Geological Uncertainty Research Articles

Surrogate-Assisted Multiobjective Optimization of a Hydraulically Fractured Well in a Naturally Fractured Shale Reservoir with Geological Uncertainty

Summary Hydraulic fracturing is the most widely used technology for the commercial exploitation of shale-gas reservoirs, yet still faces high cost and development risk caused by many inestimable and uncertain geological parameters in shale reservoirs. Optimally designing the hydraulic fracturing considering the geological uncertainty is an essential topic, and also a computationally expensive problem, which still has not obtained enough investigation. Facing this urgent requirement, a novel surrogate-assisted multiobjective optimization (MOO) framework is proposed for the hydraulically fractured well optimization in shale-gas reservoirs with geological uncertainty. A shale-gas-production numerical model is illustrated using the embedded discrete-fracture model (EDFM), in which a set of stochastic power-law distributed natural-fracture (NF) models are generated to describe the geological uncertainty. To fully grasp the expected economical revenue and the development risk, the MOO strategy is applied, in which the expected value and standard deviation of the net present values (NPVs) are regarded as the objective functions; a Pareto-ranking scheme is adopted to search for the Pareto front (PF) that satisfies both objectives. In addition, a state-of-the-art hybrid multiobjective particle-swarm optimization (HMOPSO) algorithm is developed, which couples the synthetic advantages of multiobjective particle-swarm optimization (MOPSO) and the nondominated sorting genetic algorithm II (NSGA-II) and presents a higher accuracy and efficiency than both of them. Furthermore, a Gaussian-process Kriging model is adopted as the surrogate model to replace the time-consuming numerical simulation caused by the geological uncertainty; it is then integrated with the HMOPSO to construct a surrogated-assisted HMOPSO, which helps to increase the optimization efficiency and accuracy of the computationally expensive problem. Finally, the superiority of the HMOPSO against the traditional algorithms is validated using several benchmark functions, and the surrogated-assisted HMOPSO is performed into a hydraulic-fracture (HF)-well MOO example in a shale-gas reservoir with geological uncertainty. The PF is obtained, which provides a set of tradeoff solutions for designers.

Structural, petrophysical, and geological constraints in potential field inversion using the Tomofast-x v1.0 open-source code

Abstract. The quantitative integration of geophysical measurements with data and information from other disciplines is becoming increasingly important in answering the challenges of undercover imaging and of the modelling of complex areas. We propose a review of the different techniques for the utilisation of structural, petrophysical, and geological information in single physics and joint inversion as implemented in the Tomofast-x open-source inversion platform. We detail the range of constraints that can be applied to the inversion of potential field data. The inversion examples we show illustrate a selection of scenarios using a realistic synthetic data set inspired by real-world geological measurements and petrophysical data from the Hamersley region (Western Australia). Using Tomofast-x's flexibility, we investigate inversions combining the utilisation of petrophysical, structural, and/or geological constraints while illustrating the utilisation of the L-curve principle to determine regularisation weights. Our results suggest that the utilisation of geological information to derive disjoint interval bound constraints is the most effective method to recover the true model. It is followed by model smoothness and smallness conditioned by geological uncertainty and cross-gradient minimisation.

A deep learning-accelerated data assimilation and forecasting workflow for commercial-scale geologic carbon storage

Fast assimilation of monitoring data to update forecasts of pressure buildup and carbon dioxide (CO2) plume migration under geologic uncertainties is a challenging problem in geologic carbon storage. The high computational cost of data assimilation with a high-dimensional parameter space impedes fast decision-making for commercial-scale reservoir management. We propose to leverage physical understandings of porous medium flow behavior with deep learning techniques to develop a fast data assimilation-reservoir response forecasting workflow. Applying an Ensemble Smoother Multiple Data Assimilation (ES-MDA) framework, the workflow updates geologic properties and predicts reservoir performance with quantified uncertainty from pressure history and CO2 plumes interpreted through seismic inversion. As the most computationally expensive component in such a workflow is reservoir simulation, we developed surrogate models to predict dynamic pressure and CO2 plume extents under multi-well injection. The surrogate models employ deep convolutional neural networks, specifically, a wide residual network and a residual U-Net. The workflow is validated against a flat three-dimensional reservoir model representative of a clastic shelf depositional environment. Intelligent treatments are applied to bridge between quantities in a true-3D reservoir model and those in a single-layer reservoir model. The workflow can complete history matching and reservoir forecasting with uncertainty quantification in less than one hour on a mainstream personal workstation.

