Introduction *

Base Case Design *

Optimized Design *

Heat Exchanger Details *

Economics *

Conclusion *

Bibliography *

Introduction

Miller & Associates contracted King&Co. to improve upon an acetone production process by optimizing the heat exchanger network and thereby reduce the utility costs. The process was originally developed and optimized by Group C. Williams and Kavinetor, Inc. to yield 36 MMlb/yr of acetone via the dehydrogenation of isopropanol:

Since the reaction is endothermic, Dowtherm G is used to provide the essential heat to the reactor and to the feed stream, which needs to be heated to a high temperature before it enters the reactor. Dowtherm G replaces molten salt in the reaction process because it is found to be a superior heat transfer medium with its high heat transfer coefficient of 587 Btu / hr× ft^{2× °} F. The final product leaves the separations section at 99.9 mol% purity and the azeotropic recycle stream is maintained. These conditions, as well as the temperature and pressure of all the process streams, remain unchanged in our project. The number and type of all process units excluding heat exchangers also remain unchanged. Our goal is simply to provide an improved method of heat integration based on the existing inputs and outputs of the exchangers. This improvement is necessary for the minimization of utilities, which in this case include cooling water, refrigerated water, low pressure steam, electricity to the pumps, and methane (used to re-heat the Dowtherm stream).

Base Case Design

The base case for this project was based on the winning designs for acetone production concerning the reactor and separations sections. The Carl Williams report was used to model the process until the first cooling stage of reactor effluent. At this point Kavinetor’s design was utilized starting with a second cooler for reactor effluent and continuing through the recovery of 99.9% pure acetone. Please reference the Base Case PFD found in the Appendix when reading this section of the report.

The Carl Williams report begins with an azeotropic water-IPA (88% IPA by weight) feed entering heat exchanger E-101, where it is heated from 31.9^oC to 108.9^oC by counter-current exchange with hot reactor effluent (Williams 5). Such an exchange aids in optimizing the overall system by using a process stream instead of utilities to heat the stream in question.

The reactor feed, however, is not heated enough through the counter-current contact with the reactor effluent. Thus further heating is employed in E-102 where the reactor feed at 108.9^oC is contacted with a proprietary heat transfer medium, Dowtherm G, which flows in a separate self- contained loop. According to Williams, this fluid is preferable over others due to its overall higher heat capacities and higher heat transfer coefficients. These properties also allow for Dowtherm G’s lower flow rates compared to the flow rates for molten salts (another commonly used heat transfer fluid). A fired heater E-103, reheats the Dowtherm G after it is contacted with the reactor feed. Heated with methane gas, the Dowtherm G returns to 375^oC from 350.2^oC and is ready once more to heat the feed. After this point, the feed, at 250^oC, is ready to enter the reactor system (Williams 7).

The reactor system, at which the feed enters, is composed of three shell-and-tube vertical reactors. Two of the reactors are online at all times with the third being readily available for rotation. This setup ultimately limits the downtime that this process will experience due to reactor failure.

Upon exiting the reaction system, an effluent composed of large amounts of acetone, hydrogen, and water and a small amount of IPA, enters E-101, this time as a hot stream. In this capacity, the heat exchanger acts as a chiller, cooling the effluent from 356.3^oC to 75.41^oC through the counter-current exchange previously mentioned (Williams 6). However, due to the extremely exothermic nature of the acetone reaction, even more heat must be removed than can be accomplished in E-104. Thus E-104 additionally cools the process stream to 40^oC through the use of cooling water on the shell side of the exchanger.

The 40^oC process stream, however, requires a further temperature reduction of 20^oC for entrance into the separation and purification sector of the system (Nyalakonda 8). This cooling process can only be accomplished by using refrigerated water as the cold stream since ordinary cooling water is not available at these low temperatures. This is somewhat disadvantageous since refrigerated water is very expensive. This exchange takes place in E-105.

At this point, the 20^oC reactor effluent enters a flash tank where pressure considerations allow for the most gaseous hydrogen to leave out the top along with residual IPA and water and about 10% of the acetone. This top stream exiting the flash tank is then fed to a packed absorber where the hydrogen is removed from the IPA and acetone by running water through the column. The hydrogen leaves out the top and the recovered acetone and IPA is taken out the bottom. This bottoms stream is then mixed with the bottoms of the flash tank that contains most of the IPA and acetone.

