A method for fault diagnosis of the chemical plants utilizing the Signed Directed Graph (SDG) has been developed in order to assist operators' judgment and decision-making in the system failure of the chemical plant. SDG is useful for representing the cause-effect relationship in a chemical plant. Each node in the SDG represents the state variable in the plant. Each branch in the SDG represents the direct influence between the state variables. Positive and negative influences, respectively, are distinguished by the signs "+" and "–" that are assigned to the branches. It is not so difficult to make SDG of a target plant. In addition, we can propose a kind of algorithms using fault propagation speed or fault revealing time and so on as quantitative information.

To ensure the safety of a chemical plant, counter measures against abnormal situations are examined by using a safety assessment technique such as HAZOP, FTA and so on when the plant is designed or reviewed. Then, there are such problems that these techniques require amount of man-hours and the results of assessments depend on the expertise of analysts. To overcome such problems, we have proposed the method generating a route of fault propagation using the concept of SDG. As for the operation management, we have developed a nonlinear modeling technique, which is called DB (database) modeling, using historical data during normal operations. In these studies, the boiler plant simulator is used, because we want to examine the feasibility of proposed methods to apply practical plants.

In the failure modes of a flow meter, gross errors caused by a biased indication can not be detected by the indicated value of a single flow meter. Moreover, considerable labor and cost is involved, as flow meters must be detached from a plant and be calibrated to correct such biases. In addition, it is almost impossible to calibrate all flow meters during a periodic plant maintenance period. Therefore, failure diagnosis of flow meters based on control limits has been developed as a method to identify faulty flow meters, which should be calibrated during a maintenance period, during plant operation.
Furthermore, we have developed the method estimating errors contained in the measurements of flow meters during plant operations.

In the chemical industries, it has been required to improve energy conservation for reducing greenhouse gases such as CO₂. Furthermore, energy-saving chemical processes provide reduction of the operational cost, however, they have possibilities of the increase in the equipment expenses. Therefore, it is necessary to make decision of a trade-off problem between the energy cost and the equipment cost. And the optimization of a single process is not necessarily corresponding to the optimal solution of the whole plant or the site. Then, we introduce a multiagent framework into the process design support system, in order to resolve the trade-off problems using the cooperative and/or conflicting effects between the agents.