Multi-Objective Optimization of CO2 Sequestration in Heterogeneous Saline Aquifers under Geological Uncertainty

This paper presents a Pareto-based multi-objective optimization for operating CO2 sequestration with a multi-well system under geological uncertainty; the optimal well allocation, i.e., the optimal allocation of CO2 rates at injection wells, is obtained when there is minimum operation pressure as well as maximum sequestration efficiency. The distance-based generalized sensitivity analysis evaluates the influence of geological uncertainty on the amount of CO2 sequestration through four injection wells at 3D heterogeneous saline aquifers. The spatial properties significantly influencing the trapping volume, in descending order of influence, are mean sandstone porosity, mean sandstone permeability, shale volume ratio, and the Dykstra–Parsons coefficient of permeability. This confirms the importance of storable capacity and heterogeneity in quantitatively analyzing the trapping mechanisms. Multi-objective optimization involves the use of two aquifer models relevant to heterogeneity; one is highly heterogeneous and the other is less so. The optimal well allocations converge to non-dominated solutions and result in a large injection through one specific well, which generates the wide spread of a highly mobile CO2 plume. As the aquifer becomes heterogeneous with a large shale volume and a high Dykstra–Parsons coefficient, the trapping performances of the combined structural and residual sequestration plateau relatively early. The results discuss the effects of spatial heterogeneity on achieving CO2 geological storage, and they provide an operation strategy including multi-objective optimization.

Optimisation of life-of-mine production scheduling for block-caving mines under mineral resource and material mixing uncertainty

ABSTRACT This paper presents a methodology to integrate mineral resource uncertainty from scarce orebody sampling and material mixing uncertainty due to ore flow into the life-of-mine production scheduling of block caving mines. Geological uncertainty is integrated using a Sequential Gaussian Simulation workflow and material mixing uncertainty by simulating material flow using a Cone of Movement concept. A case study is presented in which the uncertainty-based workflow is compared with a deterministic workflow. The stochastic optimisation yields a 4.97% increase in the NPV and a robust production schedule that meets production targets under both sources of uncertainty.

Iterative learning-based many-objective history matching using deep neural network with stacked autoencoder

This paper presents an innovative data-integration that uses an iterative-learning method, a deep neural network (DNN) coupled with a stacked autoencoder (SAE) to solve issues encountered with many-objective history matching. The proposed method consists of a DNN-based inverse model with SAE-encoded static data and iterative updates of supervised-learning data are based on distance-based clustering schemes. DNN functions as an inverse model and results in encoded flattened data, while SAE, as a pre-trained neural network, successfully reduces dimensionality and reliably reconstructs geo-models. The iterative-learning method can improve the training data for DNN by showing the error reduction achieved with each iteration step. The proposed workflow shows the small mean absolute percentage error below 4% for all objective functions, while a typical multi-objective evolutionary algorithm fails to significantly reduce the initial population uncertainty. Iterative learning-based many-objective history matching estimates the trends in water cuts that are not reliably included in dynamic-data matching. This confirms the proposed workflow constructs more plausible geo-models. The workflow would be a reliable alternative to overcome the less-convergent Pareto-based multi-objective evolutionary algorithm in the presence of geological uncertainty and varying objective functions.

3D geo-cellular modeling for Oligocene reservoirs: a marginal field in offshore Vietnam

This study focuses on constructing a 3D geo-cellular model by using well-log data and other geological information to enable a deep investigation of the reservoir characteristics and estimation of the hydrocarbon potential in the clastic reservoir of the marginal field in offshore Vietnam. In this study, Petrel software was adopted for geostatistical modeling. First, a sequential indicator simulation (SIS) was adopted for facies modeling. Next, sequential Gaussian simulation (SGS) and co-kriging approaches were utilized for petrophysical modeling. Furthermore, the results of the petrophysical models were verified by a quality control process before determining the in-place oil for each reservoir in the field. Multiple geological realizations were generated to reduce the geological uncertainty of the model assessment for the facies and porosity model. The most consistent one would then be the best candidate for further evaluation. The porosity distribution ranged from 9 to 22%. The original oil place of clastic reservoirs in the marginal field was 50.28 MMbbl. Ultimately, this research found that the marginal field could be considered a potential candidate for future oil and gas development in offshore Vietnam.

