Jeroen Snippe
Research Subjects
Research Interests
Jeroen Snippe has a total of 40 co-authors. Here are the top co-authors by number of shared publications:
Anirban Chandra
Shell Information Technology International Inc
3 shared publicationsAniruddha Panda
Shell India Markets Pvt. Ltd
3 shared publications- Nathaniel Forbes Inskip
The Lyell Centre, Heriot-Watt University
3 shared publications Niko Kampman
3 shared publicationsAndreas Busch
2 shared publications- Detlef Hohl
Shell Information Technology International Inc
2 shared publications
Latest Publications
- 18 February 2026
- Published by Society of Petroleum Engineers (SPE) in SPE Journal
- Vol. 31 (04) , 2174-2192
Quantifying the Uncertainty of the 3D Evolution of CO2-Mass and Pressure Plumes for Site Screening Applications with Physics-Informed Fourier Neural Operators
Summary: Identification of the most suitable set of target storage sites in terms of capacity and containment early in the project phase is one of the key elements for successful carbon dioxide (CO2) sequestration. There is a need for rapid approximate models that can forecast the migration of CO2-mass and pressure plumes. One family of surrogate models that has proved effective for simulating CO2 plume migration is Fourier neural operator (FNO)-based techniques. We discuss the development of two FNO architectures, standard FNO (3DFNO) and U-Net augmented FNO (U-FNO), to serve as fast simulation engines in a CO2 storage site screening platform using deep neural networks and operator learning. We also discuss various physics embedding training strategies (e.g., mass conservation) for improving the reliability of these artificial intelligence simulations. We provide computational results in 3D plume-migration evolution problems in terms of training costs and prediction accuracy for the 3DFNO and U-FNO protocols. We focus our investigations on layered geology problems typical of the screening stage. In inference mode, both models offer O(105) computational acceleration with respect to conventional reservoir simulation subject to an acceptable level of compromise in accuracy. While U-FNO incurs approximately 50% higher computational costs compared with 3DFNO, it improves the prediction accuracy. We demonstrate that U-FNO yields more accurate predictions than 3DFNO in layered systems for CO2-mass accumulation and CO2-rich phase (gas) saturation, while 3DFNO yields more accurate predictions for evolution of pressure change. The platform over which the FNO engines are deployed is designed for predicting the 3D spatiotemporal evolution of CO2-mass accumulation, gas phase saturation, and pressure in arbitrarily layered systems. The fast simulation platform enables Monte Carlo sampling to compute statistical distribution functions of plume-migration performance indicators. Fast FNO-based surrogate models thus enable a previously impossible Monte Carlo–based uncertainty quantification workflow for predicting CO2-mass plume-migration distance and post-injection pressure evolution.
- 18 February 2026
- Published by Society of Petroleum Engineers (SPE) in SPE Journal
- Vol. 31 (04) , 2174-2192
Quantifying the Uncertainty of the 3D Evolution of CO2-Mass and Pressure Plumes for Site Screening Applications with Physics-Informed Fourier Neural Operators
Summary: Identification of the most suitable set of target storage sites in terms of capacity and containment early in the project phase is one of the key elements for successful carbon dioxide (CO2) sequestration. There is a need for rapid approximate models that can forecast the migration of CO2-mass and pressure plumes. One family of surrogate models that has proved effective for simulating CO2 plume migration is Fourier neural operator (FNO)-based techniques. We discuss the development of two FNO architectures, standard FNO (3DFNO) and U-Net augmented FNO (U-FNO), to serve as fast simulation engines in a CO2 storage site screening platform using deep neural networks and operator learning. We also discuss various physics embedding training strategies (e.g., mass conservation) for improving the reliability of these artificial intelligence simulations. We provide computational results in 3D plume-migration evolution problems in terms of training costs and prediction accuracy for the 3DFNO and U-FNO protocols. We focus our investigations on layered geology problems typical of the screening stage. In inference mode, both models offer O(105) computational acceleration with respect to conventional reservoir simulation subject to an acceptable level of compromise in accuracy. While U-FNO incurs approximately 50% higher computational costs compared with 3DFNO, it improves the prediction accuracy. We demonstrate that U-FNO yields more accurate predictions than 3DFNO in layered systems for CO2-mass accumulation and CO2-rich phase (gas) saturation, while 3DFNO yields more accurate predictions for evolution of pressure change. The platform over which the FNO engines are deployed is designed for predicting the 3D spatiotemporal evolution of CO2-mass accumulation, gas phase saturation, and pressure in arbitrarily layered systems. The fast simulation platform enables Monte Carlo sampling to compute statistical distribution functions of plume-migration performance indicators. Fast FNO-based surrogate models thus enable a previously impossible Monte Carlo–based uncertainty quantification workflow for predicting CO2-mass plume-migration distance and post-injection pressure evolution.
