity U and the liquid head H on orifice discharge coefficient can be combined by a single dimensionless parameter, the Froude Number, which is defined as follows: Fr = = U/(gH)

0.5

2

2

Where g is gravity acceleration, which is 32.17 ft/s (9.81 m/s ). The correlation of orifice coefficient with (C f ) and without (C o) transverse flow is as follows: 2

2

-0.5

Cf /C /Co = 1.0 - [F r /(1.0+4.0Fr )](D/T) Where: D = Orifice diameter T = Orifice plate thickness

Limitations of equations: The equations used by FRI to represent the data above are empirical fits. Although the terms they they include terms that theory would would say are important, the form of the fits is is somewhat arbitrary. Accordingly, they may not not extrapolate well outside the range of conditions studied.

The range of conditions studied are: Orifice Diameter: 0.125in. (3.18 mm) - 0.75in. (19.0 mm) Plate Thickness: 0.075in. (1.90 mm) - 0.146in. (3.71 mm) D/T: 3 - 7 hL : 2in. (50.8 mm) – 30in. (762 mm) Reynolds Number: > 4500 Surface Tension: 22 dyne/cm - 72 dyne/cm Viscosity: 0.9 cP - 2.3 cP Note:

 

0.125in. (3.18 mm) orifice data is valid only for liquid head levels greater than 5in. (127 mm) because lower heads result in flows that are in the laminar flow regime. 0.125in. (3.18 mm) orifice data is not valid when used with 10 gauge (0.146in. or 3.71 mm) plate plate because the D/T < 1.

Page 31 of 67

Issued: 10/25/2011 Revised:11/11/2013



Another limitation with equations for side orifice flow, in particular, is that the endpoints of the two piecewise sections do not overlap. As the diameter increases, the correlation gives a sudden drop in the orifice coefficient at 0.5in. (12.7mm), where it changes definition. FRI staff has been made aware of this problem, but as of the writing of this section, there is not yet a formal resolution. Caution should be used when applying this correlation to design distributors with orifices that are within about 4.5in. (114 mm) of the trough wall or each other, as additional flow phenomena may apply. The orifice coefficients calculated by these correlations do not account for the any pressure drop across the distributor distributor due to upward flow flow of vapors around the distributor. distributor. Refer to section 3.3 for a method of calculating gas pressure drop. The edge shape of the orifice orifice affects the orifice discharge coefficient. coefficient. Guidance from (6) Tony Cai and Frank Rukovena of FRI suggests the following: Use punched orifices when when possible. Drilled orifices are not round round and the drill makes a ragged edge. If it is necessary to use drilled orifices, orifices, insure that any burrs are removed. removed. . For large distributors with high orifice count, it is important that the manufacturer maintain good quality control control on orifice size during punching punching Inspection after the fact, or a water test, can indicate if orifice uniformity is poor – but it is often too late to fix it at that point. Liquid flow should be in the punched direction, because the punch cuts a sharp edged when entering and then produces a torn shaped edge as it exits the metal sheet. Laser cutting can be an acceptable acceptable alternative to punching. punching. The benefit of laser cutting is is a very consistent orifice diameter diameter and edge quality. However, like punching, punching, the orifice discharge coefficient is different different between the laser entry and exit sides. sides. So the liquid flow direction and the laser pierce direction must be consistent for all components of the distributor. 3.3.4

Rectangular Slots

Rectangular slots discharging into tubes or trough sidewall are used in some distributors, particularly when the distributor distributor must be able able to operate at any point within within a wide range of flows. Because the flow is continuous over the range, they do not have the issue with poor distribution distribution quality during transition transition between tiers that circular orifices have. In high-liquid rate rate and fouling services, slots slots are relatively wide. The slots are located on the sides of troughs usually starting at the floor to allow particulates to flush through rather than accumulate. In non-fouling services, services, tall, narrow slots are more typical. typical.

Page 32 of 67

Issued: 10/25/2011 Revised:11/11/2013



Figure 3.3.4 Narrow Slot Tube Courtesy of Sulzer Chemtech

The recommended equation for flow through a narrow slot is that from Perry’s Chemical th Engineers’ Handbook (6 ed., p. 5-19, Eq. (5-32)): 1.5

q = 0.386 L ho

2g)0.5

Where q = volumetric flow rate L = width of slot ho = height of liquid above bottom of slot g = gravity acceleration It is important to insure that consistent units are used when applying applying this equation. For the purposes of applying this equation, “narrow slot” means that h o is greater than the slot width, L. Conceptually, it is likely that the orifice coefficient flow through a narrow slot is not a constant, just as is the the case for flow through orifices. orifices. However, there are no generally accepted criteria for minimum liquid level needed in a slot to insure uniform flow. 3.3.5

Triangular Notches

Triangular, or V, Notches are sometimes sometimes used in high rate and fouling fouling service. As was mentioned in Section 3.3.1, these types of metering devices are extremely sensitive to levelness of the distributor. distributor. In FRI tests, efficiency obtained from from packed beds below Vnotch distributors was considerably poorer than that of the same bed below a tubed drip pan distributor. The recommended equation for flow through a triangular notch is that from Perry’s th Chemical Engineers’ Handbook (6 ed., p. 5-19, Eq. (5-33)): 2.5

q = 0.31 ho

2g)0.5 / tan φ

Where q = volumetric flow rate ho = height of liquid above bottom of slot g = gravity acceleration φ = angle in degrees, measured up from a horizontal line drawn at the apex Experience of some members suggests that these sorts of orifices do not meter well at all. To have any hope of getting a reasonable distribution pattern, a drip guide is mandatory to make sure the liquid goes to the right spot. When you don't include the drip guides the surface tension tends to make the liquid stick to the trough and not distribute well.

Page 33 of 67

Issued: 10/25/2011 Revised:11/11/2013

3.3.6


Section 2.02.1

Other Gravity-Flow Metering Devices

The most common types of gravity flow metering devices are circular orifices, narrow slots and triangular notches, but many other types of openings are possible. One example is Y-notches, which, as illustrated in the figure 3.3.6 below, is a V-notch immediately above a rectangular notch.

Figure 3.3.6 Y Notched distributor (combined V and rectangular notches)

Another example is slots, either rectangular or triangular, in which the tip is curved, rather than sharp-edged. Figure 3.3.4 of a narrow-slot distributor illustrates this. Often these sorts of devices can be understood, at least approximately, through analogy to one of the more common types of orifices. Flow through even more unusual opening shapes can be determined by integrating the equation for the shape of the opening from the free surface to the weir crest, as follows: H

Qopening

 C d 2 g

 bh

0 .5

dh

0

Where Q = volumetric flow rate h = height of liquid H = height of liquid at the weir crest b = width of the slot opening at elevation h g = gravity acceleration Cd = orifice coefficient The orifice coefficient may need to be determined experimentally, if it is not known or cannot be inferred from that of similar orifices. 3.3.7

Pressure-Driven Metering Devices

Sometimes pressure in the feed pipe, rather than gravity, is used to drive flow through a distributor. Examples of these could include feed pipes that pre-distribute flow into the parting box of trough distributors, spray nozzles, and ladder type distributors with orifices in the bottom. Discussion of these devices is beyond the scope of this section.

Page 34 of 67

Issued: 10/25/2011 Revised:11/11/2013

3.4


Section 2.02.1

Auxiliary Distributor Features 3.4.1

Flow Guides

The general purpose of all types of flow guides is to direct the liquid close to the packing and make sure it goes exactly where desired. Placing the liquid close to the packing reduces the chance that the streams will drift significantly before they reach the packing, resulting in a distribution pattern other than what the designer intended. Flow guides can also be used to keep discharging liquid away from the bottom of the distributor (to prevent wicking) or from the entrance to a gas riser (to prevent re-entrainment). In addition to just preventing drift, tubes can be used to direct liquid to irrigate hard to reach locations necessary for uniform distribution, especially close to the wall or under a beam or truss. A liquid stream passing through a horizontally positioned orifice, such as in the wall of a trough distributor, initially has a purely horizontal component of velocity. This initial velocity exiting the orifice, v o, is a function can be estimated by the equation: vo = Cv * ( 2 * g * h )

0.5

Where h is the liquid height above the centerline of the orifice, g is acceleration due to gravity (9.81 m/s 2 (32.2 ft/s2), and Cv is the vena contracta orifice coefficient (0.98 for a sharp-edged orifice). Then the force of gravity acts on the stream, eventually bringing it to the surface of the packing. The time required for this to happen can be estimated from the equation: 0.5

t = ( 2 * y / g)

Where y is the vertical distance from the orifice to the surface of the packing. The horizontal distance, x, that a stream will drift before hitting the packed surface is: x = vo * t As discussed in section 3.3.2, a certain minimum liquid head is necessary to insure uniform flow through an orifice. Even with that minimum head, there will be some horizontal drift by the time gravity brings the stream down to the packing. As flow rate increases, the drift distance increases proportionally. As a result, the distribution pattern the liquid streams make as they impact the surface of the packing will change with operating rate. If the drift is large enough, the streams can impact neighboring troughs or streams discharging from neighboring troughs, resulting in further deterioration in distribution quality and poorer packing efficiency. Figures 3.4.1 and 3.4.1a show how the drip point locations relative to the top of the packed bed may change as the flow rate changes. Properly designed flow guides can kill the horizontal component of velocity and prevent the drift from occurring.

Page 35 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Stream Pattern at Full Rate Without V-Baffles Parting Bo x Distributor Troug hs

Packing Pattern at ful l rate. Distribution ch anges with rate. Possibil ity to hit other trough s, or wall, or to be redirected by vapor.

Figure 3.4.1 Courtesy of ExxonMobil

Figure 3.4.1a Courtesy of ExxonMobil

3.4.2

Half-Tubes and Deflector Baffles

Half tubes and various deflector baffles are attached to the outside of trough distributors to guide liquid from a side exit metering device. Half tubes may be of a semicircular, triangular or rectangular shape.

Page 36 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Half tubes can also function to channel liquid from an overflow notch or hole at the top of distributor troughs. This can help to minimize entrainment by keeping the liquid stream away from the rising vapor in the constricted space between troughs. Also, it can reduce the negative impacts of overflow on liquid distribution quality. In an overflow situation, there is seldom enough liquid head to provide uniform flow through the overflow notch. But when tubes are used to channel the overflow, the liquid is at least distributed in a roughly regular fashion onto the packed surface, rather than running randomly over the edge of troughs. The distributor shown in Figure 3.4.2 does not have half-tubes that extend up far enough to channel the liquid from the overflow notches to the desired irrigation point. This is not the preferred arrangement.

Figure 3.4.2 Typical rectangular trough with half tubes Courtesy of Nutter (Now Sulzer Chemtech)

Drip tubes extend below the floor of pan, plate or the bottom of trough distributors to channel liquid close to the packing surface. See section 3.8.1 Height Between Packing and Distributor for additional details. The drip tubes must be large enough for self-venting flow such that there is no interference with metering the liquid. Self venting flow in the drip tubes can be calculated using the FRI formula: d=0.918(Q)^0.4 d = inches, Q = hot gpm (FRI Vol 5 1.05) 3 d=1.115(Q)^0.4 d = meters, Q = hot m /s (FRI Vol 5 1.05)

3.4.3

Flow Baffles and Splitters

Another type of flow guide is one in which the splitting of streams happens in two steps. The distributor itself makes an initial, gross distribution split, and then those streams are further split or “multiplied” either by the flow guide, or in the case of structured packing by the packing itself.

Page 37 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

One advantage of this approach is that the initial split handles more flow. This allows for a larger orifice, which is more resistant to fouling. Getting that first, gross, liquid distribution split correct reduces the potential negative impact on performance if there is some irregularity in the distribution quality of the second, finer split. These types of flow guides fall into two broad types, linear flow or “splash plate” baffles, and flow multipliers using other geometries. Linear flow baffles are commonly used in trough distributors above structured packing. Orifices or other openings in the sides of the troughs discharge onto a linear baffle, which spreads the streams and results in a discharge pattern that is a line, or a close series of points, across the packed surface. If this line is positioned properly with respect to the surface of a bed of structured packing, then the individual sheets of packing themselves provide the tertiary “split” of the streams. This type of distributor is also referred to as a line distributor. It is often used to wet more sheets of structured packing than a traditional drip point distributor, especially for low liquid rates which limit pour point densities. It requires that the distributor be oriented properly with respect to the sheets of packing beneath to ensure optimal performance. Proprietary baffle designs exist with the intent to enhance liquid spreading and minimize entrainment of liquid from the baffles by the rising vapor stream. A conceptual sketch illustrating the operation of such a distributor is provided in Figure 3.4.3.

Figure 3.4.3 Courtesy of Koch-Glitsch (US patent 6,722,639)

A consequence of the multiplier effect of the spreading baffles is that the final drip point density will be higher than that of the primary orifices in the trough walls. It can be difficult to apply traditional guidelines for drip point densities to splash plate distributors. Multiplier effects in the range of 2-4 times the primary orifice density are often claimed by vendors. Linear flow baffles are not recommended for use above beds of random packing. More specialized flow multipliers/flow splitters may have a circular baffle or spider type which takes one metered discharge and divides it into several separate streams (see Figures 3.4.3a, 3.4.3b, 3.4.3c, and 3.4.3d). Figure 3.4.3d shows a square dush type flow (25,26) splitter which is located below a distributor flow guide . The designs are proprietary, but the principle is to increase the effective pour point density without increasing the number of holes and using holes too small in size to be practical. Page 38 of 67

Issued: 10/25/2011 Revised:11/11/2013


Linear Flow Baffle

Figure 3.4.3a

Section 2.02.1

Circular Baffle Flow Splitter

Figure 3.4.3b Courtesy of Sulzer Chemtech

Figure 3.4.3c Multiflow distributor with distribution fingers Courtesy of Raschig GMBH

Figure 3.4.3d Square dish flow splitter Courtesy of Koch Glitsch

Page 39 of 67

Issued: 10/25/2011 Revised:11/11/2013

3.4.4


Section 2.02.1

Directed tubes

The term “directed tubes” refers to flow guides that direct the flow somewhere other than straight down. So a liquid stream exiting from either the bottom or side orifice is diverted to a desired place on the packed surface. Directed tubes can be used to provide uniform liquid coverage across the surface area of a packed bed, even if beams or other internals block part of the area beneath the distributor. They can also be used to insure uniform irrigation of liquid around the periphery of the column, if using only drip points directly on the troughs would result in dry areas. 3.4.5

Extended Tubes

The term “extended tubes” refers to flow guides that extend from the bottom of a pan or trough floor instead of being attached to the side of a trough (see Figure 3.4.5). One justification that is sometimes used for extended tubes is to prevent liquid coming out of a horizontal orifice from adhering to the underside of the distributor, then running along for some distance before falling to the packing (the “wicking” effect). Another justification for extended tubes is that in some distributor designs, vapor flow has a significant horizontal component immediately below the deck of the distributor, as it moves towards the vapor risers. Further away from the deck, the horizontal component is less severe. Also, by moving the point at which the liquid stream starts closer to the packed surface reduces the time available for drift to occur. Extended tubes used for this purpose should generally extend below the bottom of distributors by at least 1in. (25 mm). If the vapor rates are relatively high, the tubes may be extended to 3in. (76 mm) to reduce the possibility of re-entrainment. Increasing the length of the tubes beyond 3in. (76 mm) poses a risk of damage during shipment. The tubes are not usually extended all the way to the top of the bed, especially in the case of structured packing. The installation tolerance for the height of the bed is such that damage to the tube or packing might occur if there is any interference. The tubes should be large enough for self-venting flow, so as not to interfere with the liquid exiting the hole in the floor or the raised tube above it. (See the formulas above.) 3.4.6

Raised Tubes

Raised tubes (also known as “drip tubes”) are the portion that extends upward above the floor of the distributor (trough or pan type). They are typically tubular in shape. They can have orifices or slots in the side, as shown in the sketch below. They can have holes, notches at the top or nothing at all. The top of the tube can be cut parallel with the floor or angled as shown in the sketch. Although not shown in Figure 3.4.5, when raised tubes are used on a redistributor, the openings at the top are often shielded from falling liquid in a variety of ways.

Page 40 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Figure 3.4.5 Slotted orifice drip tube Courtesy of Sulzer Chemtech

Raising the opening off the floor of the distributor can serve a number of purposes. A primary one is to provide additional fouling resistance and run time when solids are present that will settle out. Another is to extend turndown where the holes would otherwise be in the floor of a pantype distributor. They can also be used this way if the sides of the troughs are not being used to meter the flow of each drip point. A single hole in the side of the tube can be used with the top acting as a provision for overflow. This arrangement may be more stable over the entire operating range, but will not have the range of other options. Two or more holes vertically arranged may be used for a greater operating range. Some instability may be experienced as the liquid reaches the level of the upper holes since some may begin flowing before others do and the head above the upper holes may be inadequate at some rates. This effect may be minimized through careful leveling of the troughs. This is described in more detail in section 3.3.2 Circular Orifices (Holes). As pointed out in section 3.3.2 Circular Orifices (Holes), it is difficult to locate all the holes at the same elevation when using drip tubes since each tube is installed individually. And the deck may tend to warp when welding the tube in place. Slots may offer the greatest operating range. However, they will also have the highest sensitivity to out-of-levelness since the hole area will vary with the liquid level until all holes are covered. Vapor may flow up these raised tubes. Care should be taken in the design to be sure that the pressure drop across the distributor is less than the hydraulic head in the tubes to prevent the vapor from going up the tubes. 3.5

Design Features to Handle Solids 3.5.1

Plan for Solids

Solids can be present in almost any distillation system even though the system is considered clean. It is the degree of solids that requires specific designs to handle them. New processes or process changes that are piloted in lab scale equipment may not indicate the presence of solids because, in many cases, the lab scale equipment is glass such as Oldershaw columns. In these glass systems there may not be any corrosion products such as rust. In other systems that contain continuous recycles there could also be an accumulation of heavy ends that may become insoluble over time not observed in the piloting because it was not run long enough. Page 41 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

In known fouling services, it must be realized and communicated that any design to handle solids, even feed filtering, only mitigates the fouling. It usually does not stop it. The goal of the various mitigation strategies is to increase run time. Special considerations need to be considered for physical cleaning of the distributor. For example, there should be complete access to the troughs without removing the parting box (es). Please note that consideration should be given to use trays in those services that are fouling. 3.5.2

Nature and Type of Solids

The nature and type of solids must be understood if possible. The mitigation technique may be influenced by the type and mechanism of solids and solid formation. The nature of the solids includes density (do they float or sink?), particulate size, consistency (are they hard or soft or sticky?) and concentration. Types of solids include corrosion products (rust), polymers, dirt, heavy ends, biological growth, etc. These characteristics should also be discussed with the supplier as well. 3.5.3

Orifice Size

The best fouling mitigation technique, in a known fouling system, is to maximize the orifice size. Larger orifices are usually harder to plug and will last longer. In some cases, this may require a less than optimal number of drip points per unit area. Less drip points is usually better than no drip points if plugging occurs. If packing efficiency will be compromised somewhat, then a deeper packed depth or extra packed beds may be required. FRI’s guide to minimum drip points should be consulted. This will often be in direct conflict with a supplier’s approach to your distributor design. Communication with the supplier about your fouling concerns is an essential part of the design process. Two cases were reported in which packing efficiency in a fouling service was greatly (23,24) improved by going to larger holes at the expense of many fewer drip points . 3.5.4

Feed Filtering

The feed to a packed column (e.g. the reflux or pump around) should always be filtered even if there are no known fouling agents in the feed. Even in “clean” systems, debris left from construction can plug a liquid distributor. Filters must address both the size and quantity of particles. Large volumes of small particles can quickly collect in the distributor and packing.



If the distributor hole sizes or spray nozzle free passages are smaller than ½ inch in diameter, any stream that has particulates must be filtered outside the tower. Mesh opening size should be 1/10 or smaller than the size of the holes being protected.



The second element of the filter design is the solids handling capacity. It is helpful to know the percentage of particulates from sample analysis. The filter capacity must allow for a reasonable changeover and cleaning time. Dual filters with no external bypass should be the criteria, otherwise filters eventually will get bypassed and the distributors will foul. For many services that normally use carbon steel piping in even mildly corrosive conditions Page 42 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

consideration should be given to convert piping to higher alloy downstream of the filters to prevent line corrosion products from fouling the distributors. An in-line Y-strainer (Figure 3.5.4) may be placed at any accessible location in the feed or reflux line outside the tower to catch particulates and has the capability of being cleaned by blowing down material from the strainer while in operation.



Periodic inspection of the filter is highly recommended. Loss of pressure drop across the filter is a potential indication of filter “blow-out.” A mechanical engineer should carefully inspect installation and mounting of baskets inside the filters. There have been many cases in which the process liquid found a path for bypassing the basket or filter elements due to inadequate mechanical installation.

Figure 3.5.4 Typical Y-Strainer 3.5.5

Elevated Pour Points

The most common defense against heavy solids is to elevate the pour point above the bottom of the distributor. This design has several advantages. Typical height of the metering orifice above the bottom is 1 to 2 in. (25-51mm). The raised metering orifice results in a lower horizontal liquid velocity compared to holes located in the bottom of the distributor. The effect is a higher probability that particulates will settle to the bottom rather than plugging the holes. The lower horizontal velocity also reduces the horizontal velocity effect on the individual metering devices. Distributors with elevated pour points are usually in the form of drip tubes or trough distributors with holes on the side versus on the bottom. Figure 3.5.5 below illustrates the drip tube style distributor with a raised orifice and an angled overflow at the top of the tube. If a trough distributor was selected, a notch at the top of the trough above the opening is recommended.

Figure 3.5.5 Courtesy of Celanese

Page 43 of 67

Issued: 10/25/2011 Revised:11/11/2013

3.5.6


Section 2.02.1

Notches

V-Notches, Y-notches and rectangular slots are successfully used in high fouling rate services such as FCC Slurry pump-around distributors or vent scrubbers. Such designs minimize stagnant zones to keep the particulates moving and keep the trough swept clean. The notches can reach to the floor of the troughs and may be a combination of rectangular and V-Notches. The notched opening distributor typically discharges the liquid directly into the up-flowing vapor stream. The flow distribution is a function of the flow rate through the distributor. As the flow rate increases, the liquid stream is projected further away from the distributor trough. Flow guides can be employed to eliminate these concerns. Caution should be employed when pushing these type distributors to high capacity because when the flow is filling the wide part of the "V", the flow down the length of the trough is typically very high and the flow out each "V" is no longer uniform due to variance in the liquid momentum. Contingency overflow can be employed to the parting box similar to the troughs keeping in mind however, that the parting box needs varying opening sizes along its length due to the varying feed rates to each of the troughs.

Figure 3.5.6 Y-Notched distributor (combined V and rectangular notches)

3.5.7

Screens Over Holes, Slots or Predistributors

First of all, use of any type of screens inside the column must be selected with utmost caution. The use of this method is discouraged as your first "line of defense" since these screens cannot be cleaned on line. The designer has to answer the question “What happens when the screens plug?” They should be designed such that when they do plug, they do not make the distribution worse than if they were not there. Also can the screens be cleaned during turnaround or must they be replaced? The size of the openings in the screen must be related to the hole size being protected and the expected particulates. The screen openings should be no greater than 0.3 times the hole diameter. The total open area of the screen must be 100 times the hole area that is being protected. The location of the screen can be in upstream filters, predistributor boxes, liquid collector trays, entrance to laterals, at each hole, or a combination of locations.

Page 44 of 67

Issued: 10/25/2011 Revised:11/11/2013

3.6



Foaming

A system with foaming characteristics may affect liquid distributors for packings by increasing the height of the liquid in the parting boxes or troughs and thus lowering the distributor capacity before it overflows. Pan type distributors distributors may also have higher liquid liquid heights with the same same effect. However, the effect on the distributors distributors may be different than than what is experienced with trays. The suggestions that that follow may help to minimize minimize the impact of foaming or counter counter it completely. Note that the type foam that is is described here is an unstable unstable foam which breaks up more quickly quickly rather than a stable foam such as dish soap. 3.6.1

Low Velocity Inlets

Designing the distributor such that any agitation or splashing of the liquid is minimized may help reduce the tendency to initiate foaming. One way is to use momentumdiffusing devices to introduce introduce the liquid as gently as possible. This could be in the form of structured packing, random random packing, and/or perforated perforated plates. It may be placed in the parting boxes and in the the troughs. Care should be taken not to let let it rest on the parting box or trough floor as this may inhibit the free flow of liquid horizontally. One suggestion might be to minimize the distance the liquid must f all from the distributor before before reaching the top of the packing. packing. This has to be balanced against against the problems that could occur occur if the distributor is too too close to the top of the the bed (see subsection 3.8.1). Another is the use of flow guides (see subsection 3.4.1) to allow the liquid to exit the distributor as close close to the packing as possible. possible. Some have attached tubes of gauze gauze material to the bottom of the flow guides to both reduce the agitation and to limit the interaction with the upcoming vapor stream. stream. (This works well in glycol contactors where the liquid stream is very small and the vapor stream is large.) 3.6.2

Increase Distributor Height

A primary defense against the negative consequences of foaming in liquid distributors for packed beds is to provide provide extra freeboard height in the the parting box, the trough depth, depth, and in the case of pan-type distributors, distributors, the risers. This will help offset the the increased liquid (or froth) height in the distributor that may be encountered due to the foaming. 3.6.3

Feed Pipe Submergence

Submergence of feed pipes in a parting box is a good practice to minimize the volume of vapor injected or aspirated into the liquid head by by the feed stream. This may be considered for a downpipe from a chimney or other tray above the distributor for the same reasons. However, even a very small, continuous continuous input of vapor in an unvented unvented feed pipe will promote overflow whether the system tends to foam or not. Two-phase or all-vapor feeds should not be fed into parting boxes under any circumstances. This practice is a recipe for failure.

Page 45 of 67

Issued: 10/25/2011 Revised:11/11/2013

3.6.4



Types of Foam and Characteristics

Valuable information on the types and characteristics of foaming can be found in Volume 1 of the FRI Fractionation Tray Design Handbook, Section 5.2, Pages 13 through 16, where sieve tray flooding is discussed. Much of this information information can be applied to foaming in liquid distributors for packed beds. Foaming agents may actually enter the tower through a vapor feed and then accumulate in the tower itself or accumulate accumulate in a recycle loop. These gas streams could potentially potentially be treated externally (e.g. Gas Treating Treating tower where one should should knock out the hydrocarbon liquid or solid particles before it enters the tower).

