The Expected Value of Information (EVoI) is a criterion to analyze the feasibility of acquiring new information to deal with uncertainties and improve decisions at any stage of an oil field. Here, we evaluate the influence of the use of representative models (RM) on the EVoI estimation and on the decision to develop the petroleum field. These RM are used to represent a large set of models that honor production data (FM), considering uncertainties in reservoir, fluid and economic parameters, enabling the following processes: (1) optimize production strategies (specialized for each RM and robust to all RM), (2) risk analysis, (3) select the strategy to develop the field based on risk analysis, and (4) estimate the EVoI. We evaluated the influence of the number of RM on these processes, observing the impacts on the results of reducing computational costs. For the EVoI, we applied a Complete (EVoI_FM) and a Simplified (EVoI_RM) methodology, where EVoI_FM was evaluated with all models (FM) while EVoI_RM used different groups with different numbers of RM (GR1, GR2 and GR3, ranging from 9 to 150 models in each group). To assess the quality of the results, we used the complete estimate (EVoI_FM) as a reference. The study was conducted on UNISIM-I-D, a benchmark oil reservoir in the development phase, taking an appraisal well as a source of information to clarify a structural uncertainty. Using RM to optimize specialized production strategies proved useful, since optimizing strategies for all FM would require high computational costs. Moreover, the RM could be used to represent risk curves and select production strategies under uncertainty, but less precisely, affecting directly the results of the EVoI (which is the difference between the expected values of the two curves). The precision of EVoI_RM results varied according to the number and group of RM employed, also varying the best strategies selected for field development. The choice of using simplifications or not will depend on the accuracy required or available resources. Variations in EVoI_RM may be tolerable when compared to the time saved, being the decision maker free for choosing the best estimation method.

Drilling exploration wells depend on the uncertainty we perceive. With some simplification, this uncertainty is about the value of producible hydrocarbons. To drill a prospect, the expected benefits from producing and selling hydrocarbons should outweigh the costs. This principle also applies to clusters of interrelated exploration targets. Here, the benefits of drilling each prospect is information about the neighbouring prospects, proving producible hydrocarbons, or both; these benefits should outweigh the aggregate costs. In practice, estimating the expected benefit could be challenging especially when multiple smaller discoveries are developed under a joint development scheme. With these interrelationships, what is the optimal drilling strategy? In addition, what is the economic value of a group of prospects? This paper discusses the effect of joint development economics on perceived uncertainty and drilling decisions. We suggest a framework for valuation of correlated prospects.

The global oil and gas industry will experience intensive well decommissioning activities, with tens of thousands of well plug and abandonment (P&A) operations forecast to be executed worldwide, over the next few decades. In the North Sea alone, 2,624 wells are expected to be decommissioned over the next decade ( OIL&GAS_UK 2019 ). With the increasing levels of activity in P&A operations, new, fit-for-purpose, P&A design tools and operational technologies have become essential elements in the petroleum engineer's toolbox to ensure safe and cost-effective solutions, especially in today's environment. This paper introduces a novel tool to support risk-based evaluation of well P&A designs using a numerical, grid-based, fluid-flow simulation methodology. The risk-based modelling framework covers the full range of North Sea P&A design scenarios and tracks the long-term evolution of hydrocarbon movement in the P&A well over several thousand years. A visualization engine identifies the dominant hydrocarbon flow-paths, while Monte Carlo methods are used to account for uncertainties in the model input parameters; allowing multiple design options to be ranked with respect to the risk of a hydrocarbon leak. Transient flow modelling of the well P&A system allows new key performance indicators (KPIs) to be developed, e.g. evolution of hydrocarbon saturations within the P&A well over time and the time hydrocarbons reach the surface. These KPIs are not provided by steady-state P&A models. Results presented in this paper demonstrate the value derived from applications of the developed framework to: Optimally allocate resources by identifying fit-for-purpose P&A designs (e.g. the number, location and length of wellbore barriers, and the value of remedial operations on annular flow barriers). Support risk-based decision-making via investigation and comparison of how multiple P&A design options perform for given well/reservoir conditions. Identify critical modelling parameters and optimally allocate data gathering resources to reduce uncertainty. The "likelihood of occurrence" of each cement defect type is one such critical, but very uncertain, model input parameter.

Innovative reservoir seismic characterization technologies, with massive integration of Seismic and Petrophysical data, were applied for the first time, almost ten years ago, to a green field since the appraisal phase. This seismic driven static model has been continuously kept alive during development and production, effectively supporting application of innovative technologies, operations and business decision processes. Main object of this paper is to present a global overview of how different methodologies were implemented and tuned while the model evolved, taking advantages of lessons learned to improve technical knowledge and set up new opportunities. The initial model, based on five wells, was adopted to characterize the reservoir, with a robust link between lithology, elastic and dynamic properties. The innovative Petro-Elastic facies characterization, tuned at field scale for the next development phase, proved to be valid during the drilling campaign of more than twenty HA-HZ wells, with only minor adjustments required to propagate the model. An automated workflow for geostatistical simulations conditioned by seismic probability cubes of facies and petrophysical properties and multi-scenario risk analysis, including depth conversion and well calibration of seismic properties, generated Static and Dynamic models ready to be approached with a Computer Assisted History Match (CAHM)/ Ensemble based Methods. Properties distribution did not require new seismic inversion meanwhile updating the model, confirming the predictability of seismic in terms of average properties and the consistency of Rock Physics Model and Petrophysical characterization. To drill horizontal wells, geosteering activities were assessed by generating fine scale pre-job sector models. Optimization of well trajectories was not based only on seismic but on posterior probability of facies and petrophysical properties generated by a dedicated workflow. During drilling, real time integration of seismic inversion, wellbore data, and Ultra-Deep Azimuthal EM Reservoir Mapping allowed to effectively support well placement, providing real time stratigraphic interpretation of drilled sequences. The automated workflow permits now to investigate a wide distribution of realizations converging on a subset whose reservoir dynamic simulation are in agreement with the first years of production. The model represented a chance to implement approaches and workflows deployed in our reservoir department and now applied in top projects as well as in the day by day activities and still it is a platform to test new approaches and available technologies. Single innovative items have been object of more than twelve publications or conference presentations. Ten years later, we can assert that multidisciplinary integration and continuous cooperation between the subsidiary, in charge of operations, and the HQ was the key for the successful field characterization and development. For the future, this model could be a platform to test CAHM/ Ensemble based technology integrating also future 4D seismic data thanks to the initial tuned Petro Elastic facies model.

