Reprinted with permission from CEP (Chemical Engineering Engineering Progress), April 2007. Copyright © 2007 American American Institute of Chemical Engineers Engineers (AIChE).
Heat Transfer
Improving Heat Exchanger Designs This article defines and explains the factors that affect heat exchanger design margins. With the proper application of design margins, capital costs can be lowered and plant operation improve improved. d.
Christopher Christ opher A. Bennett R. Stanley Kistler Thomas G. Lestina
Heat Transfer Research, Inc. David C. King
BP p.l.c.
R
ecognizing the need for a more effective procedure for designing heat exchangers, Heat Transfer Research, Inc. (HTRI), a global research and development consortium, organized the Exchanger Design Margins Task Task Force (EDMTF) in 2005. Twenty-five companies — engineering contractors, heat exchanger fabricators and processors (box, p. TK) — are currently represented on the EDMTF. Its goal is to establish a consensus for the definition and application of design margins. This consensus is needed because design margins have traditionally been concealed in fouling factors instead of being explicitly designated, thereby resulting in inconsistent margin application and ambiguous design comparisons.
Defining “design margin” Heat exchang exchanger er design design margin margin is defined as any heat transfer area exceeding what is required by a clean heat exchanger to satisfy a specified duty, as defined by Eqs. 1–3:
⎛ U ⎞ % Excess Area from Fouling = 100 ⎜ clean − 1⎟ ⎝ U actuaal ⎠ al ⎛ U ⎞ % Overdesign = 100 ⎜ actual − 1⎟ ⎝ U required ⎠ r ⎛ U ⎞ % Total Excess Area = 100 ⎜ clean − 1⎟ ⎝ U required ⎠ 40
www.aiche.org/cep
April 2007
CEP CE P
(1) (2) ( 3)
In order of magnitude: U clean ≥ U actual ≥ U required . Modern heat exchanger design software calculates the clean overall heat-transfer coefficient incrementally by: U clean
=
1 Ao,total
n
∑ Uc lele anan, j Ao, j
(4)
j = 1
where the subscript j denotes the variable value at a specific increment, and U clean,j, based on the clean outside area, excluding fouling resistances, is given by: 1 U clean, j
=
1 ho, j
+
Ao, j x w, j Alm, j k m, j
+
Ao, j 1 Ai, j hi, j
(5)
The logarithmic mean area for increment j is: Alm, j
=
Ao, j − Ai , j
( 6)
⎛ Ao, j ⎞ ln ⎜ ⎟ ⎝ Ai, j ⎠
The actual overall heat-transfer coefficient is calculated incrementally from: U actual
=
1
n
∑ Uactu actual al, j Ao, j
Ao, total j
=
( 7)
1
For increment j, U actual,j, the actual overall heat-transfer coefficient based on the clean outside area, including fouling resistances, is therefore:
1 U actual, j
=
1 ho, j
R fo, j
+
+
Ao, j x w, j Alm, j k m, j
+
Ao, j Ai, j
R fi, j
+
Ao, j 1 Ai, j hi , j
The overall heat-transfer coefficient required by a specific heat exchanger to satisfy the process conditions is: U required =
Qspecified
( 9)
Ao, total EMTD
The effective mean temperature difference ( EMTD) is: 1 EMTD
=
1
n
Q
j ∑ LMTD
Qtotal j
=
1
(10 )
j
The logarithmic mean temperature difference over increment j ( LMTD j) is computed by applying the standard LMTD formula to the inlet and outlet temperatures of an increment. The average total fouling resistance, based on the clean outside area, is: R f
=
⎛ ⎞ Ao, j R R + ∑ ⎜ fo, j A fi, j ⎟⎠ n j 1 ⎝ i, j 1
n
( 8)
handled by adding fouling resistances, commonly known as fouling factors , to the overall heattransfer resistance, as shown in Eq. 8. These fouling resistances can be obtained from multiple sources, including company experience and the Tubular Exchanger Manufacturers Association (TEMA) standards (1). Fouling factors have become quite controversial in recent years for many reasons. Published fouling resistances often do not reflect true performance; for some services, they are too high, while for others, they are too low. Fouling factors are static values, but some fouling mechanisms are dynamic. Temperature and velocity can greatly influence fouling, but published fouling factors account for these effects in a limited manner, at best. Fouling factors often implicitly account for uncertainty in the heat transfer methods, which can result in the duplication of uncertainty effects.