Drilling-Campaign Optimization Using Sequential Information and Policy Analytics

SummaryOptimally designed drilling campaigns are essential for improving oil recovery and value creation. They are required at different stages of the hydrocarbon-field life cycle, including exploration, appraisal, development, and infill. A significant fraction of the revenue risk comes from geological uncertainty, and for this reason, subsurface teams are frequently responsible for optimizing campaign parameters such as the number of wells, the corresponding locations, and the drilling sequence. Companies use the information and learning from drilled wells to adapt the remainder of the campaign, but classical optimization methods do not account for such learning and flexibility over time. Accounting for sequential geological information acquisition and decision making in the optimization of drilling campaigns adds value to the project.We propose a method to optimize drilling campaigns under geological uncertainty by using a sequential-decision model to obtain the optimal drilling policy and applying analytics over the policy to obtain the optimal number of wells and corresponding locations. The novel contribution of policy analytics provides better access to information within the complex data structure of the optimal policy, providing decision support for different decision criteria.The method is demonstrated in two different cases. The first case considers a set of eight candidate wells on predefined locations, mimicking the situation where the method is used after a prior subsurface optimization. The second case considers a set of 12 candidate wells regularly scattered in the same area and uses the method as the first optimization approach to filter out less-attractive regions. Exploiting the geological information on a well-by-well basis improved the expected campaign value by 65% in the first case and by 183% in the second case. The value of spatial geological information and value of flexibility from having more drilling candidates are two byproducts of the method application.

Preliminary Investigations and the Financial Sustainability of Underground Project

Tunnels upgrade connectivity and mobility, with an all-out venture cost that mostly depends on the financial support given by significant lenders. Financial sustainability is related to Risk and Sensitivity Analysis, which in turn depends the level of knowledge of the geology of the project site. The target of this paper is twofold: depict the cycle of i) a well define comprehension of the geo-related model and ii) represent the impact of geo-related uncertainties into the Risk and Sensitivity Analysis. Key elements for assessing the economic sustainability of the investment are here outlined. The undeniable degree of financing preliminary investigations because of high forthright expenses is regular for the vast majority of the underground and tunnel projects. However, there are additional risks such as the geological uncertainty mainly related to the robustness of the preliminary investigations. Significant investment is needed prior to arriving at solid comprehension of the geological model and the connected hazards. Geological uncertainties can be characterized into two significant classes: epistemic and aleatory. Fundamental investigations are normally scant and don’t cover the epistemic vulnerability related to the chosen tunnel alignment or underground facility location. For an underground infrastructure project, the geological uncertainty exerts the highest impact on the project and financial sustainability. Variables must be surveyed and contrasted with the assessed cost in order the project being eligible to some kind of financing. Depending on the result of the risk analysis, solutions can differ: i) withdrawal from the project or restructuration, ii) chances ensures, iii) dividing the risks between parties and phases, and more significant, iv) a powerful undertaking readiness. Last option being the most efficient method for risk mitigation however depends on fundamental examination by specialists. Blended finance presents a risk sharing approach.

Quantitative evaluation of geological uncertainty and its influence on tunnel structural performance using improved coupled Markov chain