Table 3.6.4 This table shows some ways of dealing with foam in liquid distributors for packed beds

Foaming Type A. Marangoni

27,28

a) b)

B. Partially miscible liquids

a) b) c)

C. Surface active agents

a) b) c)

D. Contaminated salt solutions

a) b)

E. Fine solids

a) b)

F. Immobile surfaces due to high molecular weight substances

a) b)

G. High viscosity liquids

a) b)

3.6.5

Design Defense Anti-foam agent Increase parting box and trough (or riser) height Increase temperature or pressure Increase parting box and trough (or riser) height Add carbon filters to the inlet remove small amounts of hydrocarbons e.g. Amine Service Anti-foam agent Remove agent, e.g. by adsorption or filtration Increase parting box and trough (or riser) height Anti-foam agent Increase parting box and trough (or riser) height Remove solids, e.g. by filtration Increase parting box and trough (or riser) height Remove, e.g. by adsorption Increase parting box and trough (or riser) height Lower viscosity Increase parting box and trough (or riser) height

Anti-foam (With and Without) As mentioned in the references cited in subsection 3.6.3, foaming can be caused by a number of things. Foaming can be reduced or eliminated eliminated in some cases by the addition addition of an anti-foam agent. Some processes may be adversely affected by the the addition of a “foreign material” (such as food-grade products to be used for human or animal consumption). It should be confirmed that the anti-foam agent would not not react with any of the other process chemicals or otherwise adversely affect the process, or any downstream uses of the main products. Page 46 of 67

Issued: 10/25/2011 Revised:11/11/2013



It is also possible to have have too much of a good thing. thing. Adding too much anti-foam anti-foam agent can lead to more foaming. Degradation of some anti-foam anti-foam agents can potentially form form solids. When foaming is caused by the presence of fine solids, water, or hydrocarbon contaminants, then filtration or adsorption may bring relief from foaming problems. If foaming will increase pressure drop and lower capacity, then increasing the packing size or going to a high capacity style of packing may solve solve the problem. This will typically reduce the efficiency unless the bed height is increased. 3.7

Gas Risers (Chimneys) 3.7.1

Separation of Vapor (Gas) and Liquid

An orifice pan or plate distributor can cover the whole cross section of the column. The vapor path in a very small column may simply go around the periphery of a pan type distributor. However most pan or plate distributors will require additional gas risers to channel the vapor past the liquid within the distributor distributor itself. The gas risers should be evenly distributed to minimize vapor cross flow across the top of the packed bed below without interfering with drip drip point pattern. Gas risers should also not restrict liquid movement across the distributor deck (see 3.7.6) Depending on the system and pressure of the column, the interrelationship between the gas risers and the liquid discharge points must be carefully considered to minimize the probability of liquid entrainment into and through the chimneys. Any entrainment will add recycled liquid to the distributor load and may exceed the distributor distributor capacity. With the exception of top feed or reflux distributors, hats should always be used on gas risers, since the redistributor also serves as liquid collector. Care should be taken to insure insure that the liquid overflow overflow from the hats does not interfere with the vapor exiting from the risers. Refer to additional considerations in section 3.8.1. Flow guides can be employed to minimize liquid entrainment. See section 3.4.1. 3.7.2

Riser Hats

Riser hats are designed to collect liquid falling from a packed bed above and redirect the liquid to a distributor or re-distributor. Proper design of the hats must prevent the liquid from bypassing the collector collector or being re-entrained by the the vapor. Economics has typically typically shown that rectangular rectangular risers are preferred over circular risers. The most common design for either circular or rectangular risers is to put edges on the hat and close a section between the hat and riser (chimney) (chimney) for liquid rundown. rundown. The hats may be flat or V shaped extending (typically 0.5-1 in. (12.7 to 25.4mm)) over the edges of the riser. Typical total open area between the chimney and hat is 100-150% of the riser cross-sectional area. The minimum clearance between hat and riser for a circular riser is riser diameter divided by 4. For a rectangular riser this minimum minimum is riser width divided by by two. Consideration should be given to the net horizontal area available between the hats. Riser hat size should be checked with respect to the tower diameter. The net flow area between the hats and tower tower shell should not restrict restrict the vapor flow, since since this will cause excessive vapor velocities tending to carry liquids that are falling off the hats back up to the packing above. When the excessive vapor velocity between hats becomes a concern, it may be necessary to reduce the the riser area. This increases the pressure drop drop through the risers and provides more area outside the hats. Caution should be used when reducing Page 47 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

riser area so as to not create an excessive amount of pressure drop. A check of the CFactor between the hats can be made and if found to be in excess of the system limit C(22) Factor . The area between the chimney hats should be increased. Figure 3.7.2 shows a typical configuration of square riser and hat. The liquid height on riser hats should be checked to avoid the liquid’s overflowing the side edges. A conservative and simple approach is to assume that the total area of all the hats receives all the raining-down liquid evenly. When the liquid overflow is concerned, the height of side edges should be increased, or alternatively the V-hat (Figure 3.7.2a) or the slopedhat (Figure 3.7.2b) should be used. Please note that the liquid overflow sides must be blanked to prevent liquid falling into riser and interference between vapor and liquid flows

Figure 3.7.2 Riser with flat hat

3.7.3

3.7.2a Riser with V-hat

Figure 3.7.2b Riser with sloped hat

Gas Riser Area

Gas riser area for pan or plate distributors is typically 10-25% of the vessel cross sectional area. For comparison, the equivalent open area of a trough distributor may be 25-75%. Open area should always be maximized to reduce liquid entrainment by lowering the vapor velocity out of the gas riser. However, this needs to be balanced by liquid gradient across the deck or trough, orifice pitch layout, constructability of the deck plates or troughs, and pressure drop. Care should be taken when long risers are designed. Liquid gradient and mixing should be considered when designing riser layout e.g. wall-to-wall risers. Rectangular risers should be aligned with the long dimension parallel to the liquid flow direction away from the feed point. Also the effect of the gas risers on the vapor distribution into the bed above needs to be considered. 3.7.4

Gas Riser Height

Gas riser height is determined by available space and the required maximum-design-rate height of liquid above the metering device, plus an additional 2-6 in. (51-152mm) of freeboard. Foaming and high-pressure systems may require as much as 12in. (305mm) of freeboard, see Section 3.6. Keep in mind that extra liquid height may require additional mechanical strength to support any additional liquid. Refer to Section 3.8.1 and 3.8.2 for discussions on the distance between hat and the bed above.

Page 48 of 67

Issued: 10/25/2011 Revised:11/11/2013

3.7.5


Section 2.02.1

Gas Riser Pressure Drop and Velocity

For the gas riser area in each distributor or redistributor, pressure drop must be calculated and considered as the dynamic head part of the total head on the metering device. The higher the vapor velocity and density, and the lower the percent open area that the gas risers represent, the more likely that the pressure drop will be significant. The vapor pressure drop through a riser is due to 1.) entrance loss (0.5 velocity head); 2.) friction in the straight section of riser (negligible); 3.) change of direction (1.0 velocity head) and 4.) exit loss (1.0 velocity head). The total gas riser pressure drop can be calculated with the following formula 2 DP = 2.5 V V / 334.76 (inches of Water) ∆h = DP/SGL

(1,2)

:

(inches of liquid)

Where V is velocity in feet per second, V is vapor density in pounds per cubic foot, and DP is the pressure drop in inches of water. SG L is specific gravity of the hot process liquid. The formula can be expressed in SI unit as shown below: 2 DP = 1.25 V V (Pa) ∆h = 0.102 DP / SG L

(mm of liquid)

Where V is velocity in meter per second, V is vapor density in kg per cubic meter, and DP is the pressure drop in Pa. SG L is specific gravity of the hot process liquid. This formula can be used to calculate the extra head of liquid on the distributor and may not be accurate enough for deep vacuum tower pressure drop contribution to the total tower pressure drop. 3.7.6

Liquid Gradient between Risers

Gas risers should be arranged such that the space between them available for liquid flow should not cause an overflow into the risers. The Francis weir crest formula can be applied to the pseudo “weir length”, which equals to the length from shell to shell, subtracting out the riser widths at the most constricting point. The liquid rate for the “weir crest” calculation is what can be reasonably expected to flow across the pseudo weir length. Narrow rectangular risers are usually placed with the long dimension parallel to liquid flows to reduce liquid gradient. For instance, rectangular risers should not be oriented to block liquid flow into sumps. When using round risers, the riser layout should avoid aligning their centerlines into rows.

Page 49 of 67

Issued: 10/25/2011 Revised:11/11/2013

3.8


Section 2.02.1

Other Design Considerations 3.8.1

Height Between Packing and Distributor

The height between the packing and distributor is set by the need to have enough space for vapor to flow around the troughs, and the need to reduce splashing and drift of streams. Packing performance can be negatively affected by allowing the liquid to drop from an excessive height above the packing. Splashing or drifting of the liquid on the top of the packing can cause entrainment and maldistribution. Guide tubes close the distance between the distributor discharge and the top of the packing. Guide tubes may extend all the way to the top of the packed bed, but the approach to the top of the bed may be restricted by the height of the bed limiter. Typical distance D w from the bottom of the distributor trough to the top of the packing should be 4 to 6 in. (102 to 152mm) or the convergence of 30 degree angles from vertical, whichever is greater, as shown in the sketch below figure 3.8.1. In cases of very narrow troughs, distance to the packed bed below may be reduced. For different distributor deck styles, sufficient vertical distance must be provided to enable the vapor to flow into the risers without causing vapor maldistribution in the packed bed. As a general practice, a maximum angle of vapor approach of 30 degrees from vertical is acceptable.

Figure 3.8.1

The maximum recommended drop distance from guide tube discharge to the packing is 6 in. (152mm) with typical distances being 1-4 in. (25-102mm), depending on the bed limiter design. In higher pressure columns and high vapor velocity columns the distance from the guide tube liquid discharge point to the top of the packing is a very important dimension and needs to be minimized (for example tri-ethylene glycol contactors and demethanizers). Page 50 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Baffle plate or line distributors feature proprietary designs that discharge the liquid from the spreading baffles directly onto the top layer of (structured) packing. To this end the baffles may be adjustable and be integrated with the bed hold down grid. 3.8.2

Height Above the Distributor

Height above the distributor to a mist eliminator, collector tray, support plate, flash gallery or active trays may be set by access requirements for installation, maintenance and clearance for support beams above. For a packed bed above, there must be room great enough for the vapor flow to reasonably spread itself for entry into the bed above. The distance from the top of chimney hats to the bottom of a packed bed above should be at least the greater of 12 in. (305mm) or 100% of the hat width. Additional height will be needed if large bed support beams are present. 3.8.3

Interferences from Bed Limiter

The bed limiter support angles are typically 1.5 to 3 in. (38 to 76mm) wide. Depending on tower diameter and up-lift requirements, the bed limiter may be integrated with the distributor or be independently attached to support clips or intermediate beams. To achieve guide tube approaches to the packing of less than two inches, the layout of the bed limiter pattern and supports may have to be integrated with the layout of the distributor and its guide tube locations to be sure there is no interference with the liquid distribution. Vendors of packed tower internals should be required to integrate bed limiter and distributor designs to minimize interference with the liquid streams from the distributor. Spray distributors have a different issue called shadowing. Since the spray pattern has from a 60-120 degree angle of impact on the top of the packing, the vertical and horizontal members of the bed limiter grid will intercept and misdirect some of the spray pattern. Integrating the spray nozzle layout and the bed limiter layout may minimize the interference. For plate or pan distributors, anti-migration bars or screening can be used as a bed limiter, but only at the bottom of the gas risers, and only if there is not a gap around the periphery of the distributor that would allow packing to migrate past it. Anti-migration devices should not cover liquid discharge points because liquid may wick along the screen or bar and drop onto the packing at some point other than the intended location. 3.8.4

Predistribution Systems Liquid Feed Pipes Liquid feed pipes to distributors must be compatible with the distributor design. Feed pipes may be simple single pipes to the center of a distributor or complex H pattern piping with multiple down pipes, metering orifices etc. to evenly supply multiple predistributor boxes. In general feed pipes should be designed to minimize disturbance of liquid surfaces and aeration. Turndown conditions should be incorporated into all liquid feed pipe designs. An important final step of the feed pipe design is flow testing it with its parting box/distributor as a system. Carefully designed vent holes and low point drains should be included for process and safety reasons.

Page 51 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Two Phase Flashing Feed Considerations Note that since some process simulators will not flash the feed stream at tower pressure, it may have to be flashed separately at the known tower pressure of the feed location. Any feed that is at or near the bubble point should be treated as a flashing feed.

For flashing feed devices see FRI Design Practice section 2.02.2 Internal Pipe Work to Packed Distributors. Flashing feed galleries, flash chambers, flashing feed pipes, and feed pipes with chimney trays are examples of flashing feed devices which can be supplied by column internals vendors. Parting Box Design Considerations Parting boxes distribute liquid to trough type distributors by proportioning the liquid according to the number of metering devices in each trough. The Predistribution metering of liquid may be facilitated by any of the following metering devices, i.e. round orifices, slots, V-notches, etc. with considerations similar to those for the troughs. If the parting box fails, the distributor will not work properly. Another important thing here is introduction of the liquid to the troughs without aeration, such that very little liquid gradient is developed within the individual trough. Because of the high liquid rate discharging from the parting boxes, most designs have large orifices in the bottom of the parting box floor. For designs with high turndown requirements, a combination of both sidewall orifices/slots and floor orifices can be applied.

Some trough distributor designs have the parting box integrated into the troughs. The integrated box takes less vertical space, but is typically more expensive and requires the troughs to be leveled to each other. Laterals must be gasketed and are subject to leakage. The integrated parting box is not subject to pre-distribution issues. Economics will typically determine that the integrated trough type distributor be limited to towers that are 12 ft (3.7 m) or less in diameter. For towers greater than 23 ft (7.0 m) in diameter, multiple interconnected parting boxes are recommended by most vendors. 3.8.5

Hole Size Considerations

  



Mechanical Concerns – Reproducibility or % error potential increases with reduced hole sizes. Plate thickness – The thicker the plate the harder to accurately punch holes. So small holes in heavier gage metal may require drilling. See FRI DP Section 2.02.4 Punching vs. drilling – both methods are acceptable, but may give different coefficients in the flow rate equations. Also there may be more variability created by lateral flow past venturi shaped poorly punched holes resulting in failure of distribution quality tests. Drilling will reduce the quality of the distributor, as the Cv will increase since it is difficult to keep hole size and shape uniform). Carbon steel distributors are not preferred as hole size and shape will change with time due to corrosion. Carbon steel will corrode during post steam out, construction and turnaround. A metallurgy should be chosen which is zero corrosion rated for the service. No corrosion allowance should be applied to a distributor design.

Page 52 of 67

Issued: 10/25/2011 Revised:11/11/2013

3.8.6


Section 2.02.1

Overflow Protection

In an overflow situation, there is seldom enough liquid head to provide uniform flow through the overflow notch. Overflow protection can give some relief from the adverse effects of overflow, but often will not eliminate. Overflowing liquid needs to be kept out of the vapor path whether the vapor comes through in gas risers, or between troughs. The overflow liquid needs to be distributed evenly preferably reaching the same distribution point as the metered flow. Overflow protection for side exit troughs or parting box is obtained with holes or notches (example illustrated below figure 3.8.6) at the top of the trough within the flow guide(s). The overflow should cascade down the same channel controlling the metered distribution point liquid or be able to re-enter the liquid distribution baffle for splash baffle designs. See also section 3.4.1.

Figure 3.8.6 Courtesy of Koch Glitsch

Contingency overflow protection (e.g. where the deck could foul) for orifice pans/plates is not typically incorporated into the design, but may be done with an internal tube arrangement. Plate/Pan distributor contingency overflow tubes are designed such that the top of the overflow tubes is lower than the gas riser height. 3.8.7

Trough Design Trough Height Trough height is a function of: maximum and minimum rates; minimum acceptable liquid height above the hole; whether the hole is in the bottom of the trough or elevated; vertical space limitations; style of trough. The trough height may be limited by the manhole size used on the tower and manufacturing limitations. Trough width Trough width is a function of: liquid rate, pour point density, and vapor flow pressure drop through the distributor. Trough width also impacts the height above packing requirements and open area.

Hydraulic gradients are a function of distance from supply points, resistance to flow and horizontal flow velocity. Hydraulic gradients can be reduced by providing cross-sectional area below the metering holes, maximizing the width of the trough and minimizing the distance between the supply point(s) and the farthest metering hole. The prevention of standing waves is best achieved by proper diffusion of the flow into the troughs.

Page 53 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Transverse velocities past holes are generated by the cross-sectional velocity of the liquid flowing down the trough which is a function of rate and cross-sectional area. A guideline (20) for maximum transverse trough velocity is 10-12 in/s (0.25-0.30 m/s) . The more difficult issue is the local transverse velocity created by liquid entering the trough or pan distributor. Diffusion of velocity and momentum at the point of supply is the most important factor. Vapor space covers can be used to collect liquid raining down from a bed above and redirect it into the distribution troughs or pan. If no parting box predistributor is used on a trough distributor, there must be sufficient interconnections between the troughs to balance the liquid head in the troughs.

Figure 3.8.7 Pan distributor with vapor space covers Courtesy of Koch Glitsch Trough covers can be used on all or part of the distribution trough length when the shell or head is expected to severely corrode. The covers typically found on the ends of the troughs prevent the corrosion scale from falling into the troughs and plugging orifices. This also give complete collection of the liquid from above that can be mixed with the vapor space liquid to then be redistributed. 3.8.8

Distributor Support

Pan type distributors are normally installed onto clips, support rings welded to the vessel wall or are clamped between column body flanges. As such the levelness of many of these distributors will depend on the levelness of the supporting structure. Leveling devices can be incorporated into the pan's mechanical supporting structure. However, in some cases it may be necessary to consider a different type of distributor, such as a trough type. Trough type distributors are preferably supported from the vessel wall using beams and welded attachments. This allows for the distributor troughs to be leveled independently of the vessel and packing hold-down grid and ensures robustness against column upsets. In cases where welding to the vessel wall is not allowed (frequently the case with revamps), it may be necessary to support the distributor using an integrated support grid that rests on the top layer of structured packing. The support grid allows for leveling of the troughs during installation but has the distinct disadvantage that column upsets such as vapor surges could dislodge the packing and/or distributor and push it out of level. In this case a column shutdown and entry will be required to restore distributor levelness. For guidance regarding leveling distributors, refer to the FRI Design Practices Book, Section 5.05, “Leveling Trays and Distributors”. Page 54 of 67

Issued: 10/25/2011 Revised:11/11/2013

3.8.9


Section 2.02.1

Distributor Gasketing

Refer to the heading "Checklist for Visual Examination of Packed Tower Internals Installation (Field)” in the FRI Design Practices Book, Section 5.02.03 and also the heading “Checklist for Visual Examination of Packed Tower Internals (Vendor Shop)”, in Section 5.02.05, dated 6/15/2003: Gaskets properly installed - "Some liquid distributors require gaskets at all joints. Other vendors may use gaskets for distributor assembly in special situations. For example, in large diameter towers, the long distributor troughs may have been fabricated in two pieces with a gasket joint at their connection. Extra attention must be given to gasketed distributors to make sure that the gaskets are properly installed to eliminate leakage. For applications with low liquid rates, what appears to be a small leak may actually be a significant proportion of the total liquid flow." Gasket material is the vendor's responsibility, but it should be approved by the Owner's Engineer. Vendor should demonstrate that gasketing material and gasket-to-metal seal designs have been successfully used in similar services, and will meet the specified leakage rate when installed per the vendor instructions and demonstrated during the orifice pan (or similar style) distributor flow test at the vendor shop. 3.8.10

Distributors in Small-Scale Columns

Reference is made to the FRI Design Practices Book, Section 3.07, “Design of SmallScale Columns”, subsection 3.1 and 3.2 for specific additional guidance for small-scale column distributors.

4

Distributor Testing and Installation

Proper initial liquid distribution is crucial to packed column performance. Commercial distributor design technology has improved substantially in the recent past such that both suppliers and end-users have learned to appreciate its importance. However, many design problem areas still remain, such as the minimum acceptable liquid head, feed velocity dissipation, and turbulence caused by liquid cascading from the parting box to the troughs. There are also several mechanical factors such as levelness and fit-up tolerances, drill or punch reproducibility, and leaking seams which can cause maldistribution. Finally, like any custom-designed item, distributors are prone to pitfalls and design expertise can vary between suppliers or even within suppliers. Therefore, some form of distributor flow testing is recommended to ensure proper operation. In some services, such as heat transfer or scrubbing applications, industrial experience has established that highly uniform liquid distribution is unnecessary for adequate performance of the packing. Spray nozzle and notched trough type distributors are often used in these services. For these services and distributor types, visual checks of the distributor operation at minimum and maximum rates are recommended. These qualitative flow tests are typically performed at the job site, either in the tower following installation or on a temporary rig. For many distillation services, FRI data as well as commercial experience have demonstrated the sensitivity of packing to relatively small deviations from uniform flow (29, 23). In these cases, physical inspection and visual observation are not sufficient to determine adequate performance. The only way to verify proper distribution including the uniformity of liquid flow is to quantitatively test the final product.

Page 55 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Please note: All distributors should be water flow tested. It is the last chance to validate the design and fabrication, and then make corrections or improvements as necessary. 4.1

General Overview of Water Testing Liquid Distributors

I.

Purpose a. Verify that the distributor meets the performance criteria at required operating rates b. Confirm fit up of all distributor parts (especially complicated designs) c. Check for potential leak points d. Identify defects – orifices not de-burred, fit up problems, etc. e. Dimensional checks versus drawing, orifice size, number, location, etc. f. Design sensitivity at extremes – minimum to overflow g. Solve design problems before shipment. For example; i. Momentum issues ii. Swirling motion and/or vortexing iii. Predistributor issues iv. Liquid head over full operating range h. Test all distributors even if there are multiples of the same design

II.

Procedure a. Set up in supplier's shop as in the actual installation including supports i. Ensure all safety & environmental procedures are being followed ii. If unable to accommodate the entire distributor, then set up part of the liquid distributor depending upon the capability of the supplier’s facility. All sections do need to be tested. iii. Set up actual feed piping & supports as will be arranged in the column iv. Be sure that test stand supports do not interfere with the test itself v. Allow for access to each distribution point (set up the distributor high enough to get under it for sampling) vi. Ensure distributor is level to within design tolerance b. Measure liquid depth at all flows and in all key parts of the distributor c. Video the test at all flow conditions

III.

Equipment Recommended a. Stop watch b. Measuring container to catch flow from individual drip points c. Ruler or measuring tape for liquid head depth d. Data recording supplies – paper, pencil, etc e. Video camera with sufficient recording storage media f. For very large distributors, provision should be made for a multi-point collector

IV.

Data a. Quantitative i. Flow measurements from individual drip points should be based on 10% of all drip points (or 30 minimum). For small towers this could very well be all drip points. The patterns of collection points should be defined and agreed by end-user, engineering contractor and supplier before testing. Establishing an agreed upon collection pattern is easily established with a sketch of the distributor. A maximum number of points to be tested should be considered to expedite flow testing. The number of points tested are typically between 30 to 60 per distributor. ii. Check for excessive flow through weep holes Page 56 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

iii. Area measurements can be made for larger diameter towers to look for gross maldistribution and speed up a distributor flow test b. Qualitative i. Is the distributor level? ii. Does the liquid go where designed? iii. Splashing and overflowing present? iv. Hydraulic jump v. Leakage that adversely affects the acceptance criteria at tray joints, flanges, feed piping, etc. vi. Are all flows visually uniform? vii. Swirling flows at ends of troughs? viii. Is there vortexing anywhere at low flow? ix. Check for detrimental wave action. V.

4.2

Desirable Test Stand Facilities

           4.3

Acceptance Criteria a. All qualitative issues have been satisfied b. The coefficient of variation (Cv), as defined in Subsection 3.2.4.1 above, for orifice and some notched distributors will have a typical range of 4-6% at design with allowances to go as high as 10% at turndown c. The best practices for testing line (splash plate type) distributors should be discussed with the vendor– these do not produce discrete points as do other types of distributors. d. See Subsection 4.5 below to address spray distributors

Indoor Facility Pump and metering capability to cover the full range of test rates. A dead tank receiver to calibrate flow rates. Water supplies should be properly maintained and tested especially for particulates and chlorides for austenitic stainless steel. Facility to collect and recycle water. Range of collecting devices for collecting representative group or point samples over 30-60 seconds. Facility for measuring or weighing collected samples. Programming to calculate average rates and deviations. Flexibility to simulate the plant design for feed piping to distributors and pre-distributors. Facility and capability for easily leveling pre-distributors and distributors. Test stand large enough to assemble the entire distributor (rather than a partial assembly).

Orifice Distributor Test Criteria

Quantitative flow tests are normally performed at the manufacturer's facility. The following criteria have been suggested as requiring quantitative testing: 1. Critical services where non-performance of the tower has a step change influence in plant profitability, safety, run length, etcetera. Services where the loss of fractionation stages will cause a product to go off-spec. 2. Low liquid rate services (<5 m 3/hr/m2 (2 gpm/ft2)) since uniformity of distribution is more difficult to achieve under these conditions. 3. Large diameter towers (> 2.5m diameter (8 ft.)). Leveling and large scale distribution are likely to be more difficult in larger columns. Page 57 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

4. Extremely high liquid rate services (> 75 m 3/hr/m2 (30 gpm/ft2)). Liquid velocity dissipation is often a problem, particularly from feed pipes and parting boxes. 5. High turndown requirements (> 3 to 1) since large variations in heads and discharge velocities often create distribution difficulties at one or both extremes. 6. Multi-tier orifice distributor designs since these can exhibit poor performance when liquid levels are near the orifice elevations. Flow tests are available at extra cost with commercial distributors. Charges may be quoted separately or included in the overall cost of the equipment. The cost for a flow test is justified based on improved reliability. Tests conducted at the manufacturer's shop permit dependable and cost effective correction of design or manufacturing errors. Such errors are much more difficult to rectify (if they are detected) once the distributor has arrived at the operating site. In addition, manufacturers typically have the best testing facilities. Flow tests are becoming increasingly popular as end-users recognize the importance of good initial distribution. A. Objectives - Distributor flow tests are simulations of expected distributor performance. The actual behavior may depend on the physical properties of the process fluids at the operating conditions of the column and could be significantly different from the test stand operation. For example, actual process fluids may exhibit much greater froth or foam heights than the water used in the test. Individual water streams from the orifices may look like "pencils" while under actual process conditions near the critical pressure the streams may appear almost like spray cones. Nevertheless, distributor flow tests are a valuable tool which can be used to prevent potential operating difficulties. Specifically, flow tests can be used to:

1. Permit visual observation of the distributor operation. Typical points of interest include: overflow of troughs, excessive aeration, wave formation, vertical discharge of liquid. 2. Evaluate the turn-up and turn-down capabilities of the distributor (operating range without gross maldistribution). 3. Expose leaks due to inadequate fit-up of internal flanges or poor fabrication (welding). 4. Check final assembly for fit-up and dimensional tolerances. 5. Demonstrate the functionality of leveling hardware and the capability to meet the desired level tolerance. (This can only be done if the support structure mirrors the hardware of the actual installation.) 6. Identify gross maldistribution caused by turbulence, excessive liquid velocities, and/or errors in orifice sizing or fabrication. In addition to the above, quantitative flow tests can be used to: 1. 2. 3. 4.

Characterize the uniformity of distribution on both large and small scale. Uncover maldistribution which may remain hidden during visual testing. Highlight errors in orifice sizing. Retest changes proposed to overcome observed deficiencies.

