Projeto auxiliado por computador de processos industriais: análise de flexibilidade

Abstract

This work investigates the feasibility of using a virtual plant, simulated with a process simulator, in order to obtain simpler models of the real plant (meta-models), as response surfaces, and using those meta-models to investigate the influence of input variables, controllable or not, on the performance of the plant and on the optimum conditions (flexibility analysis). As a case study, it was considered a plant of air compression of Vale Fertilizantes in operation in the city of Uberaba, Minas Gerais. We used the process simulator COCO, to simulate the virtual plant, and the \"software\" Scilab® to add a unit operation, namely an adsorption column, to the simulator, using the CAPE-OPEN protocol. The simulation results were validated using plant design data as well as operational data. For the construction of response surfaces STATISTICA® software was used and the optimization of meta-models was performed using Microsoft Excel® software. Response surfaces were constructed analyzing the behavior of each air compressors relative air humidity, temperature and the behavior of the adsorption column in relation to temperature and moisture input. The optimal values, resulting in less power for the compressors, were: temperature of 15.76 ° C, mass flow of saturated steam in air of 50.81 kg / h, for the first compressor; 13.74 kg / h of saturated steam in air and 1434.74 kg / h of cooling water, for the second compressor. For the adsorption column the best inlet temperature should be 35 ° C for any amount of steam at the entrance of the column, so that outlet steam in the column is the lowest possible. The procedure used in this thesis allowed to satisfactorily reproduce the plant performance, to investigate the process flexibility with respect to changes in the inlet conditions and to find optimal operating values.