(11)
=
Nomenclature
Reasons for adding a design margin Margins are added to heat exchangers during the design stage to account for fouling, uncertainties in heat transfer methods and fluid properties, variable process or ambient conditions, lessons learned from previous experience, and risks associated with an exchanger that does not meet the process requirements. Fouling. Fouling is defined as a conductive resistance that accumulates on the heat-transfer surface. It can lead to an unacceptable pressure drop. A common misconception is that heat exchangers always foul. A few streams that usually do not foul are listed in Table 1a. Other streams, such as boiler feedwater and cooling water, can be maintained relatively clean with proper attention. However, some process streams, such as those listed in Table 1b, can foul heavily. During heat exchanger design, fouling is traditionally Table 1. Fouling tendencies of common streams. a. Streams that Typically Do Not Foul Refrigerants Demineralized Water Non-Polymerizing (Olefin-Free) Condensing Gases Liquid Natural Gas (LNG) b. Streams that Typically Foul Heavily Crude Oil Crude Oil Distillation Overhead Amines Hydrogen Fluoride (HF) Coal Gasification Improperly Maintained Cooling Water
Ai,j Alm,j Ao,j Ao,total EMTD hi,j ho,j j k m,j LMTD LMTD j n Q j Qspecified Qtotal R f R fi,j R fo,j U actual U actual,j
U clean U clean,j
U required
x w,j
= inside area of increment j, m 2 = logarithmic mean area of increment j, m2 = outside area of increment j, m 2 = total outside heat-transfer area, m2 = effective mean temperature difference, K = inside heat-transfer coefficient of increment j, W/m2-K = outside heat-transfer coefficient of increment j, W/m2-K = counting variable, dimensionless = metal thermal conductivity of increment j, W/m-K = logarithmic mean temperature difference, K = logarithmic mean temperature difference of increment j, K = number of increments, dimensionless = calculated duty of increment j, W = specified duty, W = total calculated duty, W = total fouling resistance, m2-K/W = inside fouling resistance of increment j, m2-K/W = outside fouling resistance of increment j, m2-K/W = overall heat-transfer coefficient, based on outside area, including fouling resistance, W/m2-K = overall heat-transfer coefficient, based on outside area, including fouling resistance, of increment j, W/m2-K = overall heat-transfer coefficient, based on outside area, excluding fouling resistance, W/m2-K = overall heat-transfer coefficient, based on outside area, excluding fouling resistance, of increment j, W/m2-K = overall heat-transfer coefficient, based on outside area, needed by a specific design to satisfy process specifications, W/m2-K = wall thickness of increment j, m
CEP
April 2007
www.aiche.org/cep
41
Heat Transfer
Table 2. Uncertainties in single-phase heat-transfer coefficient as a function of geometry (4). Exchanger Geometry
Uncertainty in Heat-Transfer Coefficient
Shell-and-Tube Tubeside Shellside
±10% ±20%–50%
Plate-and-Frame
±10%–30%
Plate-Fin
±20%
The main reason fouling factors are so controversial, however, is that they can result in significant overdesign, resulting in the specification of an expensive heat exchanger with unnecessary area. For many applications, fouling factors should not contribute more than 20% excess area to the heat exchanger design. Uncertainty. There are uncertainties associated with both the fluid properties (2) and the methods used to determine heat-transfer coefficients (3). These uncertainties propagate through the computations (4) and result in a calculated overdesign that can deviate from the true value. This deviation will result in a heat exchanger that performs more or less efficiently than the computations say it should. Literature values for heat-transfer coefficient uncertainties are given in Table 2 (4). Fluid property uncertainties vary dramatically depending upon the predictive method used, and the reader is encouraged to peruse the literature for values (2). Uncertainty propagation is highly dependent on the heat transfer methods utilized, as well as the relative thermal resistances. Variable process conditions. Process conditions can vary due to day-to-day changes in process operations and turn-up and turn-down conditions. Turn-up is particularly critical at present, as existing plants are pushed harder to generate additional