Translated Abstract
Petrochemical industries often involve a wide range of flammable, explosive and toxic materials. Accidents may lead to serious consequences such as personal death and injury, environmental pollution and great economic loss. Especially, the development of large scale petrochemical plants brings about much more potential risk factors. Thus, a high level of reliability and safety is a critical issue for the success of petrochemical plants. Research on reliability and long period safe operation is of great value in theory and engineering.In this dissertation, reliability analysis methods, reliability optimization theory and key technologies of safe operation are emphatically studied considering the complexity features of petrochemical plants. Taking an industrial continuous catalytic reforming (CCR) plant as a case, the main innovative achievements in this dissertation are as follows.1) Complexity features were viewed as manifesting in the following five aspects: multi-scale, complicated process, complicated control system, several failure modes, and random human error. When analyzing reliability of petrochemical plants, the researchers must take account of these complexity features for developing proper reliability analysis methods.2) Petrochemical plants’ reliability was analyzed using fault tree analysis (FTA) method. Failure rates of basic events could be obtained by the following two ways: one is to revise the known generic failure frequency according to API 581 and the other one is to estimate the failure rate using fuzzy logic based on experts’ judgments. For the CCR plant, the prime reasons causing low reliability level of critical facilities are the abnormal states resulting from unsatisfactory control and lack of or incorrect online monitoring of process parameters that is in consistence with the fact. Because of the complexity features, a new concept “function reliability” was introduced to assess reliability level of the whole petrochemical plants. The assessment results of the CCR plant in 2011 show that function reliability can be a proper index to describe the production status of petrochemical plants.3) Keeping high safety and reliability levels must consume a large amount of manpower and material resources. With regard to how to use available resources in the most effective way, a theory of reliability optimization under cost constraint was presented. Reliability cost was firstly defined as the sum of maintenance cost and management cost. A sound mathematical function between cost and reliability was established assuming that cost would increase exponentially with the feasibility and growth of reliability. Then a nonlinear programming equation was formulated taking total cost as objective function and reliability requirements as constraint conditions it could be solved by a genetic algorithm (GA). The reliability optimization results of the CCR plant demonstrate that the research can provide an optimum cost allocation plan for reliability optimization of petrochemical plants.4) Process monitoring and soft sensors are key technologies to ensure safe operation. Considering strong self- cross-relation of process variables, a statistical process monitoring model based on canonical variate analysis (CVA) was proposed to monitor operating conditions of the reforming heat exchangers (E201A, E201B) and combined furnace (H201ACD) the results show that the proposed model can detect abnormal state timely and accurately. Aiming at online detection of carbon content of spent catalyst and octane number of reformate, soft sensors using least squares support vector machines (LSSVM) and GA were proposed. The applications show LSSVM-GA soft sensors can meet online estimation requirements with good learning performance and generalization ability. 5) According to the theory of dissipative structure, it was verified for the first time that the petrochemical system was a dissipative structure failure process was explained based on description of system status using entropy and sell-organization evolution mechanism was established to confirm the feasibility of running in a high reliability level of petrochemical plants. Self-recovery engineering was conceived as a new promising development direction in petrochemical industry and should be of great significance for improving reliability level, reducing failure probability, and raising utilization rate of facilities, energy and materials.
Translated Keyword
[Petrochemical PlantsReliabilityFault Tree AnalysisStatistical Process MonitoringSoft Sensor]
Corresponding authors email