B. Facilities - The objectives for a quantitative flow test are most easily accomplished on a dedicated distributor test stand. Most major packing suppliers have distributor test facilities of some kind, see Figures 4.3.1 to 4.3.3. The test units should include the following key features:

1. Safety and environmental procedures are established and enacted at the testing site 2. A control system to ensure a constant liquid head and steady flow over the desired liquid rate range. 3. A recirculating water system to permit test runs of essentially unlimited length at Page 58 of 67

Issued: 10/25/2011 Revised:11/11/2013


4. 5.

6.

7.

Section 2.02.1

constant conditions. This is crucial if flow uniformity is to be evaluated in detail for large diameter commercial units. Filtering of the water may also be necessary to remove debris which could plug small orifices. Limiting the chloride content of the water may again be a concern. A feed arrangement to match the actual tower installation. Use actual feed piping when available. Sufficient size to permit testing of the entire distributor. Tests of partial distributors (half or quarter plans) can be useful but care must be exercised to maintain geometric similarity to the complete design at all levels of distribution. A support structure that will permit adequate mounting and leveling of the distributor without interfering with the liquid flow. The assembly must be high enough above the ground or support grating to permit sampling underneath the unit and afford a clear visual inspection of the liquid flow from the distributor. Support hardware to simulate the actual tower attachments to which the distributor will be fastened may also be useful to check the functionality of the leveling hardware and to simulate any potential for sagging of the distributor. To provide adequate lighting and shelter from wind and weather, an indoor facility is highly recommended.

At times, a qualitative flow test is done with the distributor installed in the column or on a temporary rig at the job site. The water supply system must be capable of steady flow at the maximum and minimum design rates of the distributor. Care must be exercised when using temporary water supplies (typically fire water systems) to ensure the water is free of debris which could plug the distributor. The chloride content of the water may also be a concern for stress corrosion cracking of stainless steels.

C. Visual Observations - Visual observations of the distributor action are the basis of any qualitative flow test. However, visual checks are also an important part of quantitative flow tests conducted on the manufacturer's test stand. Representatives from the end-user company and/or engineering contractor should attend manufacturer's flow tests to witness not only the data collection techniques but also record visual observations of distributor action at all operating conditions. The following items are suggested for examination:

1. 2. 3. 4. 5. 6. 7.

Distributor assembly is complete and in accordance with drawings. Feed arrangement duplicates actual tower design. Correct number and size of feed pipes, tees, elbows, etc. Fed by downcomer(s) from a chimney tray. Liquid cascades from support plate of bed above. Distributor assembly is within level tolerance. Spot check a representative number of hole diameters and that the punch or drill direction is the same for all holes (burr side up or down). 8. Measure liquid heads in all troughs/pans/parting boxes at design and turndown rates (compare to design values). 9. Measure sagging in various parts of the distributor under dry and fully liquid loaded conditions. Sagging can be measured using a straight edge or a transit (optical) method. 10. Note splashing or spilling caused by improper alignment of troughs / pans / parting boxes / feed nozzles. 11. Note trough or pan overflow. 12. Note excessive turbulence, aeration, swirling, vortexing, wave formation, or hydraulic gradients in troughs or parting boxes. 13. Note any other irregular action in troughs or parting boxes. Page 59 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

14. Note leaks from bolted / gasketed flanges or inadequate welding. 15. A good test for levelness is the "on/off" test where the underside of the operating distributor is observed as the water flow is stopped. If the unit is level, all of the pour points will stop dripping at the same time. High points will stop dripping first and low points last. This test may be repeated several times to locate the high and low areas (hence the name "on/off" test). D. Quantitative Sampling - Distributor irrigation can be quantitatively measured either by sampling individual drip points and/or using a larger area (multiple point) sample.

A multiple point sample is acceptable in most cases since FRI tests have shown that (at least random) packings are tolerant of large deviations of individual point flow. FRI tests have demonstrated a sampling area of 2 square feet (0.2 m 2) is adequate for a 4 foot (1.2 m) diameter column using 1" (25mm) metal Pall rings (FRI Progress Report, Jan-Feb 1985). Based on current radial mixing theory, it would follow that this sampling area should be adequate for tower diameters of four feet and larger using packings whose radial spreading coefficient is equal to or greater than that of 1" Pall rings (30, 31). Others have suggested a multiple point sample area equal to 5% of the tower area. Although not essential to the determination of the final liquid distribution, the predistribution stage (feed piping, parting box, or other device) is frequently the source of difficulty. Therefore, test data on the predistribution stage can be useful for troubleshooting and is often the quickest route to correcting design difficulties. These measurements are typically the first to be made since feed pipe or parting box orifices are inaccessible once the entire distributor has been erected. Regardless of the sampler size, random measurements are sufficient for most designs. Using standard statistical methods for confidence interval estimates, one can determine how many individual distributor points must be sampled to determine the performance of the entire distributor. In most cases, a sample size of 30 to 60 is considered large enough to constitute a completely random sample, regardless of the total population of drip points. For area samples, one practitioner indicates that the number of area samples equals the tower area in square feet divided by 7. Each sample size should be 2 ft2 with a maximum of 20 indvidual areas. In addition to random sampling, the flow test should always be supplemented by some systematic sampling of areas such as feed points, under parting boxes, or at the vessel wall where deviations in flow are most likely to occur. These techniques can save considerable time and expense for testing large diameter commercial units with hundreds or thousands of drip points. The actual measurements can be made using the simple "bucket and stopwatch" technique, although more sophisticated equipment such as electronic balances or continuous level instruments are also used. In either case, reproducibility tests should be conducted to determine the accuracy of the technique employed. At each test rate, the liquid head should also be recorded for both predistributor and primary troughs or pans. In addition, a video (and/or photographs) of the tests can be a very useful documentation of the installation procedure, operating behavior, sample technique, or other unique features of the distributor. The edited video can also be used as a training aid for operators, installation crews, or other personnel unfamiliar with packed columns or high quality liquid distributors. E. Acceptance Criteria - For either qualitative or quantitative flow testing, any visually detectable maldistribution must be investigated and corrected if practical (21). For quantitative testing, the deviations from uniform flow have been shown to be approximately normally distributed(29, 11). Based on a normal distribution, the flow deviations in the data may be quantified using the sample standard deviation. The relative performance of distributors can be assessed using the ratio of the standard deviation to the sample mean. This statistic is known as the Coefficient of Variation (C V) and is frequently expressed in terms of a percent. See Subsection

Page 60 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

3.2.4 above for more details on this Cv term. Distributor Quality Index and Wetting Index are useful for the design of distributors, but they provide nothing to measure on a test stand to determine whether a distributor was built correctly. Note that there is little FRI data or published information available from which to compile a recommended acceptance criteria. One source has recommended a C V<10% based on measurements of individual drip points (11). The same source recommends a C V<5% for predistributor flows. An equivalent C V criteria for multiple point sampling must be smaller than 10%; however, no published guidelines are currently available. In each case, the deviations from uniform flow must be randomly dispersed. Graphical analysis of the data indicating both geographic location and flow magnitude can be useful to spot trends (i.e. non-random deviations).

Figure 4.3.1 – Koch-Glitsch Test Facility Courtesy of Koch Glitsch

Page 61 of 67

Issued: 10/25/2011 Revised:11/11/2013


Figure 4.3.2 – Raschig Test Facility Courtesy of Raschig GMBH

Figure 4.3.3 – Sulzer Test Facility Courtesy of Sulzer Chemtech

Page 62 of 67

Section 2.02.1

Issued: 10/25/2011 Revised:11/11/2013

4.4


Section 2.02.1

Spray Test Stand

Spray header test stands must provide differential pressure across the spray nozzles of up to about 30 psi (2 bar). The special considerations include the head on the pump, pipe connections and instrumentation of sufficient pressure rating. Additional considerations are: facilities to test the individual nozzle flow pattern and to create a total flow vs. delta P curve, and video or visual means to verify the spray angle. 4.5

Testing Spray Distributors

Flow testing on a test stand or in the field may consist of checking one nozzle or a complete distribution system. The simplest arrangement is one nozzle, a pressure gage, a mount for the nozzle, a pressurized water source, a tared container, a stop watch and a weight scale. A camera or video camera is also useful to help capture and record the spray angle. Testing a complete multi-nozzle distributor includes testing each nozzle plus checking for the expected spray overlap design at various rates/pressure-drops. In addition, the testers should come to the facility fully prepared to get wet and provisions need to be made to ensure personnel safety and environmental preservation. The rate versus pressure-drop results need to be converted from water gravity basis to the system fluid gravity at operating temperature using the formula: Capacity Flowing Fluid = Capacity Water / (Sp. Gravity Flowing Fluid) 0.5 ΔPFLUID= ΔPWATER (( Sp. Gravity Flowing Fluid)/(Sp. Gravity Water)) 0.5 4.6

Field Testing Distributors

Flow testing spray distributors inside columns can demonstrate that all nozzles are unplugged and flowing properly. They can also show that the flow pattern of each nozzle and the full distributor is as expected, and that the overall pressure drop of the nozzles matches the design. It is advised that running the tests with filters in place and doing an initial test without the spray nozzles installed will help clear the lines of debris. Water testing of pan type (orifice plate) distributors in the field can only be done by observing the distribution pattern of the liquid leaving the bed. Packing may (and probably will) interfere with the distribution pattern. Keep in mind that a poor initial distribution will not be corrected by the packing. Radioisotope (Gamma) Scans may be a more fruitful approach to field testing liquid distributors. Grid Scans, as suggested in Section 3.01.3 Subsection 3.3.3, are recommended as the first step in diagnosing liquid maldistribution in packed towers. 4.7

Leveling a Distributor

The need for proper leveling is to ensure design performance. Most industry practice is to require levelness of troughs and parting boxes to be within 1/8 inch (3 mm) HIGH to LOW for individual troughs. When troughs are hydraulically connected, the troughs need to be leveled collectively. Leveling the troughs before the parting box(es) is installed is highly recommended. Levels should be referenced to the bottom of the troughs or parting box(es). The effect of out of levelness is directly related to the level above the metering hole at both operating maximum and minimum rates. For a typical turndown to 50 % and with 2 inches (50 mm) as the minimum height above the metering hole, the effect of the typical level deviation would be a flow deviation of 3% at minimum and < 1% at maximum rate. For low efficiency applications such as heat transfer zones, Page 63 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

a less stringent levelness criterion could be acceptable. Water leveling devices are of two main types: one type being a reservoir with one or more tubes or a single tube manometer style where the level is read against two rules and compared. The water level can be enhanced by using water with dark food coloring added. The coloring makes the meniscus more visible. Air bubbles in the tube are the most frequent error generators and the tube periodically needs to be checked for their presence. A standard check for accuracy and reproducibility is to first do a side-by-side check. Then do a fixed-location level check at the full span of the level tube and reverse the ends of the manometer tube to verify the same levelness reading. A scale needs to be attached to the measuring end(s). Laser Levels should be of the self-leveling type. A good laser level should be capable of reading within 1/16 inch or 1.5 mm over a 30 foot (9 meter) distance. Laser level accuracy can be checked by placing the device in a large room, marking the walls, then rotating the level and observing level reproducibility. Laser levels with bubble leveling devices are no better than a carpenter’s level. A carpenter’s level or machinist’s Level can be used to level smaller (36 inch (1 meter) or less) diameter troughs and pan distributors. A pair of accurately cut blocks can be used to reach over chimneys or tubes in the distributor to assist using a carpenter’s or machinist’s level. The level accuracy should be checked by rotating the level 180 degrees to validate the level reading. Leveling of a plate or pan type distributor (with the wall serving as the perimeter containment and mounted directly on a support ring) is dependent solely on the accuracy of the support ring installation. All other gravity distributors including self-contained pan and various trough types must have an independent means of adjusting level. Pan distributors should be hung from the wall with at least 3 leveling points. Trough distributors normally have adjustable supports at both ends of a trough or trough section. Each parting box and trough is leveled independently although one end of half troughs may be "ganged together" with the major adjustment taken at the opposite end. Some proprietary distributors may have unique leveling requirements and means.

4.8

Other Distributor Installation Issues

Troughs with laterals bolted directly to a center trough, and plate distributors require gasketing. Make sure the correct type of gasket material and style is used for the service startup (e.g. large temperature change), steam out and operating conditions of the tower. Gasketing (if required) must be continuously applied with no gaps. The design should have sufficient bolting frequency to ensure leak tight closure. Gasketing should be a last resort and not used unless absolutely necessary. Fasteners through the plate deck should be gasketed. Review vendor procedures on gasket installations as verified during the distributor test. Installation hold points need to be established for distributor levelness checks (e.g. before parting boxes are installed). All hardware should be checked for tightness. Refer to manufacturers specifications for the required tightness. Orientation. Check the installed orientation of the bed limiter to make sure it will not interfere with the distributor drip tubes. With the exception of single point feeds, all distributors must be properly oriented to the feed piping per drawings.

Page 64 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

Some gravity distributors for structured packing have side hole discharge to a vertical baffle plate to spread the point discharge to a sheet of liquid, or multiple drip points. This type distributor should NOT be oriented with the troughs parallel to the sheets of the top layer of packing elements. Multiple distributors of identical diameter must be checked to make sure the correct one is installed at the correct location. Verify hole sizes before testing and again before installation. If the distributors are the same diameter, some or all parts may be interchangeable and could be intermixed. Make sure pipe laterals are welded or bolted on in the correct direction. Seal welded plate type distributors need to be manufactured from thick enough material of at least 12 gage (2.5 mm) such that the seal welding operation will not warp the final deck to an outof-tolerance levelness. For trough distributors, verify that the holes in the predistributors (parting boxes) are aligned above troughs so that liquid will not bypass the troughs by flowing into the gaps between troughs. Before and after installation, verify that all distributor components are clean. Preventative measures should be employed to keep any dirt, gravel, wood shavings, boards, paper products, plastic film, loose hardware etc. out of the distributor during installation. A final cleaning should be made before manhole closure (using vacuum cleaners if necessary) to ensure that small distribution holes are not plugged before startup.

Page 65 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

References

1.) 2.) 3.) 4.)

FRI Topical Report 92 Kister, H. Z., "Distillation Operation", McGraw-Hill, New York (1990) Kister, H. Z., "Distillation Design", McGraw-Hill, New York (1992) Moore, F. and Rukovena, F., Liquid and Gas Distribution in Commercial Packed Towers, Chem. Plants and Process , Edition Europe 1987, pp 11-15 5.) Moore, F. D., Distributor Design and the Effects on Tower Performance, Norton Company, Akron, Ohio, 1984 6.) Rukovena, F., and Cai, T. J., Packed Tower Internals Important to Tower Efficiency, AIChE Spring Meeting 2009, Distillation 2009: Improved Design and Operation to Meet Economic and Energy Challenges”, Paper 64a, April 28, 2009. 7.) FRI Progress Report Sept-Oct 1982 8.) FRI Progress Report Jan-Feb 1983 9.) FRI Progress Report May-June 1992 10.) FRI Progress Report July-Aug 1992 11.) Perry, D., Nutter, D. E., & Hale, A., Liquid Distribution or Optimum Packing Performance, Chemical Engineering Progress , Jan 1990, pp 30-35 12.) Dhabalia, D., and Pilling, M., Distributor Design and Testing, PTQ Revamps 2006, pp 9-11 13.) Billingham, J. F., Bonaquist, D. P., Lockett, M. J., Characterization of the Performance of Packed Distillation Column Liquid Distributors, IChemE Symposium Series No. 142, Distillation and Absorption '97, Vol 2, paper 69, pp 841-852 14.) Spiegel, L., A New Method to Assess Liquid Distributor Quality, Chemical Engineering and Processing, 45 (2006) pp 1011-1017 15.) Schultes, M., Grosshans, W., Müller, S., & Rink, M., Present a Modern Liquid Distributor and Redistributors Design - Part 1, Hydrocarbon Engineering, January 2009 16.) Schultes, M., Grosshans, W., Müller, S., & Rink, M., Present a Modern Liquid Distributor and Redistributors Design - Part 2, Hydrocarbon Engineering, February 2009 17.) Wheeler, E.R., “Design Criteria for Chimney Trays”, Hydrocarbon Processing, Vol. 47, No.7 (July 1968) 18.) Chen, Gilbert K., "Packed Column Internals", Chemical Engineering , pp 40-51 (March 5, 1984) 19.) “Operating Characteristics of Corrugated Grid”, M. Pilling, M. Bannwar, and C. Campbell, AIChE 2001 Annual Meeting 20.) CEP March 1993 by J Bonilla 21.)Olsson, F. R., “Distributors that Worked and those that Didn’t”, Chemical Engineering Progress , (Oct 1999) 22.)Stupin, W.J. and H. Z. Kister, “System Limit – the Ultimate Capacity of Fractionators”, Trans. IChemE, Vol. 81, Part A, P.136-146, January 2003 23.)McMullan, B.D., A.E. Raviez, and S.Y.J. Wei, “Troubleshooting a Packed Vacuum Tower – A Success Story”, Chem. Eng. Progress, p.69, July 1991 24.)Kister, H. Z. and S. Chen, “Solving a Tower Salt Plugging Problem”, Chem. Eng., p.129, Aug 2000 25.) Klemas, L. and Bonilla, J.A., "Accurately Assess Packed-Column Efficiency", Chemical Engineering Progress, July 1995, page 30. 26.)Bonilla, J.A., "Don't Neglect Liquid Distributors", Chemical Engineering Progress, March 1993, page 58 27.)F.J. Zuiderweg and A. Harmens, “The influence of surface phenomena on the performance of distillation columns”, Chemical Engineering Science Genie Chimque, March 1958, vol. 9, nos2/3 28.)F. J. Zuiderweg and T. Yanagi, “Marangoni effects or liquid maldistribution”, AIChE National Meeting, Houston, April 1989 29.)Kunesh, J.G., Lahm, L.L., and Yanagi, T., “Commercial Scale Experiments That Provide Insight on Packed Tower Distributors”, Ind. Eng. Chem. Res., Vol. 26, 1987, p. 1845. Page 66 of 67

Issued: 10/25/2011 Revised:11/11/2013


Section 2.02.1

30.)Albright, M.A., “Packed Tower Distributors Tested”, Hydrocarbon Processing, Sept. 1984, p. 173. 31.)Hoek, P.J., Wesselingh, J.A., and Zuiderweg, F.J., “Small Scale and Large Scale Liquid Maldistribution in Packed Columns”, Chem. Eng. Res. Des., Vol. 64, Nov. 1986, p. 431.

Page 67 of 67


INTERNAL PIPEWORK TO PACKED TOWER DISTRIBUTORS

Issued:

01/15/1995

2.02.2

Revised:

INTERNAL PIPEWORK TO PACKED TOWER DISTRIBUTORS

INTERNAL PIPEWORK TO PACKED TOWER DISTRIBUTORS ...................................... 1 1.

Introduction ....................................................................................................................... 2

2.

Liquid ......................................................... ........................................................................ 2

3.

Feed Piping to Pan Distributor ................................................ .......................................... 2

4.

Ladder Predistributor .................................................. .......................................................3

5.

Two Phase Flow ................................................ ................................................................ 4

6.

Vapor Distribution .............................................................................................................5

7.

Radial Inlet Nozzle Design Criteria.................................................. ................................. 6

Page 1 of 13

Issued: 01/15/1995

INTERNAL PIPEWORK TO PACK ED TOWER

Revised:

1

DISTRIBUTORS

2.02.2

Introduction

Good liquid and vapor distribution in packed towers is necessary to develop the full potential of the packing. When either phase or both are poorly distributed, the liquid to vapor ratio will vary and may cause pinch points which limit the mass transfer. The entering pipework is important in getting the initial even distribution of the liquid and vapor.

2

Liquid

The section is concerned with Newtonian fluids; the distribution of non-Newtonian fluids is beyond the scope. The liquid should enter a distributor in a way that does not disturb the action of the distributor. Ideally there should be no velocity or pressure drop gradients and no surging in the distributor so that the distributor can uniformly distribute liquid through its outlets. The piping should not obstruct the installation, servicing or removal of packing or distributors. For small columns, e.g., < 2 feet (610 mm) in diameter, with a body flange at the top, a removable distributor pipe is often warranted. (See Figure 1.) Thus a 1 inch (25 mm) pipe with a short radius 90° elbow through a 3 inch (75 mm) flange, or a 2 inch (50 mm) pipe through a 4-6 inch (100-150 mm) flange, will permit the internal piping to be removed when servicing a top distributor from the top of the column. Often this problem is circumvented by having the top flange located at the distributor support ring instead of at the top of the vertical section of the column. For large columns manholes larger than 18 inches (450 mm) are recommended for servicing the distributors.

3

Feed Piping to Pan Distributor

Ideally the liquid should enter the distributor below the liquid level to prevent splashing. Where the liquid is fed into the center of small pan distributors, the feed pipe is usually located above the top of the gas risers; the liquid velocity should be limited to 4 feet per second to avoid splashing and disturbing the liquid head above the orifices (1.2 m/s) (36). Some vendors pass the entering liquid through a small section of structured packing to reduce turbulence. In larger columns, e.g., >3 feet (900 mm) in diameter, a predistributor is often used to distribute the liquid to a pan type distributor. The predistributor should be constructed to minimize turbulence. In many cases the liquid feed to the distributor pan or parting box is piped without due consideration to the turbulence that may be created. Because of this and other considerations, the pipework and distributors should be designed as a compatible system (37). Ideally the combined distributor and the piping system should be tested on a test stand. If testing on a test stand is not feasible, the combined system could be tested in place before putting the system into operation. Two devices were used in FRI tests to minimize turbulence. The FRI staff designed a pipe and baffle arrangement that introduces the liquid in three places in a 4 foot (1200 mm) diameter column, 12 inches (300 mm) from the pan floor. (See Figure 2, taken from Topical Report 106.) The Norton Corp. designed a predistributor for Intalox structured sheet metal packing that had a clearance of about 1-1/2 inches (38 mm) from the pan floor. (See Figure 3, taken from Topical Report 106.) Both predistributors appeared to perform equally well.

Page 2 of 13

Issued: 01/15/1995


Revised:

4

DISTRIBUTORS

2.02.2

Ladder Predistributor

A properly designed ladder type of predistributor can offer good distribution with large turndowns by utilizing multiple levels of laterals. These predistributors are used to feed parting boxes, pan type distributors and even trays. "The feedpipe must be elevated and oriented to prevent feeding liquid into vapor risers or onto riser covers." (39) Maldistribution can show up in a manifold as excess flow at the closed end, the feed end or both (3). (See Figure 4.) If the flow is relatively fast, the momentum and kinetic effects will tend to cause greater flow at the closed end. (See Figure 4-B.) If the friction effect dominates, the excess flow will tend to be at the feed end. (See Figure 4-C.) Senecal points out that a combination of both effects can cause excess flow at both ends and diminished flow in the center. This flow phenomena is quite complex and very difficult to calculate accurately. A rule of thumb is that the ratio of the specific kinetic energy of the inlet stream to the pressure drop across the outlet, and the ratio of the friction loss in the pipe to the pressure drop across the outlet, should both be equal or less than onetenth.(3) For 5% maldistribution:

h P = 0.1hO Where:

h p hO

= =

pressure drop through pipe pressure drop through orifices

and Velocity Head =

0.1 hO 2

Where: Velocity Head or Kinetic Specific Energy

V i α g

= = =

=

α V i

2 g

average inlet velocity correction factor for use of average velocity (1.0 for plug flow; 1.05-1.1 for turbulent flow; 2.0 for laminar flow Gravitational acceleration

(37)

)

Senecal points out that this criteria does not always produce an adequate quality of distribution. Zens discussed methods for calculating gas flow distribution in manifolds and branched piping systems (5). Although his paper was about gas distribution, the theory presented could be adapted for liquid systems. Most systems have some degree of fouling; hence Senecal pointed out that a minimum sized hole is required (3). The minimum size will be system dependent; a range of 1/4 inch to 1/2 inch (6.3-12.7 mm) is suggested. Increasing the hole size may require increasing the pipe size to meet the Senecal criteria. To check the quality of distribution, the manifold should be tested on a test stand at similar flowrates with the same fluid or a fluid with similar properties. Usually this is impractical and water is used as the test fluid. Then the effect of property differences should be taken into consideration. There is a tendency for the liquid flowing through orifices in the bottom of a ladder type of distributor to flow not vertically but at some angle from the vertical that takes into account the forward momentum of Page 3 of 13


INTERNAL PIPEWORK TO PACK ED TOWER DISTRIBUTORS

2.02.2

the liquid in the laterals. Therefore the liquid velocity in the final distributor pipes should be kept below 1 foot per second (.3m/s). Short nipples are often used to direct the liquid from the orifices in essentially a vertical direction (37). Where the diameter of the laterals would be excessive at a velocity of 1 foot per second (.3 m/s), a maximum velocity of 3 feet per second (.9 m/s) is acceptable. The Darcy equation can be used to calculate the length of laterals meeting the Senecal criteria for a given head of liquid. Where the liquid direction changes from horizontal to vertical, the vertical pipe length should be at least 3 times the pipe diameter to overcome the forward motion of the liquid in the horizontal pipe and obtain essentially a vertical flow. In general good distribution can be obtained from a pipe distributor when liquid enters from one end if the pipe is less than 5 to 6 feet (1500-1800 mm) long. For greater lengths the liquid should enter in the middle of the pipe distributor. When the pipe is greater than ten feet (3050 mm), the liquid should be added to the lower pipe at several points. Again the Darcy equation should be used to calculate the length of laterals meeting the Senecal criteria for a given head of liquid. The liquid entering a parting box should have a calming zone otherwise the splashing will cause uneven liquid heads through nearby orifices and down pipes. Often wire gauze packing or a box with holes is used. Internal piping may require supports to prevent vibration damage, e.g., broken flange joints. Usually the piping supports consist of a bracket or pipe with one end welded to the opposite wall and the other end to the internal piping. Sometimes the bracket consists of two or three pieces bolted together. By using vertical bolting slots, the levelling can be properly adjusted. Horizontal slots are sometimes used to adjust for horizontal distances. If the fluid entering contains vapor as well as liquid, the vibration can be worse than with a single phase. Figure 5 shows a typical support for internal piping.

5

Two Phase Flow

The design of a good two phase distributor is difficult and a potential source of serious distribution problems. Ideally the vapor and liquid phases should be separated outside the column with each phase introduced to the column through separate nozzles. Then an external separator tank is often used. Vertical two phase flow often leads to slug flow. External piping should be designed to avoid slug flow. "A flashing feed should not be delivered through a perforated pipe sized for liquid flow." (37) The designer should look at all of the pressure drops in the inlet line including the pressure drop across control valves to determine if the feed is likely to flash. A flow map of the piping can be drawn using computer software packages. This is a complicated design that could be profitably discussed with vendors.