revenue. Turn-up can result in exceeding the erosion velocity for the fluid/metal combination and presents the potential for vibration damage to tubular designs. Because turn-down often results in lower velocities, it can cause the exchanger to foul. Variable ambient conditions. When one of the heat exchanger streams is influenced by ambient conditions, it results in a variable EMTD, which affects the unit’s performance, particularly in such devices as air coolers and once-through cooling water exchangers. Process reductions occur when design temperature limits are exceeded. Previous experience. One of the most frequently cited reasons for adding margin to a heat exchanger design is that is how it has always been done. As the example covered later will demonstrate, this is not always the best practice. While accounting for the performance of previously designed units is important, it should not be the only fac42
www.aiche.org/cep
April 2007
CEP
Basic Design Algorithm Detailed design algorithms for fouling mitigation and excess margin reduction have been published elsewhere (5) and will not be reiterated here. Instead, the following is the most basic of design algorithms. 1. Check company experience with the heat exchanger to be designed. 2. Decide on fouling factors. a. If a stream is determined to be non-fouling, do not use a fouling factor for that stream in Eq. 8. b. If a stream is known to foul, use a fouling factor in Eq. 8 according to the company’s best practices. 3. Place the most heavily fouling stream on the tubeside to facilitate cleaning, if necessary, and to avoid the areas of low velocity that occur on the shellside. 4. Design for high velocities within erosion and vibration limits. If possible, a. Tubeside velocity should be ≥ 2 m/s. b. Shellside B-stream (the main crossflow stream through the bundle) (6) velocity should be ≥ 0.6 m/s. Exceptions to this general high-velocity rule for fouling mitigation include corrosion, geothermal brines, and slurries that present an erosion limit. Note the importance of metal selection on corrosion and erosion. 5. Keep overdesign (Eq. 2) between 0 and 20% where industry experience permits. Consider larger overdesigns for tubeside laminar flow, mist flow boiling, and shellside mixture condensation in deep gravity flow.
tor considered. The heat exchanger designer should initially consult the company’s design recommendations. Fouling factors and overdesign should then be assigned as the field experience and anticipated turn-up dictate. Risk. Heat exchangers are often intentionally oversized because of perceived risk. Because certain heat exchangers are particularly essential to operations, the designer will intentionally overdesign a unit to ensure that it will satisfy the duty no matter what occurs during operation. What the designer needs to realize, however, is that excessive overdesign can actually cause fouling and other problems with the exchanger and plant operations.
Problems with excessive design margin The excessive use of design margin has several drawbacks. Clearly, superfluous heat-transfer area translates directly to unnecessary capital cost. Needless heat-transfer area also results in a larger, heavier exchanger; weight and footprint are very important considerations for offshore applications. Worst of all, excessive design margin can also result in accelerated fouling — becoming a self-fulfilling prophesy. Designers often incorporate excess margin by increasing the shell diameter. This increases the cross-sectional area available for flow, resulting in lower shellside velocities for
a given flowrate. Furthermore, the number of tubes increases, which reduces tubeside velocity. Lower velocities often increase the rate of fouling. Over-performance caused by excess heat-transfer area can also accelerate fouling because the process stream temperature change will be greater than desired, requiring the flowrate of the utility stream to be reduced or other measures to be taken. Turn-down results in lower velocities, which can initiate or accelerate fouling.
Triethylene Glycol Contactor Gas/Liquid Heat Exchanger Separator
Heat Exchanger
Gas Export
Compressor
Heat Exchanger
Production Gas
CW
Metering Package
Compressor Cooling Water (CW)
CW Gas/Liquid Separator
Gas/Liquid Separator
Figure 1. Gas compression process at an offshore facility analyzed in the example.
