FLUID FLOW
Reboiler and condenser operating problems Seemingly small details related to heat exchangers can have a negative impact on distillation system heat input, cause tray flooding and affect overhead condenser and column capacity. Case studies reported here illustrate some of the difficulties Edward L Hartman and Tony Barletta Process Consulting Services Inc
H
eat exchanger operating problems are frequently the cause of distillation system limitations. Unit process design, individual equipment design and the interaction of heat exchangers and other equipment are all essential to proper unit operation. The ability of a distillation system to move heat to and from the process is directly linked to the design and operation of the reboiler and condenser, and various aspects of this can be seen in the three examples that follow follow.. Case study 1 discusses how the configuration of a condensate pot pressureequalising line and condensate line affected the heat input to the distillation system. The second study reviews the interaction between the reboiler vapour return and tray flooding. Finally, Case study 3 examines how the location and size of auxiliary pieces of equipment, such as valves, affect overhead condenser and column capacity.
Case study 1
CDU debutaniser A large crude distillation unit (CDU) was revamped to increase crude throughput by 25%. Other revamp goals were to improve preheat and product recovery. A comprehensive unit test run was performed to establish the baseline unit performance and determine equipment limitations. As with other areas of the CDU, the debutaniser required significant modifications to achieve revamp objectives. This column was essentially limited by its reboiler and condenser system capacity. The existing overhead condensing system had an air-cooler followed by a water cooler. The column had two existing reboilers to provide heat input. One reboiler used gasoil as the heating medium, while the other employed medium pressure steam. Among other modifications, a new reboiler designed for high-pressure steam replaced the existing medium-pressure steam reboiler. The purpose was to boost the logarithmic-
Debutaniser From small reboiler
600 psig steam TC
LC
To small sma ll reboiler
Steam reboiler
Pressureequalising line Condensate pot LC
Stabilised naphtha
Condensate to flash drum
Figure 1 Revamped steam reboiler mean-temperature-difference (LMTD) between the heating medium and the debutaniser bottoms, making it possible to increase the reboiler heat duty and column heat input. The new steam reboiler was also fitted with a condensate pot to stabilise tower operation. The control philosophy for the reboiler system consists of maintaining the small gasoil reboiler at a constant base-load duty by direct control of the combined gasoil flow. The large steam reboiler duty varies based on the tower bottoms temperature, which controls the steam rate. In this way, the steam to the large reboiler responds to the process heat demand, while the small reboiler duty remains constant. Figure 1 illustrates the configuration of the revamped steam reboiler reboiler.. The aim of the arrangement shown in Figure 1 was to provide direct control of the steam flow to the reboiler. The steam control valve changes the steam flow to the reboiler, which consequently changes the condensing pressure in the reboiler. Varying the condensing pressure equates to changing the con-
47 P T Q SUMMER 2003
www.e ptq.com
densing temperature, which affects the heat transfer in the reboiler. This control scheme results in a fast response time and stable operation. No condensate level should build up in the reboiler tubes. The condensate level should be kept in the condensate pot, preventing steam blow-by into the condensate system. However, not all of the details specified in the basic process design were incorporated into the final equipment design. Note that the pressure-equalising line (or balance line) from the top of the condensate pot was tied to the steam inlet line to the reboiler, downstream of the control valve. The purpose of the balance line is to ensure that any steam coming from the reboiler does not “vapour-blanket” the condensate pot. In normal operation, the flow in the balance line should be close to zero. Experience shows that the most effective way to analyse an engineering problem is to apply fundamental principles. In the case of fluid flow in a conduit or system in which no work is
FLUID FLOW
600 psig steam P1
Pressureequalising line
Z1
Debutaniser reboiler tubesheet view)
P2
⌬Z
Condensate pot
Z2
LC
Condensate to flash drum
