You are on page 1of 20

Heat Exchanger Design Project

Carbon Dioxide Cooling

Ryan Poling-Skutvik and Hindi Kornbluth

Poling-Skutvik and Kornbluth

Table of Contents
Abstract ................................................................................................................................... 2 Introduction .............................................................................................................................. 4 Design Calculations ..................................................................................................................... 5 Physical Properties of Both Fluids ................................................................................................. 5 Required Heat Duty ................................................................................................................. 5 Tube Specifications .................................................................................................................. 6 Internal Heat Transfer Coefficient ................................................................................................ 7 External Heat Transfer Coefficient ............................................................................................... 7 Overall Heat Transfer Coefficient ............................................................................................... 10 Pressure Drop in the Pipes ........................................................................................................ 11 Pressure Drop in the Shell ......................................................................................................... 12 Heat Exchanger Simulation ........................................................................................................... 14 Conclusions ............................................................................................................................. 15 Nomenclature ........................................................................................................................... 16 References ............................................................................................................................... 17 Appendix ................................................................................................................................ 18 LMTD correction factor for a one-shell pass heat exchanger ................................................................ 18 Moody Chart ........................................................................................................................ 19

Poling-Skutvik and Kornbluth

Abstract
A shell-and-tube heat exchanger was designed to process 50,000 pounds of gaseous carbon dioxide per hour, cooling it from 0F to -50F with Refrigerant-134a. Hand calculations were initially performed in order to achieve a proposal for the physical design, which was then confirmed with a model in PRO/II. The results of the PRO/II model were assumed to be more accurate and are thus used to present the specifications of the heat exchanger, although the physical construction was shown to be valid, so the hand calculations are used in that case. The overall heat duty of the exchanger is 486,300 BTU/hr. This means that 64,609 pounds of refrigerant must flow through the exchanger in order to provide the necessary heat transfer. The exchanger has 8 passes on the tube side and a single pass on the shell side, which means that there are a total of 72 tubes twenty feet in length running through the shell. The tubes are 0.5% carbon steel 16 BWG with an outer radius of 1.25 in and the shell has an inner diameter of 21.25 in, also made out of carbon steel, providing a heat transfer area of 446.8 ft2. The flow rates of the carbon dioxide and the refrigerant create an overall heat transfer coefficient of 23.534 BTU/ (hr*ft2*F), which includes the respective fouling factors and conductive resistance provided by the carbon steel. They also lead to pressure drops within the system 2.175 psi on the shell side and 2.999 psi on the tube side. In order for the design to be more hydraulically efficient, the carbon dioxide is pressurized to 75 psi before entering the heat exchanger. The refrigerant is more condense as a liquid and enters the heat exchanger at only a slightly higher than atmospheric pressure in order for it to overcome the pressure drop in the system. All of these design specifications are summarized in Table 1.

Poling-Skutvik and Kornbluth

Table 1 - Summary of Design Specifications

Shell Side Component Mass Flowrate (lb/hr) Temperatures Inlet/Outlet (F) Inlet Pressure (bar) Number of Passes Shell Inner Diameter (in) Tubes: OD/ID/Pitch (in) Tube Configuration Length (ft) Total Number of Tubes Number of Baffles Baffle Spacing (in) Fouling Factor ((ft2*hr*F)/Btu) Pressure Drop (psi) LMTD (F) F U0 (BTU/hr*ft2*F) A0 (ft2) Heat Duty (BTU/hr) Carbon Dioxide 50,000 0/-50 5.17/5.02 1 21.25 ----11 21.25 0.001 2.175 60.293 0.931 23.534 446.8 486,300

Tube Side Refrigerant-134a 64,609 -100/-71.911 1.38/1.17 8 -1.25/1.12/1.5625 Square 20 72 --0.002 2.999 ------

Poling-Skutvik and Kornbluth

Introduction
A shell-and-tube heat exchanger is needed to cool gaseous carbon dioxide from 0F to -50F using Refrigerant 134a at -100F. The refrigerant cannot rise above -60F. The heat exchanger must be able to process 50,000 pounds of carbon dioxide per hour but the flow rate of refrigerant is unspecified. The exchanger must be both hydraulically and thermally suitable, providing an adequate rate of heat transfer while only creating moderate pressure drops. Initial hand calculations are needed in order to determine the physical set-up of the heat exchanger, although the desired specifications will be confirmed using the modeling software package PRO/II.

