Decision Support System for Coal Mill Fault Diagnosis in Coal-Fired Steam Power Plant

Abstract

The coal pulverizer mill at PLTU (coal-fired steam power plant) Rembang is essential boiler equipment that processes coal raw materials into fine coal powder to perfect combustion in the furnace. In the operation of 4 coal mill units, delay and self-combustion often occurred due to using Low-Rank Call (LRC) coal with high moisture and volatile matter content. Several operation patterns are carried out to avoid delay and self-combustion, such as adjusting the primary air supply with a sufficiently high temperature and the suitable coal supply for the drying process in the coal pulverizer mill system.
Unfortunately, the adjustment frequently causes the operating parameters to fall outside the standard limits. A dynamic simulation of the plant model that represents mass flow rate, heat transfer, and energy balance is performed to understand this problem. Simulation results can accurately reveal the actual coal mill operating conditions and monitor the effect of changes in operation patterns. For the failure diagnostic, the ANFIS method is utilized to describe three coal plant failures: excess coal, coal shortage, and coal explosion. Fuzzy logic is used to determine the type and magnitude of the error, while the Bayesian network is used to solve the root cause problem. The proposed technique is validated using historical data from the PLTU Rembang. It can be observed in the simulation results that the relative error of primary air flow, inlet temperature, mill motor current, and Air-Fuel Ratio are less than 4%.