REACTOR OPTIMIZATION:

THE PRODUCTION OF ACETONE

MANAGER: RYAN KELLOGG

ENGINEERS: PATRICK LUCK

ANDREW HARPER

12 OCTOBER, 1998

 

 

ABSTRACT

 

The production of acetone via the dehydrogenation of isopropanol is optimized in this reactor design project. The computer program Hysys was utilized to simulate the process, using a Peng Robinson equation of state with UNIQUAC activity coefficients.

The dehydrogenation reaction occurs with first order kinetics and is endothermic; thus, a circulating stream of molten salt must heat the reactor. The reaction occurs in the vapor phase over a zinc oxide and zirconium oxide catalyst at temperatures between 300^oC and 400^oC. The feed stream is therefore pumped to a higher pressure and vaporized before entering the reactor. Reactor conversion is an increasing function of temperature and pressure, although the pressure used is limited by the thermodynamic reaction equilibrium and the heat transfer requirements of the reactor.

The design team seeked to minimize the capital and utilities costs of the reactor and associated pumps and heat exchangers. The single reactor process given in the book was improved upon by increasing the feed pressure, yielding cost savings of nearly 3%. These savings came primarily as a result of decreasing the reactor size. Further cost savings were made by installing an adiabatic reactor after the heated reactor. This reduced the necessary conversion in this expensive reactor, both reducing its capital cost and the utilities costs associated with the molten salt heating stream. The total cost savings were $150,000, nearly a 5% improvement on the base case.

 

CONTENTS

 

1. Introduction

   1. Process Flow Diagram
   2. Economic Assumptions
   3. Reaction Kinetics
   4. Reaction Equilibrium
   5. Thermodynamics Package

2. Safety
3. The Base Case
   1. Process Details
      2. Cost Breakdown
4. Single Reactor Optimization

   1. Methodology
   2. Reactor Size: Kinetic and Heat Transfer Requirements
   3. Cost of Optimized Single Reactor System

5. Two Reactor System: Final Optimization

   1. Basis for Addition of Second (Adiabatic) Reactor
   2. Optimized Process
   3. Cost Savings

6. Final Recommendations
7. References

   Appendix A: PFD and Stream Details for Single Reactor System

   Appendix B: Key Parameters and Cost Breakdowns for Major Test Runs

   Appendix C: PFD and Stream Details for Dual Reactor System

    

    

   INTRODUCTION

    

   This objective of this project is to economically optimize the production of acetone from a reactor via the dehydrogenation of isopropyl alcohol:

    

   (CH₃)₂CHOH ® (CH₃)₂CO + H₂

   This reaction is endothermic and is carried out in a catalyst-packed plug flow reactor. Due to the endothermic nature of the reaction, the reactor is designed as a heat exchanger with a large number of 2 inch diameter tubes, heated by a molten salt stream.

   The separation equipment is not modeled in this study. Thus, to avoid any question of the effects of our reactor manipulations on the costs of separation, the reactor conversion was held constant at 90%. Therefore, the goal of the optimization of the reaction was not to increase conversion but to find the best combination of inlet and outlet conditions that minimized capital and utility costs while achieving a 90% conversion.

    

   Process Flow Diagram

   The entire system used for this process consists of a feedstock stream, a recycle stream, a pump, a heater, the reactor, and two coolers, as illustrated in Graphic 1. Both the feedstock and recycle streams are azeotropic mixtures of isopropyl alcohol (IPA) and water (88 wt.% IPA), with the recycle stream being 10% of the flowrate of the feedstock. The pump is a centrifugal pump with an efficiency of 40%; this pumps the feedstock up to the reactor pressure. The vaporizer boils the feedstock using high-pressure steam, as the reaction occurs in the gas phase at temperatures exceeding 300^oC. Finally, the two coolers liquefy the reactor outlet stream and decrease its temperature to 20^oC so that it may be separated.

    

   

    

    

    

   Economic Assumptions

   There are a number of assumptions used in the costing of the process. All the groups analyzing this process agreed that an incremental, rather than complete, economic analysis was necessary to correctly interpret results. Instead of determining an overall profitability for the process as a whole, the net present value of the suggested modifications to the base case plant design (that is, the design presented in the text), was calculated.

   This method of analysis was chosen for several reasons. First, the separators and associated utilities could not be costed because they were not modeled. Second, the separation was assumed to be perfect; that is, all the unreacted IPA was recycled (in the text, 99% is recycled so this assumption is valid). This is significant because all the IPA entering the plant is thereby converted to acetone: there are no losses. All modifications to the reactor design could therefore have no affect on the overall 100% conversion within the plant. Thus, the prices for the product and the feed streams had no impact whatsoever on the design optimization, and were not considered. Our design modifications affected only the capital costs and the utilities costs of the reactor and its associated heaters, coolers, and pumps. Therefore, an incremental economic analysis that considered only these costs was clearly preferable to any highly inaccurate attempt to determine an overall plant profitability, when faced with the uncertainties of chemical prices and the costing of unmodeled separators.