Poling-Skutvik and Kornbluth

Design Calculations
Physical Properties of Both Fluids
Before any calculations can be started, the physical properties of both fluids need to be estimated. Certain properties, such as heat capacity, depend on the temperature of the fluid, which changes as the fluids proceed along the exchanger. In order to simplify the calculations, these properties were calculated at the average fluid temperature. Additionally, viscosities are also evaluated at the wall temperature, which is the average temperature of the other fluid. These are predominantly used to compensate for viscous changes. These different properties are listed in Table 2
Table 2 - Physical Properties of Fluids 1

Physical Property Initial Temperature (F) Final Temperature (F) Heat Capacity (BTU/lbF) Density (lb/ft3) Thermal Conductivity (BTU/hr*ft*F) Dynamic Viscosity (lb/ft*hr) Kinematic Viscosity (ft2/hr) Prandtl Number Dynamic Viscosity at Wall (lb/ft*hr) Kinematic Viscosity at Wall (ft2/hr)

Carbon Dioxide 0 -50 0.2634 2.08 0.007685 0.030 0.0143 1.02 0.027 0.0143

Refrigerant-134a -100 -65 0.2912 92.55 0.06497 1.751 0.0189 7.175 1.031 0.0118

Required Heat Duty


The required heat duty, Qreq, can be calculated by analyzing the heat transfer involved in the process for which all of the data is known, the Carbon Dioxide feed.
Equation 1- Heat Duty

The heat duty is the same regardless of which species data is used for calculations, so this equation is used to find the undecided mass flow are and temperature difference of R134-a.

= = 50000

. 2634 50 = 658500 /

Lienhard IV, John H. and John H. Lienhard V. A Heat Transfer Textbook. Cambridge: Phlogiston Press, 2011.

Poling-Skutvik and Kornbluth

Tube Specifications
The heat duty Qreq can also be defined as a function of area, the log mean temperature difference, the overall heat transfer coefficient, and the LMTD correction factor F. F can be found by using the LMTD Correction Factor charts in the appendix. U is reasonably estimated by comparing the heat exchanger to the situations in the Kern Appendix. For CO2 and R134a, it can be modeled as gas in water, which has overall heat transfer coefficient from 2 to 50, so the reasonable estimate chosen is 25.
Equation 2- Area of a Heat Exchanger

The Log Mean Temperature Difference is defined as the difference between each end of the heat exchangers temperature difference, over the difference in the logarithms of the differences.
Equation 3-Log Mean Temperature Difference

6585000 = = = 436.1 ^2 () (25 /( 2 ))(54.8 )(. 95)

In a tube and shell heat exchanger, it is best to place the more corrosive material in the tubes, because they are easier to remove and replace. It was decided that CO2 would go on the shell side, as it will do the least amount of damage or fouling to its tubing. After selecting specifications for the tube exchangers, the required length of the pipe is determined by dividing the necessary surface area by the product of the number of pipes and the external surface area. The cross sectional area is calculated by assuming an appropriate velocity in the tube, and dividing the mass flow in the tubes by the product of the density of the material and its velocity. The assumed velocity is 1 m/s = 11811 ft/hr.
Equation 4-Cross Sectional Area

() =

1 2 (100 (50)) (0 60) = = 54.8 100 (50) log 1 ln 0 (60) 2

The number of pipes can now be find by dividing the necessary cross sectional area by the cross sectional area of each pipe (from tubing characteristics table) and rounding up to a whole integer.
Equation 5-Number of pipes

