Research On Multi-Product Air Separation Gas Supply Network Production Scheduling

The air separation gas supply network is an important part of the steel industry,providing high-purity oxygen,nitrogen and argon for the steelmaking and ironmaking process.However,in the actual gas supply process,the phenomenon of insufficient gas supply or gas emission frequently occurs due to the frequent fluctuation of downstream gas consumption,resulting in waste of resources and economic loss.To reduce costs and increase efficiency,enterprises urgently require to arrange the operation of equipment appropriately through scheduling,hence to respond to changes in the demand side timely and improve economic benefits.Begin with the process of enterprise production scheduling,this thesis investigates the problem of monthly scheduling formulation,scheduling in uncertain scenarios,and closed-loop scheduling.The main research contents are as follows.1.A static scheduling scheme for the air separation gas supply network was established.To accurately depict the characteristics of the units,first a surrogate model of air,gaseous and liquid products of oxygen,nitrogen and argon is built;then the nonlinear relationship between air compressor load and energy consumption is analyzed,where a strategy to separate modes in the raw material space is proposed to avoid the computational complexity caused by the growth of scheduling decision variables;further,the scheduling models of PSA,liquefier,vaporizer,oxygen compressor,and nitrogen compressor are established.Combining the above models and the mass balance of oxygen,nitrogen,and argon pipe network,a mixed integer programming of multi-product integrated scheduling is constructed to optimize the total profit.The scheduling model is instantiated with the actual industrial air station unit as the object,and the actual case analysis is carried out,and the results of manual scheduling are compared with the solution results of the optimization proposition,which verifies the feasibility of decision-making and the economy compared with manual scheduling.2.To handle the uncertainty of startup time and demand in scheduling decision-making,a discrete two-stage stochastic optimization method is adopted,and a scheduling optimization model considering duplex uncertainty is established.The first stage decides the startstop of the air separation plant,and the second stage decides the load of the air separation plant and the start-stop of the liquefier,gasifier and other equipment.For different uncertain parameters,several probability distributions are used to describe the uncertainty,the Monte Carlo sampling method is used to generate the second stage scenarios,and the backward reduction method is used to reduce the number of scenarios and complexity of the problem.Compared with the deterministic scheduling results,the uncertain results can better adapt to the occurrence of different scenarios.In order to avoid the negative effects of small probability events,the conditional value at risk is introduced as an indicator,the results of stochastic optimization scheduling and risk scheduling are compared,and the impact on profits under different risk factors is discussed.3.As a hub connecting upstream and downstream,the stability of the pipeline network is crucial for both.This thesis proposes a closed-loop scheduling framework based on virtual control.Based on the model predictive control method,a virtual pressure control scheme for low-pressure oxygen and medium-pressure oxygen pipeline networks is established,and the controller output corrects the demand parameters of the short-term scheduling proposition;on the one hand,the short-term scheduling inherits the start-stop and nominal load from the monthly scheduling.The estimated gas demand and de-mand correction amount solve the short-term scheduling optimization proposition,give the final decision result,and finally form a closed-loop scheduling scheme in the rolling time domain,eliminating the negative effects of forecast errors and other unmeasurable disturbances.