   To calculate the net present value of plant improvements, the plant is assumed to have a 20 year life. All utilities costs are assumed to be constant throughout that timespan, and are discounted at an interest rate of 7%. All capital costs are assumed to be paid up-front in year zero and are therefore not discounted at all. However, a contingency of 15% is added to all capital costs to reflect unforeseen installation and construction problems, as well as the tendency to underestimate capital costs. Income taxes are not considered; thus, no tax write-offs from depreciation enter into this analysis.

   All capital costs are estimated using CAPCOST, while utilities costs are calculated using the correlations presented in Chapter 3 of the text.

    

   Reaction Kinetics

   The kinetics used in this process are those presented in the problem statement in the text. The reaction is first order with respect to the concentration of IPA, and occurs in the vapor phase over a catalyst:

   -r_IPA = k_oexp(-E_a/(RT))C_IPA

   with E_A = 72.38 MJ/kmol, k_o = 3.51 x 10⁵ m³gas/(m³reactor*s), C_IPA = kmol/m³gas

   In addition, several side reactions occur to a small extent, but these reactions are ignored due to their small importance at temperatures below 400^oC. Note that the reaction rate is increased by an increase in pressure, as this increases the concentration of the reactant.

   The catalyst used is a zinc oxide and zirconium oxide combination (6%-12% ZrO), which works very well over reaction temperatures from 300⁰C-400⁰C with minimal degradation¹. Metal catalysts such as copper were rejected because they operate only at temperatures exceeding 450⁰C. Cost data for ZnO/ZrO was unavailable; however, an alternative presentation of the problem suggested a catalyst cost of $1 per pound in 1976². This was doubled to roughly represent inflation, yielding a catalyst cost of $2 per pound for this problem.

    

   Reaction Equilibrium

   In order to help find the best process to obtain the desired 90% conversion, the equilibrium conversion was calculated for several different pressures and temperatures, using the Gibbs reactor in Hysys. As might be expected due to the endothermic nature of the reaction (heat of reaction = 62.9 kJ/mol), an increase in temperature led to an increase in conversion. Conversely, an increase in pressure led to a decrease in conversion, because two moles of gas produced for every mole consumed in the reaction. Therefore, a lower pressure favors the products, whereas a higher pressure favors the reactants. The effects of this can be seen in Chart 1, which plots conversion against pressure at various temperatures.

   

   Thermodynamics Package

   The Peng-Robinson EOS, with UNIQUAC activity coefficients, was used to thermodynamically model the system in Hysys. The UNIQUAC coefficients were used to model any liquid-liquid and liquid-vapor interactions that occurred between the IPA and water in the feed vaporizer and in the coolers. Sensitivity runs demonstrated that the property package selected had little effect on reactor conversion and utilities expenses. A switch to a PRSV EOS with no activity coefficients increased reactor conversion by only 0.2%, and decreased the vaporizing and cooling duties by 0.9% each. On the other hand, the use of UNIQUAC activity coefficients with an ideal EOS decreased reactor conversion by 1.9%, with negligible effect on the heating and cooling duties.

    

   SAFETY

    

   This process is extremely unsafe. Anyone who comes within mile of it has children with 3 feet, one eye, and bushy eyebrows. But seriously folks, this process is a very safe one. All streams are kept at relatively low temperatures and pressures. Therefore, no exotic equipment or safety practices must be used. One potential problem is the auto ignition temperature of acetone at 538° C.¹ This is not a large problem because no part of the designed process has a temperature anywhere near this temperature. In addition, hydrogen is one of the products of the reaction. Hydrogen and acetone vapors are both flammable if exposed to an ignition source; therefore, all streams containing product should be kept clear of all such equipment (for example, the fired heater for the molten salt). Plant ventilation should be adequate to dissipate hydrogen and acetone vapor in the event of a leak to avoid any concentrations. Dissipation will also mitigate the negative health effects of acetone exposure, namely irritation to the eyes, skin, and upper respiratory tract.³

    

   THE BASE CASE

    

   Process Details

   The process presented in the book is based on a single shell and tube reactor with 488 2" diameter, 20’ long, catalyst filled tubes which is heated on the shell side with a stream of molten salt. The salt is a eutectic mixture of various nitrite and nitrate salts which melts at 146° C: 40% NaNO₂, 7% NaNO₃, and 53% KNO₃⁴. For the base case used in our process, the number of tubes was modified to 710 in order to achieve the desired conversion. The reactor tubes are constructed of stainless steel to handle the corrosive hydrogen gas produced in the reaction, while the shell is carbon steel because the molten salt should not be a problem, as illustrated on page 64 of the text.