= .0591 2

# =

. 0591 2 = 9 . 068 2 /

Poling-Skutvik and Kornbluth

The number of tube passes can be calculated by rounding to an even number the length of tubes necessary divided by the length of each tube (chosen to be 20, a common length).
Equation 6-Number of passes

The number of tubes can be calculated by multiplying the number of passes by the number of pipes.
Equation 7-Number of tubes

# =

( ) 148.1 = = ( )8 ( ) 20

Internal Heat Transfer Coefficient


the Reynolds number.
Equation 8- Reynold's Number

# = # # = 8 9 = 72

In order to find hi, it is necessary to determine the flow regime inside the pipe. This is done by computing

For turbulent flow, the Sieder-Tate correlation holds.


Equation 9-Sieder-Tate correlation

(11811)(. 0933) = = = 58264 . 0189


1 .14 = .23 .8 3 = 310.1

wall temperature.

Here, is the viscosity of the material at the bulk temperature, and is the viscocity of the material at the This corresponds to =

External Heat Transfer Coefficient


depends on the Reynolds number of the fluid flowing through the shell and relates physical properties of the fluid to the heat transfer coefficient.
Equation 10 - Colburn factor

= 215.9

The external heat transfer coefficient, , can be calculated through the Colburn factor, . This factor

1 3

0.14

Poling-Skutvik and Kornbluth

In this equation, all of the physical properties of the fluid are evaluated at the film temperature except for heat transfer coefficient can be found using this equation. In order to determine the factor, the Reynolds number of the fluid needs to be determined because the factor depends on the Reynolds number through the following relation
Equation 11 Relation between Reynolds number and Colburn factor

, which is the viscosity of the fluid at the average wall temperature. Once the Colburn factor is known, then the

= 0.5 1 +

Although the calculation of the Reynolds number through a simple pipe is simple, calculating it for a fluid flowing through the shell and around the tubes is a bit more complicated. The Reynolds number depends on the mass flux of the fluid, the effective diameter of the shell, and the viscosity of the fluid.
Equation 12 Reynolds number in a shell

(0.08 0.6821 + 0.7 0.1772 )

= wetted perimeter. This can be calculated as follows


Equation 13 Effective diameter of a shell

The effective diameter of the shell can be calculated as four times the free area within the shell divided by the

The mass flux is also needed in the computation of the Reynolds number. It is the mass flow rate of the carbon dioxide through the shell divided by the flow area across the bundles
Equation 14 Mass flux through the shell

2 2 4 4 =

= adjacent tubes, the baffle spacing, and the tube pitch.

The flow area depends on four different parameters: the dimensions of the shell, the clearance between

Poling-Skutvik and Kornbluth

Equation 15 Flow area in the shell

The clearance between adjacent tubes is defined the difference between the tube pitch and the outer diameter of the tubes inside the shell. These equations can now be used to generate the dimensionless Reynolds number. With the design specifications reviewed in the Overview of Design section, each of these values can be determined systematically starting from Equation 6 to Equation 2. 1.5625 1.25 21.25 21.25 12 12 12 12 = = 0.6272 2 1.5625 12 = 79723 = 22.15 = 0.6272 2 2 2 50,000
2 2 1.5625 1.25 4 12 4 12

1.25 12

= 0.1031

= = 0.5 1 +

(0.1031 ) 79723

From this Colburn factor, the exterior heat transfer coefficient can easily be determined by substituting values for physical constants evaluated at the average temperature of the carbon dioxide into Equation 1 and solving for .

21.25 (0.08(2.77 105 )0.6821 + 0.7(2.77 105 )0.1772 ) = 418.77 21.25

0.0297

= 2.77 105

Poling-Skutvik and Kornbluth

10

1 3

0.14

(418.77) 0.0077

= 31.83

0.14 3 0.0297 0.2634 0.0297 1.25 0.0077 0.0270 12

the internal heat transfer coefficient because the fluid passing through the shell is a gas, with a much lower thermal conductivity than the liquid refrigerant that flows through the pipes of the exchanger.

