(246b) Scheduling and Optimization for Industrial Cogeneration Plants | AIChE

(246b) Scheduling and Optimization for Industrial Cogeneration Plants


The scheduling application of cogeneration plants presented uses a first principle, steady-state, non-linear model with the following details

  • Model validated continually with plant data and also used for online optimization
  • Accounts for equipment that gets switched off and turned on
  • Includes Process Control strategy that may use different equipment in an hierarchal manner
  • Control curves and design performance curves for equipment
  • Scheduler results are implemented using a model predictive controller

The scheduling application also has following capabilities

  • User can input future ambient temperature
  • User can switch parts of plant on or off to account for equipment contingency

The above modeling details enable the scheduler model predictions to have a close match for fuel consumption at different operating scenarios. The error from plant data is less than two percent over the entire operating range thereby allowing for efficient participation in the power scheduling market. Simpler linear models are not sufficiently accurate because they do not allow for control strategy, equipment control and performance curves, and inherent process nonlinearities.

Process Description

Scheduling of power and real-time optimization for three industrial cogeneration plants at one of Dow's Louisiana sites is presented in this paper. The three combined heat and power (CHP) plants (cogeneration plants) are capable of meeting the steam and electrical power needs of other chemical production facilities. Surplus power is produced thereby allowing Dow to participate in the day-ahead Mid-Continent Independent System Operator (MISO) power market. Power is scheduled in the MISO day-ahead market at the estimated cost to produce and adjustments to the power offer are made in real-time market thirty minutes prior to each hour.

The industrial cogeneration process at the site consists of three power plants with seven gas turbines and five steam turbines that make surplus power and four pressure levels of steam for different consumers. A steady-state model of the process that includes component material balances, energy balances and thermodynamics is developed in Aspen Plus Optimizer, Aspentech's equation oriented environment. The process model consists of equations for gas turbines, steam turbines, heat exchangers, steam headers, fuel headers, condensate system, pressures relief valves, pumps and compressors. The process model in the equation oriented environment is also appended with additional equations for process control strategy and equipment control specifications. The overall steady-state process model consists of approximately 18,000 equations. 

Scheduler Implementation

The offer curve for the power schedule includes external exported power and incremental heat rate. Heat rate is the common measure of system performance in a cogeneration power plant. It is defined as the fuel energy input divided by the energy generated and has units of million BTU/MWhr. The incremental heat rate curve is able to match within two percent error from actual plant data for different ambient temperatures and operating scenarios thereby enabling better decision making for scheduling power. The close match of the power offer curve with plant data can be attributed to the following

  • Fundamental nonlinear model with process control strategy and equipment control curve details
  • Continuous model validation with real-time plant data

Implementation of the exported power schedule is done effectively using a model predictive controller. The power scheduling application has been in continuous use since January 2014 to enable efficient participation in day-ahead and real-time power market due to its robustness and accuracy.