Qualitative Fault Detection and Diagnosis in Chemical and Plant-Wide Processes

Abstract

With increasing automation in chemical processes, occurrences of faults are very common now a days. Some time we understand the effect of faults on different process variables but we cannot control the system instantaneously. An undetected fault can often get out of proportion and cause massive losses to an industry in terms of infrastructure and personnel. Hence, detection and diagnosis of fault is necessary to overcome such types of problems. When it is observed that a process variable affects in unwanted manner, the next step is to identify the root of fault, determination of the variables affected by that particular fault by using suitable technique to diagnose them. A qualitative technique for fault detection and diagnosis (FDD) is performed by using signed diagraph (SDG) method. The signed directed graph (SDG) technique has been a widely used to diagnose the fault mainly because of its resemblance to a human way of reasoning, its in-depth analysis of abnormal situation, and its ability to present a complete description of all process variable effected by a fault. SDG represents the causality between the processes variables involved in system. Case studies including double effect evaporator, multi stream controlled CSTR and a plant wide flow sheet involving reaction and separation process have been considered. In the present work, each system is described in the form of differential equation (DE), Algebraic equation (AE) and differential algebraic equations (DAE). To construct the SDG of a system either arc value calculation or perfect matching (steady state SDG) is performed. The propagation table determines the shortest path between exogenous (root node) variable and the measured variable. Initial response is necessary to diagnose the faults. In this work SDG is constructed for systems with and without control loop. To diagnose the fault, backward reasoning approach and incipient fault diagnosis (IFD) are used. A quantitative simulation is also performed to validate the proposed SDG technique in identification of fault. The identified faults are considered as candidate faults.