The mixed stream then enters the waste water separation column where it is rigorously separated into waste components and an overhead rich in IPA and acetone. The condenser of this tower cools the distillate with cooling water operating at 10^oC temperature difference, 30^oC to 40^oC. The reboiler of this tower heats the bottoms with low-pressure steam, at 160^oC, to enable adequate boil-up ratios. These ratios, in turn, allow for greater recovery of acetone and IPA in the distillate (Nyalakonda 15).

The waste stream exiting out the bottom of the waste water tower is then fed to E-108 where it is cooled from 97.55^oC to 44.95^oC. This step is needed since the waste water must enter the waste water treatment plant at this specified temperature.

The final operation performed in this process is that of the purification of acetone to 99.9%. The distillate from the waste water tower enters the acetone purifying column at 59.15^oC. It is then rigorously distilled in order to obtain a 99.9% pure acetone stream at 56.95^oC leaving as the distillate and an 88% azeotropic mixture of IPA and water at 84.65^oC leaving out the bottom. The condenser and reboiler of this column behave similarly to those discussed in the waste water tower. Again, cooling water and low-pressure steam are used in their respective equipment (Nyalakonda 15).

Optimized Design

Pinch Analysis

The pinch analysis helps establish an arrangement of heat exchangers that minimizes the amount of utilities used in the optimized process. In this acetone plant, two hot streams are to be cooled, and four cold streams are to be heated. The following table labels the process streams as hot or cold streams and gives the initial and desired final temperatures.

Stream H1, the reactor effluent, is listed as having a desired temperature of 40° C; however, it must be cooled to a temperature of 20° C before entering the flash tank in the separations section of the process. Refrigerated water was used in the base case to cool this stream from 40° C to 20° C, and this, unfortunately, cannot be avoided. At 31° C, the coldest process stream, C1, would not be able to cool H1 beyond 41° C without violating the specified minimum temperature approach of 10 C° . Similarly, cooling water begins at a temperature of 30° C and can only cool H1 to a temperature of 40° C. Therefore, refrigerated water must be used to cool the stream from 40 ° C to 20 ° C, and H1 is specified to reach only a temperature of 40 ° C for the pinch analysis.

Table 1 includes the streams that run through the column reboilers as cold streams, but the condenser streams are not included. Because heat is to be removed from these streams, they would be classified as hot streams. However, both streams enter the condensers at relatively low temperatures, 56 ° C and 58 ° C, making it difficult to match them with cold streams. The only possible match would be with C1, the reactor feed. Unfortunately, condensing these streams requires relatively large duties, -2336190.8 and –2789086.2 Btu/hr. The reactor feed, having a flowrate of only 2674.2 kg/hr, would be incapable of condensing either stream completely, so the streams would have to be cooled further with cooling water anyway. Therefore, it seems impractical to run the condensers with a process stream as a coolant, especially since cooling water is relatively inexpensive. Steam, on the other hand, is expensive and makes up 73% of the base case cost. Because it would be very advantageous to be able to replace the low pressure steam in the reboilers with another fluid, the reboiler streams were included in the pinch analysis.

The mC_p values for each stream must be approximated, where m is the mass flowrate and C_p is the heat capacity. These values are calculated using the formula Q=mC_pD T, where Q is the heat duty and D T is the temperature change. This gives an approximate average value for mC_p and is fairly accurate for streams with nearly constant values for mC_p. However, many of the process streams included in this analysis undergo a phase change. For example, the reactor feed begins as a liquid at 31° C but enters the reactors as a vapor at 250° C. The reactor effluent leaves the reactors as a vapor at 356° C and enters the separations process as a liquid/vapor mixture at 40° C. The mC_p values of these streams cannot be assumed to be constant as the streams phase change. Therefore, these streams are broken up into dummy streams of constant mC_p, but each with different values, to approximate the mC_p profile of the phase changing streams. The reactor feed stream is broken up into three dummy streams determined by the dew point and bubble point of the stream. The reactor effluent is only broken up into two dummy streams, one representing the stream as it is cooled to its dew point and the other representing the stream as it is further cooled to its desired temperature, 40° C. This stream is not broken up into three streams because the final state of the stream is a liquid/vapor mixture, meaning that the bubble point is not reached. The dew points and bubble points of these streams are determined using Aspen sensitivity runs, which also give the duty for each temperature interval. The following table gives the duties and temperatures of the dummy streams as well as for the original, unbroken streams.

