A model-independent tool for evolutionary constrained multi-objective optimization under uncertainty

White, Jeremy T.; Knowling, Matthew J.; Fienen, Michael N.; Siade, Adam; Rea, Otis; Martínez, Guillermo Roberto Solarte

doi:10.1016/j.envsoft.2022.105316

Cited by 11 publications

(10 citation statements)

References 43 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In spite of our efforts to obtain a robust linear version of the model, the methods based on the Jacobian matrix have shown their limitations. This advocates for the use of more robust ensemble methods such as the Iterative Ensemble Smoother (IES (White et al, 2018)) for parameter estimation and uncertainty quantification, and evolutionary algorithms for optimization [PESTPP-MOU (White et al, 2022)]. The presented workflow may easily be extended to these methods recently available in the PEST + + Suite.…”

Section: Discussionmentioning

confidence: 99%

Particle tracking as a vulnerability assessment tool for drinking water production

et al. 2022

View full text Add to dashboard Cite

The simulation of concentration values and use of such data for history-matching is often impeded by the computation time of groundwater transport models based on the resolution of the advection-dispersion equation. This is unfortunate because such data are often rich in information and the prediction of concentration values is of great interest for decision making. Particle tracking can be used as an efficient alternative under a series of simplifying assumptions, which are often reasonable at groundwater sinks (wells and drains). Our approach consists of seeding particles around a sink and tracking particles backward, up to the source boundary condition, such as a contaminated stream. This particle tracking approach allows the use of parameter estimation and optimization methods requiring numerous model calls. We present a Python module facilitating the pre- and post-processing operations of a modeling workflow based on the widely used USGS MODFLOW6 and MODPATH7 programs. The module handles particle seeding around the sink and estimation of the mixing ratio of water withdrawn from the sink. This ratio is computed with a mixing law from the particle endpoints, accounting for particle velocities and mixing in the source model cells. We investigate the best practice to obtain robust derivatives with this approach, which is a benefit for the screening methods based on linear analysis. We illustrate the interest of the approach with a real world case study, considering a drinking water well field vulnerable to a contaminated stream. The configuration is typical of many other drinking water production sites. The modeling workflow is fully script-based to make the approach easily reproducible in similar cases.

show abstract

Section: Discussionmentioning

confidence: 99%

Particle tracking as a vulnerability assessment tool for drinking water production

et al. 2022

View full text Add to dashboard Cite

show abstract

“…The optimization was solved using the NSGA-II nondominated-sorting genetic algorithm (Deb et al, 2002) implemented in PESTPP-MOU (White et al, 2022), using ensemble-based constraint uncertainty. PESTPP-MOU was selected because it 255 implements a wide range of evolutionary algorithms, which are more effective than traditional optimization methods when solving highly nonlinear optimization problems typical of coastal environments (Ketabchi and Ataie-Ashtiani, 2015), its use required very little modification to input files after having previously used PEST++ software, and the reliability-based optimization could be implemented using PESTPP-MOU's "risk as an objective" option (White et al, 2022). The NSGA-II algorithm has been applied in many past coastal OUU studies (Mostafaei-Avandari and Ketabchi, 2020).…”

Section: Optimization Using Ensemble-based Uncertaintymentioning

confidence: 99%

“…5). The prediction ensemble 265 was evaluated for all the individuals of the initial population, reevaluated every 10 generations (which required a total of 30 x Npost model simulations each) and reused in the intermediate generations, as a tradeoff between uncertainty quantification and computational constraints (White et al, 2022;PEST++ Development Team, 2022).…”

Section: Optimization Using Ensemble-based Uncertaintymentioning

confidence: 99%

“…2050 sea level and recharge climate ensembles were generated and incorporated into the model parameter ensemble. Multi-objective optimization under uncertainty was then conducted via PESTPP-MOU (White et al, 2022), using the ensemble to account for uncertainty in the simulated response 75 to groundwater extraction within the optimization process. Results were compared both to the current water demand and to 2050 water demand projections.…”

Section: Introduction 30mentioning

confidence: 99%

See 1 more Smart Citation

An ensemble-based approach for pumping optimization in an island aquifer considering parameter, observation and climate uncertainty

