Stochastic and Robust Multi-Objective Optimal Management of Pumping from Coastal Aquifers Under Parameter Uncertainty

Combined simulation-optimization approaches have been used as tools to derive optimal groundwater management strategies to maintain or improve water quality in contaminated or other aquifers. Surrogate models based on neural networks, regression models, support vector machies etc., are used as substitutes for the numerical simulation model in order to reduce the computational burden on the simulation-optimization approach. However, the groundwater flow and transport system itself being characterized by uncertain parameters, using a deterministic surrogate model to substitute it is a gross and unrealistic approximation of the system. Till date, few studies have considered stochastic surrogate modeling to develop groundwater management methodologies. In this study, we utilize genetic programming (GP) based ensemble surrogate models to characterize coastal aquifer water quality responses to pumping, under parameter uncertainty. These surrogates are then coupled with multiple realization optimization for the stochastic and robust optimization of groundwater management in coastal aquifers. The key novelty in the proposed approach is the capability to capture the uncertainty in the physical system, to a certain extent, in the ensemble of surrogate models and using it to constrain the optimization search to derive robust optimal solutions. Uncertainties in hydraulic conductivity and the annual aquifer recharge are incorporated in this study. The results obtained indicate that the methodology is capable of developing reliable and robust strategies for groundwater management.     

A Stochastic Non-linear Programming Model for a Multi-period Water Resource Allocation with Multiple Objectives

A rapid increase in demand and severe droughts in recent years has increased the pressure on water supplies throughout most parts of Australia. This has resulted in the need for tools to allocate limited water across users in different regions, and explore scenarios so as to achieve economic, social and environmental benefits. A major challenge in water resource allocation is dealing with the uncertainty in the system, particularly with respect to reservoir inflow. Stochastic non-linear programming is applied to water resource allocation to accommodate this uncertainty across the time periods of the planning horizon. A large range of solutions is produced representing the distributions of uncertainty in reservoir inflow. These solutions are used in a Monte Carlo simulation to estimate the trade-off in amounts of water allocated versus risk of not achieving minimal reservoir levels. The methodology is applied to a case study in South East Queensland in Australia, a region which is currently facing a severe water shortage over the next 3 years. A new water supply initiative that the Queensland State Government is considering to overcome the water crisis is assessed using the methodology.     

汛限水位动态控制的防洪极限风险分析

综合考虑水文、水力不确定性因素对汛限水位控制下的水库防洪极限风险进行研究，采用随机模拟方法计算极限防洪风险率。应用一阶季节性自回归模型模拟多场入库洪水序列，考虑水力不确定性对泄洪能力的影响，在给定调洪规则下对不同汛限水位方案进行调洪，得到水库最高调洪水位和防洪极限风险率。实例结果表明：水文因素的随机性和防洪调度规则是水库防洪风险的主要影响因素，水力因素对防洪风险影响不大，同时得出了水库面临汛限水位所能承受的极限风险率，为决策者安全度汛提供一种参考依据。     

A Conditional Value-at-Risk Based Inexact Water Allocation Model

A conditional value-at-risk (CVaR) based inexact two-stage stochastic programming (CITSP) model was developed in this study for supporting water resources allocation problems under uncertainty. A CITSP model was formulated through incorporating a CVaR constraint into an inexact two-stage stochastic programming (ITSP) framework, and could be used to deal with uncertainties expressed as not only probability distributions but also discrete intervals. The measure of risks about the second-stage penalty cost was incorporated into the model, such that the trade-off between system economy and extreme expected loss could be analyzed. The developed model was applied to a water resources allocation problem involving a reservoir and three competing water users. The results indicated that the CITSP model performed better than the ITSP model in its capability of reflecting the economic loss from extreme events. Also, it could generate interval solutions within which the decision alternatives could be selected from a flexible decision space. Overall, the CITSP model was useful for reflecting the decision maker’s attitude toward risk aversion and could help seek cost-effective water resources management strategies under complex uncertainties.     