As shown in Table 2, the dew point of the reactor feed is at approximately 110.6° C, and the dew point is at about 150.7° C. The dew point of the reactor effluent is at about 169.7° C. The duties of the original streams are given in the reports by Kavinetor, Inc. and Group C. Williams. Summing the duties of the dummy streams gives approximately the same duty as the original stream; therefore, the duties given by Aspen for the dummy streams are reasonable.

The two streams that pass through the column reboilers also phase change. However, they are not broken up into dummy streams because they enter the reboiler already at the bubble point and exit the reboiler before they reach the dew point. Also, the temperature change of these streams while they phase change is minimal. It is, therefore, assumed that the mC_p values are nearly constant for these streams.

The following table gives the average mC_p values used in the pinch analysis.

The mC_p values were used in Matlab’s entHC program to generate the following plot.

The curve representing the cold streams is shifted by 750000 Btu/hr to satisfy the specified 10 C° minimum temperature approach. The pinch analysis gives 90° C and 80° C as the pinch temperatures for the hot and cold streams, respectively.

Process streams are matched carefully so that heat is not transferred across the pinch. Also, for a match to be feasible at the pinch, the mC_p value of the hot stream must be greater than that of the cold stream on the cold side of the pinch, and the mC_p value of the cold stream must be higher than that of the hot stream on the hot side of the pinch (Seider 257). The following diagram represents the heat exchanger network based on mC_p values and the pinch analysis.

The red lines represent the hot streams, and blue lines are the cold streams. The gray lines connecting two streams represent heat exchangers, and the blue and red dots represent cold and hot utilities, respectively. The reactor effluent, H1, is matched with the reboiler stream of the acetone column, C2. As shown in table 2, the reboiler stream requires a relatively high duty and is, therefore, able to cool H1 from 356° C to the pinch temperature. H1 is then further cooled by C1 to 78° C, and then is cooled by cooling water to the desired temperature of 40° C. The waste water stream, H2, is matched with the reactor feed, C1, on the hot side of the pinch. H2 needs to be cooled only 7 C° to the pinch temperature. It is then matched again with C1 and is cooled to the desired temperature of 45° C. The reactor feed, C1, is first heated by the waste water stream, H2, from 31° C to 46° C. It is then matched with H1, the reactor effluent, and is heated to the pinch temperature. On the hot side of the pinch, it is heated again with the waste water stream to a temperature of 82° C. It is then heated to the desired temperature of 250° C with Dowtherm G. The acetone column reboiler stream, C2, is matched with the reactor effluent, H1, at the pinch and is partially vaporized to a vapor fraction of 0.53. It is then heated with Dowtherm G to its final vapor fraction of 0.95. Finally, the waste water column reboiler stream, C3, is only run with Dowtherm G and is not matched with any other process stream. Matching C3 with another stream would require splitting one of the hot streams, adding an extra heat exchanger. Also, as shown in table 2, the required duty of C3 is relatively large. Neither hot stream would be able to vaporize C3 to its desired vapor fraction; therefore, additional heat would have to be provided by Dowtherm G.

Dowtherm G was chosen as the heating utility for this process because of the high cost of steam. Dowtherm G is readily available as it is already being used to drive the endothermic reaction occurring in the reactor section, as specified by the design by Group C. Williams. Using the Dowtherm G to heat process streams requires more duty from the fired heater that is used to heat the Dowtherm G. However, this fired heater is powered with methane, which is significantly less expensive than steam.

Figure 3 shows that two streams, H2 and C1, are matched with each other twice, once on the hot side of the pinch and once on the cold side. These two heat exchangers are combined into one so as to minimize the number of heat exchangers in the network. The elimination of one of these exchangers requires that heat be transferred across the pinch, which will increase the amount of utilities required as seen in the following table.

Because the low flow rate of the waste water stream, the duty required to cool the stream is relatively small, as can be seen in table 2. Therefore, the increase in utilities resulting from combining the two heat exchangers is minimal.

Optimized PFD

The proposed network of exchangers is modeled in HYSYS and shown in Figure 4. The feed is first heated (E-101) by the waste water stream, which is cooled to its desired temperature in this step. It is next heated by the reactor effluent (E-102) to its pinch temperature of 80C, and again by Dowtherm G (E-103) to its final temperature before entering the reactor. The reactor effluent, which leaves at 356C, is cooled (E-106) by the reboiler stream of the acetone column before it passes through E-102 to heat the feed. The effluent is then passed through another two exchangers with cooling water and refrigerated water, respectively, before it reaches the flash unit of the separations section. The reboiler stream of the acetone column, after passing through E-106, must be heated again by Dowtherm in E-104 before returning to the column. The reboiler stream of the waste water column is not used to cool another process stream, but is completely heated by the Dowtherm (E-105). There are a total of 7 exchangers shown in Figure 4, but there are actually a total of 10 necessary exchangers. Of the remaining three, two are the condensers which are cooled using cooling water, just as they are in the base case. As mentioned before, they are not included in the heat integration because of their low temperatures. The third one not shown is the exchanger that uses refrigerated water to cool the effluent.

