Risk-Averse Two-Stage Stochastic Daily Optimal Chiller Loading with Chiller Commitment in the Presence of Photovoltaic Systems and Environmental Uncertainties
1
Department of Electrical Engineering, University of Bonab, Bonab, Iran
2
Division of Mechatronics Engineering, Faculty of Electrical and Computer Engineering, University of Tabriz, Tabriz,Iran
10.61882/jgeri.2026.2097327.1130
Abstract
Most daily optimal chiller loading (DOCL) studies keep every chiller in service and optimize the part load ratios (PLRs) separately for each realization of the uncertain inputs, which leaves the operator without a schedule that can be fixed in advance. This paper formulates the problem as a risk-averse two-stage stochastic program for a grid-connected multi-chiller plant equipped with a photovoltaic (PV) array. In the first stage, decided one day ahead, the on/off commitment of the chillers (with minimum up/down times and start-up costs) and the energy bought in the day-ahead market are fixed; in the second stage, the PLRs, the real-time exchanges with the grid and any unserved cooling are adapted to each scenario of cooling demand, solar irradiance and ambient temperature. Ten scenarios obtained from 1,000 Monte Carlo samples by backward scenario reduction are used, and risk is controlled with the conditional value-at-risk (CVaR). Because the chiller power curves are non-convex, the resulting mixed-integer nonlinear program is solved through an exact commitment-state reformulation in GAMS that guarantees the global optimum. Compared with the conventional DOCL in which all chillers stay on, chiller commitment reduces the expected chiller energy by 4.29% and the expected daily cost by 1.49%. A moderately risk-averse schedule (β = 0.5) lowers the CVaR of the daily cost by 1.56% for a 0.29% increase in the expected cost. The expected value of perfect information and the value of the stochastic solution are quantified, sensitivity analyses on PV size and load level are reported, and on the deterministic benchmark of the literature the exact solution improves on the best results of three particle swarm optimization variants by 1.59–5.20%.
Ahmadian, A., Khoudeh, H. & Akrami, S. M. R. (2026). Risk-Averse Two-Stage Stochastic Daily Optimal Chiller Loading with Chiller Commitment in the Presence of Photovoltaic Systems and Environmental Uncertainties. (e746630). Journal of Green Energy Research and Innovation, (), e746630 https://doi.org/10.61882/jgeri.2026.2097327.1130
MLA
Ahmadian, A., Khoudeh, H., & Akrami, S. M. R. "Risk-Averse Two-Stage Stochastic Daily Optimal Chiller Loading with Chiller Commitment in the Presence of Photovoltaic Systems and Environmental Uncertainties" .e746630 , Journal of Green Energy Research and Innovation, , 2026, e746630. doi: 10.61882/jgeri.2026.2097327.1130
HARVARD
Ahmadian, A., Khoudeh, H., & Akrami, S. M. R. (2026). 'Risk-Averse Two-Stage Stochastic Daily Optimal Chiller Loading with Chiller Commitment in the Presence of Photovoltaic Systems and Environmental Uncertainties', Journal of Green Energy Research and Innovation, (), e746630. doi: 10.61882/jgeri.2026.2097327.1130
CHICAGO
Ahmadian, A., Khoudeh, H. & Akrami, S. M. R., "Risk-Averse Two-Stage Stochastic Daily Optimal Chiller Loading with Chiller Commitment in the Presence of Photovoltaic Systems and Environmental Uncertainties." Journal of Green Energy Research and Innovation, (2026): e746630, doi: 10.61882/jgeri.2026.2097327.1130
VANCOUVER
Ahmadian A., Khoudeh H., Akrami S. M. R. Risk-Averse Two-Stage Stochastic Daily Optimal Chiller Loading with Chiller Commitment in the Presence of Photovoltaic Systems and Environmental Uncertainties. JGERI. 2026;():e746630. doi: 10.61882/jgeri.2026.2097327.1130