Benefit and Risk Balance Optimization for Stochastic Hydropower Scheduling

Limited by inflow forecasting methods, the forecasting results are so unreliable that we have to take their uncertainty and risk into account when incorporating stochastic inflow into reservoir operation. Especially in the electricity market, punishment often happens when the hydropower station does not perform as planned. Therefore, focusing on the risk of power generation, a benefit and risk balance optimization model (BRM) which takes stochastic inflow as the major risk factor is proposed for stochastic hydropower scheduling. The mean-variance theory is firstly introduced into the optimal dispatching of hydropower station, and a variational risk coefficient is employed to give service to managers’ subjective preferences. Then, the multi-period stochastic inflow is simulated by multi-layer scenario tree. Moreover, a specific scenario reduction and reconstruction method is put forward to reduce branches and computing time accordingly. Finally, the proposed model is applied to the Three Gorges Reservoir (TGR) in China for constructing a weekly generation scheduling in falling stage. Compared to deterministic dynamic programming (DDP) and stochastic dynamic programming (SDP), BRM achieves more satisfactory performance. Moreover, the tradeoffs for risk-averse decision makers are discussed, and an efficient curve about benefit and risk is formed to help make decision.     

水文序列延长后已建水库防洪特征水位复核

为探究洪水不确定性对大坝防洪特征水位设计的影响,采用水库调洪演算的随机微分方程推求防洪特征水位的概率分布线型,以三峡水库为例,通过调洪演算将洪水不确定性转换为防洪特征水位的不确定性,以分析已建水库防洪特征水位的分布规律,并对三峡水库防洪特征水位进行了复核分析。研究结果表明:随着样本容量的增加,洪水的信息量增大,三峡大坝设计资料和原校核洪水位(180.4 m)的可靠度均有所提高;当样本容量为120 a时,设计洪水资料和原校核洪水位的可靠度分别为93.19%和99.17%。研究结果为提高水库大坝安全设计提供了理论依据。     

Long-Term Planning of Water Systems in the Context of Climate Non-Stationarity with Deterministic and Stochastic Optimization

Seasonal inflow variability, climate non-stationarity and climate change are matters of concern for water system planning and management. This study presents optimization methods for long-term planning of water systems in the context of a non-stationary climate with two levels of inflow variability: seasonal and inter-annual. Deterministic and stochastic optimization models with either one time-step (intra-annual) or two time-steps (intra-annual and inter-annual) were compared by using three water system optimization models. The first model used one time-step sampling stochastic dynamic programming (SSDP). The other models with two time-steps are long-term deterministic dynamic programming (LT-DDP) and long-term sampling stochastic dynamic programming (LT-SSDP). The study area is the Manicouagan water system located in Quebec, Canada. The results show that there will be an increase of inflow to hydropower plants in the future climate with an increase of inflow uncertainty. The stochastic optimization with two time-steps was the most suitable for handling climate non-stationarity. The LT-DDP performed better in terms of reservoir storage, release and system efficiency but with high uncertainty. The SSDP had the lowest performance. The SSDP was not able to deal with the non-stationary climate and seasonal variability at the same time. The LT-SSDP generated operating policies with smaller uncertainty compared to LT-DDP, and it was therefore a more appropriate approach for water system planning and management in a non-stationary climate characterized by high inflow variability.     

Assessment of Stochastic Operation Optimization for Reservoirs of Contrasting Scales

Deriving optimal release policies for dams and corresponding reservoirs is crucial for the sustainable water resources management of a region as they directly control the distribution of water to several users. Mathematical optimization algorithms can help in finding efficient reservoir operating strategies taking into account complex system constraints and hydrologic uncertainty. The robustness of operation optimization models may be influenced by physical reservoir characteristics such as size and scale and the effectiveness of a model for a particular case study does not always guarantee the same level of success for another application. This research focused on assessing the applicability of an implicit stochastic optimization (ISO) procedure to derive rule curves for two different dams of contrasting reservoir scales in terms of physical and operational characteristics. The results demonstrated the feasibility of the proposed technique for both small- and large-scale systems in view of the lower vulnerability provided by the ISO-derived policies in contrast to operations carried out by the standard reservoir operating policy as well as the proximity of the ISO operations with those by perfect-forecast deterministic optimization. The ISO procedure also provided operating rules similar to, and even less vulnerable than, those derived by stochastic dynamic programming.     

