Optimal condition-based cleaning of solar power collectors

Abstract

Cleaning costs of mirrors contribute a significant portion in operation and maintenance costs (O&M) of concentrating solar power (CSP) systems. The optimal cleaning policy is obtained from a tradeoff between revenue received from generating electricity and the costs of conducting cleaning operations (e.g. water, labor, etc.). However, this balance depends strongly on the local electricity market and weather conditions and currently available cleaning policies have not considered variation in electricity prices nor the potential for “natural” cleaning events (e.g. rain).In this study, a Condition-Based Cleaning (CBC) policy is developed for mirrors whose degradation is stochastic and subject to seasonal variations. The optimal policy is determined by formulating and solving a finite-horizon Markov decision process whose time-varying transition matrices describe stochastic soiling, rain events, and imperfect cleanings. The optimal cleaning policy is therefore a time-varying reflectivity threshold, below which cleaning is triggered.The methodology has been applied to a case study on a hypothetical plant in Brisbane, Australia. Using publicly available electricity price and weather data, the optimized CBC policy was found to save 5–30% of total cleaning costs compared with a fixed-time strategy. Importantly, higher CBC savings are achieved when the direct cleaning costs are high, indicating that the policy could be particularly significant for countries with high labor or resource (water, etc.) costs (e.g. Australia). Though applied to CSP in this study, the methodology is also applicable to optimal cleaning of other solar collectors (e.g. photovoltaic collectors), albeit with different efficiency models.