(140g) Real Time Optimization of Industrial Gas Plants and Networks

Industrial gases are widely used in many areas including the chemical, refining, metallurgy, environmental, and semiconductor industries. The industrial gases used in large quantities typically include oxygen, nitrogen, argon, hydrogen, and carbon monoxide (CO).

The air gases – oxygen, nitrogen and argon – are usually produced through cryogenic distillation, when a large amount of them are needed in high purity. In a cryogenic air separation unit (ASU), air is first liquefied at a very low temperature and then separated into different components through distillation. This is a very energy intensive process. For example, the Air Liquide Group runs about 400 ASUs worldwide, and its electricity consumption in 2010 was more than one thousandth of the total world electricity consumption. Steam methane reforming (SMR) is one important process to produce hydrogen and CO. In this process, natural gas is reacted with steam in a heated furnace at the presence of catalysts to produce syngas, which is a mixture of hydrogen, CO, methane, and other gases. This mixture is then further processed to produce pure hydrogen, CO, or certain mixtures. These products are often transported to customers through a pipeline, which connects multiple SMR plants and customers. SMR processes also consume a lot of energy, mainly in term of natural gas. Around half of Air Liquide’s annual energy bill is for the purchase of natural gas.

It can be seen that improving the efficiency of these industrial gas plants and pipelines is important in improving an industrial gas company’s profit and competitiveness while reducing its energy consumption and carbon footprint. Real time optimization (RTO) is an important technique to achieve such a goal. A real time optimizer takes market dynamics such as changes in energy price and client demands into account in real time, and calculates the most profitable mode of operating the plants by using the plant models with energy consumption equations and process constraint equations. Then the calculated optimal set points are downloaded to the plant control system to ramp the plant or sent to operators as references.

In this work, two RTO applications in industrial gas production are studied: one ASU plant and one SMR pipeline. The important aspects of the applications are addressed including the general IT structure and functions of its different software components, important steps in completing such projects, and major challenges and corresponding solutions in mathematical optimization.