If it is necessary to have two phase flow in the column, that flow should be introduced to an appropriate phase separator, e.g., a raceway shown in Figure 6, which is good for a high degree of vaporization, or baffles shown in Figure 7 (see also Section 1.01-5), which is suitable for a low degree of vaporization. More complicated systems are often used. Under vacuum conditions one weight percent of vapor can produce many volume percent of vapor. It is very important for the process engineer to accurately determine the degree of vaporization of a feed stream entering a column on both a weight and volume basis. Minimum, maximum and expected degrees of vaporization should be calculated. These data should be transmitted to the distributor manufacturer. The designer should also be concerned about mixing the feed vapor with the vapor rising from the packed bed below.

Page 4 of 13


6


2.02.2

Vapor Distribution The Problem - A packed tower differs from tray towers in pressure drop. Normally the pressure drop across the bottom tray is sufficient to redistribute a badly maldistributed vapor. Packed towers, however, tend to have much lower pressure drops. When trayed columns are converted to packed towers, the designer should determine if the entering vapor nozzle is adequate for the proper distribution of the vapor.

Thus at the April, 1989, American Institute of Chemical Engineers Meeting in Houston, a speaker discussed the conversion of a valve tray column to a packed column. The vapor entered the column at F S = 80 (English units, F S = 98 in metric units). The pressure drop across the bottom valve tray straightened out the maldistributed vapor profile. When the column started up with random dumped packing, the HETP was approximately 10 feet (3050 mm) instead of the expected 2.5 feet (760 mm) due to vapor maldistribution. After installing a chimney tray above the vapor inlet with a pressure drop of approximately 9 inches (230 mm) of water, the HETP dropped to approximately 2.5 feet (760 mm). In that situation pressure drop was not a limiting feature of the design and the chimney tray solved the problem. For systems where pressure drop is limiting, a properly designed vapor sparger can be a solution with a much lower pressure drop. In general low pressure drop packings are more sensitive to vapor redistribution than high pressure drop packings. Thus structured packed beds usually have the most problems because they are usually designed for very low pressure drops. While vapor maldistribution is most serious in large diameter towers, serious maldistribution has been reported in towers as small as 12 inches (300 mm) in diameter. One vendor (41) states: "When vapor enters the column less than one column diameter under the packing support and from the side of the column thru a port, the diameter of which is one third of the column diameter or less, poor distribution of the vapor into the packing results - some areas of the packing will have less than 50% of the vapor flow desirable, while other areas will have much too much flow." Porter and Ali showed that severely maldistributed inlet gas is converted to uniform flow within one-half of a column diameter with pall rings (40). Stikkelman showed that a steady distribution of gas is formed within 12 inches (305 mm) of the bottom of a 20 inch (510 mm) column which contained one inch (25 mm) diameter pall rings (42). The lateral movement of vapor is not large relative to the vertical movement after uniform flow is established and thus composition maldistribution would tend to continue up the column. Redistributors normally will mix and redistribute liquid but have little direct effect on the vapor composition maldistribution. Solutions to the Problem - Where pressure drop is not a consideration, chimney trays, vapor spargers, and inlet deflectors can be used to achieve good vapor distribution. High pressure drop packing support trays are sometimes used. The designer should beware of the risk of reducing the tower hydraulic capacity due to flooding at the support plate. It is necessary to be sure that at the maximum vapor rate, the liquid buildup on the plate does not interfere with the vapor flow. In many vacuum systems, however, the pressure drop across these devices cannot be tolerated. Then, large inlet vapor nozzles and half pipe distributors are used.

If a sparger is used, it is important to have a proper balance between the kinetic energy and momentum force of the inlet stream, the friction losses along the length of pipe and the pressure drop across the outlet holes. This type of design was investigated by V.E. Senecal (3). His article is discussed in Perry's Handbook (39). As a rule of thumb, to avoid greater than a 5% maldistribution "the ratio of a kinetic energy of the inlet stream to pressure drop across the outlet hole and of friction loss in the pipe to pressure drop across the outlet hole should be equal to or less than one-tenth". Often inlet deflectors are Page 5 of 13

Issued: 01/15/1995


Revised:

DISTRIBUTORS

2.02.2

used for radial entry (See Section 1.03-1). Greskovich and O'Bara developed a method for computing the pressure drop in perforated pipe distributors (43). Sometimes a pipe distributor is designed with the bottom center section cut out so that the vapor would flow down and around the pipe before contacting the bottom support plate. (See Figure 8) Muir and Briens found that such a half cylinder distributor produced a discontinuity of vapor above the pipe (38). Furthermore this type of design does not necessarily produce an even flow of vapor down the length of sparger if the cut is uniform. However a half pipe arrangement with diminishing donut rings offers low pressure drop and reasonable distribution. By having the half pipe contain a 75% open donut baffle 25% across the column, a 50% open donut baffle 50% across the column and a 25% open donut 75% across the column, the high velocity entering vapors tend to loose their kinetic energy and in experimental equipment were directed upward around the half pipe evenly (41). (See Figure 9) The design of half pipe vapor distributors requires further investigation. The effect of the method of vapor entry upon the coefficient of vapor variation to a packed bed was studied by Muir and Briens (38). They concluded that with a single entry point, radial entry provided more uniform vapor flow and a lower pressure drop than tangential entry. Two opposing radial entry points produced a more uniform flow than one entry point. In general, tangential entry should be avoided for packed columns except where needed to separate two phase flow.

7

Radial Inlet Nozzle Design Criteria

The following criteria can be used for calculating the minimum size nozzle for vapor entering a column in order to obtain good vapor distribution. It is based upon the velocity head times the vapor density in the vapor line balanced against the pressure drop in the bottom packing.

H V

=

V 2 ∗ ρ g 2 ∗ g H V V ρ g g

Where:

Calculate

R =

Where:

Δ P

= = = =

(kg/m2) Vapor velocity in line (m/s) Vapor density (kg/m3) Gravitational acceleration (9.81 m/s 2)

H V Δ P

=

Pressure drop in first meter of packing (mbar)

Then if R < 25 No vapor maldistribution if

25 < R < 45

Marginal vapor maldistribution. Vapor distributor needed in large columns and/or columns where critical performance is required.

if

45 < R

Vapor maldistribution is a problem. A vapor distributor is needed.

These criteria apply to radial vapor inlets.

Page 6 of 13



Figure 1. R   

Page 7 of 13

2.02.2



Figure 2. FRI TDP BAFFLE PREDISTRIBUTOR

Figure 3. LIQUID SAMPLER NORTON PREDISTRIBUTOR

Page 8 of 13

2.02.2



2.02.2

Figure 4. P   Of these perforated pipe distributors, “A” gives ideal distribution, but momentum and kinetic predominate in “B”, friction in “C”, and upstream disturbance plus momentum and kinetic enerygy in “D”.

Page 9 of 13



Figure 5. INTERNAL PIPING SUPPORTS

Page 10 of 13

2.02.2



Figure 6. T   

Figure 7. FLASHING FEED BAFFLES

Page 11 of 13

2.02.2



Figure 8. VAPOR SPARGER

Page 12 of 13

2.02.2



Figure 9. HALFPIPE DISTRIBUTOR DIMINISHING DONUT RINGS

Page 13 of 13

2.02.2


REDISTRIBUTION IN PACKED COLUMNS

Issued:

01/15/1995

2.02.3

Revised:

REDISTRIBUTION IN PACKED COLUMNS

REDISTRIBUTION IN PACKED COLUMNS ................................................................ ..... 1 1.

Introduction .................................................. .......................................................... ......... 2

2.

Redistributor Hardware ................................................................................................... 2

3.

Design Considerations ................................................ ..................................................... 5

4.

Design Calculation Options ............................................................................................. 6

Page 1 of 16


1

REDISTRIBUTION IN PACK ED COLUMNS

2.02.3

Introduction

A redistributor is a liquid or vapor distributor located between two packed beds not necessarily associated with a feed point. Redistributors act mainly on the internal vapor and liquid flows. The ir primary function is to correct maldistribution. This maldistribution may develop in deeper beds of packing due to variations in the packing uniform ity. These variations can be due to inconsiste nt packing methods (such as crushed packing in on e area or po or fit-up of stru ctured packing blocks), out-of-level structured or grid-type packing, structured packing s eams, internal support members, and the colum n wall. Th e column wall, and development of pr eferential vapor or liquid flow at the co lumn wall, is one of the primary reasons for the use of redistributors. Va riations in flow may also develop as a result of unavoidable distributor or packing plu gging due to water soluble salt pre cipitation, shell corrosion products, or polymerization products. Another function of redistributors is to promote liquid mixing and/or vapor mixing. This eliminates any radial concentration gradie nts that m ay have develo ped in either phase due to m aldistribution. More frequent redistribution can also be used to provide a safety factor, in case any one of the distributors fails for some re ason. In addition to m ass transfer desi gn considerations or safety factor cons iderations, redistributors may be ne cessary simply because the allowable maximum mechanical bed height for the packing has been exceede d. Although earlier design practice often allowed breaks to oc cur between packed beds with no redis tributors, today we realize that the liqui d flow from a typical packing su pport plate is inadequately uniform to be used for the distribution to the next packed bed. (71) Redistribution and the ne ed for it is still a very active element of investigation by research groups and within the industry. Because of the ongoing research in this area and the lack of consensus on when redistributors are required, the followin g discussion is broken into three separ ate sections: a Hardw are Section, a Design Consid eration Section, and a D esign Calculation Section. The Hardware section describes the hardware necessary to acco mplish redistribution, a subject on which there is gener al agreement. The Design Consideration Section presents the important factors for consideration regarding redistribution. The Design Calculation section provides the reader with the cu rrent calculation m ethods for answering the question: "How Often Should I Redistribute?"

2

Redistributor Hardware Liquid Redistributors - Industrial application of l iquid redistributors is far m ore common than vapor redistributors. This is so for three reasons:

1. Liquid distribution has been found to be more critical in packed bed performance than vapor distribution because vapor spreads easier. 2. Redistributors always occur between packed beds, where the high i nlet velocities which normally require good vapor distributors are not found. 3. Vapor remixing has only recently been identified as an important concern for redistribution. The final distribution elements of liquid redistributors are designed using the same fundamentals as initial liquid distributors. (See section 2.02.1 "Liquid Distributors".) In fact, it is almost always the case that the hardware design used for the initial feed distributor is duplicated with the redistributor. (With the exception that the flow element sizing may change if the redistributed liquid rate is different from the Page 2 of 16



2.02.3

top feed liquid rate.) Orifice pans or plates can be used directly as redistributors ( Figure 1), simply by adding hats over the vapor risers. Good design requires that an orifice pan (which has a gap at the wall) have a wall wiper ring to prevent liquid bypassing. It also has the advantage that a separate collector tray is not required. However, no liquid radial remixing in larger diameter columns will be achieved by allowing liquid to fall onto a standard orifice pan or plate distributor from the packed bed above. Trough type liquid distributors usually require a collector tray above to properly collect the liquid from the upper bed and distribute it to the troughs ( Figure 2). The collector tray must be liquid tight and have a riser design which prevents backtrapping of liquid. Leakage is prevented by tight gasketing or seal welded construction. Non-leakage is best confirmed by a water test of the collector where the rate of change of water level is measured. Underside viewing during this test will help pinpoint leak areas for further sealing. An alternative method is to provide liquid deflector baffles above the troughs ( Figure 3). This alternate is not preferred because of liquid bypassing the redistributor and the lack of liquid mixing. Liquid redistributors have the additional design objec tive of promoting liquid mixing that is absent fro m most initial liquid distributor designs. A common method used to promote liquid mixing is to collect all the liquid from a packed bed using a chimney tray or other type of liquid collector and effect the remixing via a single central downcomer. The liquid from the common central downcomer is then redistributed through troughs or pipes b ack to the liq uid distributor device. (See Figure 2A) Some vendors will add static mixing elements to the central downcom er or will use s tructured packing as a static m ixer to promote liquid mixing. Smaller diameter towers may not need any special redistributor design features to aid liquid remixing. Features promoting liquid remixing usually require additional tower height and will add to the cost; however they may be necessary. An alternative to the use of an intermediate collector tray with central downcomer, i s to use a n appropriate height of structured packing at the bottom of a bed of random packing ( Figure 4) to effect some radial liquid m ixing before allowing the li quid to fall int o a redistributor. The consen sus among industrial practitioners is that this does not guarantee effective radial remixing except perhaps in smaller columns. The same care to avoid maldistribution should be taken with t he redistributor design as taken with a liquid distributor. This was demonstrated clearly by the 1993 FRI tests of a poorly designed redistributor. (The liquid head was too low when considering the method of introduction of liquid from the collector tray.) These test data confirmed that a poorly designed redistributor can be worse than no redistributor at all. Redistributors should be water tested just as distributors are tested. It is very important in these tests to include the same method of introducing the liquid onto the distributor as will be used in the field. It is important that the height required for redistribution be considered as carefully as the height available for initial liquid distributi on, so that the redistributor performance is not adversely affected. In revam p situations, there is always the tendency to try to maximize packed height. If pa cked height comes at t he expense of redistributor height, and as a result redistributor performance suffers, the use of a redistributor will be counter productive. Wall wipers within packed beds ( Figure 5) have fallen out of general use in random packed towers in recent years, apparently because of concern about their effecti veness and the effect on vapor/liquid contacting and also perha ps because of the advent of packings with less tendency to allow liquid to migrate to the wall. Wall wipers will also cause so me vapor bypassing of the liquid. Wall wipers can also, if incorrectly designed, prematurely limit the capacity of the packed tower. FRI studies on capacity limits introduced by retaining tray support rings confirm this and indicate the capacity debit to be roughly proportional to the percent cross sectional area obstruc ted. Wall wipers have typicall y been used in smaller towers, typically less than 3 feet [ 910 mm] diameter, and especially in t he stripping section of Page 3 of 16



2.02.3

these towers. Their purpose is to avoid any wall fl ow liquid es caping out the bottom of the column without having its low boiling components removed. (71) It is intere sting to note that most structured packings now incorporate a ty pe of wall wiper to correct preferential wall flow. This is due to the greater liquid spreading ability of structured packing which leads to increased wall flow ( Figure 6). For smaller diameter columns, each layer of structured packing is bound with a wall wiper device to divert wall flow liquid back into the packing. This may not always be done in larger diameter columns where the packing is brought in as rectangular blocks. Vapor Redistributors - Vapor redistribution is remixing the va por to remove any radial concentration gradients. Limited work has been done investigating such devices. Radial concentration gr adients will develop in the vapor as well as the liquid whenever liquid maldistribution occurs. Molecular diffusion alone is not enough to remove these concentration gradients, and they will degrade the performance of a packed tower. (75) Theoretical studies using the "zone-stage model" have shown the importance of vapor remixing in combination with liquid redistribution. (76) The explanation for the be neficial effect of vapor remixing is that even when the liquid distribution ha s been corrected, residual radial vapor c oncentration gradients may still exist which can degrade mass transfer efficiency. (35)

FRI has patented a vapor remixing device which is shown in Figure 7. The design princ iple of this device is to promote vapor mixing through the use of circumferential vapor risers which discharge vapor radially inward and interior vapor risers which discharge vapor radially outward. Another patented vapor re -mixing device uses a static mixing element such a s swirl vanes within the vapor risers to effect radial rem ixing.(77) Th is device, shown in Figure 8, is si milar to t he concept illustrated in Figure 2 for the liquid. T he number of risers is minimized with one preferred for high pressure operation. This caus es all the vapor to fl ow into one restricted ar ea, which contains a mixing element. For vacuum operation more risers will be required and are spaced in a conventional manner. Figures 7 and 8 are general schematics illustrating two design concepts for a vapor rem ixing device. Any design of such a device must be done with careful attention t o proper hydraulic sizing of the ports and vapor risers to avoid creating a more severe problem than the one being solved.

The same chimney tray which is used to collect the liquid to a central downcomer for liquid redistribution can also promote radial vapor m ixing. This will occur "naturall y" when the vapor discharging from the chimneys has sufficient velocity so that a fraction of the column diam eter is traversed before the jet decays. Not many existing towers have internals sp ecifically designed to redistribute vapor, except via this "natural" redistribution achieved by typical chimney tray design. However, many of these colu mns work as designed. Therefore, th e need for specifi c vapor redi stribution hardware is not yet well established. Many so-called "vapor i njection plates", which are in fact standard corrugated packing support plates with 100-120% open area relative to the column area, can be expected to do little if anything to promote vapor remixing in packed beds. Owing to some favorable test data using flow visuali zation techniques, structured packing has al so been suggested as a possible radial vapor remixing device. However, at this time no design criterion is known specifying how many layers might be required to achieve a given degree of radial vapor mixing.

Page 4 of 16


3


2.02.3

Design Considerations

A major decision in the design of a pa cked column is how often to redistribute. Since a redistributor always uses significant column height and is a costly piece of equipm ent, this decision is not taken lightly. In tray to packing revamps, the requirement to redistribute often means the loss of packed height and reduced theoretical separating capabilit y. T herefore use of a redistributor is indicated when the effective HETP due to channeling becomes so high that the total bed height (i ncluding redistributor) for the separation is reduced by correcting the channeling and regaining the design HETP of the packing. However, there are often other considerations, especia lly in revamp situations. It is reco mmended that every redistributor location have an adj acent manhole to install and later inspect the necessary internals. If the column was constructed manhole deficient, it may be necessary to add a manhole and in some cases to post weld heat treat the vessel after the manhole addition. This can add significantly to the revam p cost and m ore importantly take m ore time than is av ailable during that particular plant sh utdown. In some cases the cost and/or time of manhole addition are considered to be too high and a manhole is not added. In this case it will not be possible to inspect the redistributor with out r emoving packing, a very time consuming and costl y procedure. If packing i s removed, it may not be suitable for reuse in th e column due to damage and new packing will have to be purchased. Lack of inspection access will not be noticed as long as the t ower is operating well. However, if it starts to operate poorl y, inspection access becomes extremely important. Coordinating new manholes with existing platforms will reduce their cost significantly. The need for redistributors is reduced with packings that have better radial mixing tendencies since this reduces the effect of channeling on the mass transfer efficiency of the packing. (35) Also pa ckings that tend to have reduced wall flow will also have less need for redistributors. Taller beds of packing have been used with mixed reports of success in the industry. High pressure distillation service, above about 150 psia (10.3 bar), seems to be particularly sensitive to the need for redistribution. Th e industry has found recentl y that in large diam eter high pressure distillation columns, structured packing performance can degrade significantly without redistribution. (78) The following considerations are involved in the decision to determine whether or not a redistribution is necessary: 1. Packing Type - some packings have better radial distribution than others, meaning that any radial concentration gradients t hat develop will be evened out; these packings will require less redistribution. Some packings also have less tendency to develop wall flow; these packings als o will require less redistribution. 2. Packing Size - in general, smaller, more efficient, packings will require shorter beds and more redistributors than larger, less efficient, packings. 3. Column Diameter - in general larger diameter columns require more redistribution than smaller columns. As the ratio of the radial distribution coefficient to column diameter drops the packing will have less capability on its own to correct the radial concentration gradients that develop. Literature sources indicate less concer n with redistribution in columns of 8 to 12 inches (0.2 to 0.3 m) diameter or smaller. A mitigating factor for large diameter towers is th at the wall flow area takes up less of the total bed area on a percentage basis.

Page 5 of 16



2.02.3

4. Distribution Quality - poor initial distribution quality will require more redistribution, as will a distributor which tends to plug. The same considerations used with initial liquid distributors apply to redistributors: pour point density, flow variation between pour points, distributor levelness and pour point placement (especially in the wall area, beneath supports, and near vapor risers). These are all important factors affecting the need to redistribute. 5. Difficulty of Separation - if the separation is already close to being pinched, small variations in L/V ratios such as might develop with channeling or wall flow will have m uch larger effects on the required number of stages to make the separation than for an easy separation far from a pinch. Therefore, redistribution, which seeks to insure constant L/V ratios internally , is more necessary for difficult separations. 6. Mechanical Bed Height Limitations - modern practice is to reestablish goo d liquid distribution whenever mechanical bed height lim itations require that beds o f packing be broken. Current metal random and structured packings are capable mechanically of bed heights up to 45 to 50 feet [13.7 to 15.2 m] or more depending on packing type, packing wall thickness and metal type. 7. Separation Pressure - hi gh pressure distillations, above about 150 psia (10.3 bar), have higher volumetric liquid to vapor flows; they also usua lly have closer density differences, and lower surface tensions; for these reasons the phases must be separated more often to reduce the effects of vapor backmixing. 8. Vessel Design Factors - in revamp situations the availability of manholes and the cost of adding them may play a role in the decision whether to require redistribution.

4

Design Calculation Options

Design calculation options can be broken down into the following groups: Rules-of-Thumb, Shortcut Two Tower Models, Rigorous Multi-Tower Models, Zone-Stage Models and Bed Height Correlations. The following sections describe the current status of each of these areas in turn. Rules-of-Thumb -

1. Maximum bed heights - By far the most common rule-of-thumb used in the industr y to determine how often to redistribute is to set a maximum permissible bed height. T ypical values range from 20 to 30 feet (6 to 9 m). However, successful applications of bed depths from 40 to 45 feet (12 to 14 m ) for modern metallic random packing and structured packing have been reported in the industry. (79) On the other hand, there are also reported experiences where such tall beds performed poorly. (80) Often the maxi mum bed height used may be service specific, such as amine absorbers where the standard design is often t wo 15 to 20 foot (4.5 to 6.0 m) deep beds with liquid redistribution between them. 2. Maximum number of theoretical stages - Displacing or used in conjunction with the maximum bed height rule-of-thumb, is a rule-of-thum b for maximum bed height based on the num ber of theoretical stages. This r ule-of-thumb is onl y valid for distillation services and should not be used for strippers or absorbers or quench towers. Using the number of theoretical stages has the advantage that it recognizes the process aspects of the effects of maldistribution. For example, it will automatically account for some of the factors cited above, such as packing size and pressure level up to 150 psia (10.3 bar). In ea ch case, as the packing efficiency increases, a rule-of-thumb

Page 6 of 16



2.02.3

based on the num ber of theoretical stages will also require that the beds become shorter. Unfortunately, for distillation pressure levels above about 150 psia [10.3 bar], this rule-of-thumb does not cor rectly account for the redistributio n needs of structured packings. In this case experimental measurements show HET P's increasing but redistributors m ay need to be applied more often. Typical values for this rule-of-thumb range from 10-15 theoretical stages per bed with one vendor recommending a maximum of 20 theoretical stage s. (80) One source prese nts data on a low pressure application with structured packing which shows that good performance can still be obtained in s maller diameter columns with as many as 32 theoretical stages in a be d of packing.(81) The most often used number is in the range of 12 stages per bed, which, when combined with a packing with an HETP of 24 inches (0.61 m), results in bed heights very close to the 25 foot (7.6 m) bed height guideline used by many companies. 3. Ratio of bed height to column diameter - Another common rule-of-thumb found quite often in older texts is a fixed ratio of bed heigh t to column diameter. This can vary from 6 for Raschig rings to 10 for more modern pall rings. (82) This rule-of-thumb recognizes the wall flow effect and the fact that it can vary with packing type. This rule-of-thumb also requires that guidelines for tower diameter to packing size ratio be followed (typically no less than 10-12 to 1 for Pall ring type packing and up to 30 to 1 for Raschig ring pack ing (82)). If packing size is too large relative to tower dia meter, excessive wall flow will deve lop more quickly thus requiring even shorter beds than the above rules-of-thumb indicate. One reference presents data that shows that as long as the recommended ratios of tower diameter to packing diameter are followed, there is no e ffect of the ratio of bed height to column diameter over the range of 5 to 15. (83) Shortcut Two Tower Models - Shortcut m odels can be used to estimate the effect of vapor or liq uid maldistribution on separation an d consequently help the designer deter mine the need to redistribute liquid. The McCabe-Thiele diagram can be used in the manner shown in Figure 9. Two colum ns are assumed with unequal liquid flow and equal vapor flow. The Mc Cabe-Thiele construction is carried out with a fixed number of stages for each column based on the "true" packing H ETP. (This is the HETP which could be expected with no maldistribution of vapor or liquid.) Then the apparent number of stage s is stepped off between th e top and bot tom compositions using t he external reflux operating line. The ratio of the apparent number of stages to the stages f rom the “true” HETP then gives the expected loss of efficiency for the initial assu med level of maldistribution. For separations which show sensitivity to maldistribution, more frequent redistribution wo uld be reco mmended. A similar constru ction can be carried out with the Kremser equation or ot her shortcut equations. However, with this approach the designer only learns the sensitivit y of the given separation to a certain level of maldistribution, not the actual degree of maldistribution that will be expect ed or the effect of its development as bed l engths are increased. A combination of this ty pe of approach with rules-of-thumb for expected maldistribution for packing and distributor types can lead to a consistent policy for the decision on how often to redistribute. Multi-Tower Models with a Process Simulator - Most process simulators can be s et up to handle a network of colum ns which can then b e used in a manner similar to the above two-tower approach to simulate the effect of a given level of maldistribu tion on separation efficiency. While pro viding a more exact model of the performance of the column, this approach suffers from the same requirements as the 2tower shortcut method above: some rules-of-thum b are still required for the expected developm ent of maldistribution for given packing and distributor types. Zone-Stage Model - The zone-stage model has been describ ed at length in various of FRI reports. (84) It has also been discussed in the open literature. (85) I t is a model still undergoing development as more information on flow in packed beds becomes available. In this model, random packing is subdivided into axial stages and concentri c zones with the stage he ight set equa l to the "true" HETP of the packing. Vapor and liquid leaving each "stage" are assumed to be in equilibrium. The liquid rates at each stage i n

Page 7 of 16



2.02.3

the model are obtained fro m the liquid flow profile at th e average height of the st age in column. In the area of the wall, the flow i ncludes the wall flow. Width of zones are set to be a multiple of the packing diameter in order to account for radi al mixing in the colum n. Vapor distribution is assu med to be uniform and is not affected b y the changing liq uid flow rates, b ut radial vap or mixing is taken into account. The zone-stage model incorporates the radial dispersion characteristics of the p acking and also accounts for other important variables such as th e relative volatility, concentration, the interaction of the packing and wall, an d column diameter and bed length. The initial liquid distribution can be varied to approximate the actual liquid distributor performance. The zone-stage model has also been applied to model structured packing with appropriate (i.e. empirically determined) spreading coefficients. A more realistic variation of the zone-stage model for structured packing has also been suggested. This variation uses square zones, and constrains forward and backward flow in one direction only in a single layer. Other, theoretical models based specifically on the liquid flow mechanism in structured packing have al so been developed and sho uld be considered as well for evaluating the redistribution question with structured packing. (86) Since the zone-stage model can simulate the actual packed bed hydrodynamics as well as the effect of redistributors, it can be used as a design tool for the decision as to how deep beds should be for a given performance. So me FRI member companies have experimented with the zone-stage model for this reason. It is expected that in t he future, a dist ribution model such as the zone-stage model will be used for each ne w packed tower design to optim ize packed bed height. Alternativ ely, a distribution m odel such as the zone-stage model will be used to deve lop more comprehensive rules-of-thumb that will more rigorously specify when and how often to employ redistributors. Bed-Height Correlations - There is also work in the area of d eveloping correlations for maximum allowable bed heigh t based on com binations of some of the va riables mentioned earlier. A pl ot of "stable" bed heights and "unstable" bed heights using FRI data was pr esented at the May, 1992 FRI TAC meeting. This plot used the relationship: (specifi c area of packing x colum n diameter x length of bed / radial spreading coefficient) versus (liquid velocity / gas velocity) to discriminate between the stable and unstable operating bed heights. Unstab le beds were based on be ds where the light key concentration profiles were concave, i.e. the separation efficiency of the bed deteriorated at the bottom of the bed.