Interval-parameter Two-stage Stochastic Semi-infinite Programming: Application to Water Resources Management under Uncertainty

In this study, an interval-parameter two-stage stochastic semi-infinite programming (ITSSP) method was developed for water resources management under uncertainty. As a new extension of mathematical programming methods, the developed ITSSP approach has advantages in uncertainty reflection and policy analysis. In order to better account for uncertainties, the ITSSP approach is expressed with discrete intervals, functional intervals and probability density functions. The ITSSP method integrates the two-stage stochastic programming (TSP), interval programming (IP) and semi-infinite programming (SIP) within a general optimization framework. The ITSSP has an infinite number of constraints because it uses functional intervals with time (t) being an independent variable. The different t values within the range [0, 90] lead to different constraints. At same time, ITSSP also includes probability distribution information. The ITSSP method can incorporate pre-defined water resource management policies directly into its optimization process to analyze various policy scenarios having different economic penalties when the promised amounts are not delivered. The model is applied to a water resource management system with three users and four periods (corresponding to winter, spring, summer and fall, respectively). Solutions of the ITSSP model provide desired water allocation patterns, which maximize both the system’s benefits and feasibility. The results indicate that reasonable interval solutions were generated for objective function values and decision variables, thus a number of decision alternatives can be generated under different levels of stream flow. The obtained solutions are useful for decision makers to obtain insight regarding the tradeoffs between environmental, economic and system reliability criteria.     

An Inexact Two-stage Fuzzy-stochastic Programming Model for Water Resources Management

An inexact two-stage fuzzy-stochastic programming (ITFSP) method is developed for water resources management under uncertainty. Fuzzy sets theory is introduced to represent various punishment policies under different water availability conditions. As an extension of conventional two-stage stochastic programming (TSP) method, two special characteristics of the proposed approach make it unique compared with existing approaches. One is it could handle flexible penalty rates, which are much reasonable for both of the authorities and users, and have seldom been considered in the TSP framework. The other is uncertain information expressed as discrete intervals and probability distribution functions can be effectively reflected in the optimization processes and solutions. After formulating the model, a hypothetical case is employed for demonstrating its applicability under two scenarios, where the inflow is divided into four and eight intervals, respectively. The results indicate that reasonable solutions have been obtained. They provide desired allocation patterns with maximized system benefit under two feasibility levels. The solutions present as stable intervals with different risk levels in violating the water demands, and can be used for generating decision alternatives. Comparisons of the solution from the ITFSP with that from the ITSP (inexact two-stage stochastic programming) and TSP approach are also undertaken. It shows that the ITFSP could produce more system benefit than existing methods and deal with flexible penalty policies for better water management and utilization.     

A Novel Solution for Stochastic Dynamic Game of Water Allocation from a Reservoir Using Collocation Method

In this study, a continuous model of stochastic dynamic game for water allocation from a reservoir system was developed. The continuous random variable of inflow in the state transition function was replaced with a discrete approximant rather than using the mean of the random variable as is done in a continuous model of deterministic dynamic game. As a result, a new solution method was used to solve the stochastic model of game based on collocation method. The collocation method was introduced as an alternative to linear-quadratic (LQ) approximation methods to resolve a dynamic model of game. The collocation method is not limited to the first and second degree approximations, compared to LQ approximation, i.e. Ricatti equations. Furthermore, in spite of LQ related problems, consideration of the stochastic nature of game on the action variables in the collocation method would be possible. The proposed solution method was applied to the real case of reservoir operation, which typically requires considering the effect of uncertainty on decision variables. The results of the solution of the stochastic model of game are compared with the results of a deterministic solution of game, a classical stochastic dynamic programming model (e.g. Bayesian Stochastic Dynamic Programming model, BSDP), and a discrete stochastic dynamic game model (PSDNG). By comparing the results of alternative methods, it is shown that the proposed solution method of stochastic dynamic game is quite capable of providing appropriate reservoir operating policies.     

IFTSQP: An Inexact Optimization Model for Water Resources Management under Uncertainty

Abstract

An interval-fuzzy two-stage quadratic programming (IFTSQP) method is developed for water resources management under uncertainty. The methodincorporates techniques of interval-parameter programming, two-stage stochastic programming, and fuzzy quadratic programming within a general optimization framework to tackle multiple uncertainties presented as intervals, fuzzy sets and probability distributions. In the model formulation, multiple control variables are adopted to handle independent uncertainties in the model's right-hand sides; fuzzy quadratic terms are used in the objective function to minimize the variation in satisfaction degrees among the constraints. Moreover, the method can support the analysis of policy scenarios that are associated with economic penalties when the promised targets are violated. The developed method is then applied to a case study of water resources management planning. The results indicate that reasonable solutions have been obtained. They can help provide bases for identifying desired water-allocation plans with a maximized system benefit and a minimized constraint-violation risk.     

