Search NASA⌕ Search

SEARCH · Search NASA

Results for “Computing methodologies”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,657 records · Page 92

Predictions of Transient Flame Lift-Off Length With Comparison to Single-Cylinder Optical Engine Experiments

A state-of-the-art, grid-convergent simulation methodology was applied to three-dimensional calculations of a single-cylinder optical engine. A mesh resolution study on a sector-based version of the engine geometry further verified the RANS-based cell size recommendations previously presented by Senecal et al. (“Grid Convergent Spray Models for Internal Combustion Engine CFD Simulations,” ASME Paper No. ICEF2012-92043). Convergence of cylinder pressure, flame lift-off length, and emissions was achieved for an adaptive mesh refinement cell size of 0.35 mm. Furthermore, full geometry simulations, using mesh settings derived from the grid convergence study, resulted in excellent agreement with measurements of cylinder pressure, heat release rate, and NOx emissions. On the other hand, the full geometry simulations indicated that the flame lift-off length is not converged at 0.35 mm for jets not aligned with the computational mesh. Further simulations suggested that the flame lift-off lengths for both the nonaligned and aligned jets appear to be converged at 0.175 mm. With this increased mesh resolution, both the trends and magnitudes in flame lift-off length were well predicted with the current simulation methodology. Good agreement between the overall predicted flame behavior and the available chemiluminescence measurements was also achieved. Our present study indicates that cell size requirements for accurate prediction of full geometry flame lift-off lengths may be stricter than those for global combustion behavior. This may be important when accurate soot predictions are required.

42 ENGINEERING↗

Activities of the OECD/NEA Expert Group on Assay Data for Spent Nuclear Fuel

Management of spent nuclear fuel is a key issue for many NEA member countries. In nuclear criticality safety, the decision of many countries to advance burnup credit as part of their licensing strategy has heightened recent interest in experimental data needed to validate computer codes used in burnup credit calculations. This paper discusses recent activities of an Expert Group on assay data, formed under the OECD/NEA/NSC/WPNCS (Working Party on Nuclear Criticality Safety) to help coordinate isotopic assay data activities and facilitate international collaboration between NEA member countries developing or implementing burnup credit methodologies. Recent activities of the Expert Group are described, focusing on the planned expansion of the Spent Fuel Isotopic Composition Database (SFCOMPO), and preparation of a state-of-the-art report on assay data that includes sections on recommended radiochemical analysis methods, techniques, and lessons learned from previous experiments.

07 ISOTOPE AND RADIATION SOURCES↗

Exploratory Study of 4D versus 3D Robust Optimization in Intensity Modulated Proton Therapy for Lung Cancer

Purpose: The purpose of this study was to compare the impact of uncertainties and interplay on 3-dimensional (3D) and 4D robustly optimized intensity modulated proton therapy (IMPT) plans for lung cancer in an exploratory methodology study. Methods and Materials: IMPT plans were created for 11 nonrandomly selected non-small cell lung cancer (NSCLC) cases: 3D robustly optimized plans on average CTs with internal gross tumor volume density overridden to irradiate internal target volume, and 4D robustly optimized plans on 4D computed tomography (CT) to irradiate clinical target volume (CTV). Regular fractionation (66 Gy [relative biological effectiveness; RBE] in 33 fractions) was considered. In 4D optimization, the CTV of individual phases received nonuniform doses to achieve a uniform cumulative dose. The root-mean-square dose-volume histograms (RVH) measured the sensitivity of the dose to uncertainties, and the areas under the RVH curve (AUCs) were used to evaluate plan robustness. Dose evaluation software modeled time-dependent spot delivery to incorporate interplay effect with randomized starting phases of each field per fraction. Dose-volume histogram (DVH) indices comparing CTV coverage, homogeneity, and normal tissue sparing were evaluated using Wilcoxon signed rank test. Results: 4D robust optimization plans led to smaller AUC for CTV (14.26 vs 18.61, respectively; P=.001), better CTV coverage (Gy [RBE]) (D{sub 95%} CTV: 60.6 vs 55.2, respectively; P=.001), and better CTV homogeneity (D{sub 5%}-D{sub 95%} CTV: 10.3 vs 17.7, resspectively; P=.002) in the face of uncertainties. With interplay effect considered, 4D robust optimization produced plans with better target coverage (D{sub 95%} CTV: 64.5 vs 63.8, respectively; P=.0068), comparable target homogeneity, and comparable normal tissue protection. The benefits from 4D robust optimization were most obvious for the 2 typical stage III lung cancer patients. Conclusions: Our exploratory methodology study showed that, compared to 3D robust optimization, 4D robust optimization produced significantly more robust and interplay-effect-resistant plans for targets with comparable dose distributions for normal tissues. A further study with a larger and more realistic patient population is warranted to generalize the conclusions.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Hydropower Optimization Using Artificial Neural Network Surrogate Models of a High-Fidelity Hydrodynamics and Water Quality Model

