Multi-product Optimal Scheduling Research For Gas Network With Air Separation Units

A Cryogenic air separation unit produces high-purity oxygen,nitrogen and argon by separating the components of air.These gaseous products are sent directly to the steel and iron plants through pipelines.As the demand of gaseous products fluctuates frequently in the steel and iron plants,the imbalance between supply and demand may lead to the occurrence of undersupply or release,thus resulting in massive energy wastage and economic losses.Scheduling is an effective method to reduce energy consumption and improve economic benefits,which responds to changes in the demand side by rationally arranging the operating mode and production level of the device.The scheduling problem is divided into long-term scheduling and short-term scheduling according to the time scale.The long-term scheduling time scale is much larger than the time scale of dynamic process.Generally,a static model is employed and the dynamic process is ignored.The short-term scheduling time scale is comparable to the time scale of dynamic process and the dynamic characteristics of the process is considered.The scheduling problems under uncertainty from the perspective of long-term and short-term are studied.The main research contents of this thesis are as follows1.For long-term scheduling problem,a static scheduling model of the air separation network is built to optimize the total profit,which based on multi-product air separation units.The production model for each unit is established according to the mass balance.Especially for air separation units,which are represented with a set of operating modes,and each mode is described with a piecewise linear surrogate model.The process balance of the oxygen,nitrogen and argon pipelines are established.Optimal scheduling is modeled as a mixed integer linear programming(MILP).2.A two-stage stochastic programming is used to deal with the uncertain demand of gaseous products in long-term scheduling.Modes of operation and nominal production levels are determined in the first stage.Incremental production levels and start-up of liquefiers and vaporizers are determined in the second stage.Two-stage stochastic programming can give a more reasonable scheduling strategy compared to deterministic model.Conditional value at risk(CVaR)is introduced to measure the risk.The expected profits at different levels of risk are discussed.3.For short-term scheduling problem,a dynamic scheduling model of the air separation network is built based on the dynamic characteristics of load changes A scheduling method based on two specific time scales is proposed,a low time scale over which the set point of production level is determined,and a faster time scale over which the process dynamics evolve.Other part of the dynamic scheduling model is the same with the static scheduling model.Case study shows that dynamic scheduling model is more reasonable in short-term scheduling compared to static scheduling model.4.A periodic rescheduling method based on the feedback of the pipeline pressure is proposed to deal with the uncertain demand of gaseous product in short-term scheduling.Uncertain demand causes a deviation between the estimated pressure and the actual pressure of the pipeline.A pseudo-linear regjression algorithm is used to identify the disturbance model and multi-step demand disturbance is predicted.The predictions are embedding in the dynamic scheduling model as feedback for rescheduling,thus resulting in a closed-loop.The dynamic scheduling calculations are carried out periodically over a receding horizon as new predictions are available Case study shows that periodic rescheduling can eliminate the adverse effect of unmeasured disturbance in process while meeting the gaseous demand.