Optimal Design and Operation of Batch Processes

Abstract

The increased importance of high value-added specialty chemicals has stimulated interest in the development of better design and optimization methods for batch processes. These processes have a number of particular aspects (time-dependent behavior, discrete processing and structural alternatives) which make the development of optimal designs quite difficult. Five sets of decisions must be made to design a batch process. Because the entire problem is so complex, previous workers have focused on smaller, more manageable parts of the overall problem. Much of the previous work has dealt with the optimal sizing of equipment units in order to produce a set of products in a new multiproduct batch plant with minimum capital investment. Processing conditions have generally been assumed given in the form of the product recipes and not subject to change. This thesis focuses on the design and operation of batch processes and represents a first attempt to consider process performance issues and equipment sizing decisions together in an optimization framework. Complexities introduced by using existing equipment are also included. The optimization problem consists of selecting the equipment units to use at each stage in the process and choosing values for all process operating conditions and operating times in order to optimize a suitable objective function.A problem formulation for the optimal design of a new batch process is developed. This formulation incorporates the effects of the overall multipurpose plant environment on the design of a single new process by allocating fixed costs through the use of equipment usage charges. These charges represent the opportunity cost of allocating scarce plant resources to one product rather than another. A decomposition strategy is proposed for solving the optimization problem. By partitioning the decision variables into two groups, simpler subproblems are generated.The Performance Subproblem involves optimizing the continuous variables that represent the processing conditions and operation times for a process with fixed structure. Generic performance trade-offs involving processing intensity and the distribution of performance load are identified. The problem is formulated as a nonlinear programming problem (NLP) and solved using a successive quadratic programming algorithm. Results 2 are reported for a series of test problems to illustrate the basic elements of the solution approach.The Structure Subproblem involves assigning known equipment units to stage locations for a process with fixed performance in order to minimize the total equipment usage charges. The combinatorial optimization problem can be formulated as a mixed integer nonlinear programming problem (MINLP) and solved using mathematical programming methods such as the Outer Approximation, Equality Relaxation method. To circumvent potentially large solution times, an approximate solution strategy based on local search techniques is developed. This approximate method generates near-optimal solutions and requires one...

Process performance models in the optimization of multiproduct protein production plants

“…Moreover, this approach captures the influence of the variables that establish economic trade-offs between or among stages (Barrera and Evans, 1989) and the interactions among different product requirements. None of these effects can be taken into account with the constant factor model.…”

Mixed‐logic dynamic optimization applied to batch distillation process design

“…A similar but simpler design problem has been presented by Olden-Ž . burg et al 2002 . The example is solved with the proposed MLDO formulation as well as with a MIDO solution technique.…”

Worst-case and distributional robustness analysis of finite-time control trajectories for nonlinear distributed parameter systems

“…Define as the perturbation about the nominal vector . Then, the control trajectory is (2) It is desirable to define an uncertainty representation that is general enough to include the uncertainty descriptions common in practice. This can be done using the Hölder -norm [28] defined for a general vector by…”

“…The model-based computation of optimal control policies for batch and semibatch processes is of increasing interest due to industrial interest in improving productivity [2], [6], [17], [40]. However it has been shown that all of the benefits of using optimal control can be lost with subtle variation in either the model parameters or in the performance of the feedback controller to track the optimal setpoint trajectory [16], [41], [44].…”