Building more realistic reservoir optimization models using data mining - A case study of Shelbyville Reservoir

In this paper, we promote a novel approach to develop reservoir operation routines by learning from historical hydrologic information and reservoir operations. The proposed framework involves a knowledge discovery step to learn the real drivers of reservoir decision making and to subsequently build a more realistic (enhanced) model formulation using stochastic dynamic programming (SDP). The enhanced SDP model is compared to two classic SDP formulations using Lake Shelbyville, a reservoir on the Kaskaskia River in Illinois, as a case study. From a data mining procedure with monthly data, the past month’s inflow (Q_t−1), current month’s inflow (Q_t), past month’s release (R_t−1), and past month’s Palmer drought severity index (PDSI_t−1) are identified as important state variables in the enhanced SDP model for Shelbyville Reservoir. When compared to a weekly enhanced SDP model of the same case study, a different set of state variables and constraints are extracted. Thus different time scales for the model require different information. We demonstrate that adding additional state variables improves the solution by shifting the Pareto front as expected while using new constraints and the correct objective function can significantly reduce the difference between derived policies and historical practices. The study indicates that the monthly enhanced SDP model resembles historical records more closely and yet provides lower expected average annual costs than either of the two classic formulations (25.4% and 4.5% reductions, respectively). The weekly enhanced SDP model is compared to the monthly enhanced SDP, and it shows that acquiring the correct temporal scale is crucial to model reservoir operation for particular objectives.     

Evaluation of stochastic reservoir operation optimization models

This paper investigates the performance of seven stochastic models used to define optimal reservoir operating policies. The models are based on implicit (ISO) and explicit stochastic optimization (ESO) as well as on the parameterization–simulation–optimization (PSO) approach. The ISO models include multiple regression, two-dimensional surface modeling and a neuro-fuzzy strategy. The ESO model is the well-known and widely used stochastic dynamic programming (SDP) technique. The PSO models comprise a variant of the standard operating policy (SOP), reservoir zoning, and a two-dimensional hedging rule. The models are applied to the operation of a single reservoir damming an intermittent river in northeastern Brazil. The standard operating policy is also included in the comparison and operational results provided by deterministic optimization based on perfect forecasts are used as a benchmark. In general, the ISO and PSO models performed better than SDP and the SOP. In addition, the proposed ISO-based surface modeling procedure and the PSO-based two-dimensional hedging rule showed superior overall performance as compared with the neuro-fuzzy approach.     

River suspended sediment modelling using a fuzzy logic approach

Correct estimation of sediment volume carried by a river is very important for many water resources projects. Conventional sediment rating curves, however, are not able to provide sufficiently accurate results. In this paper, a fuzzy logic approach is proposed to estimate suspended sediment concentration from streamflow. This study provides forecasting benchmarks for sediment concentration prediction in the form of a numerical and graphical comparison between fuzzy and rating‐curve models. Benchmarking was based on a 5‐year period of continuous streamflow and sediment concentration data of Quebrada Blanca Station operated by the United States Geological Survey. The benchmark results showed that the fuzzy model was able to produce much better results than rating‐curve models. The fuzzy model proposed in the study is site specific and does not simulate the hysteresis effects. Copyright © 2006 John Wiley & Sons, Ltd.     