This bed height stability criterion is based on the assumption that concave concentration profiles ar e caused by channeling of vapor through the packing. Thompson, (87) however, shows that vapor channeling is only one out of many factors that may lead to co ncave profiles. This finding questions the basis and validity of this particular height stability criterion.

Page 8 of 16



2.02.3

Figure 1. PAN-TYPE REDISTRIBUTOR

NOTE: "Courtesy Mobil Research and Development Co." Released in B. Stober letter of 8/1/94

Page 9 of 16



Figure 2. TROUGH-TYPE REDISTRIBUTOR WITH COLLECTOR TRAY AND MIXING ELEMENT


Page 10 of 16

2.02.3



Figure 3. TROUGH-TYPE REDISTRIBUTOR WITH LIQUID DEFLECTOR BAFFLES


Page 11 of 16

2.02.3



Figure 4. PAN-TYPE REDISTRIBUTOR USING STRUCTURED PACKING AS RADIAL MIXING ELEMENT


Page 12 of 16

2.02.3



Figure 5. "ROSETTE" WALL-WIPER NOTE: "Courtesy Norton Chemical Process Products Corporation" Released in K. Rowe letter of 11/18/94

Figure 6. STRUCTURED PACKING ELEMENT SHOWING WALL WIPER RINGS

Page 13 of 16

2.02.3



NOTE: "Courtesy Sulzer Chemtech Ltd." Released in R. Plüss fax of 10/27/94

Figure 7. FRI VAPOR REMIXING DEVICE (See application warning in text)

Page 14 of 16

2.02.3



Figure 8. PATENTED VAPOR-REMIXING DEVICE (See application warning in text)

Page 15 of 16

2.02.3



2.02.3

Figure 9. McCABE THIELE DIAGRAM DIAGRAM USED TO EVALUATE THE EFFECT OF LIQUID MALDISTRIBUTION ON OVERALL SEPARATING POWER

Page 16 of 16

FRI VOLUME 5: FRA CTIONATION DESIGN DESIGN HANDBOOK

LIQUID FLOW THROUGH GRAVITY DISTRIBUTOR ORIFICES

Issued:

01/31/1998

Revised:

2.02.4


1.0

Introduction ..................................................... ........................................................................................................... ................................................................. ........... 2

2.0

Dynamic Operation ................................................. .................................................... .......................................................... ......2

2.1

Downward Orifice Flow in Fully Turbulent Flow Regime (i.e. Orifice Horizontal) .. 2

2.2

Side Orifice Orifice Flow (i.e. Orifice Orifice Vertical) ................................................ ...................... 5

Page 1 of 8


1


2.02.4

Introduction

This section details the calculation of orifice coefficients of liquids flowing through circular orifices in gravity type distributors. The following discussion and equations, based on empirical studies of orifices formed in flat plates, are not applicable applicable for the design of of tubed distributors. distributors. The orifice coefficient is commonly used in the following form to determine flow rates through nozzles and orifices:

q = CA(2 gh L )

0.5

Where:

q C A g h L

= = = = =

is the volumetric flow rate is the orifice coefficient is the area is the gravitational constant is the loss of static pressure head due to fluid flow

The orifice flow mechanism can be overflow, such as with a V-notch trough, or underflow with submerged orifices in the distributor distributor wall or floor. The two basic types of gravity distributors distributors are pan distributors and trough distributors. distributors. Other variations of specialized distributors distributors are multi-pan distributors for lower flow applications and tubed distributors for fouling and high turndown applications. Distributors can be designed with holes in the horizontal plane for downward flow or with holes in the vertical plane for side flow. Proper sizing and layout of the orifices are essential for a successful distributor design. Correct calculation of the liquid flow through the orifices is a major factor in assuring proper liquid distribution. The factors which affect the orifice coefficients for various orifice orientations will be discussed below. The effect of liquid transverse velocity in the distributor pans or troughs on orifice coefficients will also be discussed. Previous research(95,96) on orifice plates in horizontal pipes has shown that the orifice coefficient is mainly a function of the Reynolds Reynolds number, orifice beta ratio, and and pipe diameter. The orifice coefficients used in gravity flow distributors differ from pipe flow because the driving force across the orifice, in terms of velocity head, is generally much lower. Also, the orifice beta ratio term is not applicable for distributors. distributors. Therefore, a different set of correlations are used to calculate the orifice coefficients. Research at FRI has shown that, once the orifice flow is in the turbulent regime (Re > 3000), the orifice coefficients for a horizontal orifice in a trough or pan type distributor are determined solely by the liquid head above the orifice plate. (97) With heads exceeding 12" (305 (305 mm), the limiting limiting orifice coefficient coefficient value is approximately 0.7.

2

Dynamic Operation 2.1

Downward Orifice Flow in Fully Turbulent Flow Regime (i.e. Orifice Horizontal)

In past practice, the most common type of orifice distributor has consisted of pans or troughs with orifices in the bottom to facilitate facilitate downward flow. The advantage of this design is its its simplicity and low cost. The disadvantage disadvantage of this design is the susceptibility susceptibility to fouling fouling due to the the accumulation of scale settling in the the bottom of the pan. A suggested design practice practice is to avoid using orifice diameters of less than 0.25"(6 mm) to 0.375"(9 mm) in any suspected fouling service.(94) Page 2 of 8



2.02.4

Instability in low liquid head region - FRI research has shown that orifice coefficients are always unstable, fluctuating between 0.5 and 1, if the liquid head is below 1"(25 mm) and is always stable above 3"(76 mm). When orifice coefficients become unstable, the flow through individual holes or sections in the distributor will vary and the distributor performance will deteriorate. If calculated head levels fall below 1"(25mm), consideration should be given to reducing the orifice diameter or number of orifices in order to increase the head level.

These results assume that the effects of transverse velocity are negligible. However, in many cases, these effects are not negligible. (33) In further FRI testing, a correlation was developed to derate the orifice coefficient in order to account for the effects of transverse trough velocity (98). This will be discussed later. Flow characteristics vs. Head for horizontal orifices - Figure 1, taken from FRI Topical Report 115, depicts varying liquid heads flowing through a horizontal orifice. Measurement points for horizontal and vertical orifices are shown in Figure 2. Unstable - head less than 1"(25 mm) : In this region, a typical orifice coefficient value is 0.75, but may fluctuate anywhere from 1.0 to 0.5. The orifice coefficient is highly unpredictable. Intermediate - 1"(25 mm) < Head < 3"(76 mm): In this region, decreasing the head from 3"(76 mm) to 1"(25 mm) will directionally increase the orifice coefficient. This change is more pronounced with lower ratios of orifice diameter to plate thickness. At higher ratios of orifice diameter to plate thickness, the orifice coefficient may actually decrease slightly with a decrease in liquid head. Stable - head greater than 3"(76 mm): In this region, the orifice coefficient follows a gradual asymptotic decrease to the lower limit of approximately 0.7 as head levels increase. Influence of orifice diameter and plate thickness - The FRI study found that both orifice diameter and plate thickness have some effect on orifice coefficient. The orifice coefficient generally increases with decreasing hole diameter and/or increasing plate thickness. The study has also shown that the orifice coefficient approaches a minimum value when the ratio of the orifice diameter to plate thickness approaches 1. Correspondingly, the orifice coefficient approaches a maximum value as the ratio of orifice diameter to plate thickness approaches a value of 3. Influence of orifice manufacture - It has been noted that the shape of the leading edge of the orifice can influence the orifice coefficient. (94) This shape is largely determined by the method of manufacture of the orifice. In either the inlet or outlet edge of the hole, it appears that burrs control the orifice coefficient and negate the rounding effect of the punch entry. The FRI research showed that the orifice coefficient is approximately 10% greater for flow with the punch direction through a deburred orifice than with any other combination. Common practice is for the orifice to be punched with the direction of liquid flow.

The following correlations for orifice coefficient are based on non-deburred holes punched in the direction of flow. Influence of surface tension - The FRI study performed tests using water and Isopar M ™ to evaluate the effects of surface tension on orifice coefficients. The test results indicated, that within the realm of this testing (22 - 70 dynes/cm), surface tension does not have any significant effect on the orifice coefficient.

Page 3 of 8



2.02.4

Influence of viscosity - As stated earlier, the correlations from the FRI research are based on the premise that the flow through the orifice is in the turbulent regime. The fluids tested in the studies had viscosities in the range of 0.9 cP to 2.3 cP. Fluids with viscosities outside of this range, especially highly viscous fluids, will have a direct effect on the Reynolds number and possibly create a laminar flow condition. When dealing with fluids outside of the above mentioned range of viscosities, additional information will be required to determine whether or not the given correlations are still applicable. Orifice coefficient correlations with horizontal orifices - Based on the FRI study, the correlation for orifice coefficient is as follows: ⎛ D ⎞ ⎡ − 0 . 129 ⎜ − 3 . 0 ⎟ 0 . 18 T ⎝ ⎠ C ( inches ) = [0 . 807 − 0 . 04141 n (h L )] ⎢ 0 . 93 + 0 . 0118 h L + 0 . 102 e ⎢ ⎣

⎛ D ⎞ ⎡ − 0 . 129 ⎜ − 3 .0 ⎟ 0 . 18 T ⎝ ⎠ + 0 . 102 e C ( mm ) = [0 . 941 − 0 . 04141 n (h L )] ⎢ 0 . 93 + 0 . 0066 h L ⎢ ⎣

Where:

C hL

D over T

2

2

⎤ ⎥ ⎥ ⎦ ⎤ ⎥ ⎥ ⎦

= the orifice coefficient = the liquid head measured from the bottom of the orifice plate = the ratio of the orifice diameter to the plate thickness.

Limitations of equations - These correlations do not account for the pressure drop across the distributor due to upward flow of vapors around the distributor. Use of these correlations is limited to the range of the orifice plate geometry used in the study. Specifically:

Orifice Diameter: Plate Thickness: D/T: hL: Surface Tension: Viscosity:

0.25" (6.4mm) - 0.75" (19.1 mm) 0.075"(1.9 mm) - 0.25" (6.4 mm) 3-7 1"(25 mm) - 30"(762 mm) 22 dyn/cm - 72 dyn/cm 0.9 cP - 2.3 cP

Effect of transverse velocity on downward orifice flow - Transverse flow describes the flow of liquid traveling horizontally across distributor deck towards the orifices. Transverse flow velocity in the distributor has two significant effects on orifice flow. First, it reduces the orifice flow. At constant liquid heads, the orifice coefficient with the transverse flow velocity is smaller than without the transverse flow velocity. The higher the transverse flow, the smaller the orifice coefficient will be. It should be noted, however, that the influence of the transverse flow velocity on the orifice flow is dampened as the liquid head increases above the orifice. The second effect is that the transverse velocity causes the liquid jet from the bottom of the orifice to deflect in the direction of the transverse flow. The higher the velocity, the greater the angle of deflection. This can be a significant problem as the distance from the distributor to the top of the packing is increased.

If the angle of deflection is a significant problem, the distributor will have to be modified to alleviate the problem. As mentioned above, the design liquid level in the distributor can be increased to limit the effect. If practical, the distributor can be located closer to the top of the packed bed to decrease the influence of the liquid deflection. Another option to eliminate orifice Page 4 of 8



2.02.4

flow deflection would be the use of a tubed distributor. The Froude number, a ratio of kinetic to potential energy forces, is a parameter that relates the effects of liquid head versus transverse velocity. It makes a significant contribution in the equation formulated to quantify the effects of transverse velocity on orifice coefficients. It is expressed as,

Fr =

Where: U g h L

= = =

U

(gh L )0.5 the transverse velocity in the trough, ft/sec (m/sec) the acceleration due to gravity, 32.2 ft/sec 2 (9.81 m/sec2) the liquid head measured from the bottom of the orifice to liquid surface, ft (m)

Orifice coefficient correlation - A dimensionless derating factor should be used to account for the effects of transverse liquid velocity on downward flow orifice coefficients. This correlation, developed by FRI, is shown below. 2 ⎡ ⎤⎛ D ⎞ Fr = 1.0 − ⎢ ⎟ 2 ⎥⎜ Co ⎣ (1.0 + 4.0 Fr )⎦⎝ T ⎠

Ct

Where: Ct Co D T

2.2

= = = =

− 0 .5

the orifice coefficient with transverse velocity the orifice coefficient with no transverse velocity the orifice diameter the plate thickness

Side Orifice Flow (i.e. Orifice Vertical)

Horizontal orifice flow is similar to downward flow with respect to the orifice coefficients, being mainly determined by the liquid head and orifice diameter (99). However, the relationship between side orifice coefficients and orifice diameter and plate thickness is different from that of horizontal orifice coefficients. With all other factors being equal, side flow orifice coefficients are generally smaller than downward flow orifice coefficients. Orifice plate thickness also has a much weaker influence on side orifice c oefficients than on downward orifice coefficients. Influence of trough bottom, trough side wall, or other orifices - Study results show that the proximity of the bottom of the trough to the orifice has no measurable influence when the center of the orifice is greater than 2"(51 mm) above bottom. The study also shows that the trough end wall has an influence only on orifices with diameters of at least 0.75"(19.1 mm). The influence is negligible with smaller orifices that are greater than 4.5"(114 mm) away from the side wall or each other. Instability in low liquid head region - The effect of liquid head on side orifices is very similar to that of downward flow orifices. Generally, side orifice flow becomes stable after the liquid head is greater than 2"(50.8 mm) above the centerline of the orifice. With smaller diameter orifices (0.125" or 3 mm), the flow does not become stable until the flow through the orifice is well into the turbulent regime (Re > 4500). For 0.125"(3 mm) holes, this occurs when the head exceeds 5" (127 mm).

Page 5 of 8



2.02.4

Orifice coefficient correlations for vertical orifices – English Units 2 ⎡ ⎛ 0.5 ⎞ ⎛ (T − 0.083) ⎞ ⎤ If D < 0.5 : C = [0.764 − 0.0322 ln (h L )] ⎢1 + 0.0266⎜ ⎟ − 0.078⎜ ⎟ ⎥ ⎝ D ⎠ ⎝ 0.083 ⎠ ⎥⎦ ⎢⎣ 2 ⎡ ⎛ (T − 0.083) ⎞ ⎤ If D ≥ 0.5 : C = [0.764 − 0.0322 ln (h L )] ⎢1 − 0.078⎜ ⎟ ⎥ ⎝ 0.083 ⎠ ⎥⎦ ⎢⎣

S.I. Units

If D < 12.7 : C = [0.868 − 0.0322 ln (h L )] If D ≥ 12.7 : C = [0.868 − 0.0322 ln (h L )]

Where: C h L D T

= = = =

2 ⎡ ⎛ 12.7 ⎞ ⎛ (T − 2.108) ⎞ ⎤ ⎟ − 0.078⎜ ⎟ ⎥ ⎢1 + 0.0266⎜ D 2 . 108 ⎝ ⎠ ⎝ ⎠ ⎣⎢ ⎦⎥ 2 ⎡ ⎛ (T − 2.108) ⎞ ⎤ ⎟ ⎥ ⎢1 − 0.078⎜ 2 . 108 ⎝ ⎠ ⎣⎢ ⎦⎥

the side orifice coefficient the liquid head from the center of the side orifice to the liquid surface the side orifice diameter the orifice plate thickness.

Limitations of equations - Use of these correlations is limited to the range of the orifice plate geometry used in the study. Caution should be used when applying this correlation to design distributors with orifices that are within 4.5"(114 mm) of the trough wall or each other. These correlations do not account for the pressure drop across the distributor due to upward flow of vapors around the distributor. Other limitations for use of these equations are:

Orifice Diameter: Plate Thickness:

0.125"(3 mm) - 0.75"(19.1) (Note 1) 0.075"(2 mm) - 0.146"(4 mm) ( Note 2)

D over T:

3-7

hL: Reynolds Number: Surface Tension: Viscosity:

2"(50.8 mm) - 30"(762 mm) (Note 1) > 4500 22 dyn/cm - 72 dyn/cm 0.9 cP - 2.3 cP

Note 1 Note 2 -

0.125"(3 mm) orifice data is valid only for liquid head levels greater than 5"(127 mm) because lower heads result in flows that are in the laminar flow regime. 0.125"(3 mm) orifice data is not valid when used with 10 gauge (0.146" or 4 mm) plate because the T/D < 1.

Page 6 of 8



Figure 1.

Page 7 of 8

2.02.4



Figure 2. LIQUID HEAD MEASUREMENT

Page 8 of 8

2.02.4


BED LIMITERS AND HOLDDOWN PLATES

Issued:

11/01/1999

2.03

Revised:

BED LIMITERS AND HOLDDOWN PLATES

BED LIMITERS AND HOLDDOWN PLATES .................................................................. .... 1 1.

General Comments .............................................................................................................2

2.

Bed Limiter .............................................. ........................................................................... 2

3.

Holddown Plate ................................................................................................................. .2

4.

Design Features Applicable to Both Types of Device........................................................ 3

Page 1 of 3


BED LIMITERS AND HOLDDOWN PLA TES

2.03

The following definitions are used for these items of equipment. A bed limiter is a device which is located immediately above the packing and which is securely attached, directly or indirectly, to the column wall. A holddown plate is a device which rests directly on top of the packing and restrains upward movement of the packing only by the weight of the device.

1

General Comments

The use of some device which provides a positive restraint on any significant upward movement of the packing is normally regarded as good design practice for random packing and is generally recommended for structured packing. For high open area grids, where the upper few layers are often linked together, and the lower layer is positively attached to the grid support, a separate holddown means is not normally used. However, such a device should be considered for situations where serious upset conditions might be expected. It is essential to avoid upward movement of packing for a variety of mostly self evident reasons. With random packing, particularly on relatively shallow beds, unrestricted upward movement can give rise to an uneven bed height leading to all manner of process inefficiencies and restrictions. Elements can become lodged into distributors, pipe nozzles, etc. and migrate to various parts of the unit. With structured packing, while such movement in normal operation would be unexpected, movement is still possible. For example, hydraulic surge during an operational upset, can dislodge structured packing elements which will not, of course, return to their former location.

2

Bed Limiter

Design will vary according to whether random or structured packing is being used and whether the device forms an integral part of the liquid distributor. In the latter case, the unit may be attached to the column wall using leveling bolts or shimmed joints.

3

Holddown Plate

This device, which rests directly on top of the packing, minimizes inter-element movement and consequential erosion. Its use is confined mainly to situations where ceramic or other brittle packings are used. Achieving the necessary weight of a holddown plate sometimes incurs the use of substantial constructional members. As exemplified by recent FRI tests, (22) particular care must be exercised to ensure that these do not interfere with the proper action of the liquid distributor. Mechanical design should be such that rotation about the horizontal axis of the plate cannot occur. It then follows that migration of elements around the periphery of the plate is unlikely to occur.

Page 2 of 3


4

BED LIMITERS AND HOLDDOWN PLA TES

2.03

Design Features Applicable to Both Types of Device

Specific design should be discussed with the supplier. The mesh size, where used, should be considered in relation to the form as well as size of random packing being used. It should not be so small as to present adefinite restriction to the passage of vapor or liquid. It should be particularly remembered that half-ring type packings can find their way through surprisingly seal l mesh holes. The bed limiter and its supporting device should be constructed in such a manner that unnecessary coverage of packing by cross-members is avoided. Angle shaped bars, channel beams or I beams should be avoided to maximize open area and to avoid interference with liquid distribution. Leveling bolts, where used, should have locking or double nuts. It is recommended that a bed limiter is used with structured packing. It is essential to ensure that each block is positively restrained and will not rotate about a fulcrum point.

Page 3 of 3


SUPPORT PLATES

Issued:

11/01/1988

2.04

Revised:

SUPPORT PLATES

SUPPORT PLATES ........................................................................................ .......................... 1 1.

Function .................................................. ............................................................................ 2

2.

Supports for Random Packings .................................................. ........................................ 2

3. Supports for Structured and Grid Packings ............................................... .......................... 2 4. Ladder Predistributor ...........................................................................................................2

Page 1 of 2


1

SUPPORT PLA TES

2.04

Function

The packing support performs the function of supporting the weight of packing including the weight of liquid holdup under flooded conditions. Allowance should be made for absorbing the shock of pressure surges, deterioration due to corrosion, and any additional weight not separately supported (such as a hold down plate or redistributor resting directly on the packing). The support, and any beams used, should be of such a design that vapor distribution is not adversely affected. Liquid leaving the base of the packed bed must be discharged in such a manner as to avoid any deterioration in vapor distribution. Sufficient free area should be provided to permit free flow of downcoming liquid and rising gas.

2

Supports for Random Packings

The so-called gas injection plate, variants of which are produced by most manufacturers, is commonly used with random packings. The corrugated nature of the plate allows liquid to drain preferentially into the troughs, while at the same time allowing vapor to enter the bed without first having to pass through the accumulated liquid. These plates are normally manufactured with perforations whose total crosssectional area exceeds that of the tower. In this way a low, or controlled, pressure drop may be obtained. This feature also results from the ability of the plate design to segregate vapor from any locally accumulated liquid. Non-corrugated plates are also used. These plates are sometimes manufactured from grids with a gauze covering to support small packing elements. While it is essential to avoid downward migration of packing elements, judgment should be exercised when considering a screen covering that it does not create a restriction to vapor/liquid flow that could otherwise lead to premature flooding or other bottleneck. The growing awareness of the significance of poor vapor distribution has focused considerable attention on the nature of the support device. This should be a consideration when deciding upon the type of support to be used.

3

Supports for Structured and Grid Packings

The nature of structured or grid packing permits the use of a simple support grid, which ensures a very large open area. Nonetheless, as with random packings, construction should be such as to minimize the interruption of vapor and liquid flow.

4

Ladder Predistributor

The packing support device should be adequately attached to the tower walls using methods similar to those used for trays. The weight of packing is not sufficient to prevent upward movement. Through bolts are not recommended.

Page 2 of 2

OUTLETS - TOP: INTERMEDIATE LIQUID DRAWOFF

FRI VOLUME 5: FRACTIONATION DESIGN HANDBOOK Issued: Revised:

01/15/1997 10/01/2006

2.05.5


OUTLETS - TOP: INTERMEDIATE LIQUID DRAWOFF .................................................. 1 1.

Introduction........................................................................................................................ 2

2.

Chimney Tray ................................................ .................................................................... 2

3.

Vane-Type Collector ............................................. ................................................... ......... 3

4.

Drawoff for Liquid Sampling ............................................................................................ 3

5.

Draw Nozzle Sizing ................................................. .......................................................... 3

Page 1 of 8

Issued: 01/15/1997 Revised : 10/01/2006

1


2.05.5

Introduction

An intermediate liquid drawoff may be either a partial draw or a total draw. With a partial draw, only a fraction of the liquid in the column is withdrawn while the remainder flows down the column as internal reflux. With a total draw all of the liquid is withdrawn. In the case of a total draw, a portion of the liquid withdrawn (or in some cases all of that liquid) may be returned to the tower below the draw point as external reflux. The external reflux may be cooled, heated, stripped, etc. before it is returned to the column. A partial draw gives only indirect control of the internal reflux which may not be acceptable in some cases, e.g. when the internal reflux rate is small compared to the draw rate and a small change in draw rate could upset the tower. Intermediate liquid drawoffs are normally made with a chimney tray or a vane-type collector. Other devices are sometimes used when only a very small fraction of the liquid is withdrawn.

2

Chimney Tray

A chimney tray is the most common means for withdrawing liquid from a packed tower. Two typical chimney tray designs are shown in Figure 1. Use of a draw sump is preferred to a nozzle flush with the chimney tray floor as this provides additional head above the outlet nozzle without increasing the riser height. Side draw sumps are shown in Figure 1. A center sump is often used in larger diameter towers. The number and arrangement of the vapor risers should be selected on the basis of providing proper vapor distribution to the packed bed above at a reasonable cost. Risers may be round or rectangular. Rectangular risers are less expensive to fabricate. The total riser area is typically 15 to 25% of the tower area, but may be much greater in vacuum service. The annular area between the top of the riser and the hat should be at least 1.25 times the riser area. Spacing between the riser hats and the packed bed above must be sufficient to allow the vapor to spread so that none of the packing is starved of vapor. With "narrow" risers, spacing between the riser hat and the packed bed should be at least 12 in. (305 mm) and a 18 in. (457 mm) spacing is preferred. A tray with fewer but wider risers will require more height to provide adequate vapor distribution. In larger-diameter towers, beams may be used to support the packing. Solid beams may interfere with vapor distribution if there is not sufficient height between the bottom of the beam and the riser hats. Tray seams generally need to be sealed to prevent leakage. Gasketing the seams is sufficient for some services, but seal welding is often required. With a partial draw, it is necessary to redistribute the internal reflux which flows from the collector tray to the packed bed below. The preferred design is to have the collector and redistributor as separate devices and to flow from the collector to the redistributor. A design with a separate collector and redistributor is shown in Figure 2 of Section 2.02.3 ("Redistribution in Packed Columns"). By adding a side downcomer(s) to the chimney collector tray, it can provide a partial liquid drawoff. Taking a partial draw from the central downcomer is not recommended as this could interfere with liquid distribution. In order to reduce the height of a new tower or to fit more packing in an existing tower, some designs combine the draw tray and the redistributor into a single device by putting orifices or drip tubes in the floor of the chimney tray so that it serves as a orifice plate distributor. One disadvantage of this combined design is that it does not promote liquid mixing. Another disadvantage is the difficulty of feeding liquid to the packing below the draw sump. Guide tubes have been used to direct flow under the sump, however, guide tubes may alter the flow characteristics of the orifices or drip tubes. Alternatively, tall drip tubes in the floor of the sump have been used for the same purpose. The additional complication of the guide tubes or tall drip tubes reinforces the need for flow testing (see Section 2.02.1, part C). .

Page 2 of 8

Issued: 01/15/1997 Revised : 10/01/2006

3


2.05.5

Vane-Type Collector

As shown in Figure 2, a vane-type collector consists of a number of inclined vanes. Liquid is collected in a channel at the bottom of each vane and drained into an annular sump at the vessel wall. In larger towers, there may be a center sump in addition to the annular sump. A vane-type collector has a large fractional open area and consequently a small pressure drop as compared to a chimney tray. By adding a draw nozzle to the sump a vane-type collector can provide a liquid drawoff. If the drawoff is only a partial draw, internal pipework is required to feed the internal reflux from the sump to a redistributor. It should be noted that a vane-type collector will not capture all the liquid falling from the packed bed above and cannot give an absolutely total draw. Nonetheless, the amount of bypassing may be small enough that such a draw could be considered essentially "total" for many services. If an absolutely total draw is required, a chimney tray or an internal head should be used. An example is a multi-service tower where the liquid introduced below the draw tray is a different material than the liquid taken from the draw tray.