Hydropower operations optimization subject to environmental constraints is limited by challenges associated with dimensionality and spatial and temporal resolution. The need for high-fidelity hydrodynamic and water quality models within optimization schemes is driven by improved computational capabilities, increased requirements to meet specific points of compliance with greater resolution, and the need to optimize operations of not just single reservoirs but systems of reservoirs. This study describes an important advancement for computing hourly power generation schemes for a hydropower reservoir using high-fidelity models, surrogate modeling techniques, and optimization methods. The predictive power of the high-fidelity hydrodynamic and water quality model CE-QUAL-W2 is successfully emulated by an artificial neural network, then integrated into a genetic algorithm optimization approach to maximize hydropower generation subject to constraints on dam operations and water quality. This methodology is applied to a multipurpose reservoir near Nashville, Tennessee, USA. The model successfully reproduced high-fidelity reservoir information while enabling 6.8% and 6.6% increases in hydropower production value relative to actual operations for dissolved oxygen (DO) limits of 5 and 6 mg/L, respectively, while witnessing an expected decrease in power generation at more restrictive DO constraints. Exploration of simultaneous temperature and DO constraints revealed capability to address multiple water quality constraints at specified locations. Here, the reduced computational requirements of the new modeling approach demonstrated an ability to provide decision support for reservoir operations scheduling while maintaining high-fidelity hydrodynamic and water quality information as part of the optimization decision support routines.

13 HYDRO ENERGY↗

Benchmark Tracking System for Performance Monitoring

Benchmarking is essential for high-performance software development, particularly for monitoring performance across code iterations. This project focused on enhancing the benchmarking process for Lamellar, an asynchronous runtime for High-Performance Computing (HPC) systems developed at Pacific Northwest National Laboratory. Prior to this work, benchmark results were difficult to track and compare across code versions, presenting significant challenges in identifying performance regressions and long-term trends. The primary objective was to establish a systematic, reproducible approach for measuring performance and detecting regressions following code commits. Our methodology involved three key components: standardizing benchmark outputs, implementing data versioning, and developing analysis tools. We standardized the benchmark output format to JSON Line records containing specific fields (execution time, hardware specifications, and environmental variables). To address data management challenges, we evaluated several options and eventually chose a git repository dedicated to benchmark data. We developed a suite of Python tools that processed benchmark results, enriched them with metadata, and facilitated search in the repository. The resulting system enables more efficient filtering and comparison of performance metrics across commit histories, hardware configurations, and benchmark variants through a unified query interface. Our implementation reduces computational overhead by first checking for existing results through configuration matching before initiating new benchmark runs, thereby conserving resources. The system has been validated by Lamellar developers. It organizes results by benchmark type and build configurations for efficient retrieval. Future developments include a planned Large Language Model interface for predicting benchmark performance, incorporating the criterion package for statistical analysis, which will enable automated detection of statistically significant performance changes, and integration with continuous integration pipelines. Despite these enhancements being reserved for future work, this project has successfully provided the Lamellar development team with a framework for maintaining consistent performance standards and identifying optimization opportunities across workloads and hardware environments.

97 MATHEMATICS AND COMPUTING↗

Multigrid Reduction in Time for Chaotic and Hyperbolic Problems (Final Report)

