Effect of the quality of streamflow forecasts on the operation of cascade hydropower stations using stochastic optimization models. (15th June 2023)

• Record Type: Journal Article
• Authors: Liu, Yuan; Ji, Changming; Wang, Yi; Zhang, Yanke; Jiang, Zhiqiang; Ma, Qiumei; Hou, Xiaoning
• Is Part Of: Energy. Volume 273(2023)
• Publication Date: 2023-06-15
• Subjects: Cascade reservoirs -- Optimal hydropower generation -- Streamflow forecast quality -- Generalized maintenance of variance extension -- Stochastic optimization; SDP Stochastic dynamic programming -- BSDP Bayesian stochastic dynamic programming -- GMOVE Generalized maintenance of variance extension -- CAOS Cumulative annual output shortage -- CAPR Cmulative annual power revenue -- SR System reliability; Power resources -- Periodicals; Power (Mechanics) -- Periodicals; Energy consumption -- Periodicals; 333.7905
• DOI: 10.1016/j.energy.2023.127298
• Languages: English
• ISSNs: 0360-5442

Abstract

Abstract: Determining the economic value of streamflow forecasts is essential to judging the operation of cascade hydropower systems and investing in improved forecasting systems. Previous analyses of the streamflow forecast value are mainly based on deterministic optimization strategies. This paper investigates the impact of long-term (10-day-ahead) streamflow forecasts on the operation of a cascade hydropower system using stochastic dynamic programming (SDP) and Bayesian stochastic dynamic programming (BSDP). Synthetic streamflow forecasts with different bias, variance, and precision are generated by the generalized maintenance of variance extension approach. A case study is performed to evaluate the performance of these strategies in terms of cumulative annual power revenue (CAPR) and system reliability (SR). The results show that, even when using the forecast with the largest uncertainty and bias, the stochastic optimization strategies increase at least 6.63 × 108 CNY in CAPR and 33.89% in SR compared with a reference strategy that uses no forecast information. The SDP performs best with forecast systems that have a negative bias and high accuracy. Compared with the SDP, BSDP increases at least 1.80 CNY in CAPR and 0.28% in SR and is better able to handle forecast uncertainty, and is insensitive to forecast bias. Highlights: Extended the GMOVE approach by adding the explanatory variable. Constructed a new framework to quantify the effect of the streamflow forecast. Analyzed how the stochastic optimization model handles the streamflow forecast. The advantage of BSDP in dealing with forecast uncertainty and bias was disclosed.