Figure 2 Steam reboiler configuration and energy balance reference points being performed, Bernoulli’s theorem expresses the law of energy conservation as follows: “The total head (energy) at any particular point in the system above a common reference level is equal to the sum of the elevation head, pressure head and velocity head”. Figure 2 details the steam reboiler configuration and the reference points for the energy balance. The pressure at Reference Point 1 is the steam pressure. Since there is essentially no flow in the
balance line and the tie point is close to the reboiler inlet, a simplification that can be assumed is that the pressure at the condensate pot (Reference Point 2) is the same as the steam pressure (p2 = p1). Also, within the accuracy of a refinery engineering problem, the velocity at both reference points can be assumed to be the same. Therefore, Bernoulli’s equation can be simplified to: z1 = z2 + hL + ³p bundle
600 psig steam Pressureequalising line
Pressure-equalising line (after correction)
Debutaniser reboiler
Level build-up Unstable level
Condensate pot LC
Condensate pot LC
From medium pressure steam reboiler
To low pressure steam
PC
LC
Steam flash drum
To condensate
Figure 3 New steam reboiler system design details
48 P T Q SUMMER 2003
where z = elevation above reference level, feet hL = line loss, feet of fluid ³pbundle = bundle pressure drop, feet of fluid The equation shows that the elevation head at Reference Point 1 must be equal to the elevation at Reference Point 2 plus the friction losses between the two points, which includes piping line loss and bundle pressure drop. In other words, a liquid head driving force (³z) must build up to move condensate between Points 1 and 2. The difference in elevation between the reboiler and the condensate pot normal liquid level was small because the pot upper tangent line was higher than the bottom of the reboiler. With the balance line being tied to the steam line, the friction losses included the pressure drop of the steam condensing in the bundle. The net result was that the condensate built up a liquid head in the reboiler bundle, reducing the surface area available for steam condensation. This limited the heat duty achievable in the reboiler and consequently either the feed rate to the debutaniser and/or the reflux rate had to be cut back. A higher steam flow will lead to higher pressure drop, which leads to higher condensate backup in the bundle. The correct tie point for the balance line is the steam side of the reboiler near the outlet. Since the steam is on the tube side, the outlet compartment of the channel head would be the correct location. Fortunately in this case, there was an idle nozzle connected to the channel head below the pass-partition plate. A simple pipe run was constructed to tie the pressure-equalising line to this nozzle, making the pressure in the condensate pot the same as the reboiler outlet pressure on the steam side. Once the pressure in the condensate pot was equalised with the reboiler outlet pressure, no condensate liquid level was required to build up in the reboiler bundle. The other detail that was overlooked in the new steam reboiler system design is illustrated in Figure 3. The condensate pot outlet line discharges into a flash drum to recover low-pressure steam. The condensate pot level control valve is located near the condensate pot rather than near the flash drum. The portion of the outlet line downstream of the level control valve has two-phase flow due to the condensate flashing at the lower pressure. This results in excessively high velocities in that part of the line. However, prior to reaching the flash drum, the outlet line
FLUID FLOW
Modified claus Thermal stage
Converter #1 claus
Hydrogenation Thermal stage
Converter #1 claus
AC 95%
97%
Converter #2 claus
Converter #3 claus
H2
H2S/SO2 AC Converter #2 claus
Converter #3 claus
re
Air Converter #4 Hi-activity 99.5%
Thermal stage
Converter #1 claus
H2S 98.8%
A
Direct oxidation
AC
Converter #3 selective oxidation or
Converter #2 claus
Air
C
Air
Figure 4 Toluene recovery column, showing average flows, temperatures and pressures before the onset of flooding combines together with another condensate pot outlet line coming from a medium pressure steam reboiler. The medium pressure condensate has also flashed after the control valve and it is at a lower temperature than the high pressure flashed condensate coming from the debutaniser reboiler. The mixing of these condensate streams at different temperatures leads to rapid condensation of flashed steam, which results in a “steam hammer” in the condensate line. The net result is that the debutaniser condensate pot level control is unstable and the level still builds up into the reboiler. The maximum reboiler capacity cannot be achieved due to less surface area being available for steam condensation.