Robust Water Supply Chain Network Design under Uncertainty in Capacity

This paper focuses on the capacity uncertainty in water supply chains that occurs when facilities face disruption. A combination of scenario-based two-stage stochastic programming with the min-max robust optimization approach is proposed to optimize the water supply chain network design problem. In the first stage, the decisions are made on locations and capacities of reservoirs and water-treatment plants while recourse decisions including amount of water extraction, amount of water refinement, and consequently amount of water held in reservoirs are made at the second stage. The proposed robust two-stage stochastic programming model can help decision makers consider the impacts of uncertainties and analyze trade-offs between system cost and stability. The literature reveals that most exact methods are not able to tackle the computational complexity of mixed integer non-linear two-stage stochastic problems at large scale. Another contribution of this study is to propose two metaheuristics - a particle swarm optimization (PSO) and a bat algorithm (BA) - to solve the proposed model in large-scale networks efficiently in a reasonable time. The developed model is applied to several hypothetical cases of water resources management systems to evaluate the effectiveness of the model formulation and solution algorithms. Sensitivity analyses are also carried out to analyze the behavior of the model and the robustness approach under parameters variations.

     

Optimal Operation of Reservoir Systems using Simulated Annealing

A stochastic search technique, simulated annealing (SA), is used to optimize the operation of multiple reservoirs. Seminal application of annealing technique in general to multi-period, multiple-reservoir systems, along with problem representation and selection of different parameter values used in the annealing algorithm for specific cases is discussed. The search technique is improved with the help of heuristic rules, problem-specific information and concepts from the field of evolutionary algorithms. The technique is tested for application to a benchmark problem of four-reservoir system previously solved using a linear programming formulation and its ability to replicate the global optimum solution is examined. The technique is also applied to a system of four hydropower generating reservoirs in Manitoba, Canada, to derive optimal operating rules. A limited version of this problem is solved using a mixed integer nonlinear programming and results are compared with those obtained using SA. A better objective function value is obtained using simulated annealing than the value from a mixed integer non-linear programming model developed for the same problem. Results obtained from these applications suggest that simulated annealing can be used for obtaining near-optimal solutions for multi-period reservoir operation problems that are computationally intractable.     

基于贝叶斯统计与MCMC思想的水库随机优化调度研究

针对马尔柯夫随机动态规划中的维数灾问题,提出一种改进马尔柯夫随机动态规划方法,基于贝叶斯统计原理,采用马尔柯夫链蒙特卡洛方法(Markov Chain Monte Carlo,MCMC)从数学角度出发,推求出一定预报级别下的实际来流概率密度函数,建立与预报级别相关的实际来流概率矩阵,在考虑预报误差发生的情况下进行不确定性优化调度,并且将该方法计算结果与有无预报时段相结合的马尔柯夫随机动态规划方法计算结果进行比较。结果表明,该方法所得到的结果比马尔柯夫随机动态规划结果更加贴近实际多年平均发电量,并且能够有效地减少计算量,缩短计算时间,从一定程度上解决了维数灾问题,本方法为不确定性优化调度提供重要理论参考。     

基于鲁棒规划方法的农业水资源多目标优化配置模型

在干旱半干旱地区,调整种植结构可以促进农业水资源的高效利用。农业水资源配置需要在多个目标间权衡博弈,对各目标的偏好和赋权直接影响着优化模型的输出和决策方案的制定,但以往研究往往忽略了权重确定过程中因主观等因素的影响而普遍存在的不确定性。针对农业水资源多目标规划中存在的权重不确定性难题,建立了基于鲁棒优化方法的农业水资源多目标优化配置模型方法(MRPWU)。该方法可以把权重中蕴含的复杂不确定性信息纳入建模过程,产生可靠的模型结果;并能提供效益值及风险值均定量化的方案集,便于决策者在权衡效益与风险后确定最优方案。模型以作物种植经济收益和碳吸收量最大化为目标、以水土资源供需平衡等为约束条件,并应用于农业水资源供需矛盾突出的甘肃省民勤县。优化结果表明,随着保护度水平的提高,生态效益上升,经济效益和综合效益下降,系统面临的风险也随之下降。相比于权重为确定参数的模型,MRPWU模型可以在综合效益下降3.7%的同时,较大地提高系统应对权重不确定性以及风险的能力。与2017年的实际情况相比,MRPWU模型可以减少种植面积1.6%、节省灌溉用水3.9%,同时提高生态效益1.6%。     

