Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
College of Engineering
Department of Chemical Engineering and
Material Science
Final Heat Exchanger Project:
Group: Flummoxed by Flux
Here we report the design and optimization process of a shell-and-tube heat exchanger which
effectively heats a stream of acetone flowing at 80,000 kg/hr from 10 °C to 30 °C using a stream
of DOWTHERM MX (DMX) flowing at 100,000 kg/hr with an inlet temperature of 100 °C as the
heating fluid. This report explains the approach used to determine key system parameters, such as
the heat load, the outlet temperature of the heating fluid, and the Log-mean Temperature
Difference (LMTD), as well as the methods used to design a configuration that satisfies these
constraints. We also include several tables comparing the different configurations studied. All
exchangers are in the E-HS-2CN form, in TEMA nomenclature (E – single shell pass, HS – hot
fluid shell side, 2CN – 2 tube pass countercurrent).
Energy balance calculations across the tube and the shell side yielded an outlet temperature of
81.02 °C for DMX and an LMTD of 70.51 °C. These values were used to design and compare
several shell-and-tube heat exchangers. The constraints under which the designs were evaluated
were the following: heat tranfer area, fluid velocity (for both sides), pressure drop (for both sides)
and LMTD correction factor, FG. Preliminary shortcut calculations using Aspen V8.2 suggested a
2-2 shell-and-tube configuration as the lowest area configuration (15.23 m2). However, specific
design constraints and optimization resulted in selection of a 1-2 shell-and-tube configuration with
an area of 22.60 m2
In the final heat exchanger configuration, the cold fluid (acetone) flows at 1.39 m/s through the
tube side. The tube side is composed of 190 2-meter long carbon steel tubes with an outer diameter
of ¾” and Birmingham Wire number (BWG) 18 arranged in a 0.938” triangular pitch. DMX flows
at 1.73 m/s in the 0.440 m diameter steel shell side.
The convective heat transfer coefficients for the shell and the tube sides were calculated to be 1440
W/(m2°C) and 2510 W/(m2°C), respectively. Fouling was also accounted for both fluids, using a
value for hdi of 5000 W/(m2°C) for (hdi-1 =0.0002 m2°C/W). Calculation of the overall heat transfer
coefficient (U0) gave a coefficient U0=621 W/(m2°C). This value is within 6.5% of the value
calculated by Aspen (584.38 W/(m2°C)).
A short discussion for future recommendations is also included at the end of the document.
2
Design approach
The heat exchanger design approach required preliminary calculation to estimate the outlet
temperature for the hot fluid, DOWTHERM MX (DMX). This temperature was calculated with an
energy balance under the assumption that all heat lost by DMX is transferred to the cold fluid (acetone),
i.e. no energy is lost to the surroundings. The properties of the fluids at several temperatures were
exported from Aspen V8.2 and verified using external sources [1,2]. An Excel spreadsheet containing
both fluids properties at different temperatures was used to interpolate to the outlet temperature (Table
1). The outlet temperature was used as an iterative basis to find the temperature for which the average
heat capacity of the heating medium and the heat capacity required by the heat balance converged. This
temperature was found to be 81.02 °C. Obtaining the outlet temperature for the heating fluid allowed
calculating the LMTD. The LMTD was equated to the heat load and used to calculate the product of
the overall heat transfer coefficient (U0) and total heat transfer area (A0).
After the set of preliminary calculations, the parameters were used as inputs for in Aspen to
obtain an estimate of the overall heat transfer coefficient as well as the area. This process was repeated
for the following shell-and-tube configurations: 1-2, 2-2, 2-4. Double pipe heat exchanger
configurations were not considered as they would require either large areas or very high velocities to
meet the large flow rates in the present system.
Decreasing heat exchanger area typically results in reducing cost [3]. Therefore, the shell-and-
tube configurations were preliminarily ranked in terms of their area. Given the importance of reducing
cost, the heat exchangers efficiency (from most economical to least economical) ranked as follow: 2-
2, 2-4, and then 1-2. The 2-2 shell-and-tube heat exchanger configuration was initially used for the
design. However, given the lack of rigorous LMTD correction data for this configuration, the
calculations were uncertain. In addition, the pressure drops across both sides of the 2-2 heat exchanger
were much larger than the 0.5 bar allowable pressure drop, so the configuration was ultimately
discarded. The next tested configuration was a 2-4 shell-and-tube heat exchanger. This type of heat
exchanger is typically very effective [4], however, the pressure drop under the design conditions was
again too large. The most appropriate configuration was obtained using a1-2 heat exchanger. Despite
the relative large area of this configuration (as compared to the previously proposed), the parameters
used in this configuration met the specific requirements. This configuration was further optimized to
minimize the area and optimize the velocity of the fluids. A comparison of all the different
configurations tried can be seen in Table 5 in the Appendix.
