Managing energy in a large petrochemical plant is an extremely complicated undertaking. It involves managing highly interdependent systems throughout the facility in real time. The plant requires electricity to run compressors and pumps. Steam is needed to drive reactions that must be supplied at almost any cost to avoid a shutdown. Energy costs are a huge expense that must be managed effectively.

Dow Chemical's Seadrift (TX) petrochemical plant has a number of gas turbines that generate power. The plant's turbines also generate waste heat that is passed through a boiler to produce steam. Dow uses this power internally to run the plant and then sells the excess into the grid. When the energy necessary to run the plant is not available internally, Dow purchases power from the grid. The price at which electricity can be purchased from or supplied to the power grid changes by the minute. Prices are generally much higher during the day when power demands are greater than at night. The management of these operations can have a important impact on the profitability of the plant.

Prior to implementing a rules-based solution, Dow engineers manually controlled the various power-generating functions. This rarely allowed time to trade off the cost and value of power when making decisions, such as at what level to most efficiently run a gas turbine, because they were so busy with other tasks. Therefore, plant energy generation was not optimized effectively.

In addition to the actual operation, other complexities complicate the modeling of energy system operation. Sensor readings may be inconsistent or noisy. Equipment behavior will vary with time. Upset conditions may exist that must be resolved before actual optimization can occur.

[click to enlarge] Rule-driven optimization achieves constant online use.

Dow engineers implemented a number of “optimizing” applications that did not yield positive results, learning that a simple optimizer model could not directly manage the complexities of their system. Dow engineers turned to Gensym's G2 because an intelligent, rules-driven system can manage these complexities, account for system inconsistencies and work with the optimization model to determine the best operational plan, all in real time. The rule-driven approach greatly simplified the incorporation of functionality that is based on a plant engineer's hands-on expertise. The application also includes tools that let users run different cases offline and measure the results. Dow engineers frequently use this tool to evaluate potential changes to the optimizer and to assist in operator training.

Dow Plans to Transfer Savings to Other Plants
Last year the Seadrift plant ran in closed loop mode 98 percent of the time, saving Dow $1.25 million dollars in energy costs. The savings that have been achieved are just the beginning. Dow believes it can more than double them by utilizing the optimization for more decisions and then multiplying those savings across other plants. Additionally, Dow operators now view the optimizer as being an important part of their success because it provides a better view of overall plant operations.