Bi-objective Optimization of Sectorial Cleaning Policy for the Solar Fields of Concentrating Solar Tower Plants

The reduction of energy production due to heliostat soiling causes generation losses and therefore reduces the contribution and competitiveness of CSP. The optimal cleaning plan for CSP plants must strike the proper balance between the cost for cleaning activities and revenue losses due to a reduced power production. However, uncertainty in future soiling rates, weather conditions, and electricity sale prices (e.g. due to electricity markets, uncertainty in future purchasing agreements) significantly complicate the cleaning optimization. To address these challenges, a bi-objective simulation optimization approach is developed to optimize the cleaning policy to simultaneously minimize cleaning costs and expected generation losses in the presence of stochastic weather, dust. The outcome of the proposed approach is the optimal number of cleaning trucks and periodic cleaning schedule for non-dominating solutions (Pareto front), which may be used to study the sensitivity of the cleaning schedule to variations in electricity prices. The approach is applied to a case study for a CSP plant located in Woomera, South Australia. The results show that for electricity sale prices between 40-100 AUD/MWh, the optimal number of trucks is 3 and the total cleaning costs are sensitive to the number of trucks purchased, hut not to the cleaning frequency. However, as electricity prices increase above 300 AUD/MWh, the total costs become quite sensitive to the cleaning frequency. Finally, a comparison of the simulation-optimization approach and a (simpler) deterministic optimization indicates that the latter approach is suitable for most practical electricity prices.