The entire optimized process, including all process units and streams, is provided in Figure 5 for a more complete representation. The optimized design consists of ten heat exchangers, one more than the base case design. However, the design successfully decreases the amount of utilities required for the process.

Heat Exchanger Details

Detail of a Floating Head Heat Exchanger

To determine the specifications required for a floating heat exchanger, a detailed analysis was performed on exchanger E-102. This exchanger transferred heat from the hot reactor effluent on the shell side to the incoming cool reactor feed on the tube side. A floating head model was appropriate for this situation since there was some condensation in the hot stream along with significant temperature differentials. The floating head’s structure also allowed for thermal differential expansion of the tubes without the fear of shell or tube-support damage.

To determine the required heat transfer area for this exchanger, a simulation was run in Hysys in order to determine overall UA values and log mean temperature differences (LMTD). The simulation reported a heat duty of 273,100 Btu/hr, a UA value of 8369 Btu/^oF-hr and at LTMD of 33.57^oF with the incoming hot stream at 194 ^oF, the incoming cold stream at 199 ^oF, the outlet hot stream at 175 ^oF, and the outlet cold stream at 176 ^oF. Flow rates and temperatures for each stream can be found in the Appendix, in the Hysys report based on the optimized heat exchanger network. These numbers were checked for accuracy by inserting them into the standard heat transport equation: Q = UAD T_ln. To calculate the heat transfer area for this operation, a heat transfer coefficient, U, was needed. This parameter was determined from heuristic values that correctly paired the counter-current streams. In this case, process streams exchanged heat with each other with the cold shell side reactor feed remaining liquid during the entire heat transfer while the hot tube side reactor effluent condensed slightly from a vapor fraction of 1 to 0.9364. These conditions allowed for the choice of a heat transfer coefficient of 150 Btu/hr-ft². This number depicted a situation where a condensing vapor was crossed with a flowing liquid (Perry 10-39).

The area that resulted was 55.79 ft², which could be further dissected into tube number. Using a heuristic value of ¾" outer diameter for each tube and a characteristic length of

8 ft (Seider 316), the number of tubes was calculated to be 36 by the equation:

#tubes=A/[(2p (.03125")8’].

Tube pitch was chosen to be 1" in a triangular arrangement. This configuration was suitable since it adequately withstands medium to high pressure drops and as well as is generally suited for non-fouling fluids (Seider 318).

Baffles were arranged with a baffle cut of 25% (a value given by heuristics) and a shell to baffle clearance of 1/8" (Seider 318). Plain tube joints were used since there were no extremely high pressures in this exchanger that would require the more durable grooved or serrated variety (Ludwig 26).

This heat exchanger as well as all the others in the acetone production system was constructed out of carbon steel for both the shell and the tubes. This material was allowable since acetone and its constituents are not very corrosive and therefore do not pose a significant threat for structural damage. Also carbon steel less expensive than stainless steel and its use in the heat exchangers of this process helps to reduce overall capital cost (Turton 728-736).

Table 5 provides a list of heat exchanger specifications.

Other Exchangers

Six of the other heat exchangers in the process were also chosen as floating head shell-and-tube exchangers due to the phase changes that occur within each of them (Turton 728-730). The floating head model was especially good for the three reboilers of this system E-104, E-105, and E-106. The reboilers exhibited not only phase changes but also the high temperature differentials that often caused tube expansion. By choosing the floating head model for this application, the potential damage to equipment due to this thermal phenomenon was greatly reduced (Ludwig 9).

The condensers of both columns were modeled as fixed tube sheet shell-and-tube exchangers since they did not exhibit the thermal expansion so often seen in reboilers and other process exchangers (Ludwig 9). Thus the need for a floating head model in this situation was eliminated even though phase changes were still present. For fixed tube sheet exchangers can handle phase changes appropriately, but cannot accommodate high temperature differentials.