Optimizing Wellfield Operation in a Variable Power Price Regime

Wellfield management is a multiobjective optimization problem. One important objective has been energy efficiency in terms of minimizing the energy footprint (EFP) of delivered water (MWh/m³). However, power systems in most countries are moving in the direction of deregulated markets and price variability is increasing in many markets because of increased penetration of intermittent renewable power sources. In this context the relevant management objective becomes minimizing the cost of electric energy used for pumping and distribution of groundwater from wells rather than minimizing energy use itself. We estimated EFP of pumped water as a function of wellfield pumping rate (EFP‐Q relationship) for a wellfield in Denmark using a coupled well and pipe network model. This EFP‐Q relationship was subsequently used in a Stochastic Dynamic Programming (SDP) framework to minimize total cost of operating the combined wellfield‐storage‐demand system over the course of a 2‐year planning period based on a time series of observed price on the Danish power market and a deterministic, time‐varying hourly water demand. In the SDP setup, hourly pumping rates are the decision variables. Constraints include storage capacity and hourly water demand fulfilment. The SDP was solved for a baseline situation and for five scenario runs representing different EFP‐Q relationships and different maximum wellfield pumping rates. Savings were quantified as differences in total cost between the scenario and a constant‐rate pumping benchmark. Minor savings up to 10% were found in the baseline scenario, while the scenario with constant EFP and unlimited pumping rate resulted in savings up to 40%. Key factors determining potential cost savings obtained by flexible wellfield operation under a variable power price regime are the shape of the EFP‐Q relationship, the maximum feasible pumping rate and the capacity of available storage facilities.     

A 2D well-balanced,coupled model of water flow,sediment transport,and bed evolution based on unstructured grids with efficient variable storage strategy

In recent decades,a few Godunov-type,finite volume two-dimensional(2D)unstructured grid,coupled flow,and sediment models(GF2DUCM)have been developed for flows over erodible beds.These kinds of models are generally analyzed as a Vertex Model(VM)that define topography at the cell vertex,which can lead to the non-conservation of mass regarding flow,sediment,and bed evolution.Here,a full cellcantered variable storage method(Central Model or CM)is applied as the solution of the GF2DUCM.In this method,terrain elevation is defined at the cell centroids;this accurately describes the physical relations between the water depth and topography deformation.This approach can fully eliminate calculation errors in topography deformation at local cells caused by the interpolation of topography deformation at the cell vertex,and reduced uncertainty in the computation of the GF2DUCM.The model performance is systematically tested using a series of laboratory experiments,which demonstrate the mass conservation feature and high accuracy in reproducing hydrodynamic and morphological processes.     

Operations optimization of multireservoir systems using storage moments equations

Fletcher–Ponnambalam presented a new model for considering the balance equation of the storage volume of the reservoir using indicator functions. For stochastic inflows, the two storage moments of this balance equation, namely, the mean and variance, calculated using a random release policy were found to be quite accurate unlike any known models. Significantly, this model required no discretization of storage volumes or releases. In this paper, this work has been extended to two new cases: for multireservoir systems which require further consideration of the stochastic releases and for arbitrary distribution of inflows using the Beta-equivalent Kumaraswamy distribution which has a simpler form than Beta. The randomized release policies are easy to use even in a multireservoir problem. The Parambikulm-Aliyar Project from India is used as a case study. Results show accurate predictions of mean of storages in the multireservoir case but show the need for further improvement in the standard deviations of storages. The optimal benefits and the policies obtained are shown to be at least as good as obtained using Monte Carlo-based methods.     

Modeling ground water flow in alluvial mountainous catchments on a watershed scale

In large mountainous catchments, shallow unconfined alluvial aquifers play an important role in conveying subsurface runoff to the foreland. Their relatively small extent poses a serious problem for ground water flow models on the river basin scale. River basin scale models describing the entire water cycle are necessary in integrated water resources management and to study the impact of global climate change on ground water resources. Integrated regional-scale models must use a coarse, fixed discretization to keep computational demands low and to facilitate model coupling. This can lead to discrepancies between model discretization and the geometrical properties of natural systems. Here, an approach to overcome this discrepancy is discussed using the example of the German-Austrian Upper Danube catchment, where a coarse ground water flow model was developed using MODFLOW. The method developed uses a modified concept from a hydrological catchment drainage analysis in order to adapt the aquifer geometry such that it respects the numerical requirements of the chosen discretization, that is, the width and the thickness of cells as well as gradients and connectivity of the catchment. In order to show the efficiency of the developed method, it was tested and compared to a finely discretized ground water model of the Ammer subcatchment. The results of the analysis prove the applicability of the new approach and contribute to the idea of using physically based ground water models in large catchments.     