Case study 2
Aromatics fractionator The toluene recovery column (TRC) in the aromatics complex of a large refinery experienced flooding problems. The feed to the TRC is the bottom stream from the reformate splitter. The TRC splits the feed into three streams: overhead, side-draw and bottoms. The toluene overhead stream is used as com-
mercial grade toluene or blended into gasoline. The side-draw stream produces a higher purity toluene than the overhead stream. It is normally routed to a toluene disproportionation unit where it is transalkylated to yield high purity benzene and xylenes. The TRC bottoms are pumped to the xylene recovery column (XRC). The TRC had two reboilers, with one of them being heat integrated with the XRC. The heating medium for the TRC primary reboiler consists of the XRC overhead condensing vapours. The TRC secondary reboiler operates with hot oil from a closed loop system. The TRC operation was troublesome from its initial startup. Flooding was not unusual at feed rates higher than 25 000bpd and was evidenced by high pressure drop, temperature excursions and products going off specification. This limited the refinery’s ability to recover toluene. The TRC limitation forced the operators to pull more toluene in the reformate splitter overhead, which increased the extraction unit operating costs. At times, the reformer charge rate was cut due to the TRC flooding problems.
50 P T Q SUMMER 2003
Extensive field testing was carried out to find the cause of the flooding. Figure 4 shows the average flows, temperatures and pressures around the TRC prior to the onset of flooding. In order to gather additional information, the column was deliberately pushed into flooding by increasing the reflux and boil-up. The data and trends indicated that the tower started to flood at a reflux rate close to 28 000bpd. After analysing the test data, it was concluded that the trays were flooding prematurely. The data also revealed that different causes contributed to the tower’s poor performance. One of the root problems was the seal pan arrangement of the bottom tray with respect to the secondary reboiler return, which was poorly designed. The TRC primary reboiler is a reboiler-condenser that exchanges heat against the XRC overhead condensing vapours. Under normal operating conditions, this reboiler provides 60% to 70% of the total reboiler duty. The secondary reboiler uses hot oil and acts as a “swing” reboiler, providing the remaining heat to satisfy the total process heat demand. Control of the TRC bottom purity is via the reset of the hot oil flow controller at the outlet of the secondary reboiler. The flow reset is provided by either a differential temperature controller measured between trays 67 and 59, or by a direct temperature controller on tray 67. The bottom tray, 81, was a four-pass
Secondary reboiler return nozzle 30" dia.
Seal pans
Primary reboiler return nozzles 36" dia.
12
5/8"
15 1/2" #81 24" 42"
13'-6" I.D.
Figure 5 TRC secondary reboiler with a single return line
FLUID FLOW
122' 124'
T77 Ring T78
PDC
PC
-1.0
TC
T79
219
126'
1
8.3
T80
2
128'
T81
LC
33020 FC
10 11
DTC
130'
FC
1128
57
132'
58
TRC overhead
14.9
59
134'
N3 N4 N5 nozzles
FC
PC
7210
25920
TRC sidedraw
From reformate splitter
136'
DTC 80
138'
Note: Toluene recovered as overhead and sidedraw=99.5% BPD
FC
140'
18.5
Bottoms liquid level
psig
LC
F
°
19720
Hot oil
XRC overhead
FC
17772
To XRC
Figure 6 Scan lines of active areas adjacent to centre downcomers of bottom six trays and bottom of tower design with side and centre downcomers and seal pans. The primary reboiler has two return lines with the corresponding tower nozzles located off-centre, between the side and centre seal pans at a lower elevation. The secondary reboiler has a single return line with the nozzle located at the tower centreline directly underneath the centre seal pan (Figure 5). The primary reboiler return nozzles follow good design practices by introducing the vapour/liquid mixture parallel to the edges of the seal pans. However, the arrangement of the secondary reboiler return nozzle caused the reboiler vapour to directly impinge on the centre seal pan. This resulted in the reboiler vapour return entraining the liquid overflowing the centre seal pan. This liquid entrainment to bottom tray 81 reduced its capacity and contributed to the premature flooding. A column scan confirmed the problem with the secondary reboiler return configuration. Figure 6 shows the scan lines of the active areas adjacent to the centre downcomers of the bottom six trays and the bottom of the tower, including the liquid level. The scan lines in the space between the reboiler return nozzles and the bottom tray do not indicate a clear vapour space. In a
Figure 7 Process conditions at maximum reflux and boil-up rates after revamp
tower functioning properly, clear vapour exists in that area. This confirms the entrainment of liquid overflowing the centre seal pan to the bottom tray. The other factors that contributed to the column premature flooding were: — The two-pass trays above the feed (1–57) had geometrical and hydraulic characteristics conducive to vapour cross-flow channelling; this hydraulic phenomenon occurs when vapour flows preferentially through the valves or perforations near the outlet weir of the tray — The four-pass trays below the feed (58–81) had a reduced number of valves, resulting in a high dry-tray pressure drop — The centre and off-centre downcomers on all trays did not have anti-jump baffles, which prevent the liquid entering a downcomer from flowing from one pass to another without actually entering the downcomer; this results in loss of tray capacity. During a planned turnaround, the centre seal pan was revamped to prevent the liquid overflowing the seal pan from being entrained to the bottom tray. Tray 81 was revamped to a vapour distribution tray, while the remaining 80 trays were upgraded to high capacity trays.
