Sie befinden Sich nicht im Netzwerk der Universität Paderborn. Der Zugriff auf elektronische Ressourcen ist gegebenenfalls nur via VPN oder Shibboleth (DFN-AAI) möglich. mehr Informationen...
With the sale of large-scale chemical plants, the plant constructor undertakes to fulfil certain guaranteed values. For a nitric acid plant one of the essential guarantee figures is the acid yield. Consequently, a check of this value is a central item during the plant acceptance test procedure (test and guarantee runs).
Until now stationary material balances have been used to determine the yield. However, the result may be falsified for two major reasons. First the measurements entering the calculation may exhibit systematic errors where the investigation is difficult. Another source of error is the time lag in the measurements caused by the plant dynamics.
In the present report the material balances were determined with the plat dynamics taken into account, in addition the measurements made during the process were monitored. This was done on the one hand by utilizing the redundancy of the available measurements by so-called voters (which form weighted means) and on the other hand by the use of mathematical models (input-variable, output-variable and parameter observers). In order to design the observers mathematical models of the various sources of error (error hypotheses) were introduced and the size of the postulated error was estimated. In order to establish the dynamic model of the overall plant the process was shared into subsystems. Through correlation of the individual subsystems and utilization of hardware redundancy it was possible to recognize faulty measurements and obtain an improved material balance.