Optimal Operation Of Inte-Rbasin Water Transfer Projects By Considering Hydrological Forecast Uncertainty

The highly nonlinear hydrological systems and the complex mechanisms of runoff formation make it difficult to accurately forecast natural runoff into reservoirs.Using runoff information with inaccurate forecasts to guide water project scheduling decisions will inevitably create operational risks.The initial investment in inter-basin water transfer(IBWT)projects is huge.In the actual operation and management process,optimal scheduling of IBWT projects considering forecast uncertainty can further tap into the operation of the project and improve its operation benefit.How to achieve effective integration of forecasting and operating is a key challenge that needs to be addressed.This paper takes mid-long-term optimal operation of the Han to Wei IBWT project was selected for a case study,analyses the impact of forecast uncertainty on scheduling performance indicators,and presents an effective method for deriving multi-objective operating rules by incorporating ensemble forecasting.The main work and conclusions of this paper are as follows:(1)Based on historical runoff data,deterministic optimal operating rules for IBWT projects were developed.Here,multi-objective optimization was applied to develop effective operating rules for IBWT with the obj ectives of minimizing the water shortage index,maximizing the energy production and minimizing the energy consumption.A metaheuristic algorithm named cuckoo search was used to optimize the decision variables in the framework of parameterization-simulation-optimization.The results indicate that there are mutual restriction between the above objectives;analytic hierarchical process was used to optimize the operating rules with the multi-year average water transfer of 1.0 billion m~3 and 1.5 billion m~3,the corresponding water rule curves change smoothly and the optimal operating results are reasonable.(2)Based on the above deterministic optimal operating rules,the impact of runoff forecast uncertainty on operating performance indicators is analyses.Using four methods to select the forecast factors,constructing three different models for forecasting incoming runoff(BP neural network model,extreme learning machine model,and two-parameter monthly water balance model),runoff forecasts are fed into the above deterministic optimal operating rules to quantify the impact of forecast uncertainty on water transfer index,energy production and energy consumption.The results indicate that when the annual water diversion target is 1 billion m~3,the rates of change in water shortage index are+435.3%,+133.8%and+573.5%,the rates of change in energy production are-3.3%,-1.8%and-9.8%,and the rates of change in energy consumption are+20.1%,+3.4%and+5.7%,respectively;when the annual water diversion target is 1.5 billion m~3,the rates of change in water shortage index are+111.4%,+65.1%,+180.1%,the rates of change in energy production are-1.1%,-0.6%and-1.4%,and the rates of change in energy consumption are +6.2%,+3.3%and +9.3%,respectively.When there is forecast uncertainty in the model inputs,the operating performance of IBWT is significantly reduced.(3)Based on the ensemble forecast information,deriving optimal operating rules by coupling ensemble forecasting information for IBWT projects.The stochastic combination of multiple models is used to generate the ensemble forecast information,and the ensemble forecasting samples are reduced to several typical scenarios and their corresponding occurrence probabilities by using a simultaneous backward reduction method,based on the typical scenarios,a multi-objective operation optimization model is constructed to optimize the expected values of reservoir performance indexes.The results indicate that:forecast intervals based on the stochastic combination of multiple models are smaller at low flows and larger at high flows,capable of reflecting forecast uncertainty with a high degree of accuracy.When the scenarios were reduced from 1000 to 20 using the simultaneous backward reduction method,the rate of change of the statistical parameters of the forecast scenario was within 1%and the computational time was reduced from 49.17 h to 1.05 h.Compared to the deterministic operating rule,when the annual water diversion target is 1 billion m~3,the average rate of change in the water shortage index with the operating rules considering forecast uncertainty,is-19.4%,the average rate of change in energy production is+2.1%,and the average rate of change in energy consumption is-10.1%;when the annual water diversion target is 1.5 billion m~3,the average rate of change in the water shortage index with the operating rules considering forecast uncertainty,is-9.2%,the average rate of change in energy production is+1.9%,and the average rate of change in energy consumption is-5.0%.When forecasts are inaccurate,the operating rule considering forecast uncertainty is significantly more effective in supplying water,with a slight increase in power generation and a slight decrease in power consumption,outperforming the deterministic operating rule overall.