The final type of heat exchanger employed in this design was the double pipe model used to cool the waste water stream and heat the reactor effluent in E-101. Double pipe exchangers are effective in that they create relatively little heat transfer area due to the presence of individual shelled tubes (Turton 732). This reduction of heat transfer area ultimately results in a reduction in capital cost. Double pipe heat exchangers are not adequate for handling phase changes and thus could only be used in this section of the design since all streams flowing into and out of the exchanger are liquid.

For all of the heat exchangers in this process, heat transfer coefficients were estimated from Perry’s handbook. These coefficients were chosen based on the which materials exchanged heat. For example, all reboilers and condensers were given a U value of 150 Btu/^oF-hr based on phase change considerations. This reasoning was also employed when choosing a U value of 150 Btu/^oF-hr for a system with condensing vapor paired with flowing liquid as seen in exchanger E-102.

Table 6 tabulates the types of exchangers present in the optimized design.

* This exchanger uses refrigerated water and was not used in the optimized heat exchanger network since it was kept at the specifications given by the base case and no alternate to refrigerated water could be found.

** These condensers were not included in the optimized heat exchanger network since they were also kept at the specifications given in the base case.

^† Values taken from Perry’s Chemical Engineering Handbook.

Economics

Capital Costs

Base case costs for both C. Williams and Kavinetor’s design consist of heat exchanger prices as well as the costs for the various utilities used within the process. The capital costs for the optimized case are slightly higher than that of the base case due to the additional heat exchanger required in the reboiler section of the acetone purification column. Additionally, even though the optimized fired heater has a larger duty than does the one in the base case, it still is smaller than the minimum size required by CapCost. Thus both heaters are priced equally at this minimum value.

The breakdown of capital costs is as follows:

Operating Costs

The operating costs for each of these designs are where the significant differences are noted. Due to the removal of low pressure steam from the optimized reboilers, the optimized case has much lower YOC’s when compared to the base case. A comparison of these two designs is seen in Table 8.

There is an annual savings of $300,000, which translates into a 55% decrease from the base case operating costs. The significant difference in utility distribution can be seen in Figure 6.

As shown in the bar graph, the costs of electricity, cooling water, and refrigerated water are approximately the same for both the base case and the optimized case. Methane costs increase in the optimized case due to the increased usage of Dowtherm G, but the major change is the complete elimination of low pressure steam.

The net present value of operating costs for the optimized case, at a 10% discount rate, is $2,700,000 for a plant life of 20 years. This value is 58% less than the net present value of costs for the base case.

 Conclusion

King&Co. has successfully accomplished the goal of this project and has proposed an optimized design of the acetone production process that will significantly lower annual utility costs. Our method of heat integration is not only cost-efficient but also adheres to thermodynamic rules while maintaining important specifications detailed in the base case. In comparison to the base case exchanger network, there is only one more exchanger unit in the optimized case. This translates into a higher capital cost, but the increase is very small relative to the significant decrease in utilities. A large portion of this decrease is due to the replacement of low pressure steam with Dowtherm G, which is already readily available in the reaction process. The advantage here is that more Dowtherm need not be purchased; it only requires more methane in order to be re-heated before returning to reactor. The remaining savings in utilities is due to the matching of process streams with each other in heat exchangers, thereby requiring less heating and cooling from other sources such as cooling water and steam. Overall, the teamwork of the engineers of King&Co, Group C. Williams, and Kavinetor, Inc. has resulted in an optimized design that provides a much improved acetone production process.

Bibliography

Dowtherm G: Heat Transfer Fluid. DOW Chemicals: September, 1997.

Ludwig, Ernest E. Applied Process Design for Chemical and Petrochemical Plants,

2^nd Ed. Houston, Texas: Gulf Publishing Company. 1983, pp. 9-31.

Nyalakonda, Kavita. "Optimized Acetone Separations Design." 1999.

Perry, Robert H. Chemical Engineer’s Handbook, 5^th Ed. New York: McGraw-Hill Book

Company. 1973, pp.10-37-10-41.

Seider, Warren D., J.D. Seader and Daniel R. Lewin. Process Design Principles.

New York: John Wiley & Sons. 1999, pp. 303-332.

Turton, Richard, Bailie, Richard C., Whiting, Wallace B., and Joseph A. Shaeiwitz.

Analysis, Synthesis, and Design of Chemical Processes. New Jersey:

Prentice Hall. 1998, pp. 728-736.

Williams, Carl. "Acetone Production via Isopropanol Dehydrogenation: Reactor Design."

1999.