Probabilistic-fuzzy health risk modeling

Health risk analysis of multi-pathway exposure to contaminated water involves the use of mechanistic models that include many uncertain and highly variable parameters. Currently, the uncertainties in these models are treated using statistical approaches. However, not all uncertainties in data or model parameters are due to randomness. Other sources of imprecision that may lead to uncertainty include scarce or incomplete data, measurement error, data obtained from expert judgment, or subjective interpretation of available information. These kinds of uncertainties and also the non-random uncertainty cannot be treated solely by statistical methods. In this paper we propose the use of fuzzy set theory together with probability theory to incorporate uncertainties into the health risk analysis. We identify this approach as probabilistic-fuzzy risk assessment (PFRA). Based on the form of available information, fuzzy set theory, probability theory, or a combination of both can be used to incorporate parameter uncertainty and variability into mechanistic risk assessment models. In this study, tap water concentration is used as the source of contamination in the human exposure model. Ingestion, inhalation and dermal contact are considered as multiple exposure pathways. The tap water concentration of the contaminant and cancer potency factors for ingestion, inhalation and dermal contact are treated as fuzzy variables while the remaining model parameters are treated using probability density functions. Combined utilization of fuzzy and random variables produces membership functions of risk to individuals at different fractiles of risk as well as probability distributions of risk for various alpha-cut levels of the membership function. The proposed method provides a robust approach in evaluating human health risk to exposure when there is both uncertainty and variability in model parameters. PFRA allows utilization of certain types of information which have not been used directly in existing risk assessment methods.     

A fully implicit model for simulating dynamo action in a Cartesian domain

We present a fully implicit numerical method to solve the incompressible MHD equations in a strongly rotating Cartesian domain. The equations are solved in a primitive variable formulation using a finite volume discretization. In order to use massively parallel computers, we applied a domain decomposition approach in space. The performance of this model is compared with an earlier model, which treated the convective terms of the equations in an explicit manner. Our results indicate that although the fully implicit method needs about three times the memory of the implicit–explicit method, it is superior in terms of computational efficiency. As an application of this model, we investigated the influence of the Prandtl number in the range of 0.01–1000 on the dynamics of the dynamo.     

A comparison of two physics-based numerical models for simulating surface water–groundwater interactions

Problems in hydrology and water management that involve both surface water and groundwater are best addressed with simulation models that can represent the interactions between these two flow regimes. In the current generation of coupled models, a variety of approaches is used to resolve surface–subsurface interactions and other key processes such as surface flow propagation. In this study we compare two physics-based numerical models that use a 3D Richards equation representation of subsurface flow. In one model, surface flow is represented by a fully 2D kinematic approximation to the Saint–Venant equations with a sheet flow conceptualization. In the second model, surface routing is performed via a quasi-2D diffusive formulation and surface runoff follows a rill flow conceptualization. The coupling between the land surface and the subsurface is handled via an explicit exchange term resolved by continuity principles in the first model (a fully-coupled approach) and by special treatment of atmospheric boundary conditions in the second (a sequential approach). Despite the significant differences in formulation between the two models, we found them to be in good agreement for the simulation experiments conducted. In these numerical tests, on a sloping plane and a tilted V-catchment, we examined saturation excess and infiltration excess runoff production under homogeneous and heterogeneous conditions, the dynamics of the return flow process, the differences in hydrologic response under rill flow and sheet flow parameterizations, and the effects of factors such as grid discretization, time step size, and slope angle. Low sensitivity to vertical discretization and time step size was found for the two models under saturation excess and homogeneous conditions. Larger sensitivity and differences in response were observed under infiltration excess and heterogeneous conditions, due to the different coupling approaches and spatial discretization schemes used in the two models. For these cases, the sensitivity to vertical and temporal resolution was greatest for processes such as reinfiltration and ponding, although the differences between the hydrographs of the two models decreased as mesh and step size were progressively refined. In return flow behavior, the models are in general agreement, with the largest discrepancies, during the recession phase, attributable to the different parameterizations of diffusion in the surface water propagation schemes. Our results also show that under equivalent parameterizations, the rill and sheet flow conceptualizations used in the two models produce very similar responses in terms of hydrograph shape and flow depth distribution.     