The coming massive parallelism of exascale computing presents a pressing challenge for the many DOE simulations of time-dependent partial differential equations (PDEs), which typically use traditional sequential time stepping methods. Since this traditional approach is inherently serial, it presents a sequential bottleneck when moving to exascale computing, because future performance gains will come through greater concurrency, not faster clock speeds. Thus, the goal of this work is to research parallelism in time, i.e., methods that compute multiple time values simultaneously, not sequentially. The focus will be on hyperbolic and chaotic problems of interest to DOE, with the goal of enabling scalable simulations of time-dependent hyperbolic and chaotic problems on future architectures. The chosen methodology for solving these problems parallel-in-time is multigrid, because multigrid (when it works) is a powerful, optimal, and scalable solver for discretized PDEs. Multigrid is already commonly used in many DOE simulations for scalably and optimally solving space-only PDE problems. The areas of hyperbolic and chaotic problems are chosen because of their relevance to problems of programmatic interest to DOE. However, these problems are also well-known to be difficult for parallel-in-time methods, with the most common method, parareal, diverging in many cases. The current state of-the-art for parallel-in-time at LLNL is the multigrid reduction in time (MGRIT) XBraid package, which also struggles for such problems, while still showing some improvement over parareal. In summary, new methods are needed for an efficient parallel-in-time scheme for hyperbolic and chaotic problems, and this work shall research promising new multigrid methods in this area. In particular, this work shall continue researching the directions from the current collaboration with Dr. Falgout, which are laid out in the work Toward Parallel in Time for Chaotic Dynamical Systems and showed the first known results of a parallel-in-time speedup for a chaotic problem. This work outlines two key improvements to XBraid for chaotic problems, the so-called “theta” and “delta-correction” methods. Here, these two improvements will be further researched and improved (including with a new relaxation method inspired by on Least Squares Shadowing (LSS)) and explored for more complicated problems.

97 MATHEMATICS AND COMPUTING↗

The CMSN team working on computational-design-Fe-based-superconductors (Final Report)

This CMSN collaboration comprises four nodes. Rutgers (G. Kotliar K. Haule) , U Arizona(M. Van Schilfgaarde), UCDavis (S. Savrasov W. Pickett) and Ames National Labs-Iowa State University (J. Schmalian- K.M. Ho- V. Antropov). The goal of this CMSN was to develop an unified methodology for electronic structure calculations to estimate from first principles the physical properties of iron pnictides which should serve as a test of the methodologies and the codes. This goals requires the integration of methodologies to treat electron electron interactions ( which can be treated by DMFT GW and Gutzwiller methods ) and electron phonon interactions and magnetic interactions (to be treated in linear response) which were developed by the different members of this CMSN team. I summarize below the collaborative work carried out until now below stressing the difficult but all important task of joint code development. In this area significant advances have taken place during the last year. These include a) an integration of the DMFT code of the Rutgers group with the Gutzwiller codes developed in Iowa. The result is the best available code to carry out LDA+G and is having impact on correlated materials with f electrons. and b) A Wannier interface between the QP GW code developed in Arizona and the Rutgers DMFT codes. These developments will be continued and will open new capabilities to treat correlated materials in general, and will be applied to the iron pnictides in the final year of this CMSN.

Kotliar, Gabriel↗

Coupled Physics Environment (CouPE) library - Design, Implementation, and Release

Over several years, high fidelity, validated mono-­physics solvers with proven scalability on peta-­scale architectures have been developed independently. Based on a unified component-­based architecture, these existing codes can be coupled with a unified mesh-­data backplane and a flexible coupling-­strategy-­based driver suite to produce a viable tool for analysts. In this report, we present details on the design decisions and developments on CouPE, an acronym that stands for Coupled Physics Environment that orchestrates a coupled physics solver through the interfaces exposed by MOAB array-­based unstructured mesh, both of which are part of SIGMA (Scalable Interfaces for Geometry and Mesh-­Based Applications) toolkit. The SIGMA toolkit contains libraries that enable scalable geometry and unstructured mesh creation and handling in a memory and computationally efficient implementation. The CouPE version being prepared for a full open-­source release along with updated documentation will contain several useful examples that will enable users to start developing their applications natively using the native MOAB mesh and couple their models to existing physics applications to analyze and solve real world problems of interest. An integrated multi-­physics simulation capability for the design and analysis of current and future nuclear reactor models is also being investigated as part of the NEAMS RPL, to tightly couple neutron transport, thermal-­hydraulics and structural mechanics physics under the SHARP framework. This report summarizes the efforts that have been invested in CouPE to bring together several existing physics applications namely PROTEUS (neutron transport code), Nek5000 (computational fluid-dynamics code) and Diablo (structural mechanics code). The goal of the SHARP framework is to perform fully resolved coupled physics analysis of a reactor on heterogeneous geometry, in order to reduce the overall numerical uncertainty while leveraging available computational resources. The design of CouPE along with motivations that led to implementation choices are also discussed. The first release of the library will be different from the current version of the code that integrates the components in SHARP and explanation on the need for forking the source base will also be provided. Enhancements in the functionality and improved user guides will be available as part of the release. CouPE v0.1 is scheduled for an open-­source release in December 2014 along with SIGMA v1.1 components that provide support for language-agnostic mesh loading, traversal and query interfaces along with scalable solution transfer of fields between different physics codes. The coupling methodology and software interfaces of the library are presented, along with verification studies on two representative fast sodium-­cooled reactor demonstration problems to prove the usability of the CouPE library.

