| A supervisory process control system for the Biomass Indirect Liquefaction Pilot Plant has been developed. A microcomputer, the IBM 9000 system, was utilized for data acquisition, reduction, and adjustment, process modeling, simulation, and off/on-line optimization. The IBM 9000 is interfaced with the process through a front end data acquisition system.; Several supervisory control programs were set up to run concurrently on the IBM 9000. A user interface task controls the execution and resource sharing activities of the multitasking system. A data fitting package was programmed to handle instrument calibration and/or model development. Process flow and composition data are reduced and tabulated by the mass balance program. Mass balance closures are calculated for different loops. Two process coaptation approaches, sequential and non-sequential, for data adjustment and estimation were investigated.; The steady state Kalman filter, a sequential estimator, was found to be suitable for steady state process flow estimation on the microcomputer. A reduced balance approach, using optimization routines, was utilized for the non-sequential data coaptation problem. Reliable liquefaction process model parameters, subject to balance constraints, are also calculated by the off-line data adjustment routines. Gross measurement error detection tests suitable for the two data adjustment procedures were identified.; Mathematical models of the pyrolysis and Fischer-Tropsch reactors were developed using linear regression. Theoretical equilibrium studies support the experimental evidence of the dependence of the pyrolysis product gas composition on the feedstock composition, and especially on the hydrogen/oxygen ratio. Computer aided design packages, FLOWTRAN on the mainframe and CHESS on the microcomputer, are used to perform steady state simulations. A search routine, the Complex method of Box, was utilized to find the optimum process conditions for maximum yield. A simplified process plant model and a nonlinear optimization routine were set up for future on-line supervisory optimizing control applications. |