- 21 January 2026
- Published by Elsevier in International Journal of Greenhouse Gas Control
- Vol. 150
Accelerated CCS site screening using Fourier Neural Operator based surrogates for flow simulations
- 21 January 2026
- Published by Elsevier in International Journal of Greenhouse Gas Control
- Vol. 150
Accelerated CCS site screening using Fourier Neural Operator based surrogates for flow simulations
- Open Access
- 1 January 2025
- Published by Elsevier in SSRN Electronic Journal
Accelerated CCS Site Screening Using Fourier Neural Operator Based Surrogates for Flow Simulations
Large-scale deployment of carbon capture and storage (CCS) technology is imperative for mitigating greenhouse gas emissions to combat the effects of climate change. One of the key elements for the success of CCS is the selection of appropriate sites to store the captured CO2. To achieve this goal, various factors such as plume migration, storage capacity, and containment need to be validated over long-time horizons. Flow simulators, which accurately model the complex interplay of buoyancy, viscous, and capillary forces, are typically used for modeling plume migration, and thereby optimizing storage efficiency subject to subsurface constraints. However, large reservoirs and long-time horizons associated with CCS often render conventional flow simulators computationally expensive and time-consuming to apply. The computational efficiency challenge is further intensified for site screening applications in which typically large ensembles of forward simulations are necessary due to the uncertainty associated with subsurface variables such as formation dip, geological properties, and a plethora of CO2 trapping parameters. We develop a novel FNO-based surrogate modeling framework for the rapid prediction of plume-migration quantified by spatio-temporal evolution of CO2 mass accumulation and saturation fields. The FNO is a neural network architecture that can approximate underlying physics such as multiphase/multicomponent reservoir simulation equations by using large, annotated datasets of input-output pairs, rather than explicitly relying on the underlying PDEs in the optimized loss as in the case of physics-informed neural networks (PINNs). Finite-volume method based subsurface fluid flow simulation is used to generate a comprehensive data set for a wide range of realistic subsurface parameters that are encountered in CCS sites globally. The outcome of the surrogate model is spatial distributions of CO2 accumulation and saturation for multiple time snapshots from end-of-injection to the end of a long-term migration period (1000 years post end-of-injection). To evaluate the performance of FNO-based surrogate models in the context of CCS site screening applications, we propose a range of novel physics-based metrics, such as plume distance to the injection location. Results on a set of blind data show very comparable accuracy results vis-à-vis numerical simulation. We also obtain three to four orders of magnitude computational acceleration with trained FNO-based surrogate models with minimal sacrifice of prediction accuracy of CO2 mass accumulation and saturation. The FNO based surrogate model is designed to accelerate large-scale deployment of CCS via more rapid and effective quantification of subsurface uncertainties for site screening purposes.