The geo-structures embedded in the multiple variable strata could be significantly affected by the geological uncertainty. The quantitative evaluation of geological uncertainty and its influence on the structural safety of embedded tunnels are seldom studied in the past. This paper aims to analyse the effect of geological uncertainty on the structural performance of tunnel using the proposed stochastic geological modelling framework. The geological uncertainty is characterized using an improved coupled Markov chain model based on sparse limited boreholes. A mapping approach is presented to solve the mesh asymmetry problem between the simulated strata and the numerical tunnel model. The tunnel structural performance analysis is then conducted based on the combined model considering the geological uncertainty and tunnel structure. A geological uncertainty index (GUI) is proposed to quantitatively evaluate the level of uncertainty of each borehole and the whole site. The effect of the borehole layout scheme on uncertainty evaluation of factor of safety of tunnel structure is investigated by a large number of stratigraphic realizations. Boreholes collected from Norway with relatively more considerable variability and from Shanghai with relatively more minor variability are adopted as case studies to illustrate the proposed probabilistic analysis framework. The results show that the boreholes with larger GUI values and closer to tunnel locations have a greater weight to affect the embedded tunnel structural performance in uncertain geological strata.

Moraine crest or slope: An analysis of the effects of boulder position on cosmogenic exposure age

Terrestrial cosmogenic nuclide dating of ice-marginal moraines can provide unique insights into Quaternary glacial history. However, pre- and post-depositional exposure histories of moraine boulders can introduce geologic uncertainty to numerical landform ages. To avoid geologic outliers, boulders are typically selected based on their depositional context and individual characteristics but while these criteria have good qualitative reasoning, many have not been tested quantitatively. Of these, boulder location is critical, as boulders located on moraine crests are prioritised, while those on moraine slopes are typically rejected. This study provides the first quantitative assessment of the relative utility of moraine crest and moraine slope sampling using new and published 10Be and 36Cl ages (n = 19) and Schmidt hammer sampling (SH; n = 635 moraine boulders, ∼19,050 SH R-values) in the northern and southern Pyrenees. These data show that for many of the studied moraines, the spatial distribution of “good” boulders is effectively random, with no consistent clustering on moraine crests, ice-proximal or -distal slopes. In turn, and in contrast to prior work, there is no clear penalty to either moraine crest or moraine slope sampling. Instead, we argue that landform stability exerts a greater influence on exposure age distributions than the characteristics of individual boulders. For the studied landforms, post-depositional stability is strongly influenced by sedimentology, with prolonged degradation of matrix-rich unconsolidated moraines while boulder-rich, matrix-poor moraines stabilised rapidly after deposition. While this pattern is unlikely to hold true in all settings, these data indicate that differences between landforms can be more significant than differences at the intra-landform scale. As ad hoc assessment of landform stability is extremely challenging based on geomorphological evidence alone, preliminary SH sampling, as utilised here, is a useful method to assess the temporal distribution of boulder exposure ages and to prioritise individual boulders for subsequent analysis.

Reservoir Characterization, Scenario-Based Models Optimize Development Planning

This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 202273, “Reservoir Characterization and Scenario-Based Modeling To Optimize Development Planning of the Jurassic Plover Formation in the Ichthys Field, North West Shelf of Australia,” by Kazuyuki Yamamoto, SPE, Shuji Yamamoto, and Toby Jones, INPEX, et al., prepared for the 2020 SPE Asia Pacific Oil and Gas Conference and Exhibition, originally scheduled to be held in Perth, Australia, 20–22 October. The paper has not been peer reviewed. The Plover Formation is one of two reservoirs in the Ichthys field of the Australian North West Shelf. The objective of this study is to build multiple scenario-based models to optimize development planning in preparation for the upcoming production phase. The authors have integrated data and interpretations of thin sections, cores, well logs, and seismic data to create multiple geological concepts for the field and to identify key geological uncertainties. Introduction The Ichthys liquefied natural gas project is one of the world’s largest and involves the development of a gas-condensate field in the Browse Basin. The field is approximately 220 km offshore Western Australia and covers an area of approximately 800 km2 with an average water depth of approximately 250 m. The field is currently under preparation for the development of the Plover Formation. The authors conducted an integrated subsurface evaluation and built reservoir models with newly reprocessed 3D seismic data to optimize Plover development planning. Considering the geological uncertainty given the limited production data gathered before the production phase, the multiple deterministic approach was selected as the best option to optimize development planning. In this approach, it is important to capture all geological scenarios that may occur in the field and to build reservoir models in which the scenario concepts are explicitly integrated. A multidisciplinary team was organized to conduct this study. Geology of the Plover Formation The Plover Formation consists of sandstones, shales, igneous rocks, and a minor amount of coal. The depositional environment is fluvial to shallow marine. The average thickness of the formation is approximately 360 m. Based on the stratigraphic correlation anchored by mainly palynological biostratigraphic data, the formation has been subdivided into five stratigraphic members. The formation is overlain by the Ichthys Formation, which is composed mainly of argillaceous sandstone and shale deposited in a tidally influenced, lower-delta-to-shelf environment. Although the nine existing exploration wells broadly cover the entire field, the spacing between the wells is still large (8–10 km). Further-more, core coverage is low, especially in the reservoir sandstone intervals. Igneous rock, predominantly basaltic in composition and likely extrusive in origin, occurs more in the eastern part of the field. Igneous activity has complicated the distribution and connectivity of the reservoir sandstones in the Ichthys field. Key Geological Uncertainties Rock Type (RT). The results of petrographic analysis of rock samples from the Ichthys field and other surrounding fields indicate that the sandstones of the Plover Formation can be separated into multiple RTs: RT1 and RT2, with RT1 subdivided into RT1a and RT1b. Very little data exist from RT2 in the fluvial/distributary channel sandstones in the Ichthys field in comparison with other fields; this is considered to be the result of limited core. The assumption that RT2 exists in the field is critical to prepare for the possibility that a future well may be drilled that might have some nontrivial quantity of RT2.