Coulon

White²,

Pryet

et al. 2023

Preprint

View full text Add to dashboard Cite

Abstract. In coastal zones, a major objective of groundwater management is often to determine sustainable pumping rates avoiding well salinization. Understanding how model and climate uncertainties affect optimal management solutions is essential to provide groundwater managers information about salinization risk, and is facilitated by the use of optimization under uncertainty (OUU) methods. However, guidelines are missing for the widespread implementation of OUU in real-world coastal aquifers, and for the incorporation of climate uncertainty into OUU approaches. An ensemble-based OUU approach was developed, considering parameter, observation and climate uncertainty, and was implemented in a real-world island aquifer in the Magdalen Islands (Quebec, Canada). A sharp-interface seawater intrusion model was developed using MODFLOW-SWI2 and a prior parameter ensemble was generated, containing multiple equally plausible realizations. Ensemble-based history matching was conducted using an iterative ensemble smoother and yielded a posterior parameter ensemble conveying both parameter and observation uncertainty. 2050 sea level and recharge ensembles were generated and incorporated to generate a predictive parameter ensemble conveying parameter, observation and climate uncertainty. Multi-objective OUU was then conducted, aiming to both maximize pumping rates and minimize probability of well salinization. As a result, the optimal tradeoff between pumping and probability of salinization was quantified, considering parameter, historical observation and future climate uncertainty simultaneously. The multi-objective, ensemble-based OUU led to optimal pumping rates that were very different from a previous deterministic OUU, and close to the current and projected water demand for risk-averse stances. Incorporating climate uncertainty in the OUU was also critical since it reduced the maximum allowable pumping rates for users with a risk-averse stance. The workflow used tools adapted to very high-dimensional, nonlinear models and optimization problems, to facilitate its implementation in a wide range of real-world settings.

show abstract

“…It is worth mentioning that testing NSGA‐II based optimization algorithms, like PESTPP‐MOU 55 or the constrained multiobjective Bayesian optimization method 56 for the same problem or more complex cases can give us important insights about the best optimization approach for such problems in terms of accuracy and performance.…”

Section: Optimization Approachmentioning

confidence: 99%

Controlling potential far‐field brine leakage from CO₂ storage formations using deep extraction wells: Numerical and experimental testing

Askar¹,

Illangasekare

2022

Greenhouse Gases

View full text Add to dashboard Cite

Injecting CO 2 into deep geologic formations for storage purposes induces large pressure build-up that risks caprock integrity. Naturally occurring faults or pressure-induced fractures in the caprock can act as conductive leakage pathways resulting in potential contamination of the overlying shallow aquifers. Previous studies explored using brine extraction to manage such elevated pressure in the storage formation. In this paper, we extended the use of this technique to control far-field brine leakage. Extraction wells are placed in the storage zone to reduce the leakage flow through reversing the pressure-gradient locally, while minimizing the brine concentrations in the escaped-fraction by utilizing the dilution capacity of the overlying formations. The developed approach incorporated the Genetic Algorithm with transport model simulations to optimize well-placements and extraction-rates. An approximately 8m long intermediate-scale tank designed to mimic brine leakage migration in the field was used to validate this approach as field data are not available. We further evaluated the approach numerically using a hypothetical leakage scenario at the Vedder storage formation in San-Joaquin basin to assess its practicality for field implementation. The results showed that storage zone heterogeneity and fractures' permeabilities can significantly affect the optimum locations and pumping rates of the extraction wells. Brine leakage can be controlled by extracting a native-brine volume less than 50% of the injected CO 2 volume. The target concentrations in the shallow aquifer determines the extraction rates required to control a leakage through a fracture or a buried thrust fault. The study is useful to develop remediation strategies for carbon storage operations.

show abstract

A model-independent tool for evolutionary constrained multi-objective optimization under uncertainty

Cited by 11 publications

References 43 publications

Particle tracking as a vulnerability assessment tool for drinking water production

Particle tracking as a vulnerability assessment tool for drinking water production

An ensemble-based approach for pumping optimization in an island aquifer considering parameter, observation and climate uncertainty

Controlling potential far‐field brine leakage from CO₂ storage formations using deep extraction wells: Numerical and experimental testing

Contact Info

Product

Resources

About

A model-independent tool for evolutionary constrained multi-objective optimization under uncertainty

Cited by 11 publications

References 43 publications

Particle tracking as a vulnerability assessment tool for drinking water production

Particle tracking as a vulnerability assessment tool for drinking water production

An ensemble-based approach for pumping optimization in an island aquifer considering parameter, observation and climate uncertainty

Controlling potential far‐field brine leakage from CO2 storage formations using deep extraction wells: Numerical and experimental testing

Contact Info

Product

Resources

About

Controlling potential far‐field brine leakage from CO₂ storage formations using deep extraction wells: Numerical and experimental testing