3
The three main constraints that must be maintained within their appropriate ranges are area,
velocity, and pressure drop. These constraints are controlled by optimizing parameters such as the tube
length. Qualitatively, velocity and pressure drop on the tube side increase with tube length while area
decreases. On the shell side, decreasing the baffle spacing or increasing the number of passes increases
pressure drop while decreasing area. As a result, all of these factors must be balanced in order to keep
the system economically feasible with both a reasonable area and a reasonable pressure drop. A large
pressure drop, for example, could require higher inlet pressure or incorporating a pump downstream.
Other parameters such as tube diameter, pitch, and bundle spacing (between tube bundle and shell)
similarly affect area and pressure. Using iterative calculations in Excel, a set of geometric parameters
was determined that fall within optimal ranges (pressure drop less than 0.5 bar, tube velocity within 1-
2 m/s, shell velocity within 1-2 m/s, [7]) while minimizing required area. For the preliminary Excel
calculation, an initial guess for Uo was selected based on Aspen shortcut calculations. This initial
Uo was used to set an initial heat transfer Ao and thereby calculate the number of tubes (with a
fixed length and tube diameter). An output Uo was calculated based on film coefficients determined
from Nusselt number correlations. For each iteration, the initial guess of Uo was replaced with the
calculated value until the two values converged. During these Excel calculations, it was also
determined that pressurizing the acetone stream to 2 bar would be preferred as it would reduce the
percent pressure drop. Fouling was selected based on typical values for organic heat transfer fluids [3].
After these parameters were selected, the design was simulated with Aspen Detailed
Calculations. Initial values of baffle cut, nozzle diameters, and other geometric parameters were
determined using common values [3,4]. Within the detailed simulation, several iterations were made
to improve pressure drop and reduce area. These iterations included changes of nozzle diameters (to
determine appropriate shell-side pressure drops), length, and Birmingham wire gauge (without
changing heat transfer area this results in a of the decrease pressure drop due to the large cross sectional
area). After the iterations determined appropriate values, these values were inputted into excel to
illustrate the differences between the Aspen and excel calculations. The simulation was then exported
to Aspen Exchanger Design and Rating (EDR) for economic optimization. Aspen EDR compared
several designs, choosing one that minimized cost to $19,744 (see Appendix C). For specific geometry
parameters see Table 6, Appendix D and the Aspen output in Appendix C.
4
Results
Component
DOWTHERM MX[1] Acetone[2]
Properties
Interpolated at:
Evaluated at T (°C): 40 100 20
90.51
ρ kg/m3 948.2 905.2 912.0 786
Sources of Error
5
Between the hand calculations and the Aspen outputs, there is consistently an error of
approximately 5% for the overall heat transfer coefficient, Uo, and the heat load, q. It is suspected
that the material properties are the main source of error. When evaluating the properties of DMX
at its arithmetic average temperature (90.1°C), a basic interpolation method was used based on two
data points at 40°C and 100°C. With such a wide range, interpolation can become inaccurate,
particularly if the properties do not vary linearly in this range.
The pressure drop calculation for the shell side also had a consistently high error of about 300%
(see Table 2) between Excel calculation and Aspen calculations. It is suspected that the empirical
relationships between geometry and pressure drop used in Aspen are more refined then those used
in the Excel calculations (See Appendix B). For example, Aspen takes into account the affect of
spacing between the baffles and the shell, whereas the relationship used for Excel calculations is
only dependent on equivalent diameter, pitch, and baffle spacing.
Based upon analysis from Aspen Plus, Aspen EDR, and preliminary Excel calculations, we
recommend that the heat exchanger follow the geometry parameters specified in Table 6,
Appendix D. This design offers the lowest heat transfer area of 22.6 m2 while still satisfactorily
meeting heat load requirements. The economically optimized design suggested by Aspen EDR is
a possible option as well, though has a higher heat transfer area of 24.5 m2. If a factor of safety is
desired, taking into account leaking and additional fouling that may develop over years of use,
then the Aspen EDR design may be preferred as it is overdesigned by 14.5%. However if volume
and total sizing is a limiting factor, then the design in Table 6 would be preferred.