- 1 January 2025
- Published by Elsevier in SSRN Electronic Journal
Open AccessAccelerated CCS Site Screening Using Fourier Neural Operator Based Surrogates for Flow Simulations
Large-scale deployment of carbon capture and storage (CCS) technology is imperative for mitigating greenhouse gas emissions to combat the effects of climate change. One of the key elements for the success of CCS is the selection of appropriate sites to store the captured CO2. To achieve this goal, various factors such as plume migration, storage capacity, and containment need to be validated over long-time horizons. Flow simulators, which accurately model the complex interplay of buoyancy, viscous, and capillary forces, are typically used for modeling plume migration, and thereby optimizing storage efficiency subject to subsurface constraints. However, large reservoirs and long-time horizons associated with CCS often render conventional flow simulators computationally expensive and time-consuming to apply. The computational efficiency challenge is further intensified for site screening applications in which typically large ensembles of forward simulations are necessary due to the uncertainty associated with subsurface variables such as formation dip, geological properties, and a plethora of CO2 trapping parameters. We develop a novel FNO-based surrogate modeling framework for the rapid prediction of plume-migration quantified by spatio-temporal evolution of CO2 mass accumulation and saturation fields. The FNO is a neural network architecture that can approximate underlying physics such as multiphase/multicomponent reservoir simulation equations by using large, annotated datasets of input-output pairs, rather than explicitly relying on the underlying PDEs in the optimized loss as in the case of physics-informed neural networks (PINNs). Finite-volume method based subsurface fluid flow simulation is used to generate a comprehensive data set for a wide range of realistic subsurface parameters that are encountered in CCS sites globally. The outcome of the surrogate model is spatial distributions of CO2 accumulation and saturation for multiple time snapshots from end-of-injection to the end of a long-term migration period (1000 years post end-of-injection). To evaluate the performance of FNO-based surrogate models in the context of CCS site screening applications, we propose a range of novel physics-based metrics, such as plume distance to the injection location. Results on a set of blind data show very comparable accuracy results vis-à-vis numerical simulation. We also obtain three to four orders of magnitude computational acceleration with trained FNO-based surrogate models with minimal sacrifice of prediction accuracy of CO2 mass accumulation and saturation. The FNO based surrogate model is designed to accelerate large-scale deployment of CCS via more rapid and effective quantification of subsurface uncertainties for site screening purposes.
- Top Cited PapersOpen Access
- 6 March 2024
- Published by Elsevier in International Journal of Greenhouse Gas Control
- Vol. 133
Modelling geological CO2 leakage: Integrating fracture permeability and fault zone outcrop analysis
- 6 March 2024
- Published by Elsevier in International Journal of Greenhouse Gas Control
- Vol. 133
Top Cited PapersOpen AccessModelling geological CO2 leakage: Integrating fracture permeability and fault zone outcrop analysis
- Open Access
- 28 January 2024
- Published by Elsevier in International Journal of Greenhouse Gas Control
- Vol. 132
Viscous fingering in CCS - A general criterion for viscous fingering in porous media
- 28 January 2024
- Published by Elsevier in International Journal of Greenhouse Gas Control
- Vol. 132
Open AccessViscous fingering in CCS - A general criterion for viscous fingering in porous media
- Open Access
- 27 April 2022
- Published by Elsevier in International Journal of Greenhouse Gas Control
- Vol. 118
Modelling of long-term along-fault flow of CO2 from a natural reservoir
- 27 April 2022
- Published by Elsevier in International Journal of Greenhouse Gas Control
- Vol. 118
Open AccessModelling of long-term along-fault flow of CO2 from a natural reservoir
- Open Access
- 31 March 2021
- Published by Elsevier in SSRN Electronic Journal
Modelling of Long-term Along-fault Flow of CO2 From a Natural Reservoir