97 MATHEMATICS AND COMPUTING↗

Nuclear Data Adjustment with Whisper for Criticality Safety Applications [Slides]

Nuclear data adjustment is used by Whisper to compute residual nuclear data uncertainties within a full suite of benchmarks; in order to do so, a complete set of nuclear data covariances, benchmarks and k-effective sensitivity profiles (for both benchmarks and applications) is required. For applications that are similar to the full benchmark suite, the residual uncertainties add a small margin to the overall USL calculation. For applications that are dissimilar to the full benchmark suite, the residual uncertainties add a reasonable margin to the overall USL calculation (more conservative). This methodology helps to overcome any issues in the released covariance libraries where nuclear data uncertainties appear to be too large.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Enhanced mesoscale climate projections in TAR and AR5 IPCC scenarios: a case study in a Mediterranean climate (Araucanía Region, south central Chile)

Here, climate change scenarios are computed on a large scale, not accounting for local variations presented in historical data and related to human scale. Based on historical records, we validate a baseline (1962–1990) and correct the bias of A2 and B2 regional projections for the end of twenty-first century (2070–2100) issued from a high resolution dynamical downscaled (using PRECIS mesoscale model, hereinafter DGF-PRECIS) of Hadley GCM from the IPCC 3rd Assessment Report (TAR). This is performed for the Araucanía Region (Chile; 37°–40°S and 71°–74°W) using two different bias correction methodologies. Next, we study high-resolution precipitations to find monthly patterns such as seasonal variations, rainfall months, and the geographical effect on these two scenarios. Finally, we compare the TAR projections with those from the recent Assessment Report 5 (AR5) to find regional precipitation patterns and update the Chilean `projection. To show the effects of climate change projections, we compute the rainfall climatology for the Araucanía Region, including the impact of ENSO cycles (El Niño and La Niña events). The corrected climate projection from the high-resolution dynamical downscaled model of the TAR database (DGF-PRECIS) show annual precipitation decreases: B2 (-19.19 %, -287 ± 42 mm) and A2 (-43.38 %, -655 ± 27.4 mm per year. Furthermore, both projections increase the probability of lower rainfall months (lower than 100 mm per month) to 64.2 and 72.5 % for B2 and A2, respectively.

54 ENVIRONMENTAL SCIENCES↗

Multigrid Reduction in Time for Chaotic and Hyperbolic Problems (Final Report)

The coming massive parallelism of exascale computing presents a pressing challenge for the many DOE simulations of time-dependent partial differential equations (PDEs), which typically use traditional sequential time stepping methods. Since this traditional approach is inherently serial, it presents a sequential bottleneck when moving to exascale computing, because future performance gains will come through greater concurrency, not faster clock speeds. Thus, the goal of this work is to research parallelism in time, i.e., methods that compute multiple time values simultaneously, not sequentially. The focus will be on hyperbolic and chaotic problems of interest to DOE, with the goal of enabling scalable simulations of time-dependent hyperbolic and chaotic problems on future architectures. The chosen methodology for solving these problems parallel-in-time is multigrid, because multigrid (when it works) is a powerful, optimal, and scalable solver for discretized PDEs. Multigrid is already commonly used in many DOE simulations for scalably and optimally solving space-only PDE problems. The areas of hyperbolic and chaotic problems are chosen because of their relevance to problems of programmatic interest to DOE. However, these problems are also well-known to be difficult for parallelin-time methods, with the most common method, parareal, diverging in many cases. The current state-of-the-art for parallel-in-time at LLNL is the multigrid reduction in time (MGRIT) XBraid package, which also struggles for such problems, while still showing some improvement over parareal. In summary, new methods are needed for an efficient parallel-in-time scheme for hyperbolic and chaotic problems, and this work shall research promising new multigrid methods in this area. In particular, this work shall continue researching the directions from the current collaboration with Dr. Falgout, which are laid out in the work Toward Parallel in Time for Chaotic Dynamical Systems and showed the first known results of a parallel-in-time speedup for a chaotic problem. This work outlines two key improvements to XBraid for chaotic problems, the so-called “theta” and “delta-correction” methods. Here, these two improvements will be implemented in a high-performance but general way in XBraid and explored for more complicated problems. We will additionally research, as time allows, improvements to these techniques, as well as multigrid relaxation techniques based on Least Squares Shadowing (LSS by Wang) and a nonintrusive block tridiagonal solver based on MGRIT, called TriMGRIT.

