Integration of design and control: A robust control approach

In general, process systems should be designed to handle all possible changes during operation. However, processes are typically designed based on steady state models without incorporating control considerations. However, the optimal steady state design may not achieve the best closed loop performance when the controllers are implemented.; This research proposes a new method to integrate control considerations at the design step together with steady state costs, i.e. capital and operating costs. Control considerations were incorporated into the design in terms of a variability cost term. This term includes the closed loop performance and process model uncertainty. The variability cost term was obtained as a function of design and control variables. A single optimization was formulated with a single objective function which consists of the steady state cost function and variability cost. The process models can be correlated to the design variables and implemented into the optimization problem as equality constraints. The control performance is guaranteed by integrating the robust stability and manipulated variable constraints as inequality constraints.; Optimization using the traditional and a new integrated method was performed. The optimization results show that the when integrating control into the design step the total cost is always lower than the cost calculated from traditional method. As the product price increases, the savings associated with the integrated method increase. For example, as the product price was increased by factors 5, 10, 15 and 20 the savings were obtained as 16%--41% for IMC SISO, 18%--48% for MPC SISO and 8%--94% for MPC MIMO.; The designs with and without automatic control were examined using the integrated method. It was found that the design without control has a much higher capital cost and variability cost. The high variability cost (at least 26 times higher than the variability cost with control) is due to the large intermediate storage tanks needed to attenuate the product variability.; The process parameters such as process dead time and the limit of the manipulated variable were also studied. It was found that when dead time is present the integrated method saves more compared to the case without dead time. Also the limits on manipulated variable can affect the cost of the process. As the limit of manipulated variable was tighten the more process cost increases.