4

Drawoff for Liquid Sampling

Drawoff for liquid sampling is discussed in Section 2.08.

5

Draw Nozzle Sizing Partial Draw - With a partial drawoff, elevation of the draw nozzle must be such that liquid preferentially flows to the draw nozzle rather than flowing as internal reflux. This maintains a liquid level above the draw nozzle and keeps the draw piping liquid full. The liquid withdrawn will generally be a bubble-point liquid and a conservative design approach is to design the draw piping such that the pressure at all points is greater than the source pressure to avoid flashing/degassing at any point in the draw piping. Then the point where the draw piping turns downward (point 1 of Figure 3) becomes controlling for nozzle sizing.

An energy balance requires that:

P 0 − P 1

+

V 02

2

− V 1

2 g c

ρ

+

( Z 0

− Z 1

g c

) g

− E =

0

The losses ( E ) include an abrupt contraction at the nozzle and friction losses in the draw piping. The contraction loss is calculated as half a velocity head. For P 1 = P 0 and V 0 assumed to be essentially zero, the minimum required head above the centerline of the nozzle is:

h=

1.5(V 1 ) 2 g

2

+ Δ P f

The available head above the centerline of the draw nozzle must be greater than the required head. Since the liquid in the collector is aerated, the designer must take care that the available and required heads are compared in the same terms, either height of clear liquid or height of aerated liquid. An aerated liquid Page 3 of 8

Issued: 01/15/1997 Revised : 10/01/2006


2.05.5

height is multiplied by an aeration factor to convert to a clear liquid head. An aeration factor of 0.4-0.5 is commonly used. The head available for the liquid drawoff will be set by the liquid height required for the flow of internal reflux. For example, with the chimney tray/redistributor shown in Figure 3, the maximum liquid height is set by the head required for the internal reflux to flow through the openings in the chimney tray floor. Turndown in internal reflux needs to be considered in setting the available liquid height. As noted, this procedure assumes that the liquid withdrawn is aerated. The liquid may disentrain gas downstream of the draw point. Design of the equipment downstream of the draw point needs to recognize that this vapor may be present. Some designers size the draw nozzle for self-venting flow as described below under "Uncontrolled Total Draw". This results in a larger nozzle than the method above. Controlled Total Draw - This refers to a total draw where a liquid level is maintained above the draw nozzle by means of a level controller. The level controller prevents the draw box from being pulled dry. The draw nozzle and piping are sized such that the head available at the lowest controlled liquid level is sufficient to overcome the nozzle entrance loss and friction losses as with a partial draw. The allowance for aeration needs to reflect the reduced residence time at that level. Uncontrolled Total Draw - This refers to a design without a level controller where the liquid flows directly from the draw box to another vessel. Under some conditions, this type of total draw can result in unstable flow as the liquid falling in the vertical draw piping siphons the draw box level down, trapping slugs of vapor which results in surging flow. Surging flow should generally be avoided and it can be avoided by designing the piping for free-fall self-venting flow. With self-venting flow, the liquid velocity in the draw piping is low enough that large vapor bubbles will rise rather than being swept down by the falling liquid. Experimental work has shown that self-venting flow occurs when (74):

( N Fr )l =

V l

ρ l ρ l − ρ g

gD

≤

0.31

If the liquid density is much greater than the vapor density, the density term can be dropped from the equation giving the following equation for the minimum pipe diameter for self-venting flow:

D = 0.0765(Q ) D = 1.115(Q )

0.4

0.4

US Eng. Units SI Units

Kister (49) recommends that for foamy systems the pipe diameter be increased in proportion to the foaming factor. If the outlet nozzle is sized for self-venting flow, the nozzle entrance loss and friction losses will be very small.

Page 4 of 8

Issued: 01/15/1997 Revised : 10/01/2006


2.05.5

NOMENCLATURE

D E g g c h (N fr )l Q P V V' Z Δ P f ρ g Ρ l

Inside diameter of draw pipe Friction or head loss Gravitational constant Conversion factor Head relative to centerline of draw nozzle Liquid Froude number Liquid rate (hot, nonaerated) Pressure Superficial velocity Clear liquid superficial velocity Eleveation Friction loss Gas density Liquid density

Page 5 of 8

US Eng. Units

SI Units

ft ft-lb f /lb 32.17 ft/s 2 32.17 ft-lb/lbf -s ft dimensionless US gpm lbf /ft2 ft/s ft/s ft ft lb/ft 3 lb/ft 3

m m-N/kg 9.8066 m/s2 1 kg-m/N-s2 m dimensionless m 3/s Pa m/s m/s m m kg/m3 kg/m3

Issued: 01/15/1997 Revised : 10/01/2006


2.05.5

12 in (300 mm) Min. 18 in (450 mm) Preferred

For wide risers, the distance between the bott om of t he packed bed and the top of the hat should be at least ½ the hat width.

Figure 1. LIQUID DRAWOFF FROM CHIMNEY TRAY

Page 6 of 8

Issued: 01/15/1997 Revised : 10/01/2006


DESIGN A

DESIGN B Figure 2. VANE-TYPE COLLECTOR

Page 7 of 8

2.05.5

Issued: 01/15/1997 Revised : 10/01/2006


Figure 3. AVAILABLE HEAD FOR PARTIAL LIQUID DRAWOFF

Page 8 of 8

2.05.5


PACKING DATA SHEETS

Issued:

12/15/1990

2.07

Revised:

PACKING DATA SHEETS

PACKING DATA SHEETS ...................................................................................................... 1 1.

Packing Data Sheets .......................................................................................................... .2

Page 1 of 14


1

PACKING DATA SHEETS

2.07

Packing Data Sheets

The FRI “Standard” Packing Data Sheets were developed to enhance and improve communications between purchasers and suppliers of tower internals. Purchasers will benefit by receiving quotations that are more uniform and more likely to be fully responsive to their needs. This can reduce the bid evaluation effort and recycle with suppliers. The standard sheets will also help ensure that suppliers receive the information they require to quote each project to the customer's wishes. Purchasers must be aware that there is considerable disagreement within the industry regarding the basis for flooding calculations for packed towers. The primary disagreement is over the appropriate definition of flood. Traditional definitions have been based on some form of hydraulic measure, such as pressure drop or visible foam layer on top of the bed. More recently however, a new definition, referred to as “Maximum Operational Capacity” or “Useful Capacity”, has been established which is the maximum achievable throughout which retains acceptable fractionating efficiency. The second area of dispute is whether to express flooding on a constant L/V or constant liquid rate basis. Since most distillation applications would operate at nearly a constant L/V ratio, the former approach is probably more useful. However, some services such as refinery crude tower pumparounds and gas treating absorbers operate at constant liquid rate. These variations in definition and basis can be important and users should check which definitions are being applied. Please refer to additional discussion of these calculations in Section 8.2 of the FRI Tray Design Handbook. An optional input for packing efficiency calculations is included on sheet 2. The requested input is based on the data required for an FRI type of efficiency model. The lambda value is the minimum possible input. Lambda is the ratio of the slope of the operating line and the equilibrium line. The diffusivities are optional input and if not supplied are frequently calculated from approximations for normal paraffinic hydrocarbons. Refer to Section 8.4 in Volume 2 of the FRI Tray Design Handbook for further information on efficiency calculations. Also on sheet 2 is an input for packing volume to be shipped. This volume should be greater than the geometric volume of the bed to ensure the desired packed height is achieved. Excess packing (typically 5-15% of bed geometric volume) is required because the density of packing pieces is greater inside the column than in the shipping containers due to wall effects and compression. The excess volume required is frequently expressed as a function of the tower diameter to packing diameter ratio although other factors may also be involved. This value should be included in all vendor bids to ensure quoted volumes are equal. Packing manufacturers are a source of data for this value if you do not have in-house experience. Sections 2.01.1 and 2.01.2 of this manual also have information on excess packing volume and methods of installing packing. Sheet 3 contains information on the design of the liquid distributors and accumulator trays. This information is very important, given the critical dependence of packing efficiency on distributor performance and the wide range of costs for the different distributor types. As with other sections of these sheets, the more complete the input including associated sketches, the more uniform the quotations. The FRI Standard Data Sheets are for the most part self-explanatory. In general, the more complete the data, the more uniform and “correct” will be the designs offered. Some of the process data defining vapor and liquid rates are redundant. This was intentionally done to provide flexibility, particularly since tabulated data from the various process simulators are not standardized. Redundant data need not be

Page 2 of 14


PACKING DATA SHEETS

2.07

supplied but are useful to suppliers to catch inconsistencies or mis-copied numbers. Process data, whether from simulations or otherwise, must be the internal loads. Care must be taken where there are subcooled or superheated streams entering the tower. Pumparounds are a common case of a subcooled stream which can dramatically influence tower loadings. Alternate feed locations must be clearly indicated. The data sheets should define more than one set of conditions to show the design range for trays with alternate services. These data sheets are not intended to be stand-al one documents. The purchaser should include with these data sheets any additional information which could impact on the design of the tower internals. It is good engineering practice to include sketches and/or drawings of the tower and any special features, such as existing internal supports, preferred or existing feed arrangements, drawoffs, etc. The locations of all inlets (including alternate feeds) and outlets should be clearly shown, including both elevations and orientations. The locations of vessel manholes should be shown and care must be taken that the manhole size noted on the data sheet is the inside diameter, not the nominal nozzle size. In addition to sketches and/or drawings, more pages of text may be required. This would almost always be true when the purchaser is making a “hardware purchase” and has chosen to define details of the packed tower geometry that are normally established by the supplier. The presence of solids or a two-phase (vapor/liquid) feed will impact the selection and/or design of tower internals. Either occurrence should be accompanied by further explanation. Pertinent information on solids includes the amount and size distribution of the solids, their source, density, and their sticking tendency. Two phase (vapor/liquid) feed conditions should be described on the Feed Data Sheet. The key design information is the mass flow and density of the vapor and liquid phases which are normally obtained from simulation program stream summary printouts. Also, included on the data sheet are a number of mechanical considerations often of specific interest to the process engineer preparing the data sheet. Some of these may also be included in the purchaser's mechanical specification(s) covering tower internals, but some may not. Each point should be considered. It should also be understood that completely filling out this data sheet does not negate the need for including the purchaser's general mechanical specification(s) or otherwise defining what general mechanical specification(s) apply. The data sheet was developed as a service to the membership and all other users, designers, and suppliers of fractionation equipment. It can be used as is, it can be used with any modifications the user desires; or it can be used as input to the user's existing form. The intent is to improve communications, and different users have different approaches as to how that should be done.

Page 3 of 14


PACKING DATA SHEETS

2.07

Sheet 1 of 3 PACKED TOWER DATA SHEET

Client _______________________

Plant Location __________________

Engineer ________________

Job No. _____________________

Inquiry No. ____________________

Date ___________________

Item No. __________________________________

Service _________________________________

Theo. Stage No. 1 = Top/Btm

Section (Name/Description) Theo. Stage Numbers Included Loading at (Top/Mid/Btm) NORMAL VAPOR TO:

Rate, lbs/h Density, lbs/ft 3 Rate, actual ft 3/s Molecular Weight, lb/lb mole Viscosity, cP Pressure, psia Temperature, °F Design Range, % of Normal NORMAL LIQUID FROM:

Rate, lbs/h Density, lbs/ft 3 Rate, US GPM (hot) Molecular Weight, lb/lb mole Surface Tension, dynes/cm Viscosity, cP Temperature, °F Design Range, % of Normal

Page 4 of 14


PACKING DATA SHEETS

2.07

Sheet 2 of 3 PACKED TOWER DATA SHEET Item No. ____________________________________ Service ___________________________________

Section (Name/Description) Loading at (Top/Mid/Btm) PERFORMANCE REQUIREMENTS:

Max. ΔP per Section, in H 2O Max. Allowable Flood, % Derating Factor, fraction Purpose for Derating (Foaming, System, Safety) PACKING EFFICIENCY (optional):

Number of Theo. Stages Req’d. Lambda (mG/L) Vapor Diffusivity, ft 2/h Liquid Diffusivity, ft 2 MECHANICAL REQUIREMENTS:

Bed Diameter, inches Bed Height, ft Packing Volume shipped 1, ft3 Packing Type & Size Packing Material/Thickness Vessel Manhole I.D., inches SUPPORT PLATES:

Type Materials & Thickness Total Corrosion Allowance BED LIMITERS:

Type (Bed Limiter/Hold Down) Materials & Thickness 1

Typically 5-15% greater than geometric volume.

Page 5 of 14

Issued: 12/15/1990

PACKING DATA SHEETS

Revised:

2.07

Sheet 3 of 3 PACKED TOWER DATA SHEET Item No. ____________________________________ Service ___________________________________

Section (Name/Description) LIQUID DISTRIBUTORS/REDISTRIBUTORS:

Preferred Type (Pan/Trough/Pipe/Spray) Service (Distributor/Redistributor) Gravity Types:

Distr’n Density, points/ft 2 Minimum Orifice Size, inches Minimum Liquid Level, inches Installed Level Tolerance, inches Allowable Operating Deflection, inches Pressurized Types:

Number of Nozzles Pressure Drop, psi Materials & Thickness:

Pans, Troughs, Paarting Boxes Pipework Spray Nozzles Bolting Feed Pipe Nominal Size, inches Solids Present? (Yes

No _______)

ACCUMULATOR TRAYS:

Type (circular chimney, rect. trough, chevron, etc.) Open Area, % of column area Riser Height, inches Materials / Thickness Installation: (seal weld / gasket) Design Load: ______PSF with _____ inch deflection at _____F. (for all internals except gravity distributors)

Page 6 of 14

Issued: 12/15/1990

PACKING DATA SHEETS

Revised:

2.07

Sheet 1 of 3 PACKED TOWER DATA SHEET METRIC

Client _______________________

Plant Location __________________

Engineer ________________

Job No. _____________________

Inquiry No. ____________________

Date ___________________

Item No. __________________________________

Service _________________________________

Theo. Stage No. 1 = Top/Btm

Section (Name/Description) Theo. Stage Numbers Included Loading at (Top/Mid/Btm) NORMAL VAPOR TO:

Rate, kg/h Density, kg/m 3 Rate, actual m3/s Molecular Weight, kg/kg mole Viscosity, mPa s Pressure, kPa (bar a) Temperature, °C Design Range, % of Normal NORMAL LIQUID FROM:

Rate, kg/h Density, kg/m 3 Rate, actual m3/h Molecular Weight, kg/kg mole Surface Tension, mN/m (dynes/cm) Viscosity, mPa s Temperature, °C Design Range, % of Normal

Page 7 of 14


PACKING DATA SHEETS

2.07

Sheet 2 of 3 PACKED TOWER DATA SHEET METRIC Item No. ____________________________________ Service ___________________________________

Section (Name/Description) Loading at (Top/Mid/Btm) PERFORMANCE REQUIREMENTS:

Max. ΔP per Section, mbar Max. Allowable Flood, % Derating Factor, fraction Purpose for Derating (Foaming, System, Safety) PACKING EFFICIENCY (optional):

Number of Theo. Stages Req’d. Lambda (mG/L) Vapor Diffusivity, m 2/s Liquid Diffusivity, m 2/s MECHANICAL REQUIREMENTS:

Bed Diameter, mm Bed Height, mm Packing Volume shipped 1, m3 Packing Type & Size Packing Material/Thickness Vessel Manhole I.D., mm SUPPORT PLATES:

Type Materials & Thickness Total Corrosion Allowance BED LIMITERS:



Page 8 of 14

Issued: 12/15/1990

PACKING DATA SHEETS

Revised:

2.07

Sheet 3 of 3 PACKED TOWER DATA SHEET METRIC Item No. ____________________________________ Service ___________________________________

Section (Name/Description) LIQUID DISTRIBUTORS/REDISTRIBUTORS:


Distr’n Density, points/m 2 Minimum Orifice Size, mm Minimum Liquid Level, mm Installed Level Tolerance, mm Allowable Operating Deflection, mm Pressurized Types:

Number of Nozzles Pressure Drop, kPa (mbar) Materials & Thickness:

Pans, Troughs, Paarting Boxes Pipework Spray Nozzles Bolting Feed Pipe Nominal Size, mm Solids Present? (Yes

No _______)

ACCUMULATOR TRAYS:

Type (circular chimney, rect. trough, chevron, etc.) Open Area, % of column area Riser Height, mm Materials / Thickness Installation: (seal weld / gasket) Design Load: ______kPa with _____ mm deflection at _____C. (for all internals except gravity distributors)

Page 9 of 14

Issued: 12/15/1990

PACKING DATA SHEETS

Revised:

2.07

Sheet 1 of 3 PACKED TOWER DATA SHEET

Client

EXAMPLE

Job No.

Plant Location

A-124

Item No.

ANYTOWN

Inquiry No.

89-123456

T-101

Service

Engineer Date

DP 10/16/89

HIGH PRESSURE DISTILLATION

Theo. Stage No. 1 = Top/(Btm)

Section (Name/Description)

Stripping

Rectification

Theo. Stage Numbers Included

1-20

21-40

Loading at (Top/Mid/Btm)

Btm

Top

NORMAL VAPOR TO:

Rate, lbs/h

100000

70000

1.25

1.10

41

30

Viscosity, cP

0.01

0.01

Pressure, psia

80

78

Temperature, °F

247

150

20-100

20-100

150000

80000

39.7

40.1

Molecular Weight, lb/lb mole

57

30

Surface Tension, dynes/cm

25

27

Viscosity, cP

0.8

1.0

Temperature, °F

240

100

25-100

20-100

Density, lbs/ft 3 Rate, actual ft 3/s Molecular Weight, lb/lb mole

Design Range, % of Normal NORMAL LIQUID FROM:

Rate, lbs/h Density, lbs/ft 3 Rate, US GPM (hot)

Design Range, % of Normal

Page 10 of 14

Issued: 12/15/1990

PACKING DATA SHEETS

Revised:

2.07

Sheet 2 of 3 Item No.

T-101

PACKED TOWER DATA SHEET Service ___________________________________


Stripping

Rectification

Loading at (Top/Mid/Btm)

Btm

Top

PERFORMANCE REQUIREMENTS:

Max. ΔP per Section, in H 2O

15

20

Max. Allowable Flood, %

80

80

Derating Factor, fraction

----

----

Purpose for Derating (Foaming, System, Safety)

----

----

PACKING EFFICIENCY (optional):

Number of Theo. Stages Req’d. Lambda (mG/L)

20

20

1.17

0.93

Vapor Diffusivity, ft 2/h

7.5 x 10-3

Liquid Diffusivity, ft 2

4.6 x 10 -4

MECHANICAL REQUIREMENTS:

Bed Diameter, inches

78

54

20 x 2 Beds

25

1400

415

1" Pall

Structured 250 m 2/m3

Packing Material/Thickness

304

304

Vessel Manhole I.D., inches

18

18

Gas Injection

Vertical Bars

Materials & Thickness

304 L

304 L

Total Corrosion Allowance

1 mm

1 mm

Bed Limiter

Bed Limiter

304 L

304 L

Bed Height, ft Packing Volume shipped 1, ft3 Packing Type & Size

SUPPORT PLATES:

Type

BED LIMITERS:



Page 11 of 14

Issued: 12/15/1990

PACKING DATA SHEETS

Revised:

2.07

Sheet 3 of 3 Item No.

T-101

PACKED TOWER DATA SHEET Service_________________________________


Stripping

Rectification

Trough

Pan

Redistrib.

Distrib.

9

9

Minimum Orifice Size, inches

1/4

1/4

Minimum Liquid Level, inches

1

1

Installed Level Tolerance, inches

±1/8

±1/8

Allowable Operating Deflection, inches

1/16

1/16

Number of Nozzles

----

----

Pressure Drop, psi

----

----

304 L

304 L

Pipework

----

----

Spray Nozzles

----

----

Bolting

304

304

8

4

No

No

LIQUID DISTRIBUTORS/REDISTRIBUTORS:


Distr’n Density, points/ft 2

Pressurized Types:

Materials & Thickness:

Pans, Troughs, Paarting Boxes

Feed Pipe Nominal Size, inches Solids Present? (Yes

No _______)

ACCUMULATOR TRAYS:

Type (circular chimney, rect. trough, chevron, etc.) Open Area, % of column area

Circular 20-25

Riser Height, inches

8

Materials / Thickness

10 ga 304 L

Installation: (seal weld / gasket) Design Load:

144

Seal Weld

PSF with 1/16 inch deflection at 300 F. (for all internals except gravity distributors)

Page 12 of 14


PACKING DATA SHEETS

TOWER FEED DATA SHEET

Item No.

Service

Feed (Name/Description) Feed Location Temperature, °F Pressure1, psia

NORMAL VAPOR:

Rate, lbs/h Density, lbs/ft 3 Rate, actual ft3/s Molecular Weight, lb/lb mole Viscosity, cP Design Range, % of Normal

NORMAL LIQUID:

Rate, lbs/h Density, lbs/ft 3 Rate, US GPM (hot) Surface Tension, dynes/cm Viscosity, cP Design Range, % of Normal

1

Feed Conditions at tower operating pressure are required for proper design of feed piping or distributors.

Page 13 of 14

2.07

Issued: 12/15/1990

PACKING DATA SHEETS

Revised:

2.07

TOWER FEED DATA SHEET

Item No.

T-101

Feed (Name/Description)

Service

HIGH PRESSURE DIST.

Main Feed

Reflux

Between Beds 1 & 2

Tower Top

Temperature, °F

270

90

Pressure1, psia

82

80

37000

0

Feed Location

NORMAL VAPOR:

Rate, lbs/h Density, lbs/ft 3

1.18

Rate, actual ft3/s Molecular Weight, lb/lb mole Viscosity, cP Design Range, % of Normal

35 0.01 50 - 100

NORMAL LIQUID:

Rate, lbs/h Density, lbs/ft 3

100000

80000

39.2

40.1

26

27

0.87

1.0

50-100

25-100

Rate, US GPM (hot) Surface Tension, dynes/cm Viscosity, cP Design Range, % of Normal

1

Feed Conditions at tower operating pressure are required for proper design of feed piping or distributors.

Page 14 of 14


SAMPLING IN PACKED COLUMNS

Issued:

12/15/1998

2.08

Revised:

SAMPLING IN PACKED COLUMNS

SAMPLING IN PACKED COLUMNS ................................................................. ................... 1 Samples From Within the Packed Bed ............................................................................. ......... 2 Samples From Other Zones ..................................................... .................................................. 2 Sample Lines .................................................................................................................. ........... 2

Page 1 of 3


SAMPL ING IN PACK ED COLUMNS

2.08

This section aims to establish guidelines for removal of liquid samples from a packed column. It covers sampling from both the packed bed itself and from other locations associated with a packed column. These guidelines are specifically related to packed columns. Reference should be made to AIChE (draft) document on this subject. (29) Also the subject of sampling from trayed columns has been extensively covered,(28) and the reader is particularly advised to consult Section 405 of the latter reference. The general requirement of a liquid sample is that it represents the average composition over the crosssectional area of the column at any given elevation. However, as theory and practice have demonstrated that the composition will frequently not be uniform at a given elevation, sampling from within the bed is not generally practiced. Although comments on such sampling are given below, sampling from inter bed locations, such as redistribution zones, is more satisfactory. The form of sampler used by FRI are shown in Figure 1 and are self explanatory.

1

Samples From Within the Packed Bed

Sampling from within a packed bed is regarded primarily as a research, rather than a production procedure. FRI has long taken samples from within randomly packed beds and more recently from structured packed beds. Tests have shown that the presence of a sampler has no measurable effect on performance.

2

Samples From Other Zones

Care must be taken to ensure that as far as possible the sample is representative. A collector box located at the base of a downcomer is normally satisfactory. For industrial columns the most satisfactory and convenient sampling location is where liquid is collected from the packed bed. This may be at a redistribution zone, or below the packed section. In all cases it will be necessary to ensure that the liquid is well mixed before sampling. Samples withdrawn from below the packed bed should be taken from a collector tray at a location free from vapor.

3

Sample Lines

In all cases sample lines should be as small as possible to avoid the need for prolonged flushing, but large enough to avoid blockage; 5 mm diameter tube is adequate and some form of gauze filter is recommended at the collector pan. Sample coolers should be used where appropriate both for safety reasons and avoid the loss of light components. Note that it is not advisable to withdraw samples from systems which are subject to polymerization.

Page 2 of 3


SAMPL ING IN PACK ED COLUMNS

Figure 1. FRI SAMPLER FOR RANDOM PACKED BEDS NOTE: For structured packing, a single arm only is used, inserted after the packing has been installed.

SECTION A-A

Page 3 of 3

2.08


DE-RATING FACTORS – PACKED COLUMNS

Issued:

01/15/1994

2.10

Revised:

DE-RATING FACTORS – PACKED COLUMNS

DE-RATING FACTORS – PACKED COLUMNS .................................................................. 1 1.

Introduction ........................................................................................................................ 2

2.

De-Rating Staging Calculations ......................................................................................... 2

3.

De-Rating Packing Hydraulics Calculations ...................................................................... 5

Page 1 of 6


1

DE-RATING FACTORS – PACK ED COLUMNS

2.10

Introduction

Many facets of distillation column design involve some degree of uncertainty. Inaccurate, or possibly estimated, system properties (physical properties and vapor/liquid equilibrium, or VLE, data) and errors associated with design correlations for hydraulics and efficiency are among the more obvious sources of uncertainty; other sources of error are listed in the FRI Annual Report from 1973. To account for these uncertainties, the column design is often de-rated by applying correction factors, known as de-rating factors, to the design calculations. Three de-rating factors have been defined by the Design Practices Committee: Foaming Factor, System Factor, and Safety Factor. Extensive definitions of these de-rating factors are given in Section 0.03 of Volume 5 of the FRI Design Handbook. Brief definitions are provided below for convenience: Foaming Factor-allows for the tendency of the system to foam; System Factor-provides a margin of safety when designing outside the range of the physical property database of a design correlation; Safety Factor-allows for uncertainty in the design correlations.

De-rating factors can be applied at many points in the design procedure, both in the staging and hydraulic calculations. Often, a single de-rating factor is used to account for uncertainties in more than one area. Sometimes de-rating factors are applied implicitly (when rules-of thumb are used, for example), so the designer may not even be aware that the design has been de-rated. Occasionally, a design is de-rated in several places. For designs involving great uncertainty, particularly new designs, it is very tempting to de-rate every step of the design. Such indiscriminate application of de-rating factors will result in fractionation equipment that is oversized, and probably will provide much less turndown than is actually desired. This is a problem of particular concern when several individuals are assigned various responsibilities in the design procedure. For example, a thermodynamicist may add safety factor to the physical properties and VLE by predicting the separation to be more difficult than it really is. The process engineer may perform the staging calculations using a reflux ratio greatly in excess of minimum reflux and product specifications much more stringent than actually required. The engineer performing the hydraulics calculations could then set the tower diameter such that the column operates at, say, 70% of its predicted flooding velocity. A system factor, if appropriate, could be applied to further de-rate the hydraulic design. While de-rating a design is prudent to ensure successful operation, it is important to realize when de-rating factor has been applied and to avoid excessive application such as that illustrated above. Excessive de-rating of a design can result in a greatly oversized tower, which will in turn lead to oversized auxiliaries. The application of de-rating factors to staging calculations is discussed briefly below, and the de-rating of hydraulics calculations is discussed in the following section.