52 P T Q SUMMER 2003
The objective of the revamp was not only to eliminate the premature flooding but also to increase toluene recovery. The column capacity after the revamp was audited by testing at maximum reflux and boil-up. Figure 7 depicts the process conditions at the maximum rates. The maximum reflux rate achieved was close to 33 000bpd, which is significantly higher than the maximum reflux achieved prior to the revamp. At these conditions, the column was still operating satisfactorily without flooding but the pressure was starting to increase. This test demonstrated that the column was now limited by the overhead condenser’s heat removal capability. The reboiler return arrangement and the trays were no longer the constraint to column capacity.
Case study 3
Submerged condenser The depropaniser in the FCC unit of another large refinery has a submerged (or flooded) condenser, as depicted in Figure 8. The depropaniser overhead condensing configuration is also known as the hot vapour bypass system. With this design, the tower pressure is controlled by varying the condenser surface area exposed for condensation of the
FLUID FLOW
55% open
Vent PC
249 psig 115 F
Closed
°
Reflux drum
220 psig 96 F
LC
°
C3's to treater
Depropaniser CWR CWS HC
Condenser 20% open
Reflux pump
Figure 8 Depropaniser with submerged or flooded condenser column overhead vapour. This is accomplished by changing the condensate level in the condenser. This method is used when there is no net vapour product leaving the reflux drum at normal operation. It is also advantageous when the condensers are large and/or require frequent cleaning because the condensers can be located at grade. The reflux drum is elevated above the condenser to provide enough net positive suction head (NPSH) to the reflux and product pump. The condensate leaving the flooded part of the condenser must be subcooled, meaning that at the condenser outlet temperature the condensate pressure is above the bubble point pressure. This additional pressure above the bubble point pressure is converted to elevation as the liquid is transferred to the drum. On the other hand, as hot bypass
vapour condenses in the reflux drum, it establishes a thin layer of liquid at the level interface which is hotter than the bulk liquid pool in the drum. The interface liquid is in equilibrium with the vapour space. If the liquid from the condenser is not subcooled enough to absorb the total latent heat of the hot vapour bypass, vapours will start to accumulate in the reflux drum. This increases the drum pressure, which in turn causes the liquid level in the condenser to back up. Higher liquid backup means more condensate subcooling, allowing a new equilibrium to be reached in the drum. Conversely, if too much subcooling occurs at a constant hot vapour bypass rate, the drum pressure starts to fall. This causes the liquid level in the condenser to fall, reducing the subcooling
Overhead vapour in
Condensing zone Two-phase Liquid level Liquid
Subcooling zone
Subcooled liquid out
Figure 9 Surface area exposed to condensation and sub-cooling
54 P T Q SUMMER 2003
and allowing a new equilibrium to be achieved in the drum. Therefore, as long as the condenser is not surface area limited, the system will adjust itself by exposing more or less surface area for condensation and subcooling (Figure 9). Column pressure control is performed by the control valve on the hot vapour bypass line. For example, if the drum pressure falls, the pressure controller opens the hot vapour control valve. This raises the top liquid layer temperature and consequently vapour pressure in the drum. Therefore, the liquid level in the condenser rises, reducing the surface area exposed for condensation and increasing the condensate subcooling. This lowers the condensation rate and increases the system pressure, which is the intended end result. On the other hand, if drum pressure increases, the hot vapour bypass control valve closes. This reduces the top liquid layer temperature, which lowers the drum vapour pressure. As a result, the liquid level in the condenser drops, reducing the condensate subcooling. More surface area is exposed for condensation, which decreases the system pressure as intended. In summary, the surface area for condensation