Uncertainty and interpretability analysis of encoder-decoder architecture for channel detection

We have adopted a method to understand uncertainty and interpretability of a Bayesian convolutional neural network for detecting 3D channel geobodies in seismic volumes. We measure heteroscedastic aleatoric uncertainty and epistemic uncertainty. Epistemic uncertainty captures the uncertainty of the network parameters, whereas heteroscedastic aleatoric uncertainty accounts for noise in the seismic volumes. We train a network modified from U-Net architecture on 3D synthetic seismic volumes, and then we apply it to field data. Tests on 3D field data sets from the Browse Basin, offshore Australia, and from Parihaka in New Zealand prove that uncertainty volumes are related to geologic uncertainty, model mispicks, and input noise. We analyze model interpretability on these data sets by creating saliency volumes with gradient-weighted class activation mapping. We find that the model takes a global-to-local approach to localize channel geobodies as well as the importance of different model components in overall strategy. Using channel probability, uncertainty, and saliency volumes, interpreters can accurately identify channel geobodies in 3D seismic volumes and also understand the model predictions.

Partial Least Squares Regression of Oil Sands Processing Variables within Discrete Event Simulation Digital Twin

Oil remains a major contributor to global primary energy supply and is, thus, fundamental to the continued functioning of modern society and related industries. Conventional oil and gas reserves are finite and are being depleted at a relatively rapid pace. With alternative fuels and technologies still unable to fill the gap, research and development of unconventional petroleum resources have accelerated markedly in the past 20 years. With some of the largest bitumen deposits in the world, Canada has an active oil mining and refining industry. Bitumen deposits, also called oil sands, are formed in complex geological environments and subject to a host of syn- and post-depositional processes. As a result, some ores are heterogeneous, at both individual reservoir and regional scales, which poses significant problems in terms of extractive processing. Moreover, with increased environmental awareness and enhanced governmental regulations and industry best practices, it is critical for oil sands producers to improve process efficiencies across the spectrum. Discrete event simulation (DES) is a computational paradigm to develop dynamic digital twins, including the interactions of critical variables and processes. In the case of mining systems, the digital twin includes aspects of geological uncertainty. The resulting simulations include alternate operational modes that are characterized by separate operational policies and tactics. The current DES framework has been customized to integrate predictive modelling data, generated via partial least squares (PLS) regression, in order to evaluate system-wide response to geological uncertainty. Sample computations that are based on data from Canada’s oil sands are presented, showing the framework to be a powerful tool to assess and attenuate operational risk factors in the extractive processing of bitumen deposits. Specifically, this work addresses blending control strategies prior to bitumen extraction and provides a pathway to incorporate geological variation into decision-making processes throughout the value chain.