2

De-Rating Staging Calculations

The depth of packing required to make a given separation can be determined by a variety of procedures, generically referred to here as staging calculations. There are at least three places in the staging calculations where the design can be de-rated: the System Properties used in the calculation, the Internal Traffic determined by the calculation, and the HETP used to convert the number of ideal stages to an actual depth of packing. De-rating the design to account for uncertainty in one area may be equivalent to de-rating in another area. Further comments on de-rating these areas of the staging calculations are provided below. As always, the designer should realize when de-rating factors have been applied and be wary of double de-rating. Page 2 of 6



2.10

System Properties - The number of stages in found by some staging calculation, usually performed by computer. The accuracy of the simulation is highly dependent upon the accuracy of the system properties. For situations where the VLE and physical properties are known, the results of the staging calculation are often used directly, without modification. In some situations, however, the staging results may be de-rated to account for the inability of binary interaction models to predict the true behavior of the fluid mixture. This may be done by assuming worst-case system properties such that the most difficult separation, within reason, is predicted.

Often, preliminary designs (and sometimes final designs) must be based on estimated system properties. In these situations, one way of de-rating the staging calculations is to use more stringent product purity specifications than are actually required. This effectively increases the number of stages and the reflux rate required to make the separation. When the design is de-rated in this manner, additional de-rating factors are usually not applied. An alternative method of de-rating the staging calculations is to perform the calculations with the proper specifications and/or non-conservative predictions of system properties, then arbitrarily increase the packed height. The potential effects of this de-rating on tower performance can be assessed by resimulating the tower with the additional packing. If the resimulation is performed with the reflux held constant, then adding packed height is equivalent to using a more stringent purity specification - how much more stringent can be determined from the results of the resimulation. The resimulation could also be performed with reduced reflux (within the constraints of minimum reflux, of course) in order to hold the separation constant. If the original system properties (and estimate of HETP) were correct, then the actual tower may very well be operated in this way. It is important that the designer recognize the implications of the additional packed height and ensure that the packed tower internals can operate properly at the reduced reflux rates if the system properties and HETP allow. Refer to the discussion of efficiency de-rating for additional comments and warnings on de-rating designs in this way. Internal Traffic - The liquid and vapor rates used to perform the hydraulic calculations are obtained from the column staging calculations and are influenced by the number of stages used in the design. For a given separation, reflux (and, hence, the liquid and vapor rates in the tower) and the number of ideal stages are related; a typical reflux versus stages curve is shown in Figure 1. The asymptotes on the curve represent the minimum reflux and the minimum number of stages needed to make the separation. Minimum reflux and stages can be determined by using rigorous computer simulations to construct a diagram similar to Figure 1, or by using Chien's method (53). Determination of these limiting values by shortcut methods is discussed by Henley and Seader (54). The actual reflux is usually selected to be 1.1 to 1.5 times the minimum; in the face of uncertainty, one may increase the selected ratio. For instance, if 1.15 times the minimum is the economic optimum, the presence of uncertainty may favor the selection of a higher ratio such as 1.2. Often, the number of stages is set according to the optimum ratio (1.15 in this example), while reflux and reboil according to the higher ratio (1.2 in this example). This practice leads to vapor and liquid rates, reboiler and condenser duties and pump duties which are all de-rated. It is important to remember that additional de-rating factor will probably be applied to the hydraulic calculations (the tower might be designed at, say, 80% of the flooding velocity). Double de-rating the design in this way will result in a column much larger than actually necessary, and could affect the hydraulic design or performance of the distributors.

If the reflux ratio is selected to be some multiple, say C, of minimum reflux, then the required number of stages can be determined directly from the reflux versus stages curve, as shown in Figure 2. The number of stages used for design may then be increased such that the reflux requirements are fairly insensitive to the number of stages that are actually developed in the tower (N1 on Figure 2). If the design reflux ratio remains at C*R MIN, as shown, then the design point has effectively, but arbitrarily, been moved to a curve based on a higher purity product specification. De-rating the design by using higher product purities than required was described in the discussion on de-rating system properties. The designer should be Page 3 of 6



2.10

cognizant of this de-rating if the tower is designed by selecting N and R in this way. As mentioned previously, additional de-rating is probably not justified. A similar situation may arise when designing internals for tower revamps. If the tower is being revamped with higher capacity internals, then there may be some uncertainty as to how many stages will be lost. In this event, it may be necessary to design on the portion of the reflux versus stages curve where the number of stages developed is insensitive to reflux ratio (N2 on Figure 2). If the internals are designed at a higher reflux ratio than necessary (R2 on Figure 2), then the design has effectively been de-rated by designing to a higher purity than is required. The designer should be aware of the effect of such a derating, and should ensure that the internals selected can operate properly at the most optimistic, as well as the most pessimistic, prediction of reflux rate. Efficiency (HETP) - Most designers feel that packing efficiency cannot be predicted with a great degree of confidence. FRI estimates that the efficiency prediction technique described in Section 8.4 of Volume 2 of the Design Handbook is ± 15%, but this model applies only to second generation random packing and is based primarily on hydrocarbon systems with good liquid distribution (64). Empirical rules-ofthumb are often used to estimate an HETP (65),(66). Such rules are based on limited data, however, and may or may not include data that reasonably approximate the system of interest. Furthermore, the quality of the liquid distribution in the test systems on which the rules of thumb are based may or may not reasonably approximate the design conditions. The best efficiency estimates can be obtained from carefully collected and analyzed pilot plant data, research data (such as that produced by FRI), or high quality plant data on a system similar to the one of interest. Plant efficiency data should be of sufficient quality if they satisfy the guidelines set forth by AIChE in their column testing procedure: material balances should close to within 3% and energy balances to within 5% (refer to the AIChE Equipment Testing Procedure for Packed Columns for additional details (29)). Information of this quality is not always available, however, forcing the designer or troubleshooter to rely on some estimation method or rule-ofthumb. Because of the uncertainty associated with the prediction methods, it may be desirable to apply some de-rating factor to the estimate. The inherent uncertainty in the efficiency prediction methods, the scarcity of recommended means for de-rating HETP estimates, and the large de-rating factors that are recommended in some open literature sources should serve as a warning to the designer or troubleshooter to exercise great caution. These uncertainties also underscore the need for pilot testing when working with new or unusual systems.

In some cases, the efficiency estimate may be implicitly de-rated. For example, the HETP used to design a packed column may be estimated from some rule-of-thumb. Most common rules-of-thumb, such as those surveyed by Kister (67), have implicitly incorporated some de-rating to allow for scaling up small tower data, or to compensate for differences in the quality of liquid distribution. When using such rulesof-thumb, additional de-rating factors should be applied with caution. Indiscriminate application of rulesof-thumb may result in excessively conservative estimates of packing efficiency. For designs involving a fair degree of uncertainty, it is tempting to add a few feet of packing to the final design in order to increase the chances of the column making the desired separation. The incremental cost of the packing and some additional shell height is not great, and should certainly be considered along with other alternatives for ensuring successful operation of the column. However, the designer must be aware that this practice is, in effect, an arbitrary de-rating of the HETP. The design could have similarly been de-rated by arbitrarily increasing the diameter to allow additional reflux; this may limit turndown, however. If either method of arbitrarily de-rating efficiency (either extra packing or diameter) is used, the tower performance should be assessed, by simulation, at turndown conditions using the maximum expected efficiency to ensure that the desired range of operation can be obtained in the event the packing performs at the efficiency predicted without de-rating. The results may lead to selecting distributors with greater flexibility than might otherwise have been considered.

Page 4 of 6

Issued: 01/15/1994


Revised:

3

2.10

De-Rating Packing Hydraulics Calculations

The FRI procedure for designing packed columns, described in Section 8 of Volume 2 of the Design Handbook, allows the calculation of a vapor velocity at flooding (V SF), a vapor velocity at loading (V SL), and the column pressure drop at the design conditions (68). FRI also provides a procedure for modifying the flooding calculation to evaluate the possibility of flooding by system limitation. By comparing the design vapor rate and the flooding or loading vapor rate, the percentage of the flooding or loading velocity at which the tower would operate can be calculated:

% Flooding

% Loading

=

=

V S V SF V S

V SL

x 100

x 100

FRI does not make specific recommendations as to the percentage of the loading or flooding velocity at which packed towers should be designed (values recommended in the open literature are typically in range of 70% to 80% of the flooding velocity). FRI does provide packing correlation data comparisons in Section 8.5 of Volume 2 of the Design Handbook (69),(70). These values can be used to account for errors in correlating the experimental data. The FRI packing models can only be applied to first and second generation random dumped packings. As an alternative to designing at a given percentage of the flooding velocity, some modern design procedures are based on the Maximum Operational Capacity (MOC), that is, the maximum rates at which the tower performs the desired separation. The use of MOC has arisen due to the fact that packed towers often lose efficiency before hydraulic flood occurs. The velocity at MOC will be somewhat less than the flooding velocity, so a de-rating factor that would normally be applied to a design based on flooding would be excessive if applied to a design based on MOC (Strigle recommends designing at 80% to 90% of the MOC)(71). Many packed towers are designed so that a specified pressure drop, or pressure drop per unit packed height, is not exceeded. A great number of criteria for establishing upper pressure drop values have been published; most of these are rules-of-thumb listing a range of acceptable values for various services. A great deal of discretion is required of the designer when applying these criteria. Kister has summarized the published values (67). Tray design procedures usually involve a foam factor to account for the foaming tendency of the system of interest. A foam factor, per se, is usually not included in packed column design procedures, although published guidelines for designing for foaming systems implicitly incorporate a de-rating factor. For example, these guidelines might recommend designing foaming systems at a lower percentage of the flooding velocity, say 40%, or lower pressure drop, say 0.25 inches of water per foot, than would otherwise be used. There is insufficient information for evaluating these guidelines. In light of the good foam-handling characteristics of random packings, one would expect the de-rating factors to be somewhat closer to unity that those shown for trays in Table II, Section 1.20-7. Pilot testing of foaming and suspected foaming systems is highly recommended. There is little information about foaming in random packed columns and even less about towers containing structured packing.

Page 5 of 6



Figure 1. REFLUX VS. STAGES CURVE

Figure 2. REFLUX VS. STAGES CURVES FOR DIFFERENT PURITY SEPARATIONS

Page 6 of 6

2.10


PACKING FIRES

Issued:

06/15/2003

2.11

Revised:

PACKING FIRES

PACKING FIRES .................................................................................................................................. 1 1.

Disclaimer ....................................................................................................................................... 2

2.

Introduction and Background .......................................................................................................... 3

3.

Causes of Packing Fires................................................................................................................... 3 3.1

Oxygen.................................................................................................................................................. 3

3.2

Combustible Materials .......................................................................................................................... 4

3.3 Sources of Ignition ................................................................................................................................ 4

4.

Physical Issues for Packed Columns – Hydrogen Sulfide ............................................................... 6

5.

Traditional Fire Prevention Procedures ........................................................................................... 7

6.

5.1

Removing Combustible Deposits and Iron Sulfide .............................................................................. 7

5.2

Removing Packing ............................................................................................................................... 7

5.3

Fire Blankets and Other Isolation Methods ......................................................................................... 7

5.4

Wetting ................................................................................................................................................ 7

5.5

Monitoring ........................................................................................................................................... 8

5.6

General Comments ............................................................................................................................. 8

Recommended Fire Prevention/Mitigation Methods ...................................................................... 8 6.1

Design ................................................................................................................................................... 8

6.2

Operating Procedures .......................................................................................................................... 11

6.3 Maintenance Procedures ..................................................................................................................... 12

7.

Conclusions ................................................................................................................................... 13

8.

Examples With Lessons Learened ................................................................................................. 14

Page 1 of 21


1

PACK ING FIRES

2.11

Disclaimer

Fractionation Research Inc (FRI) and its Design Practices Committee make no explicit or implied guarantee that following the procedures and recommendations outlined in this document will prevent the occurrence of fires in packed towers or mitigate their consequences. The users of these procedures and recommendations apply them at their own risk. FRI accepts no responsibility for losses or damages incurred by users who choose to follow the procedures and recommendations in this document. The information and recommended practices included in this document are not intended to replace individual company standards. The information and recommendations in this document are offered as guidelines for the development of individual company standards and procedures.

Page 2 of 21


2

PACK ING FIRES

2.11

Introduction and Background

Most columns shut down without fires. Yet most refineries and chemical plants at some time or another have experienced spontaneous or hot work ignitions in at least one column. When this occurs inside a column during or after shutdown, the results can be devastating and dangerous. This paper concentrates on the fires related to packing in columns. Many refineries and chemical plants have installed packing in columns as original equipment or conversion from trays. Fires within packed beds have been reported in various services (1,4). Some have occurred in operating incinerator scrubbers. Many have occurred under carefully controlled conditions for safe shutdown maintenance (13). If the ignition occurs while personnel are working in the column, the potential safety hazard can be life threatening. The increased use of thin, high-surface-area structured packing in the last decade has increased the fire hazard of packed columns. Fires have caused:

• • •

Destruction of packing sections (1,6) Distortion and ultimate replacement of column wall sections Complete structural failure of a column (12)

(2,13)

This paper serves to summarize:

• • • • •

Causes of packing fires Physical issues for packed columns Traditional fire prevention methods Recommended fire prevention/mitigation methods Examples with lessons learned

Full understanding of the risks and benefits of packed columns will lead to better design and maintenance practices and will lower the risk of future fires.

3

Causes of Packing Fires

The unique properties of packings offer unique challenges to safe work within columns. Packing has higher surface area than trays (9,10) and the thin metal doesn’t dissipate heat. Packing is harder to clean and inspection is very limited without total removal. Distributors and Pre-distributors limit visual and physical access and retrofit designs may be even more crowded than new designs. Therefore packing is more likely to collect combustibles, but is harder to clean and inspect. Packing fires are related to two different sources. The most obvious fire is a combustible plus ignition plus air. The less known and understood fires are initiated or sustained by chemical reactions (15,16). Ultimately the prevention of fires requires eliminating at least one of the following:

• • • • 3.1

Oxygen Combustible materials Sources of ignition Reactive chemicals or metals

Oxygen

Most columns operate under pressure and are therefore unlikely to encounter a flammable Page 3 of 21

Issued: 06/15/2003

PACK ING FIRES

Revised:

2.11

condition due to air invasion during operation. Vacuum columns are the exception. However, even vacuum columns would have to incur a major mechanical failure to leak enough air into the system to cause a fire or explosion during operation. Some scrubber systems operate with, or have the potential for, both oxygen and combustibles. Air separation systems contain columns that operate with high purity oxygen. Virtually all column internal work will be done in the presence of atmospheric oxygen. In anticipation of a possible fire, facilities and procedures need to be designed to eliminate oxygen by providing fast and safe closure of air entrances. Additionally some chemical processes and some chemical cleaning or neutralizing processes can put oxygen rich deposits on the packing (1,3,6). Proper treatment of the oxygenated deposits will reduce the possibility of an uncontrollable fire (1,3,6) . 3.2

Combustible Materials

The combustible materials in columns can be vapor hydrocarbons, liquid hydrocarbons, polymers, many types of system specific deposits, coke and iron sulfide. Under certain circumstances metal or plastic packing may also be a combustion fuel source (11,14). Most vapor hydrocarbons are stripped out during standard steam-out, wash-out procedures. Most liquid hydrocarbons are flushed out by light oil washes, steam out and wash out procedures. Various chemical system polymers and byproduct depositions may be inert to various wash and neutralization procedures. Coke is inert to normal wash procedures and therefore is a potential combustible residual in most high temperature column operations. Additionally coke is usually porous and can retain hydrocarbons even with steam and water washes. The worst combustible is iron sulfide because it will undergo spontaneous exothermic oxidation under warm dry conditions. 3.3

Sources of Ignition 3.3.1

Fire and Heat

Most fires are started by some combination of spark and or heat in the presence of air above combustibles auto-ignition temperature. Some of the common sources of ignition are as follows:

• • • • •

Sparks from electrical lights and equipment Sparks from grinding equipment Sparks and spatter from welding Flame and spatter from a cutting torch Static electricity spark

A fire or explosion caused by pyrophoric material is ignited by material that is on average at, or close to, ambient temperature, but that heat locally to an autoignition temperature by exposure to air. 3.3.2

Spontaneous Combustion of Iron Sulfide

An atmosphere conducive to forming iron sulfide at low temperatures would include the following (1):

• •

A layer of iron oxide The iron oxide layer must be wet Page 4 of 21

Issued: 06/15/2003

PACK ING FIRES

Revised:

• • •

2.11

The process environment must have at least 0.2 mole percent hydrogen sulfide (H2S) The oxygen concentration must be below about 1.0 mole percent Iron sulfide formation can be accelerated by the presence of cyanides or ammonium hydrogen sulfide (NH4HS) or 1 ppmw dissolved oxygen

An atmosphere conducive to forming iron sulfide at high temperatures would include the following (1):

• • • •

Hydrogen sulfide and/or organic sulfur Low chrome carbon steel Temperatures above 500°F High velocity and/or two phase flow

In an oxygen free atmosphere where hydrogen sulfide gas is present, or where the concentration of H2S exceeds that of oxygen, iron oxide (rust) converts to iron sulfide. Additionally H2S reacts directly with the iron in the shell, piping, exchangers, etc., to form iron sulfide. Most packed systems are some form of stainless steel or other nonsteel metallurgy. However, many such systems are contained within carbon steel vessels and/or piping systems. The iron sulfide may stay at its point of creation or it may migrate throughout the system. Likely collection locations include packing, distributors and collectors. Iron sulfide exposed to air may oxidize back to iron oxide and forms either free sulfur or sulfur dioxide gas. This reaction is exothermic and as a result individual particles of iron sulfide (iron oxide) can become incandescent. This exothermic iron sulfide oxidation can ignite associated carbon deposits, hydrocarbon deposits, or even packing itself. Since no amount of steaming out will entirely remove all the oil from coke or remove any iron sulfide, the potential for a reactive environment will exist. 3.3.3

Chemistry of Iron Sulfide Spontaneous Combustion

Iron sulfide is the most common substance found in refinery distillation columns that is subject to spontaneous ignition. The reaction of rust with hydrogen sulfide forms iron sulfide (1,3): Fe2O3 + 3H2S →2FeS + 3H2O + S Iron at high temperature in the presence of hydrogen sulfide and acids forms iron sulfide. Fe + H2S→FeS + H2 The iron sulfide lies dormant in the column until shutdown when incoming air can cause the iron sulfide to react (1,3). 4FeS + 3O2 → 2Fe2O3 + 4S + heat 4FeS + 7O2 → 2Fe2O3 + 4SO2 + heat These reactions are exothermic and the iron oxide may become incandescent and ignite any flammable material in its vicinity. The white smoke commonly associated with pyrophoric iron sulfide fires (and often mistaken as being only steam) is composed SO2 gas (3) or droplets of sulfurous acid resulting from the reaction of sulfur dioxide with moisture or steam present (1). Page 5 of 21

Issued: 06/15/2003

PACK ING FIRES

Revised:

2.11

SO2 + H2O →H2SO3 3.3.4

Other Metal/Chemical Reactions

Formation of titanium dihydride, TiH2. TiH2 ignites at 224 C

(11,13)

.

Cobalt oxide (CoxOy) decomposes exothermally when exposed to air

(13)

.

A thermite reaction involving CoxOy and Ti to form TiO2 and elemental Co Zirconium metal will burn in air environment

(11,13,16)

.

(14)

.

Aluminum packing will burn in oxygen-enriched environments (21). Many studies have been made of aluminum burn properties in high purity oxygen gas, liquid and mixed gas/liquid environments. The general conclusion is that aluminum packing is no more prone to burn than aluminum trays which have been used for decades (19). Both trays and packing require initiation of the combustion, but can result in a Violent Energy Release (VER) (19).

4

Physical Issues for Packed Columns – Hydrogen Sulfide

All distillation columns with carbon steel walls and piping contain rust. Any presence of H2S or organic sulfur in carbon steel equipment will assure iron sulfide is present. As such, the danger of fires is always present. The reactivity of iron sulfides formed from rust appears to be a function of the concentration of H2S present. There is a greater likelihood of spontaneous ignition when the process involves higher sulfur feeds. Fires caused by the spontaneous ignition of pyrophoric materials usually fade out quickly unless there is an additional source of combustible material to sustain combustion. The presence of coke provides such a source, particularly if it also contains oil that was not removed during steamout. Note that coke will always contain some hydrocarbon “oil” regardless of the time spent steaming and washing. Vacuum column wash zones, heavy oil pumparound zones and packing in high temperature service have the highest probability of coking. Also, within packed beds rust or iron sulfide scale more easily accumulates at the interface between two dissimilar types of structured packing, at the interface of two identical packing layers or in any liquid holdup zone. Not all scale deposits are flammable. Gamma scan and pressure drop surveys are pre-shutdown alternate troubleshooting investigative tests that can assist in the diagnosis of fouled packing. A simple visual observation of the upper and lower surfaces of a packing bed is not an adequate way to determine whether or not the bed is plugged with coke, iron sulfide, or other scale. These surfaces are washed with distributor liquid (upper surface) or entrained material (lower surface) and can remain visually clean for several inches into the bed. If a visual inspection through a gap in the top layer of structured packing to the next layer below reveals any amount of coke or scale deposits, then it is almost certain that the bed is substantially coked or plugged. The following pictures in Figure 1 represent the bottom and top of the same brick of packing from the top layer of a Vacuum Column Wash Oil Bed. The bottom couple of inches of the brick had significant coke while the rest of the packing was clean (1).

Page 6 of 21

Issued: 06/15/2003

PACK ING FIRES

Revised:

2.11

Figure 1

Bottom of packing brick

5

Top of packing brick

Traditional Fire Prevention Procedures 5.1

Removing Combustible Deposits and Iron Sulfide

Light oil, kerosene and various solvent flushes followed by steam out and water washes traditionally have been used to remove hydrocarbons and chemical deposits from various columns. The procedures remove most of the combustibles. However, coke, iron sulfide and insoluble chemical residues are not removed by such techniques. Normally the time honored way of removing coke is by physically breaking up the coke and removing manually. Pneumatic machines ease the task. High pressure water and even blasting have been used. Any carbon steel column handling feeds with any significant amount of sulfur will have iron sulfide scale at various locations in the column. Various oxidizers and neutralizers, including Zyme-Flow⎢ and potassium permanganate, have been used to convert H2S and iron sulfide to personnel and fire-safe components (1,3,7). 5.2

Removing Packing

It is impossible to give a 100% guarantee for removal of all combustibles and pyrophoric materials from structured packing. For both safety and liability concerns, a number of companies and contractors are now requiring structured packing removal before any hot work is started. Even this extreme procedure must be done with advance planning. When pyrophoric deposits on the packing are possible, wetting and safe handling procedures are necessary before, during and after removal. 5.3

Fire Blankets and Other Isolation Methods

Normal precautions during hot work on a packed column would include fire blankets between the hot work and the top of the packing bed. This isolation method has certainly prevented many potential fires over the years. However, fires have still occurred below fire blankets. More specific isolation designs are discussed in the Recommendations section. 5.4

Wetting

The typical methods for preventing the pyrophoric fires have included keeping the iron sulfide and coke deposits wet. Procedures should include repeated wetting as necessary. Since only dry iron sulfide causes spontaneous ignition, continuous wetting has been a good approach. Distributor water washes and through-the-manhole fire hose auxiliary water are used for re-wetting for safety Page 7 of 21


PACK ING FIRES

2.11

and for fire extinguishing as necessary to minimize fire damage. However, it is often difficult or impossible to completely wet, and keep wet, all deposits due to plugging of the packed beds and partially blocked distributors. Since most vacuum columns and some other columns have large diameters, water wash and fire fighting hoses must have enough pressure to reach the far side of the bed from the external manhole. Even the physical presence of pre-distributors and distributors may restrict proper wetting by auxiliary methods. Carbon steel packing should not be used because it will rust in a water-wetted air system. Often the use of fire blankets and other isolation methods eliminates the repeated wetting option. As a result fires still occur. 5.5

Monitoring

The other coincident prevention method is to reduce the temperature of the deposits below 104°F (40°C), which eliminates spontaneous oxidation of the iron sulfide (1,8) . Thermocouples can be fooled by localized cooling from water washing. Hot spots can survive in an unwashed, un-cooled section. Summertime sun can also produce metal temperatures above 104°F (40°C). As a result fires still can occur. Safety monitoring of CO levels at the manholes and trending of column temperatures has been used as an indication of embryonic fire conditions. Beds with fire blanket and other isolation partitions need to be monitored for temperature, sulfur oxides and carbon oxides beneath the partition. Monitoring assures that a smoldering fire below an isolation partition is detected (5,6). 5.6

General Comments

Depriving the fire of oxygen is one method of quenching many fires. Some refineries prepare wooden manhole doors that can quickly seal the manhole entrances in case of fire. Quick closing designs must accommodate any chords, ropes or hoses that may partially obstruct the passageway. Reducing the number of open manholes and minimizing airflow through the tower are other methods of reducing fire damage (6). Any wash-only method of cleaning or neutralizing will have an additional concern. The chemical may not reach all the pyrophoric scale in packed beds that have plugged distributors and/or partially plugged packing. Therefore any chemical wash procedure should consider a complementary steam atomized chemical wash. The potential pyrophoric hazard varies with the level of sulfur and operating conditions of each system. No chemical procedure can guarantee that all pyrophoric scale in a column is neutralized. No procedure can eliminate human error. Therefore, fire monitoring and fire response procedures and equipment must be planned into each opening of a packed column.