The exterior heat transfer coefficient was determined to be 31.83 . This is a much smaller value than 2

Overall Heat Transfer Coefficient


The overall heat transfer coefficient relates the three stages of heat transfer present in the heat exchanger convection in the refrigerant, conduction through the wall of the pipes, and convection in the gaseous carbon dioxide. which these different modes of heat transfer act like parallel resistances. With these three modes of heat transfer, a cylindrical geometry, and ignoring fouling, the coefficient has the form
Equation 16 - Overall heat transfer coefficient in a clean exchanger

This coefficient, , can be calculated by making an analogy between the heat exchanger and a thermal circuit, in

ln 2

means that everything in Equation 7 is known except for . Solving for this gives

Since the pipes are made of 0.5% carbon steel, they have a thermal conductivity of 27.73
1

. This

1.25 1.25 ln 1.25 1 12 1.12 + = + (1.12 ) 2 27.73 31.83 2 215.9 2

= 27.18

This is less than 10% off from the initial guess of the overall heat transfer coefficient. Therefore, the heat exchanger that has been designed is most likely thermally suitable for the process, especially since this calculated value is slightly above that assumed earlier. The higher value means that there is a greater rate of heat transfer and that the designed area is greater than the required amount of area. However, this is only for the clean exchanger. In order to determine the overall heat transfer coefficient of the dirty exchanger, the fouling factors of the liquids need to be

Poling-Skutvik and Kornbluth

11

determined. For the refrigerant, the fouling factor is approximately 0.001 ft2hrF/Btu and the fouling factor of the carbon dioxide is approximately 0.002 ft2hrF/Btu 2. The final coefficient can be determined using the following equation.
1 1

1 + 2 + 134

This heat transfer coefficient is even closer to the original guess so the heat exchanger that has been designed will be thermally suitable. Since the design is most likely suitable to provide the necessary heat transfer, the hydraulic suitability must also be tested.

+ 0.002 27.18 2

+ 0.001

= 25.13

Pressure Drop in the Pipes


The pressure drop of the refrigerant in the pipes can be evaluated rather straightforward by determining the friction factor in the process and a couple other factors. The total pressure drop experienced by the refrigerant after passing through the exchanger can be determined from the following relation
Equation 17 Pressure drop from frictional pipes through the pipes of the exchanger

The friction factor can be determined from Moody charts, which relate pipe roughness and Reynolds number. A chart is in the appendix for reference. The Reynolds number of the refrigerant was already calculated in order to determine the internal heat transfer coefficient. It was determined to be 58264. Since the pipes are made of mild steel, they have a predicted roughness of 0.00015 ft. Dividing the roughness by the inner diameter of the pipes gives an effective roughness of 0.0016. Reading across on this chart gives a Darcy friction factor of approximately 0.023. The only other variable in Equation 8 that must be computed is the mass flux, which is the velocity of the fluid multiplied by its density. In this case it is computed as follows 3.28 = 303.64 3 2

2 0.14 2

= = 92.55

(Kreith and Goswami)

Poling-Skutvik and Kornbluth

12

With these variables determined, the pressure drop across the pipes can be calculated. 2 (20 ) 2

, =

1.751 1.12 2 32.174 92.55 3 12 2 0.9876

8(0.023) 303.64

0.14

= 563.38

= 3.91 2 2

From this calculation, it can be seen that pressure drop due to frictional forces in the pipe is predicted to only be 3.91 psi. There is also a pressure drop due to the change in direction experienced by the fluid as it travels through the series of hairpins within the shell. This pressure drop can be expressed as
Equation 18 Pressure drop from changes in momentum in the pipes of the heat exchanger

, = 1.2 changes is

All of the quantities in this equation are known. Therefore, the pressure drop from these momentum

2 3.28 = 1.2 8 92.55 3 = 297.2 2 = 2.06 2 32.174 2

The total pressure drop across the pipes is the sum of the drops from momentum changes and from frictional forces, totaling 5.97 psi. This is well within the capabilities of a pump and shows that the exchanger is hydraulically suitable on the pipe side. In order to determine if the design is fully suitable, the pressure drop in the shell must also be determined.