From rainfall to spring discharge: Coupling conduit flow,subsurface matrix flow and surface flow in karst systems using a discrete–continuum model

Physics-based distributed models for simulating flow in karst systems are generally based on the discrete–continuum approach in which the flow in the three-dimensional fractured limestone matrix continuum is coupled with the flow in discrete one-dimensional conduits. In this study we present a newly designed discrete–continuum model for simulating flow in karst systems. We use a flexible spatial discretization such that complicated conduit networks can be incorporated. Turbulent conduit flow and turbulent surface flow are described by the diffusion wave equation whereas laminar variably saturated flow in the matrix is described by the Richards equation. Transients between free-surface and pressurized conduit flow are handled by changing the capacity term of the conduit flow equation. This new approach has the advantage that the transients in mixed conduit flow regimes can be handled without the Preissmann slot approach. Conduit–matrix coupling is based on the Peaceman’s well-index such that simulated exchange fluxes across the conduit–matrix interface are less sensitive to the spatial discretization. Coupling with the surface flow domain is based on numerical techniques commonly used in surface–subsurface models and storm water drainage models. Robust algorithms are used to simulate the non-linear flow processes in a coupled fashion. The model is verified and illustrated with simulation examples.     

Uncertainty Analysis for a Dynamic Phosphorus Model with Fuzzy Parameters

A simplified method based on fuzzy set theory is presented to incorporate uncertainty of parameters into a dynamic total phosphorus model. Uncertainty may arise from difference between calibrated conditions and projected condition as well as from inconsistency of available data in the literature. The uncertainty in parameters was represented by fuzzy numbers that can be generated through various ways such as model calibration process, soft interpretation of literature data, and subjective opinions of experts. The proposed fuzzy approach decomposed fuzzy parameters into interval numbers at different level cuts, and solved for interval solutions through very simple calculation instead of solving nonlinear programming models. The interval solutions at each level cut were could be combined to obtain fuzzy solutions. This method has been applied to the phosphorus load-response model of the Triadelphia Reservoir near the Washington, DC area. Two pollution control scenarios have been simulated with fuzzy parameters. The measures of necessity and possibility have been used to analyze the potential risk of the two scenarios. The research results indicated that uncertainty is a very important factor in water quality modeling. By incorporating uncertainty into model framework, the fuzzy model identified the highly risky scenario that was considered preferable based on solutions of the deterministic model.     

Correction of Discretization Errors Simulated at Supply Wells

Many hydrogeology problems require predictions of hydraulic heads in a supply well. In most cases, the regional hydraulic response to groundwater withdrawal is best approximated using a numerical model; however, simulated hydraulic heads at supply wells are subject to errors associated with model discretization and well loss. An approach for correcting the simulated head at a pumping node is described here. The approach corrects for errors associated with model discretization and can incorporate the user's knowledge of well loss. The approach is model independent, can be applied to finite difference or finite element models, and allows the numerical model to remain somewhat coarsely discretized and therefore numerically efficient. Because the correction is implemented external to the numerical model, one important benefit of this approach is that a response matrix, reduced model approach can be supported even when nonlinear well loss is considered.     

Dual state–parameter estimation of hydrological models using ensemble Kalman filter

Hydrologic models are twofold: models for understanding physical processes and models for prediction. This study addresses the latter, which modelers use to predict, for example, streamflow at some future time given knowledge of the current state of the system and model parameters. In this respect, good estimates of the parameters and state variables are needed to enable the model to generate accurate forecasts. In this paper, a dual state–parameter estimation approach is presented based on the Ensemble Kalman Filter (EnKF) for sequential estimation of both parameters and state variables of a hydrologic model. A systematic approach for identification of the perturbation factors used for ensemble generation and for selection of ensemble size is discussed. The dual EnKF methodology introduces a number of novel features: (1) both model states and parameters can be estimated simultaneously; (2) the algorithm is recursive and therefore does not require storage of all past information, as is the case in the batch calibration procedures; and (3) the various sources of uncertainties can be properly addressed, including input, output, and parameter uncertainties. The applicability and usefulness of the dual EnKF approach for ensemble streamflow forecasting is demonstrated using a conceptual rainfall-runoff model.     