   The process begins with an IPA/water feed stream with a flow rate of 51.96 kmol/h. This stream is mixed with a recycle stream, and then passes through a centrifugal pump that pumps the mixed stream from 1 to 3 bars. Next, the stream passes through a vaporizer that increases the temperature from 32° C to 234° C, vaporizing the entire stream using high pressure steam. The stream then passes through the reactor and exits at 350° C with a 90% conversion of IPA. Finally, two coolers decrease the temperature of the reactor outlet to 20° C. The PFD for this process is the same as Graphic 1 and the detailed flow charts and worksheets can be found in Appendix A (though these charts are for the optimized case presented below).

    

   Cost Breakdown

   The PV of all plant costs for the base case presented in the book is $3.26 million. The yearly operating cost is $235,000, and the initial capital cost is $769,000 ($669,000 before the addition of the 15%). An inspection of the breakdown of the costs reveals the primary cost drivers of the process. Among the operating costs, the main drivers are the two heaters, demonstrated clearly in Chart 2. The feed vaporizer accounts for 66% of the operating cost while the salt heater accounts for 31% of the cost, for a total of 97%. Therefore, the pumps and coolers all together only account for 3%. Among the capital costs, the reactor itself represents 34% of the capital cost, while the salt heater, the size of which is related to the reactor heat input, is 35% of the capital cost.(Chart 3).

   

    

   

    

   SINGLE REACTOR OPTIMIZATION

   Methodology

   From the breakdown of the costs of the base case, one can see several factors that should be especially emphasized in minimizing the cost of the single reactor system. First, the cost of increasing the pressure is miniscule. Therefore, in any optimization the highest possible pressure should be used in order to maximize the kinetic benefits of increasing pressure, thereby decreasing the number of tubes needed in the reactor. However, care must be taken that the pressure does not go so high as to significantly decrease the equilibrium conversion. To assure a 90% conversion in the reactor, no pressure and temperature combinations were permitted such that the equilibrium conversion dropped below 95%.

   Second, the reactor inlet temperature should be kept to a minimum, with much of the heating taking place in the reactor. The cost of heating the fluid once it is vaporized using high pressure steam is much higher than heating using the salt stream; therefore, heating using the salt stream should be maximized and the vaporizer should merely vaporize the feed and nothing more.

   The optimal reactor outlet temperature can best be found by a trial and error method of guessing various outlet temperatures combined with various inlet conditions. The reason why there is no clear rule of thumb for this is that there are off-setting considerations: a higher outlet temperature decreases the necessary number of reactor tubes, but it also increases the size of the salt heater and its utility costs. In addition, the higher the outlet temperature, the higher the pressure that can be used in the inlet stream, because a higher temperature increases the equilibrium conversion. Therefore, the variables involved in determining the outlet temperature are numerous, and trial and error is the best method for finding the optimal temperature.

    

   Reactor Size: Kinetic and Heat Transfer Limitations

   Before a final optimum process can be designed, one final consideration must be dealt with. In order to obtain the 90% conversion at a certain temperature and pressure, a reactor has to be of a certain a volume, and therefore have a certain surface area. But, in order for the requisite amount of heat to be transferred to the reactor from the molten salt stream, a different surface area is required. Therefore, for each set of reaction parameters, a reactor surface area based on kinetics and one based on heat-transfer had to be calculated. The heat transfer coefficient used for these calculations was 20 Btu/h*ft²*° F², and the equation used to calculate the necessary heat-transfer area was that for a cocurrent heat exchanger given in Perry’s⁵:

   A = Q / (U*delT_lm), where delT_lm = [(t_1T -t_1S)-(t_2T -t_2S)] / ln[(t_1T -t_1S)/(t_2T -t_2S)]

   t_1T = inlet tube temp; t_1S = inlet shell temp; t_2T = outlet tube temp; t_2S = outlet shell temp

   One can see on Chart 4 that, as pressure increases, the number of tubes required to maintain heat transfer remains constant, while the number of tubes required to maintain the 90% kinetic conversion decreases. Because the larger of the two values must always be used, there arises a limit to the extent that increasing pressure helps reduce reactor size. Chart 4 demonstrates that, at 375⁰C, any increase in pressure beyond about 430 kPa will not decrease the necessary number of reactor tubes, as a heat-limiting constraint will have been reached. Therefore, at low pressures, the reactor is kinetically limited, while at high pressures the reactor is heat transfer limited.