It should be noted that both designs adequately meet all requirements but are not entirely
optimized. A further economic optimization route would involve increasing tube length to 2.44 m
to match industry standards and reduce waste in cut materials. Increasing the length would also
decrease number of tubes and required shell diameter, also cutting costs. We recommend that more
detailed calculations be made in Aspen Plus in combination with Aspen EDR in an effort to
decrease total heat transfer area. The current design was more conservative than perhaps necessary
due to the high pressure drop calculated by hand for the shell side. Recognizing that the Aspen
outputs show a lower pressure drop than the hand calculations, this design could be made more
6
efficient by decreasing the heat transfer area. One potential route for improvement could be an
increased number of baffles in order to increase cross flow and the overal heat transfer coefficient.
References
1. Dow Chemical. Dowtherm MX Heat Transfer Fluid Product Technical Data. The Dow
Chemical Company, 1999.
4. Macabe, W., J. Smithm and P. Harriott. Unit Operations of Chemical Engineering. McGraw-
Hill, 2001.
5. Briedis, Daina. Course Handouts, CHE 311: Fluid Flow and Heat Transfer. Michigan State
University. Fall 2013.
7. Than, Su Thet Mon, Khin Aung Li and Mi Sandar Mon. "Heat Exchanger Design." World
Academy of Science, Engineering and Technology (2008): 604-611.
7
Appendix A: Design Calculation Examples
B C D
2 Properties
evaluated at 20 90.5095
T(°C)
3 Component
Acetone DMX
Properties
4 ρ kg/m3 786 912.00
5 k (W/mK) 0.17 0.115
6
2181.81818
Cp (J/kgK) 1839.155875
2
7 μ (kg/m*s) 3.60E-04 3.25E-03
8 ν (μ/ρ) (m2/s) 4.580E-07 3.566E-06
9 Pr (Cpμ/k) 4.62E+00 5.20E+01
10 MW (g/mol) 58.08 238
11 Values Cell references
12 Intial: Ao(m2)
=q/(U0*LMTD*F 22.60 (Heat Load)/(LMTD*D$13*D$14)
G)
13 Initial Guess Uo
621 Initial guess retrieved from Aspen
(W/m2°C)
14 Fg 0.98 Retrieved from McCabe (Figure 15.6) [4]
15 Calc: Ao (m2) 22.60 (Heat Load)/(D$17*LMTD*D$14)
16 Cal: Ao (ft2) 243.264 (Heat Load)/(D$17*LMTD*D$14)* 10.7639
17 Calc 1/Uo 1/D$60+1/D$61+D$24*LN(D$24/D$25)/(2*D$21)+(D$24/D
1.61E-03
(W/m2°C) $25)*(1/D$39)+(D$24/D$25)*(1/D$38)
18 Calc Uo
6.21E+02 1/D$16
(W/m2°C)
19 1/D$16/20248
Calc Uo
0.030399337
(BTU/ft2°F)
8
22 Tube Material Steel Steel
23 kw (W/m°C) 52 Retrieved from Engineering Toolbox [2]
24 v (m/s) 1.390 mdot/($C$4*D32)
25 BWG 18 Retrieved from McCabe (Appendix)[4]
26
27 Retrieved from McCabe (Appendix) and converted
do (m)
0.01905 (0.75in*0.0254m/in)
28 di (m) 0.0165608 Retrieved from McCabe (Appendix) [4]
29 t (m) 1.24E-03 (D24-D25)/2
30 L (m) 2 2
31 Total Nt 189 D12/D30
32 Np 2 2
33 Heat Atube (m2) 1.20E-01 PI()*D24*D27
34 Flow Atube (m2) 2.15E-04 PI()*D25^2/4
35 Total Aflow (m2) 2.03E-02 D28*D31/D29
36 Pitch Type TRIANGULAR Selected
37 PT (m) 2.38E-02 D24*1.25
38 C=PT - do (m) 4.76E-03 D34-D24
39 Re 5.03E+04 ($C$4*D25*D22)/$C$7
40 Nu 2.45E+02 0.023*D36^0.8*$C$9^0.4
41 hi (W/m2°C) 2.51E+03 D37*$C$5/D25
42 hi (BTU/ft2°F) 442 D37*$C$5/D25*0.1761
43 hdi (W/m2°C) 5000 Retrieve from Sinnott (Table 12.2)
44 hdi (BTU/ft2°F) 5000 5000*0.1761
45 ft 9.00E-03 Retrieved from Table 5.53 (McCabe)
46 Δpt (Pa) 7.73E+03 ($C$4*D22^2)/(2)*((D40*D27)/D25+4)*D29
47 Δpreturn (Pa) 6.08E+03 4*D29*($C$4*D22^2)/(2)
48 Total Δpt (bar) 0.14 (D41+D42)/10^5
49 Total Δpt (psi) 2.0305 (D41+D42)/10^5*14.5037
50 Pin (bar) 2.026 1.013*2
51 %p drop 6.82% D43/D44
9
Appendix B: Equations Used*
Tha - Thb T -T
Z= ; hH = cb ca ; FG determined from figure 15.6 [5]
Tcb - Tca Tha - Tca
rVt Di rV D A flow
Heat transfer Reynolds #: Re tube = ; Re shell = sh e ; De = 4