An Integrated Approach of System Dynamics,Orthogonal Experimental Design and Inexact Optimization for Supporting Water Resources Management under Uncertainty

An integrated approach of system dynamics (SD), orthogonal experimental design (OED) and inexact optimization modeling was proposed for water resources management under uncertainty. The developed method adopted a combination of SD and OED to identify key scenarios within multiple factors, through which interval solutions for water demands could be obtained as input data for consequential optimization modeling. Also, optimal schemes could be obtained in the combination of inexact two-stage stochastic programming and credibility constrained programming. The developed method was applied to a real-world case study for supporting allocation of multiple-source water resources to multiple users in Dalian city within a multi-year context. The results indicated that a lower credibility-satisfaction level would generate higher allocation efficiency, a higher system benefit and a lower system violation risk. The developed model could successfully reflect and address the variety of uncertainties through provision of credibility levels, which corresponds to the decision makers’ preference regarding the tradeoffs between system benefits and violation risks.     

A Fuzzy Stochastic Dynamic Nash Game Analysis of Policies for Managing Water Allocation in a Reservoir System

In this paper a fuzzy dynamic Nash game model of interactions between water users in a reservoir system is presented. The model represents a fuzzy stochastic non-cooperative game in which water users are grouped into four players, where each player in game chooses its individual policies to maximize expected utility. The model is used to present empirical results about a real case water allocation from a reservoir, considering player (water user) non-cooperative behavior and also same level of information availability for individual players. According to the results an optimal allocation policy for each water user can be developed in addition to the optimal policy of the reservoir system. Also the proposed model is compared with two alternative dynamic models of reservoir optimization, namely Stochastic Dynamic Programming (SDP) and Fuzzy-State Stochastic Dynamic programming (FSDP). The proposed modeling procedures can be applied as an appropriate tool for reservoir operation, considering the interaction among the water users as well as the water users and reservoir operator.     

Operating Rules of an Irrigation Purposes Reservoir Using Multi-Objective Optimization

A multi-objective optimization technique for the operation of an irrigation reservoir is presented in this paper. The study deals with two different objective functions (OF): the minimization of reservoir release deficit from the irrigation demand (OF1) and the maximization of net benefit by the demand sector (OF2). In the first step, monthly optimization of each individual objective was performed with a deterministic non-linear programming (NLP) algorithm, that gave the lower and upper bounds for the multi-objective analysis. In the second step, multi-objective optimization was performed through the Constraint method that operates by optimising the objective function OF1, while the other (OF2) was constrained to satisfy release strategies generated by the optimization. Non-dominated set of release strategies is generated by parametrically varying the bounds of the constraints obtained from the individual optimal solutions. In the third step, the interactive analytical Step method was applied to find the best compromise solution, between the two OFs, by minimizing the distance of each non-dominated solution to an ideal solution that represents the utopian optimum for both OF1 and OF2. Furthermore, the interactive approach allows to improve the performance of the reservoir in terms of compromise irrigation releases, by changing the OF values until the satisfaction of predetermined criteria fixed by the planners and decision makers. The proposed water allocation model was applied to the Pozzillo reservoir operation, that supplies the Catania Plain irrigation area (Eastern Sicily).     

Optimal Short-term Reservoir Operation with Integrated Long-term Goals

A methodology to incorporate long-term goals within the short-term reservoir operation optimization model is proposed. Two conflicting objectives for the management of hydropower generation in two different power plants are incorporated. A chance-constrained optimization model is used to derive long-term (annual) operation strategies. With the time horizon of operation for the short-term optimization model kept equal to a single time-step of the long-term optimization model, the optimum end storages derived from the long-term model are incorporated as constraints (storage lower bounds) within the short-term model. The long-term benefits accrued from such an operation model are illustrated for a small reservoir, in South India. The solutions are compared with the historic operation. These are also compared with the solutions of a short-term optimal operation model ignoring long-term goals. The optimization model is solved using a multi-objective genetic algorithm.