Design and Optimization of Multiple Interconnected Utility Systems in An Integrated Refining and Petrochemical Complex

In an integrated refining and petrochemical complex,a centralized utility system(CUS)is introduced to integrate the steam demands of production plants.Besides,two sub-utility systems(SUSs)located inside the alkene and refinery plants,respectively,can satisfy the shaft demands.It is difficult to determine the steam production of the CUS because the steam demands of the alkene and refinery plants also depend on the design and operation of the SUSs.To explore the complicated interaction between the CUS and SUSs,we proposed a mixed-integer nonlinear programming(MINLP)model for the design and optimization of multiple interconnected utility systems to minimize the total annualized cost(TAC).An extended superstructure was suggested to contain multiple inter-plant connected steam pipe alternatives between the CUS and SUSs.A more accurate model of the complex steam turbine was proposed.Then the proposed MINLP framework is applied to a new integrated refining and petrochemical complex.Two scenarios are investigated in the case study to explore the effect of steam main temperatures on system configurations and operating parameters.By optimizing the main temperatures,a TAC of$2.7 million can be saved.Judging from the results of the two scenarios,the feasibility and effectiveness of the proposed framework for the design and optimization of multiple interconnected utility systems have been demonstrated.

A novel time-span input neural network for accurate municipal solid waste incineration boiler steam temperature prediction

A novel time-span input neural network was developed to accurately predict the trend of the main steam temperature of a 750-t/d waste incineration boiler.Its historical operating data were used to retrieve sensitive parameters for the boiler output steam temperature by correlation analysis.Then,the 15 most sensitive parameters with specified time spans were selected as neural network inputs.An external testing set was introduced to objectively evaluate the neural network prediction capability.The results show that,compared with the traditional prediction method,the time-span input framework model can achieve better prediction performance and has a greater capability for generalization.The maximum average prediction error can be controlled below 0.2°C and 1.5°C in the next 60 s and 5 min,respectively.In addition,setting a reasonable terminal training threshold can effectively avoid overfitting.An importance analysis of the parameters indicates that the main steam temperature and the average temperature around the high-temperature superheater are the two most important variables of the input parameters;the former affects the overall prediction and the latter affects the long-term prediction performance.