The role of the geological uncertainty in a geotechnical design – A retrospective view of Freeway No. 3 Landslide in Northern Taiwan

The importance of the geological model in a geotechnical engineering project has long been recognized. However, the uncertainty associated with the geological model has rarely been quantified and explicitly considered in the geotechnical design. This paper explores the role of the geological model uncertainty and the benefit of reducing such uncertainty in a geotechnical design. To this end, the landslide that occurred on 25 April 2010 at 3.3 K of Freeway No. 3 in Northern Taiwan, referred to herein as NH-3 Slope, is re-analyzed with an assumed geological model (a dip-slope model in this case) under various uncertainty scenarios. Traditionally, the design (i.e., a choice of design parameters such as slope height, slope angle, and supporting anchors) of a rock slope seeks to satisfy a target factor of safety (FS). In the case of NH-3 Slope, the apparent dip angle of the slip surface along the dip direction of slope constitutes the geological model of concern. Different survey techniques produce data with varying degrees of error (or uncertainty) in the measured apparent dip angle. For example, data from a Regional Geological Map (RGM) typically yields an estimate of the bedding plane attitude with a significant level of uncertainty due primarily to low data density. In contrast, LiDAR can significantly reduce the uncertainty of the derived bedding plane attitudes. By lowering this uncertainty, the variation (or uncertainty) in the computed FS may be reduced, which tends to yield a lower failure probability (Pf), given other conditions being the same. The benefit of lowering the geological model uncertainty in a geotechnical design is demonstrated through a retrospective analysis of the NH-3 Slope.

CO2 storage capacity estimation under geological uncertainty using 3-D geological modeling of unconventional reservoir rocks in Shahejie Formation, block Nv32, China

Underground CO2 storage is a promising technology for mitigating climate change. In this vein, the subsurface condition was inherited a lot of uncertainties that prevent the success of the CO2 storage project. Therefore, this study aims to build the 3D model under geological uncertainties for enhancing CO2 storage capacity in the Shahejie Formation (Es1), Nv32 block, China. The well logs, seismic data, and geological data were used for the construction of 3-D petrophysical models. The target study area model focused on four units (Es1 × 1, Es1 × 2, Es1 × 3, and Es1 × 4) in the Shahejie Formation. Well logs were utilized to predict petrophysical properties; the lithofacies indicated that the Shahejie Formation units are sandstone, shale, and limestone. Also, the petrophysical interpretation demonstrated that the Es1 reservoir exhibited high percentage porosity, permeability, and medium to high net-to-gross ratios. The static model showed that there are lateral heterogeneities in the reservoir properties and lithofacies; optimal reservoir rocks exist in Es1 × 4, Es1 × 3, and Es1 × 2 units. Moreover, the pore volume of the Es1 unit was estimated from petrophysical property models, ranging between 0.554369 and 10.03771 × 106 sm3, with a total volumetric value of 20.0819 × 106 sm3 for the four reservoir units. Then, the 100–400 realizations were generated for the pore volume uncertainties assessment. In consequence, 200 realizations were determined as an optimal solution for capturing geological uncertainties. The estimation of CO2 storage capacity in the Es1 formation ranged from 15.6 to 207.9 × 109 t. This result suggests the potential of CO2 geological storage in the Shahejie Formation, China.

Play fairway analysis for geothermal exploration in north-western Argentina

Play fairway analysis (PFA) is a statistically based data integration and exploration tool originating in the oil and gas industry that has recently been applied to geothermal exploration. Here, we apply geothermal PFA to a ~40,000 km2 area in the Argentine Puna of the South American Andes (average elevation >4,000 m aMSL) that has geothermal potential but no geothermal resource development to date. We develop a PF workflow customized to the local geological setting and availability of regional geoscience datasets. Evidence layers were grouped to model the spatial distribution of three primary geological attributes of a geothermal system: heat, permeability, and fluid. These layers included hot springs, geothermometry, age and distribution of volcanism, hydrothermal alteration, regional faults along with their kinematics and approximate age of most recent rupture, structural settings, and earthquakes. Spatial statistical methods were used to model and weight the distribution of these evidence layers, including distance buffers, point statistics, and weighted sum functions. Resulting favorability models were created using the veto (product) and voter (sum) methods, and both models indicate highest geothermal potential in Cerro Tuzgle/Aguas Calientes/Tocomar and west Coranzulí. Additional locations in the Puna study area may be prospective but are not resolved given the current data availability. Thus, further data collection to reduce the geological uncertainty is recommended, including acquisition of relatively low-cost datasets such as spring temperatures, fluid chemistry, detailed geological mapping, and structural characterization.