Geological sequestration of CO2 requires the presence of at least one competent seal above the storage reservoir to ensure containment of the stored CO2. Most of the considered storage sites are overlain by low-permeability evaporites or mudrocks that form competent seals in the absence of defects. Potential defects are formed by man-made well penetrations (necessary for exploration and appraisal, and injection) as well as (for mudrocks) natural or injection-induced fracture systems through the caprock. These defects need to be de-risked during site selection and characterisation. A European ACT-sponsored research consortium, DETECT, developed an integrated characterisation and risk assessment toolkit for natural fault/fracture pathways. In this paper we describe the DETECT experimental-modelling workflow, which aims to be predictive for fault-related leakage quantification, and its application to a field case example for validation. The workflow combines laboratory experiments to obtain single-fracture stress-sensitive permeabilities; single-fracture modelling for stress-sensitive relative permeabilities and capillary pressures; fracture network characterisation and modelling for the caprock(s); upscaling of properties and constitutive functions in fracture networks; and full compositional flow modelling at field scale. We focus the paper on the application of the workflow to the Green River Site in Utah. This is a rare case of leakage from a natural CO2 reservoir, where CO2 (dissolved or gaseous) migrates along two fault zones to the surface. This site provides a unique opportunity to understand CO2 leakage mechanisms and volumes along faults, because of its extensive characterisation including a large dataset of present-day CO2 surface flux measurements as well as historical records of CO2 leakage in the form of travertine mounds. When applied to this site, our methodology predicts leakage locations accurately and, within an order of magnitude, leakage rates correctly without extensive history matching. Subsequent history matching achieves accurate leak rate matches within a-priori uncertainty ranges for model input parameters.
- 31 March 2021
- Published by Elsevier in SSRN Electronic Journal
Open AccessModelling of Long-term Along-fault Flow of CO2 From a Natural Reservoir
Geological sequestration of CO2 requires the presence of at least one competent seal above the storage reservoir to ensure containment of the stored CO2. Most of the considered storage sites are overlain by low-permeability evaporites or mudrocks that form competent seals in the absence of defects. Potential defects are formed by man-made well penetrations (necessary for exploration and appraisal, and injection) as well as (for mudrocks) natural or injection-induced fracture systems through the caprock. These defects need to be de-risked during site selection and characterisation. A European ACT-sponsored research consortium, DETECT, developed an integrated characterisation and risk assessment toolkit for natural fault/fracture pathways. In this paper we describe the DETECT experimental-modelling workflow, which aims to be predictive for fault-related leakage quantification, and its application to a field case example for validation. The workflow combines laboratory experiments to obtain single-fracture stress-sensitive permeabilities; single-fracture modelling for stress-sensitive relative permeabilities and capillary pressures; fracture network characterisation and modelling for the caprock(s); upscaling of properties and constitutive functions in fracture networks; and full compositional flow modelling at field scale. We focus the paper on the application of the workflow to the Green River Site in Utah. This is a rare case of leakage from a natural CO2 reservoir, where CO2 (dissolved or gaseous) migrates along two fault zones to the surface. This site provides a unique opportunity to understand CO2 leakage mechanisms and volumes along faults, because of its extensive characterisation including a large dataset of present-day CO2 surface flux measurements as well as historical records of CO2 leakage in the form of travertine mounds. When applied to this site, our methodology predicts leakage locations accurately and, within an order of magnitude, leakage rates correctly without extensive history matching. Subsequent history matching achieves accurate leak rate matches within a-priori uncertainty ranges for model input parameters.
- 27 March 2012
- Published by EAGE Publications bv
Reactive Transport Modeling for CO2 Storage in a Depleted North Sea Gas Field
- 27 March 2012
- Published by EAGE Publications bv
Reactive Transport Modeling for CO2 Storage in a Depleted North Sea Gas Field
- 1 April 1994
- Published by Elsevier in Nuclear Physics B - Proceedings Supplements
- Vol. 34, 222-227
On the top of the energy barrier
- 1 April 1994
- Published by Elsevier in Nuclear Physics B - Proceedings Supplements
- Vol. 34, 222-227
On the top of the energy barrier