Design and Control of the Cumene Process William L. Luyben* Department of Chemical Engineering, Lehigh UniVersity, Bethlehem, PennsylVania 18015 The chemistry of the cumene process features the desired reaction of benzene with propylene to form cumene and the undesirable reaction of cumene with propylene to form p-diisopropylbenzene. Both reactions are irreversible. Since the second has a higher activation energy than the first, low reactor temperatures improve selectivity of cumene. However, low reactor temperatures result in low conversion of propylene for a given reactor size or require a large reactor for a given conversion. In addition, selectivity can be improved by using an excess of benzene to keep cumene and propylene concentrations low, but this increases separation costs. Therefore, the process provides an interesting example of plantwide economic design optimization in which there are many classical engineering trade-offs: reactor size versus temperature, selectivity versus recycle flow rate, and reactor size versus recycle flow rate. Design optimization variables affect both energy costs and capital investment. They also affect the amount of reactants required to produce a specified amount of cumene product. The economic effect of reactant consumption is very large, an order of magnitude greater than the impact of energy or capital. The process is presented in the design book by Turton et al. (Analysis, Synthesis and Design of Chemical Processes, 2nd ed.; Prentice Hall: Saddle River, NJ, 2003) and consists of a cooled tubular reactor and two distillation columns. The liquid fresh feeds and the benzene recycle stream are vaporized, preheated, and fed into the vapor-phase reactor, which is cooled by generating steam. Reactor effluent is cooled and fed to the first column that produces a distillate stream of benzene that is recycled back to the reactor. The second column separates the desired cumene product from the undesired p-diisopropyl- benzene. The purpose of this paper is to develop the economically optimum design considering capital costs, energy costs, and raw material costs and then to develop a plantwide control structure capable of effectively handling large disturbances in production rate. 1. Introduction The synthesis of a chemical process involves science, art, innovation, intuition, inspiration, experience, and common sense. Success also requires a lot of hard work and at times a little luck. The old saying that “There are many ways to skin a cat” certainly applies to the development of a process flowsheet. We start with raw materials (fresh feed streams of reactants) and want to produce the desired product(s) while, at the same time, achieving a low capital investment, low operating costs, and safe nonpolluting operation. The design engineer needs to consider a multitude of issues and objectives that are often conflicting and frequently qualitative in nature. Engineering trade-offs are important parts of chemical process synthesis. The classical example is the trade-off between costs in the reaction section of the plant versus costs in the separation section of the plant. Bigger reactors are more expensive (vessel and catalyst capital investment) but can produce higher conver- sion that requires a less expensive separation section (smaller recycle flow rates that require smaller diameter distillation columns with smaller reboilers and condensers). Discussions of these issues and examples of their application to typical chemical processes are presented in most chemical engineering design textbooks. Turton et al. 1 provide a number of useful examples, one of which is the cumene process studied in this paper. The authors do not consider optimization or control of this process. The purpose of this paper is to use the cumene process to illustrate some interesting design optimization features and principles. The two dominant design optimization variables are reactor size and benzene recycle. Increasing either one reduces the amount of undesirable byproduct that is produced, but increasing either of these variables raises capital and/or energy costs. Of more importance is the effect of these design optimization variables on conversion and selectivity, which translates into changes in the amounts of the reactants required to produce a fixed amount of cumene. For example, if a bigger reactor is used, reactor temperature can be lower, which improves selectivity while achieving the same conversion and also reduces the amount of raw materials. If the reactor is inexpensive (cheap catalyst), the optimum design is a large reactor. However, if the reactor is expensive, the optimum design is a small reactor. As Douglas 2 pointed out (Douglas Doctrine) two decades ago, the costs of raw materials and products are usually much larger than the costs of energy or capital in a typical chemical process. Therefore the process must be designed (investing capital and paying for energy) so as to not waste feed stocks or lose products (particularly in the form of undesirable products). Process economics dictate that the conversion of reactant must be quite high. These principles are clearly illustrated in the cumene process in several ways. (1) The fresh propylene feed steam contains some propane impurity, which is inert in the reactor and must have a place to get out of the process. Since the separation of propylene and propane is difficult, the economics strongly favor designing the reactor for a very high conversion of propylene. The propane and any unreacted propylene are flashed off and burned, so they only have fuel value. High propylene conversion can be achieved by either running at high temperatures or increasing the size of the reactor. The former increases the production of undesirable byproduct. The latter increases capital cost. (2) The undesirable byproduct is also burned so it only has value as fuel. Since it takes reactants to produce this product * E-mail: WLL0@Lehigh.edu. Tel.: 610-758-4256. Fax: 610-758- 5057. Ind. Eng. Chem. Res. 2010, 49, 719–734 719 10.1021/ie9011535  2010 American Chemical Society Published on Web 11/23/2009