Error covariance calculation for forecast bias estimation in hydrologic data assimilation

To date, an outstanding issue in hydrologic data assimilation is a proper way of dealing with forecast bias. A frequently used method to bypass this problem is to rescale the observations to the model climatology. While this approach improves the variability in the modeled soil wetness and discharge, it is not designed to correct the results for any bias. Alternatively, attempts have been made towards incorporating dynamic bias estimates into the assimilation algorithm. Persistent bias models are most often used to propagate the bias estimate, where the a priori forecast bias error covariance is calculated as a constant fraction of the unbiased a priori state error covariance. The latter approach is a simplification to the explicit propagation of the bias error covariance. The objective of this paper is to examine to which extent the choice for the propagation of the bias estimate and its error covariance influence the filter performance. An Observation System Simulation Experiment (OSSE) has been performed, in which ground water storage observations are assimilated into a biased conceptual hydrologic model. The magnitudes of the forecast bias and state error covariances are calibrated by optimizing the innovation statistics of groundwater storage. The obtained bias propagation models are found to be identical to persistent bias models. After calibration, both approaches for the estimation of the forecast bias error covariance lead to similar results, with a realistic attribution of error variances to the bias and state estimate, and significant reductions of the bias in both the estimates of groundwater storage and discharge. Overall, the results in this paper justify the use of the traditional approach for online bias estimation with a persistent bias model and a simplified forecast bias error covariance estimation.     

Nonlinear model reduction of unconfined groundwater flow using POD and DEIM

Nonlinear groundwater flow models have the propensity to be overly complex leading to burdensome computational demands. Reduced modeling techniques are used to develop an approximation of the original model that has smaller dimensionality and faster run times. The reduced model proposed is a combination of proper orthogonal decomposition (POD) and the discrete empirical interpolation method (DEIM). Solutions of the full model (snapshots) are collected to represent the physical dynamics of the system and Galerkin projection allows the formulation of a reduced model that lies in a subspace of the full model. Interpolation points are added through DEIM to eliminate the reduced model's dependence on the dimension of the full model. POD is shown to effectively reduce the dimension of the full model and DEIM is shown to speed up the solution by further reducing the dimension of the nonlinear calculations. To show the concept can work for unconfined groundwater flow model, with added nonlinear forcings, one-dimensional and two-dimensional test cases are constructed in MODFLOW-OWHM. POD and DEIM are added to MODFLOW as a modular package. Comparing the POD and the POD-DEIM reduced models, the experimental results indicate similar reduction in dimension size with additional computation speed up for the added interpolation. The hyper-reduction method presented is effective for models that have fine discretization in space and/or time as well as nonlinearities with respect to the state variable. The dual reduction approach ensures that, once constructed, the reduced model can be solved in an equation system that depends only on reduced dimensions.     

A stochastic conflict resolution model for water quality management in reservoir–river systems

In this paper, optimal operating rules for water quality management in reservoir–river systems are developed using a methodology combining a water quality simulation model and a stochastic GA-based conflict resolution technique. As different decision-makers and stakeholders are involved in the water quality management in reservoir–river systems, a new stochastic form of the Nash bargaining theory is used to resolve the existing conflict of interests related to water supply to different demands, allocated water quality and waste load allocation in downstream river. The expected value of the Nash product is considered as the objective function of the model which can incorporate the inherent uncertainty of reservoir inflow. A water quality simulation model is also developed to simulate the thermal stratification cycle in the reservoir, the quality of releases from different outlets as well as the temporal and spatial variation of the pollutants in the downstream river. In this study, a Varying Chromosome Length Genetic Algorithm (VLGA), which has computational advantages comparing to other alternative models, is used. VLGA provides a good initial solution for Simple Genetic Algorithms and comparing to Stochastic Dynamic Programming (SDP) reduces the number of state transitions checked in each stage. The proposed model, which is called Stochastic Varying Chromosome Length Genetic Algorithm with water Quality constraints (SVLGAQ), is applied to the Ghomrud Reservoir–River system in the central part of Iran. The results show, the proposed model for reservoir operation and waste load allocation can reduce the salinity of the allocated water demands as well as the salinity build-up in the reservoir.     

