Desenvolvimento de um sistema para validação ampla de uma caldeira com múltiplos pontos de operação

Abstract

The development of a wide validation system for a boiler, unit of a thermoelectric power plant, that operates on multiple setpoints is presented. The role of such systems is to certify that a process operates within its declared functional specifications, mainly in terms of performance and quality. To compose the system, multivariate statistical process control tools (Hotellings 2 chart, SPE, combined index ) and other statistical methods (PCA, contribution plots) are used, as well as the boilers efficiency calculation and its control loop performance indexes. This boiler, employed as complimentary power source, burns different fuels, such as blast furnace gas and natural gas, operates with constant changes in the mixture of fuels and in the operating setpoints, and also presents residual fluctuations in the processes means. Therefore, adaptations in the usage of these tools are proposed to allow their application over non-stationarity conditions and multiplicity of setpoints. Such adaptations include the use of moving average on the design step of the statistical control charts and a scheme of adaptive means during the validation procedure. All the developing stages of the validation system are programmed in Matlab over historical process data. Recursive versions of the performance indexes calculation are proposed to simulate the posteriori just-in-time execution of the system in the PI System, the PIMS installed and available in the power plant studied. Finally, the results presented show the capacity of the system to provide evidences that indicate the performance of the boiler, to point out changes in the process and to provide support for fault identification.The implementation of the validation system in the PI System platform is also presented, as well as its test results, compared with the results obtained in Matlab.