(529h) Reactor Network Development for Multiple Rigid Polyol Productions Under Uncertainty

In many polymer processes semi-batch reactors are used, and only a few examples of continuous reactors appear in the patent literature. As the product demand of a polymer increases, larger or more semi-batch reactors are built to meet the demand, and the capital cost per production capacity is predictably lowered. However, this capital cost is still too high for re-investment, and alternative lower capital cost solutions, such as switching the production from semi-batch to continuous processes, could help sustain growth. Hence, continuous processes can reduce the capital cost and the cost required for heat transfer equipment, since the heat load is more evenly distributed over time. In addition, continuous reactors are easier to optimize at a steady state to maximize the overall production rate. However, continuous reactors are not as flexible regarding multiple products, product transition can be difficult, and scale-up of the products requires extra investment. The motivation of this project is to switch the current production from the existing semi-batch reactors to continuous reactors, such as plug flow and continuous stir tank reactors, to lower the capital cost. Furthermore, it is not efficient to use the reactor network to produce a single rigid polyol. Therefore, we aim to construct a reactor configuration that is able to produce multiple rigid polyols. In addition, we are interested in developing the reactor network under uncertainty in kinetic parameters.

In this talk, we first focus on developing continuous reactor network models able to produce multiple rigid polyols under strict product and safety specifications. At the same time, we determine the optimal decision profiles that lead to minimum capital cost. Decision variables include the feed rates of initiator, monomers and catalyst, reactor temperature, residence time, number and location of monomer injection points. Moreover, we narrow down the types of continuous reactors that can be part of the network to two: plug flow reactor (PFR) with multiple feed injection points and continuous stirred tank reactor (CSTR). The PFR model is a differential algebraic equation (DAE) optimization problem. The simultaneous collocation method is applied to transform the DAE into a mixed integer nonlinear programming (MINLP) problem. An iterative algorithm is proposed to solve the MINLP, where binary variables are manually fixed.

After obtaining the optimal capital cost, we move on to handle uncertainty, which comes from eight kinetic parameters in the reaction process. Compact problem formulations with modifications on the constraint (back-off constraints) and multi-scenario formulation with each scenario corresponding to one discretized uncertainty level are adopted to develop the reactor network and operation recipe. Back off terms that are obtained from Monte Carlo simulations tighten the constraint and shrink the feasible region of the optimization problem to such a level that variations of the constraints in the worst case can still be handled and thus feasibility is ensured. The multi-scenario formulation is also tolerant to the uncertainties and is has better performance than the back off method, since it allows different operation recipes (recourse variables) for different scenarios. On the other hand, multi-scenario approach increases the problem size dramatically. In this talk we demonstrate the effectiveness of both uncertainty approaches and compare results on the multi-product reactor network.