Pressure Drop in the Shell


The pressure drop in the shell results from frictional forces and momentum changes when the gaseous carbon dioxide hits a baffle. These different forces have been correlated into a single equation to determine the pressure drop in the shell side of the heat exchanger.

Poling-Skutvik and Kornbluth

13

Equation 19 Pressure drop in the shell

The only two variables that must be determined are the number of baffles present and the Fanning friction factor. The Fanning friction factor can be found using the same Moody chart as was used when determining the Darcy friction factor because the Fanning factor is just one-fourth of the Darcy factor. The Reynolds number of the carbon dioxide was previously calculated to be 2.77*105 and the effective roughness was determined to be 0.00145, giving a Darcy friction factor of 0.023 and a Fanning friction factor of 0.00575. The number of baffles can be calculated by dividing the length of the reactor by the baffle spacing and rounding to the nearest integer. Substituting these into Equation 10 gives the following

2 ( + 1) 0.14 2

20 2 21.25 12 + 1 2 21.25 12 = 0.14 = 4.45 2 = 0.0298 0.0297 2.0789 3 0.1031 2 32.174 2 0.0270 (0.00575) 22.15 This pressure drop is very low and means that the exchanger will be hydraulically suitable for both fluids. Since the heat exchanger meets both hydraulic and thermal demands, it is a viable design, although this must first be confirmed by modeling it in PRO/II.

Poling-Skutvik and Kornbluth

14

Heat Exchanger Simulation


The proposed design of the heat exchanger was modeled in PRO/II in order to confirm that the design will meet both hydraulic and thermal demands. The simulation utilized the rigorous heat exchanger model present in the software package and used the Peng-Robinson thermodynamic model to predict different thermodynamic properties. This model was chosen because it is typically more accurate in predicting liquid properties than either the Soave or Redlich-Kwong methods while still being as accurate as those two in predicting gas-phase data. In the model, the refrigerant is placed in the tubes at a flow rate of 64609 lb/hr, as this was the calculated flow rate needed to achieve the desired changes in temperature for both the refrigerant and the gaseous carbon dioxide. The refrigerant enters the tubes at -100F and at 14.7 psia. The carbon dioxide is placed in the shell with the desired flow rate of 50000 lb/hr and enters at 0F and 75 psia. The reason the carbon dioxide is run at a relatively high pressure is because there is a relatively small flow area through the bundles, making a pressurized feed more hydraulically suitable. As determined from the hand calculations, the exchanger will need 8 passes per shell. The tubes will be made of carbon steel and will have inside diameters of 1.12 in and outer diameters of 1.25 in. They will be placed in a square pattern with a 1.5625 in pitch. The shell will have an inside diameter of 21.250 in, which will be the same as the baffle spacing. The baffles have a cut of 0.200. Additionally, the shell outlet temperature is specified to be -50F because this is the output that is desired. The model predicts that the carbon dioxide will leave the exchanger at the desired -50F and that the refrigerant will leave at -72F. This is a slightly lower temperature than the predicted -65F and it most likely derives from the use of an average heat capacity rather than integrating the heat capacity with respect to temperature. These final temperatures led to a mean temperature difference correction factor of 0.932 and a mean temperature difference of 56.23F. Additionally, the heat exchanger has a heat duty of 4.842*105 Btu/hr. This is lower than what was calculated, probably due to the inaccurate approximation of an average heat capacity in the carbon dioxide. The heat transfer area is also slightly higher due to the rounding of the number of tubes and number of passes. This makes the area equal to 560.7 ft2 whereas it was predicted to be close to 485 ft2. The heat transfer coefficient was modeled to be 24.551 Btu/(ft2hrF), which is very close to the predicted value.