Investigating the capability of data-driven proxy models as solution for reservoir geological uncertainty quantification

Quantifying and analyzing the geological uncertainties associated with the geostatistical models is critical to design a reliable production strategy. A large number, hundreds, even thousands, of model realizations should be run in a reservoir simulator to quantify the uncertainties in terms of a target flow response. However, this becomes a time consuming and computationally demanding process. In practice, different methods are used to select only a few representative models for further analyses. This study aims to take a fresh look at uncertainty quantification solutions by means of a proxy model based on artificial neural networks (ANN). The idea is replacing the flow simulator with an ANN-based proxy model by which the target responses of all realizations are calculated much faster and with a lower computational demand. The ability of computing the combined effect of several uncertain parameters along with successful experiences of utilizing ANNs for reservoir simulation, motivated us to propose our ANN-based methodology as a solution for uncertainty quantification problem. We apply our methodology on a realistic 3D synthetic model which is a channelized model with uncertainties in porosity, permeability, and facies properties. The results are compared with those of two widely used methods, i.e., the traditional ranking and distance-based clustering techniques. It is shown that the ANN-based proxy is able to accurately quantify the geological uncertainties in a much shorter time. Hence, it can be used as a reliable tool for uncertainty quantification and representative models selection. • A new solution for quantifying the geological uncertainties of reservoir models is presented. • A data-Driven proxy model based on Artificial Neural Networks is built to replace the comprehensive reservoir simulator. • The Data-Driven proxy model calculates the desired target response much faster. • The proposed solution outperforms the widely-used uncertainty quantification techniques.

Stochastic inversion of gravity, magnetic, tracer, lithology, and fault data for geologically realistic structural models: Patua Geothermal Field case study

Financial risk due to geological uncertainty is a major barrier for geothermal development. Production from a geothermal well depends on the unknown location of subsurface geological structures, such as faults that contain hydrothermal fluids. Traditionally, geoscientists collect many different datasets, interpret the datasets manually, and create a single model estimating faults' locations. This method, however, does not provide information about the uncertainty regarding the location of faults and often does not fully respect all observed datasets. Previous researchers investigated the use of stochastic inversion schemes for addressing geological uncertainty, but often at the expense of geologic realism. In this paper, we present algorithms and open-source code to stochastically invert five typical datasets for creating geologically realistic structural models. Using a case study with real data from the Patua Geothermal Field, we show that these inversion algorithms are successful in finding an ensemble of structural models that are geologically realistic and match the observed data sufficiently. Geoscientists can use this ensemble of models to optimize reservoir management decisions given structural uncertainty.

Adaptive open-pit mining planning under geological uncertainty

This research project developed an adaptive stochastic optimisation approach for multi-period production scheduling in open-pit mines under geological uncertainty, and compared it to an existing two-stage optimisation method. This new rolling-horizon optimisation approach updates the geological model each time period as new information becomes available. Numerical tests carried out earlier on open-pits of different sizes showed that, on average, the rolling-horizon adaptive policy gave better results than the non-adaptive two-stage approach. The metric used was the percentage gap between the results for each policy and those that would be obtained if the true block grades were perfectly known.This paper extends this earlier work in two ways: firstly, by introducing a second metric — the dollar-value difference between the NPV generated with perfect knowledge of the orebody and those given by the other two optimisation methods. The rolling-horizon approach is better on average than the two-stage approach, but not for all of the geostatistical simulations used to model the geological uncertainty. The second innovation in this paper is to analyse when the new rolling-horizon approach outperforms the non-adaptive one. This depends on the drill-hole spacing. For widely spaced grids, the rolling-horizon approach statistically outperforms the two-stage approach at the 95% confidence level. For very close spacings, both approaches converge toward the results for perfect knowledge.