6

Recommended Fire Prevention/Mitigation Methods 6.1

Design Trays Instead Of Packing

Traditionally new and revamp designs of columns have been quite straightforward. Today, we squeeze in the most hardware with minimum access for routine maintenance and inspection. The best technology of late has pressed packing, especially structured packing, into more services. Economics and vendor competition have driven metal thickness of structured and random packings to the limit of minimum structural requirements. Since essentially no corrosion is acceptable on a foil-thin packing, more exotic metallurgy has been used in packing service. Page 8 of 21


PACK ING FIRES

2.11

The thin metal of packing has a lower heat capacity than the thick metal of trays. The thin packing is less likely to conduct heat away from a hot spot generated by pyrophoric material. Therefore, the lower the metal thickness, the higher will be the risk for pyrophoric fire. The higher surface area of thin textured metal packings can be depositories of hard to remove coke, polymers, sulfides, oxides etc. The fact is that certain metals and alloys have lower ignition temperatures than carbon and stainless steels and are more difficult to extinguish as the design becomes more surface and self-insulating intensive. These risks of safety, liability, and physical loss must be factored into the engineered solution to any new or revamp design opportunity. Because of some significant losses in the industry, future engineering and management decisions should consider the use of trays instead of packing in high-risk environments (13). The last decade has produced further evolution of high capacity tray technology which in some services provides an economic, safer alternative to structured packing. The hard part of the tray vs. packing decision is discerning how much risk is acceptable with proper engineering design and safe operating/maintenance procedures. Staff and contractors must be dedicated to following safe operating procedures even when pressured by costs and deadlines. New Packing Designs

Some packing fire incidents are related to a collection of coke, iron sulfide or polymers at the interface of packing types or layers (1). Recent design improvements in the industry have eliminated, or minimized, the liquid holdup at the interface of stacked layers. Use of the new design in critical applications may reduce the fire risk by reducing combustible deposits on the packing. Non-reactive Metals

Historically, packing metallurgy has been chosen for resistance to the process operating system. Some metals that are non-reactive to the process can be very reactive under abnormal operating conditions or shutdown conditions. Certain metals like zirconium and titanium are known to sustain a metal fire in air (14,15,16). Even carbon steel in a finely divided state allegedly will sustain a burn under extreme conditions (12). Carbon steel packing should be avoided completely due to its low resistance to oxidation, especially in the presence of water and air. Any oxidation of a thin structured packing will affect its structural integrity as well as providing scale which can migrate and cause system fouling and pluggage problems. The oxidation of iron to iron oxide is exothermic. (Delta Hf, Fe3O4 = -267,000 cal/gmole) 231.55 g/gmole (18). In air stainless steel packing has not been shown to sustain a metal fire. High nickel content alloys are known to have the highest resistance to air-based metal fires (13). Much testing has been done on the use of aluminum in high purity oxygen service for the air separations industry. Many aluminum sieve trays have been replaced by aluminum structured packing to reduce pressure drop. Models have been developed under specified environmental conditions that characterize metals flammability (20). The following is a summary of some of the conclusions that have been published concerning aluminum in high purity oxygen service: 1. Practically all the studied industrial aluminum fire incidents involved liquid oxygen (21). 2. Aluminum sieve trays have been used throughout the industry for over 30 years with no reported combustion incidents (19,21). 3. Structured packing is no more susceptible to burn or Violent Energy Release (VER) than trays (19,21). 4. The presence of some liquid oxygen promotes VER (21). Page 9 of 21

Issued: 06/15/2003

PACK ING FIRES

Revised:

2.11

5. Probability of a burn increases with the partial pressure of oxygen (20,22). 6. The burn rate increases as the test rod diameter (or thickness) decreases and as the oxygen pressure increases (20). 7. One paper suggests the use of copper-structured packing in regions where oxygen concentrations are high (above 97.4%) and aluminum in other areas (22). 8. If a very heavy film of lubricating oil on structured packing were ignited, the burning oil would not be sufficient by itself to ignite aluminum packing (21). Water Flood

One practice is to water flood the column and cover the packing to just below the hot work area (13) . The full column of water provides a heat sink that neither the packing nor wetting can provide. An older column with corroded shell condition and original foundation design will have to be checked to determine that both can accommodate this practice. Under this procedure, hot work can be performed on only one bed at a time. The area above hot work cannot be protected by this method. Note: Certain 300 series stainless steels are susceptible to stress corrosion cracking, if the chloride content of the source water is too high. A metallurgical engineer should be consulted on the compatibility of the packing metallurgy with the water flood chloride concentration. Pressured Water Hose Connections At Manholes

Where there is risk of a packing fire, water should be readily available. So that packing can be kept wet or quenched by the spray of a fire hose through an external manhole, a permanent standpipe with appropriate valves and connections at each external manhole should be installed. The permanent design assures that a temporary arrangement does not have to be re-invented for every shutdown. The permanent design also emphasizes the importance of prevention and preparation. As part of the design, adequate supply pressure must be considered for both the maximum elevation and the maximum reach of the spray into the column. The water source and pump needs to be independent of the rest of the facilities that are shutdown. Note: Certain 300 series stainless steels are susceptible to stress corrosion cracking, if the chloride content of the source water is too high. The firewater source must be tested for chloride content as part of the pre-shutdown procedures. A metallurgical engineer should be consulted on the compatibility of the packing metallurgy with the water flood and firewater chloride concentration. Segregation

Normal practice has been to put fire blankets below hot work and above the packed beds. Fires have been reported with fire blankets in place (13). Even placing a plywood barrier under the fire blanket has not always been successful protection against hot work. Complete and continuous coverage may be hindered by distributor or column internals attachments. Therefore, 100% guarantee against fire is not possible even with the best segregation techniques. The following combination is recommended for maximum fire-safe security (13) :

• • •

Overlapping bolted metal plate covering the full cross-section of the column. Layer one fire blanket overlapping 4-12 inches (100-300 mm) at any seams. Set a second layer fire blanket at 90 degrees to blanket one also overlapping 4-12 inches (100-300 mm) at any seams.

Page 10 of 21

Issued: 06/15/2003

PACK ING FIRES

Revised:

• • •

2.11

Expansion ring to hold blanket edges against the wall. Initial and periodic wetting of the blanket. Continuous monitoring of air temperature and air quality below the fire blanket.

Note: Removal of fire blankets and other segregation equipment can cause a sudden movement of air through the protected section. A smoldering fire can burst into full flame with the new air supply. Continuously monitoring the air quality below the barrier is recommended. For maximum safety, checking the air quality under the barrier is absolutely necessary before removing the barrier. Access For Emergency Exits

Co-current with designing and planning for fire-free installation and planning for fire-free installation and maintenance is the need to design for safe personnel egress in case of fire. Many column designs of the past have addressed access and egress issues only on a minimum basis subordinate to the process and mechanical optimization of space. The time required to evacuate all personnel from an internal job location must be part of the overall planning for the work. The safe evacuation of workers will influence the amount of internals that need to be removed and re-installed during a shutdown. Safe evacuation will also influence the number of workers that can be placed in one location. The safe access provision needs to be part of the shutdown and critical path planning. Once agreed upon, the plans must be rigorously reviewed, invoked and maintained for the duration of the shutdown/turnaround. Fire Snuffing Design

Since oxygen is a necessary ingredient for most fires, the column design should allow for one or more means of eliminating oxygen from the column. Certain metal and chemical reaction fires are an exception to normal combustion. Metal Manhole Inserts - Shutting the external manholes is one obvious way of sealing the column to let the fire self extinguish by exhausting the oxygen supply. However, most external manholes are not designed for quick, emergency closure. Some plants have designed wooden or metal inserts that can instantly seal the opening in case of internal fire. Such designs must allow for any hoses and wires that will be present in the manhole during the shutdown. Inert Gas Purge - Steam, nitrogen and carbon dioxide have been used to purge and remove air from columns for process start-up preparation. However, all such connections to the column would normally be blanked for safety while personnel are inside the column. Locations that have had packing fires have considered procedures that would allow rapid re-connection of steam or other purging medium. Inert gas purges must include sealing of the manholes to prevent re-entry of air to the column. A steam purge may condense in the column and have little or no immediate effect, especially in cold weather. Note: Any inert gas purge designed for fire control must have a fail-safe design that safeguards against personnel exposure at all times. 6.2

Operating Procedures

Contractors do most shutdown/turnaround maintenance. However, the plant operators usually have the responsibility to provide a safe working environment and to define any limitations required to maintain that safe environment. For fire safety there always has to be a close working

Page 11 of 21


PACK ING FIRES

2.11

relationship between plant operators and contractors. Responsibilities must be planned, defined and implemented. Internal Clean-up

The operating department usually has the responsibility to turn the column over for maintenance in a “safe for entry” condition. Removal of toxic or flammable gases, hydrocarbons or process chemicals is covered by routine operating procedures. Heavy oils, H2S, iron sulfide and polymer deposits are more difficult to remove and require special treatment. “A Synopsis of Oxidizers for Industrial Oxidation - Use, Dangers, Limitations, and Advantages” is provided as a table in Appendix 1. Two of the chemical treatments, Zyme-Flow ⎢(Appendix 2) and potassium permanganate (Appendix 3), have detailed descriptions for treating H2S and iron sulfide based deposits. Other similar products may exist. Communication

Communication between the Control Room, Hole Watch and the workers inside the column must be maintained continuously. Training for the Hole Watch persons for packed columns should include fire warning signs and proper responses for personnel safety and emergency response. Radio communication equipment needs to be planned such that the Control Room, Hole Watch, workers inside the column and even Safety Personnel can communicate emergency information. Monitoring Temperatures

Temperatures are key indicators for minimizing risk of spontaneous combustion and as indicators for actual fire. Temperatures may be the first indication of a problem, especially during shift or break times when no personnel are in a particular section of a column. Most temperature indicators have alarm set points. To properly use the existing instrumentation it is necessary to reset the temperature alarms to low limits. Depending on the level of fire risk determined for a particular column, there may be justification to install temporary thermocouples at various locations. One key location is under any fire blanket that seals a section of a column. Monitoring Air Composition

A typical confined space safety procedure is to use a portable analyzer on a periodic basis to assure adequate oxygen and absence of hydrocarbons and other toxic gases. For columns at risk for spontaneous combustion of iron sulfide or other fires there may be justification for auxiliary continuous air composition monitors with alarms. The analysis points should be at each packing bed and/or at each external manhole location. The analyzers should be checking for sulfur oxides and other possible combustion products. 6.3

Maintenance Procedures Removal of Packing

In a column that contains packing, the safest procedure includes removing the packing before beginning hot work. This drastic and expensive process may require purchase of new, replacement packing. The projected costs of removing packing may drive design changes to a more expensive

Page 12 of 21


PACK ING FIRES

2.11

but non-hot work solution. Some operating companies and contractors have already adopted packing removal as an absolute safety criterion before doing hot work in a packed column. Segregation of Hot Work

Other than the spontaneous combustion of iron sulfide, the most consistent commonality of packing fires is the presence of hot work in or around the column. Hot work is defined as anything that will produce or may produce a spark, hot ember or flame. The producer may be as innocent as a flash camera or a steel wrench, or as obvious as a grinder or cutting torch. The use of a cutting torch above a packed bed is a more obvious issue to deal with than adding a new external thermocouple nozzle into the side of a column. Since every column and maintenance circumstance is slightly different from others, each job must have a thorough review and plan. The plan must be appropriate, understood and agreed upon by the plant supervision and safety representative, and the contractor. Water flooding, fire blankets, monitoring and/or other safe practices and even the removal of packing must be considered for each application. The plan and its implementation details must be approved both by an informed contractor supervisor and plant supervision before hot work commences. Hole Watch

In addition to normal Hole Watch training, specific Hole Watch training for packed columns should include the additional risks, hazards and required actions for both prevention of, and response to, incolumn packing fires. An additional level of safety would include a plant safety person on Hole Watch with high-level knowledge of the risks, hazards and required actions plus the plant authority to shutdown a job instantly.

7

Conclusions

• • • • • • • • •

Fires in columns are dangerous and costly. Strict adherence to informed engineering practice and proper procedures reduces the probability of fires. The probability of packing fire damage and injury can be reduced with constant monitoring, egress assessment and fire snuffing procedures. Oxygen, combustible material and an ignition source, or chemical reaction, are required for a fire. Water wash and steamout do not remove or neutralize pyrophoric scale. Zyme-FlowΤΜ or potassium permanganate oxidize pyrophoric iron sulfide scale to sulfates and sulfites. Packing fire safety and economic risk reduction requires extensive Prevention Procedures and Response Procedures. Constant and complete wetting of packing has been shown as an effective fire preventative measure. Packing fire prevention covers Design, Installation, Shutdown Maintenance Planning and Instruction, and strict Implementation of Fire Prevention Plans.

Page 13 of 21


8

PACK ING FIRES

2.11

Examples With Lessons Learened Case 1 Service: Crude Column Light Gas Oil Pumparound Packing Metallurgy: Two stacked grades of SS structured packing Fire Details: The packing bed had not been water washed or treated in any way. The fire started at the interface of the two sizes of structured packing. The fire caused hot spots and bulging around the circumference of the column. The fire was put out rapidly, but the wall section was replaced several years after the event. Lessons Learned: • Washing and neutralization of iron sulfide is required or the section must be kept wet. • Any high temperature sulfur environment can have an iron sulfide scale fire problem. • The interface between layers of structured packing holds up liquid and collects solid coke and scale deposits. Case 2 Service: Unspecified Packing Metallurgy: Grade 2Titanium Fire Details: A battery operated grinder was used for a materials identification procedure (NFPA 481, Appendix B). Sparks from the grinder ignited titanium structured packing which burned. Molten titanium melted through beds of packing and the bottom head of the column. Lessons Learned: • The Thermite Reaction can occur between a metal oxide and another metal. • Accumulations of iron oxide (or other materials such as organic residues, pyrophoric substances, etc.) on titanium structured packing can promote or accelerate combustion of titanium. • Use of water to mitigate an active titanium metal fire can produce negative as well as positive effects. • Entry procedures for towers containing structured packing made from thin sheet titanium should account for situations where the packing is coated with combustible materials, pyrophoric substances, or materials that can react with the packing (e.g., iron oxide). Case 3 Service: Chemical Plant Packing Metallurgy: Top and middle beds were titanium pall rings. Bottom bed Inconel pall rings. Fire Details: An operational upset was caused by a plug in the tower. Titanium packing in the top bed had corroded to 70% TiH2, broken into small pieces and was falling through the top bed support tray. Before the fire, significant quantities of titanium were observed on the liquid redistribution tray above the middle packing bed. A flash fire occurred followed by a titanium fire which burned through the shell in two places.

Page 14 of 21


PACK ING FIRES

2.11

Lessons Learned: • The presence of small particles of titanium contributed to the ease of ignition of the titanium and accounted for burn-through at the two locations. • The most likely mechanism for ignition is the presence of pyrophoric residue (cobalt oxide). • The presence of metal oxides accelerated the oxidation of the titanium via a mechanism known as the Thermite Reaction. • Replace the titanium with Inconel. • Material selection should consider factors such as hydride formation and accumulation of pyrophoric deposits on the packing surface. Case 4 Service: Unspecified Vacuum Crude Column Packing Metallurgy: LVGO section with stainless steel structured packing Fire Details: Packing was removed and steam cleaned outside the column. Overhead nozzle work was added to the scope after packing was re-installed. Despite fire blanket protection, sparks ignited residual hydrocarbons in the bed. Lessons Learned:

• Needed better turnaround planning • The fire blanket as supplied was not enough protection. • Steam cleaning is not effective at removing all hydrocarbons. Case 5 Service: Unspecified Packing Metallurgy: Titanium Random Packing Fire Details: Most likely ignition source was pyrophoric residue on the packing. Lessons Learned:

• •

Switched to Inconel® 625 material Used thicker wall design (nominal 0.018” vs. 0.010”)

Case 6 Service: Unspecified Packing Metallurgy: Unspecified Fire Details: Shift forman did not wait until fire hose was set up to weld a clip to tower wall above the packing. Despite fire blanket protection, hot slag still ignited a small fire within the bed. Lessons Learned:

• •

Don’t short-cut hot work procedures. A fire blanket was not enough protection. Page 15 of 21

Issued: 06/15/2003

PACK ING FIRES

Revised:

2.11

Case 7 Service: Unspecified Packing Metallurgy: Unspecified Structured packing Fire Details: Installation subcontractor cut a new 2” nozzle for a thermocouple from the outside of the vessel. The elevation given was within the packed bed. The cutting torch ignited a large fire of hydrocarbon-laden structured packing. Lessons Learned:

• •

Follow proper hot work procedures. Maintain up-to-date tower elevation and internals records.

Case 8 Service: Ethylbenzene Recycle Tower. Packing Metallurgy: Carbon Steel Structured Packing Fire Details: The fire first initiated around 10:30 pm on Feb. 11, 2001. The tower had been open for 3 days when the fire occurred. The initial fire was extinguished with water. However, fire recurred and moved from bed to bed. Water and N2 purge did not put the fire out. Seven hours later the column fell over. Hot slag or molten metal could have ignited combustible materials on the packing surface below. Lessons Learned:

• • • • •

This fire is alleged to be a metal fire with the CS structured packing acting as the fuel. The packing has a large surface area that is difficult to clean and inspect. There was no way to know what combustible deposits might have been on the packing since no packing was removed for inspection. Hot work should be eliminated above packing or the packing should be removed prior to hot work. If hot work is a must, a better hot work isolation assembly should be constructed.

Case 9 Service: Benzene-Toluene Fractionator in Styrene Plant. Packing Metallurgy: Top bed 316SS wire mesh structured packing, bottom bed 304SS sheet metal structured packing. Fire Details: The tower (14 ft. diameter, 105 ft tall) had been open a week prior to starting hot work to add an additional packing bed on top. The fire started in the top bed around 8 pm at night and was over by 10am the next morning. All crew left at 6 pm and there was no night shift working on the tower. A hole was burned through the top packing bed the size of the hole was about 2 feet at the top, 10-12 feet at ten feet down, and 4 feet at 20 feet down. The shell was deformed from the fire. An 8ft. by 8ft. area on the shell was replaced. No one was hurt. Root cause was not determined. One hypothesis was that sparks from welding somehow ignited the plywood deck. Traces of polystyrene on the packing surface fed the fire. Fire water applied might have resulted in the release of H2 with subsequent H2 explosions. The fire occurred in spite of significant mechanical preparation which included the

Page 16 of 21

Issued: 06/15/2003

PACK ING FIRES

Revised:

2.11

following:

• • • • • •

Covered the top bed with plywood Covered the plywood with fire resistant tarp Put a fire blanket over the plywood and tarp Wedged ex-tray rings against wall to complete the seal between the work area and packing Used suspended buckets to catch slag from hot work Ran water hose to keep fire blanket wet

Lessons Learned:

• • • • • •

Structured packing should be considered contaminated enough to burn even if it looks clean Solvent wash will not completely remove all contaminants because of the tremendous surface area Avoid the use of combustible material such as plywood for isolation. Use metal sheets instead. All packing should be removed prior to hot work. If hot work inside the tower is necessary, use a continuous metal sheet on top of the packing bed and fill with water to create a barrier. Consider a sand barrier as an alternative. Continuous monitoring of the tower temperature. Keep all instruments on the tower working. Consider using temporary fiber optic thermal couples. Emergency planning should consider the handling and impact of fire water. (Is there enough pressure to reach the top of the tower?)

References and Further Readings

1. Bouck, Douglas S., BP Amoco p.l.c , Vacuum Tower Packing Fires, API Operating Practices Symposium, April 27, 1999. 2. Reay, Derek, Pyrophoric fires and column shutdown, BP Technical Bulletin, 24 th September 1996. 3. Sahdev, Mukesh, Pyrophoric Iron Fires, www.ChEResources.com/ironfires.shtml 4. Koch-Glitsch, Inc., The History of Distillation Column Fires Involving Structured Packing , Houston, November 15, 2001. 5. Koch-Glitsch, Inc., Safety Bulletin 05/2001. 6. Ender, Christopher, and Laird, Dana, Koch Glitsch, Inc., Minimize the Risk of Fire During Distillation Column Maintenance, 2002 AIChE Spring National Meeting, New Orleans, LA, March 10-12. 7. Farr, Smith and Steichen, The Clorox Company, Bleaching Agents, Survey, Kirk-Othmer Encyclopedia of Chemical Technology, John Wiley and Sons, Inc., 1992, abstract, www.mrw.interscience.wiley.com/kirk/articles/survfarr.a01/abstract.html. 8. Sulzer Chemtech, Ltd., Sulzer Chemtech, Points for Consideration Regarding Installation and Revamp of Mass Transfer Columns Involving Structured Packing,VCT 460.002-e Rev.1 13.12.01. Page 17 of 21


PACK ING FIRES

2.11

9. Kister, Henry, Distillation Design, McGraw-Hill, Inc., Figure 8.7 10. Kister, Henry, Distillation Design, McGraw-Hill, Inc., Table 8.1 11. Shell Chemicals, Packing Fires Occur!, Presentation at FRI Meeting, May 2002. 12. Chevron Phillips Chemical Company, Safety Alert – St. James Incident Summary, April 18,2001. 13. Markeloff, Bob, Shell Chemical Company, Safety Bulletin on Packed Towers – Three internal tower fire incidents documented, email October 18, 2001. 14. Perry and Poole, Saint-Gobain Norpro, Multi-Company Review of Distillation Column Incidents, November 15, 2001. 15. Mannan, Dr. Sam, Texas A&M University, Mary Kay O’Connor Process Safety Center, Safety Alert: Incident at a Chemical Plant, http://mkopsc.tamu.edu/safety_alert/04_16_01.htm. 16. Mary Kay O’Connor Process Safety Center, A Fire In Titanium Structured Packing Involving Thermite Reactions, http://mkopsc.tamu.edu/safety_alert/08_03_01.htm. 17. 17. Vella, Phiilip A., Carus Chemical Co., Improved Cleaning Method Safely Removes Pyrophoric Iron Sulfide, Oil and Gas Journal, February 24, 1997. 18. Perry’s Engineering Handbook – 4 th Edition, Perry(editor), McGraw-Hill Book Company, Table 3-201. 19. Fano, E., Barthelemy, H., Lehman, J., “Tests of Combustion of Aluminum Packing and Trayed Columns”, Flammability and Sensitivity of Materials in Oxygen-Enriched Atmospheres: Ninth Volume, ASTM STP 1395, T. A. Steinberg, B. E. Newton and H. D. Beeson, editors, American Society for Testing And Materials, West Conshohocken, PA, 2000, page 389 20. Wilson, D. B. and Stoltzfus, J. M., “Metals Flammability: Review and Model Analysis,” Flammability and Sensitivity of Materials in Oxygen-Enriched Atmospheres: Ninth Volume, ASTM STP 1395, T. A. Steinberg, B. E. Newton and H. D. Beeson, editors, American Society for Testing And Materials, West Conshohocken, PA, 2000, pp. 469-496. 21. CGA G-2.8, Safe Use of Aluminum Structured Packing for Oxygen Distillation, Compressed Gas Association, Inc., 1725 Jefferson Davis Highway, Suite 1004, Arlington, VA, First Edition 1993. 22. Dunbobbin, Hansel and Werley, “ Oxygen Compatibility of High Surface Area Materials,” Flammability and Sensitivity of Materials in Oxygen-Enriched Atmospheres: Fifth Volume, ASTM STP 1111 , Joel M. Stoltzfus and Kenneth McIlroy, Editors, American Society for Testing and Materials, Philadelphia, 1991. 23. US Patent, 5551989

Page 18 of 21

Issued: 06/15/2003

PACK ING FIRES

Revised:

2.11

APPENDIX 1 A Synopsis of Oxidizers for Industrial Oxidation (Adapted from Ref. 3) Use, Dangers, Limitations, Advantages Oxidizer Dangerous Exotherm Possible? Dangerous Gas Emission Possible? Dangerous Residue Possible? Safe with Light Hydrocarbons? Liquid Application? Steam Application? pH Restrictions? Create Additional or Hazardous Sludge? Penetration Ability? Secondary Waste Treatment Required? Corrosive? Long Shelf Life? Handling Hazards*

Potassium Permanganate KmnO4

Bleach

Zyme-Flow

Zyme-Ox

Hydrogen Peroxide

Yes

Yes

No

No

Yes

Yes

Yes

No

No

Yes

Yes

Yes

Yes

No

No

No

Yes

No

No

Yes

Yes

No

Not Always

Yes No Yes

Yes No Yes

Yes Yes Yes

Yes No No

Yes No No

Yes No Yes

Yes

Yes

No

No

No

Sometimes

Moderate

Moderate

Excellent

Very Good

Poor

Good

Yes

Yes

No

No

No

Yes

Yes No High

Yes Yes High

No Yes Low

No No Low

Yes No High

Yes Yes High

* Note: Always consult MSDS sheets before using or handling any chemical

Page 19 of 21

Common Sulfide Absorbers Not Normally


PACK ING FIRES

2.11

APPENDIX 2 Use of Zyme-FlowTM

United Laboratories of Houston markets a cleaning service based on the use of a chemical solution (ZymeFlowTM) which can remove certain categories of deposits. A team from the manufacturer applies the system, although United sells the raw chemical directly as well. The system is normally applied in the liquid phase but also can be applied in the vapor phase. In the liquid method, which is the one that has been used most extensively, a mixture of the Zyme-Flow TM and water is pumped to the vessel. This mixture starts at about a 1% concentration of the chemical. The fluid mixture is circulated to provide agitation and heated to around 180 - 200 F (82-93 C) usually by a combination of the residual heat in the vessel being cleaned and direct steam injection heating. During the circulation the mixture is monitored for residual chemical strength. The chemical is consumed in the cleaning process, partly by the mild oxidation of H2S and FeS to Sulfates and Sulfites and partly by the cleaning. The chemical is basically an amine oxide which is a surfactant and a mild oxidizer. CH3-(CH2)nN(CH3)2=O where n = 6-20. A proprietary paper is used to measure the ppm active amine oxide in the circulating solution. If an intermediate storage or surge tank is used, the oil can be removed periodically from the circulating system by shutting off the circulation for about 10 m inutes. The oil will separate and can be decanted for separate disposal or recycle (23). A rough estimate for the quantity of chemical required is 0.5% of column volume, but is dependent on the quantity of iron sulfide and oil or sludge in the system. A United Laboratories representative will provide a more accurate estimate for a specific application. Where possible, the normal process pumps are used for chemical circulation. The whole column, or a part of the column may be cleaned by this method. Limited experience reports indicate that the chemical process cleans faster than steaming alone and does a better job. The vapor phase application uses the chemical dispersed into the steam-out steam at a rate of about 0.5%. The chemical does not vaporize but the droplets are carried through the equipment. This reportedly works well to reduce gas levels and decontaminates for quick entry, but does not flush oil as well as the liquid method. The main advantage of the vapor method is that the steam will carry the Zyme-Flow ΤΜ to all the vapor contacting surfaces within packed beds. The vapor method does not depend on the quality of the liquid distribution and will contact and neutralize iron sulfide on all surfaces that the steam can reach. Both the liquid and vapor methods of cleaning have been used separately and together. A combination of light oil flush, steamout, liquid Zyme-Flow ⎢ followed by vapor Zyme-Flow TM provides the most complete column cleaning and scale oxidation. However, iron sulfide scale that cannot be reached by the liquid wash or the steam wash will not be converted and still can be a hazard (1). United Laboratories sells the chemical and provides assistance and technical support for developing chemical quantities, pumping arrangements, etc., but does not provide its own equipment. A United Laboratories agent will usually work with either the refinery operators or a third party. United provides the site technical support for determining the quantity of chemicals, layout of injection and circulation, testing, chemical addition rates, etc. Environmentally, United claims the resulting material breaks down very quickly. It has an MSDS rating of 0,0,0. The effect on any specific biological water treatment plant needs to be checked. It may be expedient to store the slop oil and water and feed the wastewater treatment system slowly. For further information contact United Laboratories directly in Houston. Telephone: 281-443-0300, Fax: 281-443-0373, Toll Free: 888-898-0300, Web: www.zymeflow.com

Page 20 of 21

FRI internal design vol 2

Recommend Documents