    

    

   

    

   Cost of Optimized Single Reactor System

   The minimum cost single reactor system has an inlet temperature of 234° C, pressure of 5 bars, and an outlet temperature of 350° C. It contains 404 tubes and the pressure is low enough that the reactor is kinetically limited, rather than heat transfer limited. Pressure cannot be increased beyond 5 bars because this would decrease the equilibrium conversion below 95%, dangerously close to the desired 90%. The PV plant cost is $3.17 million dollars, saving $90,000, or 2.8% over the base case. Its yearly operating cost is $235,000, and its capital cost is $675,000 ($587,000 without the 15% increase). The majority of the savings result from a decrease in the capital expense for the reactor cost ($150,000 vs. $230,000). These savings can be traced back to the increase in the pressure from 3 to 5 bars, which allows the reaction to reach 90% conversion in a smaller number of tubes. Detailed flow parameters and compositions for the optimal setup may be found in Appendix A, while key parameters and cost data for all test runs may be found in Appendix B.

    

    

   TWO REACTOR SYSTEM: FINAL OPTIMIZATION

    

   Basis for Addition of Second (Adiabatic) Reactor

   As shown before, the largest portion of the costs associated with this system are the utilities expenses of the two heaters and the capital cost of the salt heater and reactor. Therefore, the best approach to decreasing the cost further from the single reactor system would be to find a way to further decrease the size of the reactor and salt heater. One way to do this is to include a second reactor that is adiabatic. The way this system would work is that some large portion of the reaction would occur in the heated reactor, and then the adiabatic reactor would complete the reaction to 90% conversion. The obvious advantage of such a system is that the size of the expensive heated reactor could be reduced in exchange for a cheap adiabatic reactor (essentially a packed tube).

    

   Optimized Process

   The set up of the two-reactor optimized process is essentially the same as that for the base case except that an adiabatic reactor has been added after the heated reactor, as demonstrated in Graphic 2. The optimal heated reactor has 327 tubes, and is kinetically limited. The adiabatic reactor is 3 meters long and has a diameter of 0.74 meter. Across the adiabatic reactor, the temperature drops from 360⁰C to 326⁰C as the endothermic reaction uses up energy. The heated reactor outlet temperature was chosen to be 360⁰C, as this is the minimum temperature that yields a reasonable decrease in heated reactor size. At lower temperatures, the adiabatic reactor simply could not convert significant amounts of IPA to acetone before the temperature dropped below 325⁰C. The feed is pumped to a pressure of only 3.5 bar because the pressure at the adiabatic reactor outlet can be no more than 3 bar due to equilibrium constraints. Detailed flow data can be found in Appendix C.

   

    

   Cost Savings

   The PV plant cost for the optimized two reactor system is $3.11 million, a savings of $150,000 (4.6%) over the base case and $60,000 over the best case single reactor. The yearly operating cost is $231,000 and the capital cost is $671,000 ($583,694 without the 15%). The bulk of the savings in the two reactor system over the base case occur in the reactor–80%, the salt heater–7%, and the initial catalyst charge-8% (Chart 4). Savings over the optimal single reactor system described above primarily result from a decrease in the utilities costs of the molten salt heater. Operating and capital costs for all equipment may be found in Appendix B.

    

    

   

    

    

   FINAL RECOMMENDATIONS

    

   The final recommendation of this design group is that this process should be carried out with a two reactor system consisting of a heated multi-tube reactor with 327 tubes, and an adiabatic reactor that is 3 meters long and has a diameter of .74 meter. The feed may only be pumped to a pressure of 3.5 bar, and enough heat must be provided to the heated reactor to provide an outlet temperature of 360⁰C.

   Further study must be done to obtain a better grasp of the heterogenous kinetics of the ZnO/ZrO catalyst. The appropriate pellet size and packing arrangement are yet to be determined. Further experimental data on the molten salt stream and its heat transfer properties would also be very helpful.

    

    REFERENCES

    

   1 Kirk-Othmer: Encyclopedia of Chemical Technology, 4^th Edition, vol. 1, 179-191,

   John Wiley and Son, 1976

   2 Encyclopedia of Chemical Processing and Design, Ed. J.J. McKetta, and W.A.

   Cunningham, vol. 1, 314-362, 1976.

   3 Acetone: Representative MSDS Summaries. http://www.clean.rti.org/an_msds.html

   4 Perry’s Chemical Engineer’s Handbook, Sixth Edition, 9-77.

   5 Ibid., 10-24,10-25