Example To illustrate the significant impact that excess margin can have on heat exchanger design, consider a gas compression process of an offshore facility as depicted schemati Figure 2. One-shell-pass, four-tube-pass (1-4) CEU heat exchanger in service at the example cally in Figure 1. Production gas first offshore facility. enters a partial condenser, where the heavy ends are condensed and the light Table 3. Salient details of the heat exchanger designs. ends cooled. The condensate is subsequently removed in a separator. Then the gas is BFU with In-service Fouling compressed and enters another partial conParameter CEU Factors BFU denser, where the remaining heavy ends are condensed. Effluent from the second partial Heat-Transfer Area, m 2 187 154 93.5 condenser enters another separator and then Tube Material Titanium Titanium Titanium proceeds through a triethylene glycol (TEG) Relative Cost 1.5 1.3 1 contactor to strip water from the process Estimated Weight, kg 5,700 5,350 5,000 stream. The process stream proceeds Estimated Footprint, m × m 0.81 × 7.0 0.78 × 5.9 0.78 × 4.6 through the final separator, is compressed again, and is cooled in a final heat exchangTotal Fouling Resistance, m2-K/W 0.000429 0.000429 0 er before being metered and exported. U clean , W/m2-K 1,540 1,510 1,510 The final heat exchanger in the train is U actual , W/m2-K 928 917 1,510 the focus of this discussion because experiU required , W/m2-K 748 757 1,250 ence has revealed that neither stream fouls under normal operating conditions. The Overdesign, % 24.1 21.1 21.0 shellside fluid is chlorine-treated onceExcess Area From Fouling, % 65.9 64.9 0 through seawater and the tubeside fluid was Total Excess Area, % 106 99.8 21.0 modeled as supercritical methane. Q specified , MW 4.29 4.29 4.29 Three configurations for this final heat EMTD , °C 30.7 36.8 36.8 exchanger were analyzed, and the salient details are presented in Table 3. The base B-Stream (6) Fraction 0.382 0.685 0.691 case is the one-shell-pass, four-tube-pass Tube-Side Velocity, m/s 2.95 2.88 2.89 (1-4) CEU TEMA type that is currently in B-Stream (6) Velocity, m/s 0.86 1.10 1.05 service (illustrated in Figure 2). The cen-
CEP
April 2007
www.aiche.org/cep
43
Heat Transfer
Literature Cited 1.
2.
3.
4.
5.
Figure 3. Two-shell-pass, four-tube-pass (2-4) BFU heat exchanger studied in the example.
6.
Tubular Exchanger Manufacturers Association, “Standards of the Tubular Exchanger Manufacturers Association,” 8th ed., TEMA, New York (1999). Reid, R. C., J. M. Prausnitz and B. E. Poling, “The Properties of Gases and Liquids,” 4th ed., McGraw-Hill, New York (1987). Lestina, T., and K. Bell, “Thermal Performance Testing of Industrial Heat Exchangers,” Advances in Heat Transfer, 35, pp. 1–55 (2001). American Society of Mechanical Engineers, “Single-Phase Heat Exchangers,” ASME Performance Test Code 12.5, ASME, New York (2001). Nesta, J., and C. A. Bennett, “Reduce Fouling in Shell-andTube Heat Exchangers,” Hydrocarbon Processing, 83 (7), pp. 77–82 (2004). Palen, J. W., and J. Taborek, “Solution of Shellside Flow Pressure Drop and Heat Transfer by Stream Analysis Method,” Chem. Eng. Progress Symposium Series, 65 (92), pp. 53–63 (1969).
tral baffle spacing was 22.6% of the shell inside diameter, resulting in a low B-stream (6) fraction and inefficient heat transfer. Individual fouling factors of 0.000176 m 2K/W were used for both streams (which combined via Eq. 11 to yield a total fouling resistance of 0.000429 m 2adjusting heat transfer area. U clean, duty, EMTD, and the K/W). These specifications resulted in a heat-transfer area velocities were effectively the same between the two BFU 2 requirement of 187 m and a total excess area of 106%. designs, confirming the comparability of this approach. The second heat exchanger design investigated was a Comparing the BFU design with no fouling factors two-shell-pass, four-tube-pass (2-4) BFU TEMA type with the other two designs reveals striking differences. For 2 (Figure 3), with the same fouling factors (0.000176 m example, the heat transfer area is reduced to only 93.5 m2, K/W) for both streams as in the CEU exchanger. A lowerresulting in an exchanger that is 23% less expensive than cost front head (TEMA Type B) was used because fouling the BFU with fouling factors and 33% less expensive than will not be a problem, thereby negating the need for easy the in-service CEU. Because no fouling factors were used, access to the tubesheet. An F-shell was selected to reduce the overdesign and total excess area are identical at 21%, a the exchanger footprint and weight via increased EMTD reasonable value that gives flexibility to the process. The across the exchanger. U-tubes were chosen to prevent therweight and footprint of this exchanger are also less than mal expansion problems resulting from the large terminal the other designs, which is