Field Test of a Hybrid Finite‐Difference and Analytic Element Regional Model

Regional finite‐difference models often have cell sizes that are too large to sufficiently model well‐stream interactions. Here, a steady‐state hybrid model is applied whereby the upper layer or layers of a coarse MODFLOW model are replaced by the analytic element model GFLOW, which represents surface waters and wells as line and point sinks. The two models are coupled by transferring cell‐by‐cell leakage obtained from the original MODFLOW model to the bottom of the GFLOW model. A real‐world test of the hybrid model approach is applied on a subdomain of an existing model of the Lake Michigan Basin. The original (coarse) MODFLOW model consists of six layers, the top four of which are aggregated into GFLOW as a single layer, while the bottom two layers remain part of MODFLOW in the hybrid model. The hybrid model and a refined “benchmark” MODFLOW model simulate similar baseflows. The hybrid and benchmark models also simulate similar baseflow reductions due to nearby pumping when the well is located within the layers represented by GFLOW. However, the benchmark model requires refinement of the model grid in the local area of interest, while the hybrid approach uses a gridless top layer and is thus unaffected by grid discretization errors. The hybrid approach is well suited to facilitate cost‐effective retrofitting of existing coarse grid MODFLOW models commonly used for regional studies because it leverages the strengths of both finite‐difference and analytic element methods for predictions in mildly heterogeneous systems that can be simulated with steady‐state conditions.     

Fuzzy-probabilistic calculations of water-balance uncertainty

Hydrogeological systems are often characterized by imprecise, vague, inconsistent, incomplete, or subjective information, which may limit the application of conventional stochastic methods in predicting hydrogeologic conditions and associated uncertainty. Instead, predictions and uncertainty analysis can be made using uncertain input parameters expressed as probability boxes, intervals, and fuzzy numbers. The objective of this paper is to present the theory for, and a case study as an application of, the fuzzy-probabilistic approach, combining probability and possibility theory for simulating soil water balance and assessing associated uncertainty in the components of a simple water-balance equation. The application of this approach is demonstrated using calculations with the RAMAS Risk Calc code, to assess the propagation of uncertainty in calculating potential evapotranspiration, actual evapotranspiration, and infiltration—in a case study at the Hanford site, Washington, USA. Propagation of uncertainty into the results of water-balance calculations was evaluated by changing the types of models of uncertainty incorporated into various input parameters. The results of these fuzzy-probabilistic calculations are compared to the conventional Monte Carlo simulation approach and estimates from field observations at the Hanford site.     

Seismic spectra on soil with uncertain properties

Expected spectral ordinates and their standard deviations are computed for ground motions produced by SH waves at the surface of a homogeneous soil formation resting on a half-space of rock. Wave transmission is idealized as linear and one-dimensional. Uncertainties about soil properties are taken into account using various approximations: a Taylor series expansion of the Fourier spectrum magnification factors with different numbers of terms and different intervals for numerical differentiation; discretization of the joint probability density function of the soil properties so as to match its moments up to those of third order including the crossed moments; a discretization that matches up to some fifth-order moments; and a fine numerical integration for the marginal density function of the modulus of rigidity, discretizing the rest of the density function. The latter approach gives sufficiently accurate results for practical purposes. Effects of even moderate uncertainty about soil properties on the statistics of spectral ordinates are found to be decisive.