97 MATHEMATICS AND COMPUTING↗

Bias Correction and Statistical Downscaling of Future Solar Irradiance Projections Using the NSRDB

Assessing renewable energy resources under future climate scenarios has been highlighted to understand potential impacts of future climate change in renewable generation on the power sector. Climate model projection has been recognized by the renewable energy community as a useful data set to analyze the impacts of future climate change on renewable resources. However, future climate projections generated from general circulation models (GCMs) contain inherent biases that need to be corrected for accurate analysis of future projections of climate variables. In addition, the coarse spatiotemporal resolution of GCMs needs to be improved for regional climate studies. In this work, we develop statistical methods to downscale future projections of global horizontal irradiance (GHI) in a computationally efficient way. Our approach builds statistical downscaling models that correct bias of climate projection of GHI and downscale the future GHI projection from daily-scale to hourly-scale. The National Solar Radiation Database (NSRDB) is used to calibrate the statistical models and validate the downscaled GHI projections across the contiguous United State (CONUS). Preliminary results show that the statistical approach efficiently downscales climate projections of GHI with a nBIAS of 3%, nMAE of 34 % and nRMSE of 46% calculated against NSRDB for CONUS. This study describes the implemented methodology and initial results as well as future research to create high-resolution climate data sets for solar energy applications.

analytical models↗

Input-Output Model for MACCS Nuclear Accident Imp Acts Estimation

Since the original economic model for MACCS was developed, better quality economic data (as well as the tools to gather and process it) and better computational capabilities have become available. The update of the economic impacts component of the MACCS legacy model will provide improved estimates of business disruptions through the use of Input-Output based economic impact estimation. This paper presents an updated MACCS model, bases on Input-Output methodology, in which economic impacts are calculated using the Regional Economic Accounting analysis tool (REAcct) created at Sandia National Laboratories. This new GDP-based model allows quick and consistent estimation of gross domestic product (GDP) losses due to nuclear power plant accidents. This paper outlines the steps taken to combine the REAcct Input-Output-based model with the MACCS code, describes the GDP loss calculation, and discusses the parameters and modeling assumptions necessary for the estimation of long-term effects of nuclear power plant accidents.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Solar Power Ramp Events Detection Using an Optimized Swinging Door Algorithm: Preprint

Solar power ramp events (SPREs) are those that significantly influence the integration of solar power on non-clear days and threaten the reliable and economic operation of power systems. Accurately extracting solar power ramps becomes more important with increasing levels of solar power penetrations in power systems. In this paper, we develop an optimized swinging door algorithm (OpSDA) to detection. First, the swinging door algorithm (SDA) is utilized to segregate measured solar power generation into consecutive segments in a piecewise linear fashion. Then we use a dynamic programming approach to combine adjacent segments into significant ramps when the decision thresholds are met. In addition, the expected SPREs occurring in clear-sky solar power conditions are removed. Measured solar power data from Tucson Electric Power is used to assess the performance of the proposed methodology. OpSDA is compared to two other ramp detection methods: the SDA and the L1-Ramp Detect with Sliding Window (L1-SW) method. The statistical results show the validity and effectiveness of the proposed method. OpSDA can significantly improve the performance of the SDA, and it can perform as well as or better than L1-SW with substantially less computation time.

14 SOLAR ENERGY↗

Inductive predictions of hydrologic events using a Long Short-Term Memory network and the Soil and Water Assessment Tool

We present machine learning methods to predict hydrologic features such as streamflow and soil moisture from spatially and temporally varying hydrological and meteorological data. Here, we used a temporal reduction technique to reduce computation and memory requirements and trained a Long Short-Term Memory (LSTM) network to predict soil moisture and streamflow over multiple watersheds. We show LSTM networks can be trained in a fraction of the time required by complex process-based and attention-based models such as Soil and Water Assessment Tool (SWAT) and GeoMAN without sacrificing accuracy. We also demonstrate that outside data - sourced from a watershed other than the target - can be used to train LSTM to comparable or even superior prediction accuracy. The success of LSTM in such spatially-inductive settings shows hydrologic features can be predicted with minimal prior knowledge of the watershed in question. Finally, we make all methodologies of this work publicly available as an end-to-end software pipeline that facilitates rapid prototyping of hydrologic learners.