an important consideration for temperature difference of this exchanger. Titanium was this offshore application. Because the in-service exchanger utilized to avoid corrosion problems. does not foul and this design has similar temperatures, This BFU configuration reduced the heat-transfer area velocities, and metallurgy, we are confident in the CEP requirement by 18%, with the exchanger cost dropping conviability of this low-cost design. comitantly. The total excess area is still quite high at 99.8%, and the total excess area is not simply the Companies Represented on the Exchanger Design Margins Task Force sum of the overdesign and the Alfa Laval Lund AB Joseph Oat Corp. excess area from fouling factors; APV North America, Inc. Kellogg Brown & Root, Inc. this demonstrates the compounding BASF Aktiengesellschaft Koch Heat Transfer Company, L.P. of fouling factors in the overdesign. Bechtel Ltd. Mitsubishi Chemical Engineering Corp. The third exchanger design conBP p.l.c. Nooter/Eriksen, Inc. sidered was an identical BFU Celanese Ltd. Reliance Engineering Associates (P) Ltd. Chevron Energy Technology Co. Shell Canada Ltd. except that no fouling resistance ConocoPhillips Co. Shell Global Solutions International B.V. was used in Eq. 8. An equivalent Eastman Chemical Co. Shell Global Solutions (U.S.), Inc. overdesign of 21% was achieved Ecodyne MRM, Inc. Statoil ASA by shortening the tubes. Changing E.I. du Pont de Nemours & Company, Inc. Technip the tube length is normally the ExxonMobil Research and Engineering Co. Toyo Engineering Corp. High Performance Tube, Inc. most economical approach for 44
www.aiche.org/cep
April 2007
CEP
CHRISTOPHER A. BENNETT is a researcher specializing in fouling at Heat
Transfer Research, Inc. (HTRI; 150 Venture Dr., College Station, TX, 77845; Phone: (979) 690-5069; Fax: (979) 690-3250; E-mail:
[email protected]). He holds a BS in chemical engineering from the Univ. of Toledo and an MS and PhD in chemical engineering from the Univ. of Michigan. His diverse research experience in wet chemistry, surface chemistry, materials science, mathematical modeling and aquatic biology has proven very useful to understanding heat exchanger fouling. Bennett co-chairs the HTRI Exchanger Design Margins Task Force and chairs the HTRI Crude Oil Fouling Task Force. R. STANLEY KISTLER is vice president, research and software
development at HTRI (Phone: (979) 690-5070; Fax: (979) 690-3250; Email:
[email protected]). He obtained his undergraduate and master’s degrees, as well as his PhD in chemical engineering with an emphasis on boiling, from the Univ. of Missouri-Rolla. Since joining HTRI in 1973, he has primarily focused on software development. He has also conducted experimental research on shellside single-phase flow. Kistler has helped develop many HTRI workshops and has taught dozens of courses and workshops around the world. He also serves as a guest lecturer for academic courses and has been involved in various engineering events in academia. An AIChE Fellow, Kistler is past chair of AIChE’s Heat Transfer and Energy Conversion division, and has chaired numerous sessions at National Heat Transfer Conferences. He co-chairs the HTRI Exchanger Design Margins Task Force.
THOMAS G. LESTINA, P.E., vice president, engineering services at HTRI
(Phone: (979) 690-5063; Fax: (979) 690-3250; E-mail:
[email protected]), has 20 years of engineering project management experience. As the person responsible for HTRI training, he develops, customizes and teaches at HTRI events and member companies. In addition, he has developed and taught the course, Heat Exchanger Design and Operation for ASME/AIChE. He also manages HTRI’s growing contract services and technical support. Prior to joining HTRI, he worked as a lead engineer for MPR Associates, Inc. A licensed Professional Engineer in Texas, Lestina earned a BS in mechanical engineering from Union College (Schenectady, NY) and an MS in mechanical engineering from Rensselaer Polytechnic Institute (Troy, NY). He is a member of ASME and serves as chair of the technical committee for the ASME Performance Test Code 12.5, Single Phase Heat Exchangers. DAVID C. KING is a senior heat transfer consultant for BP p.l.c (Chertsey
Rd., Sunbury-on-Thames, Middlesex, TW16 7LN, UK; Phone: +44-1932775621; Fax: +44-1932-738414; E-mail:
[email protected]). With 32 years of experience working in the refining, petrochemical, and exploration and production sectors, he is currently responsible for developing and leading a global heat exchange community and for providing leadership in heat transfer to the exploration and production sector. He developed the case for establishing HTRI’s Exchanger Design Margin Task Force and has actively participated in task force activities. He holds a BSc Honors in fuel and combustion engineering from Leeds Univ.
CEP
April 2007
www.aiche.org/cep
45
