(132c) Real-Time Water Management in Power Plants and Implications in Electricity Markets | AIChE

(132c) Real-Time Water Management in Power Plants and Implications in Electricity Markets

Authors 

Salazar, J. M. - Presenter, Vishwamitra Research Institute, Center for Uncertain Systems: Tools for Optimization and Management
Diwekar, U. - Presenter, Vishwamitra Research Institute, Center for Uncertain Systems: Tools for Optimization and Management
Constantinescu, E. - Presenter, Argonne National Laboratory
Zavala, V. M. - Presenter, Argonne National Laboratory


We present a real-time water management framework for power plants. Power plants are the second largest consumers of fresh water in the world. In these plants, huge amounts of water are lost to the environment in the cooling towers as a result of evaporation. For example, a baseline plant (550 MW) may lose up to 220,000 gallons per hour by evaporation during the summer. Cooling capacity is constrained by high ambient temperature and humidity, which force power plants to drop their power outputs, thereby limiting their participation in day-ahead and spot electricity markets. Consequently, the ambient conditions can have a critical effect on electricity prices if the power plant provides a significant portion of the base load or if the plant is located in an area with significant transmission congestion. It is thus critical to quantify the uncertainty of the day-ahead ambient conditions and to optimize the operating conditions in economic dispatch operations.

We analyze the effects of cooling capacity constraints and forecasts of the ambient conditions on the market participation of power plants. Specifically, we construct bidding curves (hourly power output vs. price) using a rigorous model of the power plant and of the cooling towers. The uncertainty of the ambient conditions is quantified in the form of ensembles using the state-of-the-art numerical weather prediction model WRF running at Argonne National Laboratory. The ensembles are used for stochastic optimization employing the BONUS (better optimization of nonlinear uncertain systems) algorithm, coupled to a steady-state power plant model implemented as a CAPE-OPEN?compliant capability. BONUS uses a reweighting scheme to update the objective function's probability of occurrence when the probability distributions of the decision and uncertain variables change. This scheme avoids unnecessary rigorous simulations. We use a pulverized coal power plant case study to demonstrate the developments.

Topics