97 MATHEMATICS AND COMPUTING↗

Memory Errors in Modern Systems: The Good, The Bad, and The Ugly

Several recent publications have shown that hardware faults in the memory subsystem are commonplace. These faults are predicted to become more frequent in future systems that contain orders of magnitude more DRAM and SRAM than found in current memory subsystems. These memory subsystems will need to provide resilience techniques to tolerate these faults when deployed in high-performance computing systems and data centers containing tens of thousands of nodes. Therefore, it is critical to understand the efficacy of current hardware resilience techniques to determine whether they will be suitable for future systems. In this paper, we present a study of DRAM and SRAM faults and errors from the field. We use data from two leadership-class high-performance computer systems to analyze the reliability impact of hardware resilience schemes that are deployed in current systems. Our study has several key findings about the efficacy of many currently deployed reliability techniques such as DRAM ECC, DDR address/command parity, and SRAM ECC and parity. We also perform a methodological study, and find that counting errors instead of faults, a common practice among researchers and data center operators, can lead to incorrect conclusions about system reliability. Lastly, we use our data to project the needs of future large-scale systems. We find that SRAM faults are unlikely to pose a significantly larger reliability threat in the future, while DRAM faults will be a major concern and stronger DRAM resilience schemes will be needed to maintain acceptable failure rates similar to those found on today's systems.

97 MATHEMATICS AND COMPUTING↗

Positron Emission Tomography in Water Resources and Subsurface Energy Resources Engineering Research

Recent studies have demonstrated that positron emission tomography (PET) is a valuable tool for in-situ characterization of fluid transport in porous and fractured geologic media at the laboratory scale. While PET imaging is routinely used for clinical cancer diagnosis and preclinical medical research—and therefore imaging facilities are available at most research institutes—widespread adoption for applications in water resources and subsurface energy resources engineering have been limited by real and perceived challenges of working with this technique. In this study we discuss and address these challenges, and provide detailed analysis highlighting how positron emission tomography can complement and improve laboratory characterization of different subsurface fluid transport problems. The physics of PET are reviewed to provide a fundamental understanding of the sources of noise, resolution limits, and safety considerations. We then layout the methodology required to perform laboratory experiments imaged with PET, including a new protocol for radioactivity dosing optimization for imaging in geologic materials. Signal-to-noise and sensitivity analysis comparisons between PET and clinical X-ray computed tomography are performed to highlight how PET data can complement more traditional characterization methods, particularly for solute transport problems. Lastly, prior work is critically reviewed and discussed to provide a better understanding of the strengths and weakness of PET and how to best utilize PET-derived data for future studies.

02 PETROLEUM↗

Reduced order models for assessing CO 2 impacts in shallow unconfined aquifers

Risk assessment studies of potential CO 2 sequestration projects consider many factors, including the possibility of brine and/or CO 2 leakage from the storage reservoir. Detailed multiphase reactive transport simulations have been developed to predict the impact of such leaks on shallow groundwater quality; however, these simulations are computationally expensive and thus difficult to directly embed in a probabilistic risk assessment analysis. Here we present a process for developing computationally fast reduced-order models which emulate key features of the more detailed reactive transport simulations. A large ensemble of simulations that take into account uncertainty in aquifer characteristics and CO 2 /brine leakage scenarios were performed. Twelve simulation outputs of interest were used to develop response surfaces (RSs) using a MARS (multivariate adaptive regression splines) algorithm (Milborrow, 2015). A key part of this study is to compare different measures of ROM accuracy. We then show that for some computed outputs, MARS performs very well in matching the simulation data. The capability of the RS to predict simulation outputs for parameter combinations not used in RS development was tested using cross-validation. Again, for some outputs, these results were quite good. For other outputs, however, the method performs relatively poorly. Performance was best for predicting the volume of depressed-pH-plumes, and was relatively poor for predicting organic and trace metal plume volumes. We believe several factors, including the non-linearity of the problem, complexity of the geochemistry, and granularity in the simulation results, contribute to this varied performance. The reduced order models were developed principally to be used in probabilistic performance analysis where a large range of scenarios are considered and ensemble performance is calculated. We demonstrate that they effectively predict the ensemble behavior. But, the performance of the RSs is much less accurate when used to predict time-varying outputs from a single simulation. If an analysis requires only a small number of scenarios to be investigated, computationally expensive physics-based simulations would likely provide more reliable results. Finally, if the aggregate behavior of a large number of realizations is the focus, as will be the case in probabilistic quantitative risk assessment, the methodology presented here is relatively robust.

54 ENVIRONMENTAL SCIENCES↗