Utilização de uma ferramenta para seleção automática de estruturas de controle para plantas de processos.

Abstract

In the chemical industry is very common, and essential, the use of control systems to keep the process running continuously and so that the final product is within the predetermined specifications and safely. However, the growing demand and competitiveness in the industry of processes requires an optimized operation and better use of raw materials and energy, and the key question is how to achieve these objectives without the need for large capital investments. The difficulty in maintaining the controlled plant also gives due also to the application of control loops observing the process locally rather than in an integrated manner. The problem increases when you have a process plant with large mass and energy integration, where small disturbances in the operating cycle, when spread to the whole plant, can destabilize a process previously stabilized, requiring, in this way, a more robust control structure. When it comes to the application of a control structure where there is an approach in the plant as a whole, a plantwide structure. Methods of determination of control structures have been presented in the literature. On the other hand, there are methodologies that have a very deep mathematical basis, such as presented by Skogestad where the user must apply a series of steps according to a certain order to select, in addition to the control structure, the "best" variables that should be controlled and manipulated in order to make the plant selfoptimized. The procedure can be applied with the aid of commercial simulators (e.g. AspenPlus ) converging or results obtained more suitable if compared with other methodologies, like applied by Lyuben. In this way, motivated by the context, this work addresses the construction and application of a tool (Applied Programming Interface) in VBA language with interface in Excel® using the methodology and mathematical concepts found in literature. The goal is to provide a means of obtaining possible facilitator variable controlled arrangements based on the procedures available in the literature. The tool communicates through Component Object Model (COM) technology from Windows" system with the stationary chemical process simulator PRO/Il". The array gain, as well as the calculation of the Hessian matrix, which are necessary for the application of the method, is obtained through the use of the Akima Spline method (linear algebra package, Alglib'*) or through the method of Richardson extrapolation. To obtain the best sets of variables, after construction of the array gain, based on local minimum singular value method proposed by Skogestad, optimization procedure is applied the branch-and-bound. The results indicate the feasibility of application of complex theoretical contents in a simple manipulation tool for the user, which will be obtained the best possible sets of variables tracked for building control plantwide structures for process plants.