(679b) Modeling and Optimization of Supercritical Pulverized Coal Power Plants Under Part Load Operation | AIChE

(679b) Modeling and Optimization of Supercritical Pulverized Coal Power Plants Under Part Load Operation


Ghouse, J. - Presenter, McMaster University
Eslick, J. C., National Energy Technology Laboratory
Burgard, A. P., National Energy Technology Laboratory
Lee, A., National Energy Technology Laboratory
Zamarripa, M. A., National Energy Technology Laboratory
Ma, J., National Energy Technology Laboratory
Eason, J. P., Carnegie Mellon University
Nicholson, B., Sandia National Laboratories
Laird, C. D., Sandia National Laboratories
Biegler, L., Carnegie Mellon University
Bhattacharyya, D., West Virginia University
Miller, D., National Energy Technology Laboratory
Despite advances in commercial simulation packages for developing process models, the ability to use such models for optimizing complex plant flowsheets remains challenging. The Institute for the Design of Advanced Energy Systems (IDAES) is developing an open-source, advanced process system engineering framework [1] built on Pyomo [2] - a Python based algebraic modeling language. The primary objective is to leverage the capabilities of a high-level programming language to not only optimize and improve the efficiencies of existing power plants but also to accelerate the development of the advanced energy systems of the future.

The increase of renewable energy sources for power generation presents significant challenges for existing coal fired power plants that were primarily designed to provide constant, baseload power. In particular, these plants much often operate at part load due to the requirement of the evolving grid. Thus, it is important to consider how to best operate these plants under part load to ensure high efficiency and environmental compliance.

In this work, an equation-oriented plant wide model of a supercritical pulverized coal (SCPC) power plant is developed. The IDAES framework and its tools are leveraged to construct, initialize, and solve the resulting large-scale optimization problem. Due to the significant mass and energy integration in SCPC plants, we show that the entire plant must be considered when evaluating the impacts of several intricately-related operational and design decisions. For example, optimization considering only the boiler can lead to suboptimal conditions from the perspective of the entire plant.

Models for the different unit operations are developed by considering the tradeoff between accuracy and robustness to make them amenable for flowsheet optimization under varying operating scenarios. For example, a hybrid model that considers radiation effects in 3D and mass change in 1D for the fire side of the boiler is more preferable to a detailed CFD model for simulation studies [3]. However, the simplified first principles model still poses significant challenges for optimization studies. On the other hand, a reduced-order model can be utilized but the resulting inaccuracy risks leading the optimizer to invalid solutions. To this end, a trust region methodology [4] is applied to develop an adaptive surrogate model whose validity is checked at every iteration in the neighborhood of values of the decision variables. This model is coupled with a first-principles model of the water side inside the boiler tubes. This hybrid approach greatly reduces the computational expense with minor loss in the accuracy.

Models of other equipment items include the steam turbine that can handle choked flow conditions and steam condensation, feed water heaters and condenser with the ability to estimate the tube wall temperatures, and a selective catalytic reduction reactor. A smooth flash formulation is used that enables reliable vapor liquid equilibrium calculations over a broad range of operation that spans both subcritical and supercritical conditions [5]. It is coupled to an external function call that implements IAPWS steam properties [6], returns exact 1st and 2nd derivatives, and is embedded directly into the optimization problems via the AMPL Solver Library. The plant-wide model is used to optimize the operating conditions under part load operation subject to critical operating and materials constraints so that the plant efficiency under these off-design conditions can be maximized without damaging the plant hardware.


[1] D. C. Miller et al., “Next generation multi-scale process systems engineering framework,” in Proceedings of the 13th International Symposium on Process Systems Engineering, San Diego, California, USA, 2018.

[2] W. E. Hart et al., Pyomo - Optimization Modeling in Python, Second. Springer, 2017.

[3] J. Ma, J. P. Eason, A. W. Dowling, L. T. Biegler, and D. C. Miller, “Development of a first-principles hybrid boiler model for oxy-combustion power generation system,” Int. J. Greenh. Gas Control, vol. 46, pp. 136–157, 2016.

[4] J. P. Eason and L. T. Biegler, “Reduced model trust region methods for embedding complex simulations in optimization problems,” Comput. Aided Chem. Eng., vol. 37, no. June, pp. 773–778, 2015.

[5] A. P. Burgard et al., “A smooth, square flash formulation for equation-oriented flowsheet optimization,” in Proceedings of the 13th International Symposium on Process Systems Engineering, San Diego, California, USA, 2018.

[6] IAPWS, “International Association for the Properties of Water and Steam, IAPWS R6-95(2016),” 2016.