varies according to the hot vapour bypass pressure controller, allowing the condenser to meet the heat duty demand and maintain tower pressure. The response of the pressure control loop is highly non-linear. When pressure must be increased, the response is fast because the condensate can flow back from the drum to the condenser without delay when the hot vapour bypass opens. However, when the hot vapour bypass closes to reduce the pressure, the response is slow. The vapour in the receiver must be condensed to lower the pressure, which then leads to a lower level in the condenser. The submerged condenser method requires several key aspects to be designed correctly in order to ensure a trouble-free operation. The level in the receiver should be held as steady as possible because it directly affects the condenser level. The liquid level interface in the drum cannot be agitated. This would cause the vapour pressure in the drum to fluctuate erratically, affecting the liquid level in the condenser and consequently the system pressure. Another key aspect is the hot vapour bypass line and control valve sizing. This should be based on the vapour flow required for condensation in the accumulator and the differential head between the liquid level in the drum and condenser. The amount of sub-
FLUID FLOW
cooling required is also an important design point. Finally, the correct location and configuration of the condensate and vapour bypass piping, as well as provision for purging non-condensable gases are also important for a successful application of this overhead condenser configuration. Originally, the undersized hot vapour bypass control in the depropaniser in question led to poor tower pressure control. The pressure differential between the reflux drum and the condenser was relatively low (approximately 7psi). The hot vapour bypass control valve must impose essentially the same pressure drop (minus vapour line friction losses). However, the control valve and line were undersized for the required vapour flow and differential pressure. Therefore, the control valve had to operate wide open, which led to a deficient drum and tower pressure control. To remedy the situation, a hand-operated control valve was installed in the condensate line from the submerged condenser to the overhead receiver, as noted in Figure 7. The purpose of this valve is to impose pressure drop between the condenser and the reflux drum. This would give the undersized hot vapour bypass more driving force to operate in a more stable range and improve the tower pressure control. However, this brought about the unfortunate effect of reducing the overhead condenser capacity. The additional 20psi of pressure drop in the line imposed by the hand operated control valve requires further subcooling of the condensate so that no flashing occurs in the reflux drum. In order to achieve the additional subcooling, the liquid backup in the condenser increases, reducing the available surface area for vapour condensation. For the same boil-up rate, the tower pressure increases and the condenser duty capacity is reduced. At high column charge rates, the hot vapour bypass closes in order to keep the reflux drum pressure at the set point. The liquid level in the condenser rises up accordingly, until a point is reached where there is not enough surface area to condense the vapours and the tower starts to pressure up. The seemingly simple solution to the undersized hot vapour bypass system actually resulted in a column capacity limitation. Unit charge rate and/or reaction selectivity towards olefins was compromised due to this problem. The correct solution would have been to increase the size of the hot vapour bypass.
Edward L Hartman is a process engineer for Process Consulting Services Inc, Houston, Texas. His work includes revamp design and field trouble-shooting of refinery processes, and he has written numerous technical papers on refinery revamps and troubleshooting. E-mail:
[email protected] Tony Barletta is a chemical engineer with Process Consulting Services Inc. His primary responsibilities are conceptual process design (CPD) and process design packages (PDP) for large capital revamps. CPD work involves heater and other major equipment modifications.
55 P T Q SUMMER 2003