m m LHT perimeter
rVsh DH 3.44PT2
Flow Reynolds: Re shell = ; Triangle Pitch: DH = - do
m p do
1 1 r0 1 r0 Dr r0 1 1
= + + + + ;hd fouling
U 0 hi ri hdi ri k rlm hd 0 h0
hi di
Nutube = = 0.023Re 0.8 Pr 0.3
kf
ho De
Nushell = = 0.36 Re 0.55 Pr1/3
kf
æ fL ö rV 2
Dptubes = N p ç t t + 4 ÷ t
è Dit ø 2gc
Ds rVsh 2 L
Dpshell = fs (N b + 1); fs = exp(0.576 - 0.19 ln(Re sh )); N b = t - 1;
De 2 B
1/2.207
æ Nt ö
For 1-2 HX: Ds = Dbundle + 0.056; Dbundle = do ç
è 0.249 ÷ø
DsCB
Shell Flow Area As = ; Pitch PT = 1.25do ; C = PT - do ; B = Baffle Spacing
PT
*Allequations and variable definitions used in this report are taken from course handouts (e.g.
Heat Transfer Help Sheet [8]) or can be found in the course textbook, McCabe [5].
10
Appendix C: Spreadsheets and Aspen Outputs
90
80
TS Bulk Temp. (C) + SS Bulk Temp (C)
70
60
50
40
30
20
10
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Distance f rom End (mm)
11
Aspen Simulation Output
STREAM VELOCITIES:
12
SHELLSIDE MAX. CROSSFLOW REYNOLDS NO. 5046.7091
PRESSURE DROP:
13
Appendix D: Mechanical Summary
71
19. 05
81
23. 8125
Tube O. D. mm 19. 05
T2 S1
379
T1
S1
379
379
416
T1 S2
S2
223 461 1074
Pulling Length 1460
75 75
99
99
396
150 150
2 Bolts 2 Bolts
Fixed Sliding
The specifications of this design are considered to be realistic based upon typical values
published in several heat exchanger design guides [3,6,7]:
Optimal tube-side velocities for acetone are typically between 1.5 and 3 m/s, as
described in Table 8.2 in the course handout [5]. The velocity in this design is 1.4 m/s,
which is slightly lower than this suggested range. However, another design guide
suggested a reasonable range of 1-2 m/s [7], and it was found in this case that a slightly
lower tube-side velocity resulted in a more reasonable pressure drop.
Optimal shell-side velocities for benzene are typically between 1.4 and 2.8 m/s [Table
8.2, ref. 5]. As DMX is an alkylated aromatic, it likely has similar behavior and viscosity
as benzene, so this range should serve as a good approximation. The shell side velocity
for this design is 1.7 m/s, which is taken to be reasonable.
As these streams are at relatively low pressure, a reasonable pressure drop for both the
shell and tube side should be below 10 psi or about 0.5 bar. In this design, pressure drop
is maintained well below 0.5 bar for both streams (see Table 2).
This tube uses a nominal ¾” outer diameter tube of carbon steel. Carbon steel tubes are
inexpensive and the ¾” diameter is common and preferred for compact exchanger
designs with low-fouling fluids [7].
The optimum ratio of tube length to shell diameter is typically between 5 and 10 [7]. For
this design the ratio is 4.5. This is less than optimal, but still reasonable.
The Shell diameter is typically about 56 mm larger than the bundle diameter [7], which
is determined by the number and size of tubes. This spacing is specified to both reduce
shell-side pressure drop and to increase the ease of construction and maintenance of the
exchanger [6].
The preferred tube lengths are: 1.83 m, 2.44 m, 3.66 m, 4.88 m and 7.32 m [7]. The tube
length selected is 2 m. This is not ideal due to the English basis of many tubes. A future
design optimization for economics would likely take this into account and adjust the
length to 2.44 m. However, the length is reasonable from a sizing perspective.
For the heating requirements of this system, a 1-2 heat exchanger is appropriate. A 2-4
heat exchanger may decrease area requirements, but the complexity of construction and
maintenance are likely economically prohibitive and unnecessary for the relatively low
heat load.
15