Poling-Skutvik and Kornbluth

15

Conclusions
While simulation software is useful, and can be more accurate than hand calculations, it is only a supplement to knowledge of heat transfer and ability to compute and analyze data. In calculating the specifications of a heat exchanger designed to cool 50,000 lb/hr of CO2 from 0F to -50F with Refrigerant 134-a supplied at -100F, there were some things that had to be calculated by hand before any simulation software could be used. Determining the necessary heat duty, the resulting surface area, the number of pipes necessary, and preliminary calculations such as those must be computed for estimates before the simulation is useful. The specifications of the design assignment stated that the refrigerant couldn't leave the exchanger at a temperature greater than -60F, and furthermore the mass flow wasn't known- trying to build a heat exchanger without calculating a reasonable estimate for the outlet temperature and mass flow would have taken a lot longer and wouldn't have necessarily been more accurate than hand calculations. However, regarding the outlet temperature, one place where the software was superior was in its calculation was in its knowledge of Cp data, where its software easily computed the integral of the constant over the temperature range, instead of using the average value assigned in chart data. This caused inaccurate guesses, though they were not too far off, for the outlet temperature of R-134a, the log mean temperature difference, and necessary heat duty Additionally, the rounding of the number of tubes and passes made the predicted hand-calculated value of the heat transfer surface area much smaller than its real value. The data leads to the suggestion that the heat exchanger should be built with 1tube, 8 shell-side passes, a shell inner diameter of 21.25 in, tubes with outer diameter 1.25 in, tube inner diameter 1.12 in, and square pitch of 1.5625 in. It will have seventy two 20 ft tubes, and 11 baffles with 21.25 in baffle spacing.

Poling-Skutvik and Kornbluth

16

Nomenclature
Symbol Description Shell-side heat transfer coefficient Effective shell diameter Heat capacity Baffle spacing inside shell Mass flux Outside diameter of the pipes flowing through the shell of the heat exchanger Overall heat transfer coefficient in a clean exchanger Thermal conductivity of the solid in the pipe walls Number of tube passes in the exchanger Darcy friction factor Density Total cross-sectional flow area of the pipes Pressure drop experienced by the fluid flowing in the shell of the exchanger Fanning friction factor Bulk velocity of the fluid in pipes Log mean temperature difference Kinematic viscosity Symbol Description Colburn factor Thermal conductivity of the fluid Dynamic viscosity Inner diameter of the shell Tube pitch Clearance between adjacent tubes in the shell Overall heat transfer coefficient in a dirty exchanger Pressure drop through the pipes of the exchanger due to frictional forces Length of each tube Conversion factor between pounds-force and pounds-mass Mass flow rate Pressure drop through the pipes of the exchanger due to momentum changes Number of baffles in the shell Dynamic viscosity of the fluid evaluated at the wall temperature Required heat duty Log mean temperature difference correction factor for shell and tube heat exchangers Internal diameter of the pipes flowing through the shell of the heat exchanger

, ,

Poling-Skutvik and Kornbluth

17

References
Kreith, Frank and D. Yogi Goswami. Handbook of Energy Efficiency and Renewable Energy. Boca Raton: CRC Press, 2007. Lienhard IV, John H. and John H. Lienhard V. A Heat Transfer Textbook. Cambridge: Phlogiston Press, 2011.

Poling-Skutvik and Kornbluth

18

Appendix
LMTD correction factor for a one-shell pass heat exchanger

Reference: Lienhard IV, John H. and John H. Lienhard V. A Heat Transfer Textbook. Cambridge: Phlogiston Press, 2011.

Poling-Skutvik and Kornbluth

19

Moody Chart

Reference: Lienhard IV, John H. and John H. Lienhard V. A Heat Transfer Textbook. Cambridge: Phlogiston Press, 2011.

You might also like