The objective of this work is to develop a methodology to estimate the fraction of Reserves assigned to each Reserves category (1P, 2P, and 3P) of the PRMS resources classification matrix using a cumulative distribution function (CDF). Previous published work has often used Swanson's Mean (SM) as the basis for allocating Reserves to individual categories, but we found that this method, which relates the Reserves categories through a CDF for a normal distribution, is an inaccurate means to determine the relationship of the Reserves categories with asymmetric distributions, and our work identified a better method, Gaussian Quadrature (GQ). Production data are lognormally distributed, regardless of basin type, and thus are not compatible with the SM concept. The GQ algorithm provides a methodology to estimate the fraction of Reserves that lie within the 1P, 2P, and 3P categories — known as their weights . GQ is a numerical integration method that uses discrete random variables and a distribution that matches the original data. For this work, we associate the lognormal CDF with a set of discrete random variables that replace the production data, and determine the associated probabilities. The production data for both conventional and unconventional fields are lognormally distributed, thus we expect that this methodology can be implemented in any field. We selected 38 wells from a Permian Basin dataset available to us, and we performed probabilistic decline curve analysis (DCA) using the Arps Hyperbolic model and Monte Carlo simulation (MCS) to obtain a probability distribution of the 1P, 2P, and 3P volumes. We considered this information to be our "truth case," to which we compared relative weights of different Reserves categories from the GQ and SM methodologies. We also performed probabilistic rate transient analysis (RTA) using the IHS Harmony software to obtain the 1P, 2P, and 3P volumes, and calculated the relative weights of each Reserves category. Once we completed these first two steps, we implemented a 3-point GQ to obtain the weights and percentiles for each well. We analyzed the GQ results by calculating the percentage differences between the probabilistic DCA, RTA, and GQ results. The probabilistic DCA results indicated that the SM method is an inaccurate method for estimating the relative weights of each Reserves category. Our results show that the GQ method was able to capture an accurate representation of the Reserves weights, we note that we assumed an expected lognormal CDF for Reserves. We believe that 1C, 2C, 3C, and 1U, 2U, and 3U Contingent and Prospective Resources are distributed in a similar way ( i.e. , as a lognormal CDF) but have greater variance. Based on our results, we conclude that the GQ method is accurate and can be used to approximate the relationship between the relative weights of resources in PRMS categories. This relationship will aid entities in reporting Reserves of different categories to regulatory agencies because it can be recreated for any field, play, or region. These distributions of Reserves and Resources Other than Reserves (ROTR) are important for planning and for resource inventorying. The GQ method provides a measure of confidence in our prediction of the Reserves weights because of the relatively smaller percentage differences between the probabilistic DCA, RTA, and GQ weights than those implied by the SM method. For reference, our proposed methodology can be implemented in both conventional and unconventional reservoirs.

This paper describes a new integrated workflow that successfully bridges the gap between different measurement scales to characterise a clastic reservoir. This workflow combines thin sections, well logs and elastic curves (Vp, Vs and Rho or any combination of these) to generate a petro-elastic log facies classification that is not only calibrated to core data, but at the same time constitutes a shared key input for seismic inversion classification and reservoir modelling. The application of this workflow is deemed crucial whenever seismic attributes are required to be used as a driver for properties distribution within a geological model. Given the inherent constraint of the seismic resolution, the recommended logfacies model has to honour a robust petrophysical reservoir characterisation (porosity, permeability) while assuring the maximum discrimination within the elastic inversion space (e.g. P-Impedance vs. Vp/Vs ratio) with a minimum number of classes. In this paper, we outline a framework for building an integrated log facies model that includes the following steps: first, a facies model is defined at the scale of thin sections by means of a clustering technique. This represents the reference for a core-supported facies classification that in turn is linked to the sedimentological facies. A porosity-permeability relationship is analysed as it represents a common petrophysical domain for both facies models: thin sections and core data. The "poro-perm" relations are preserved when moving to log scale, where a petro-elastic log facies model is generated with a small number of classes. Secondly, Formation Evaluation (FE) is carried out to provide a simple, general and robust petrophysical model paramount not only for reservoir characterisation but also to be used as input for a dedicated Rock Physics Model (RPM), which links petrophysics and seismic velocities. FE and RPM are tuned one to the other until optimised. Thereafter, in order to generate classes that are not affected by hydrocarbon effects, fluid replacement modelling is performed to produce synthetic elastic curves in brine condition, that are input - together with porosity and volumes of minerals - to the facies classification. The procedure iterates until a few log facies are simultaneously discriminated in the petrophysical (e.g. Porosity vs. Permeability), petro-elastic (e.g. Porosity vs. Vp/Vs ratio) and in the elastic space of seismic inversion (e.g. P-Impedance vs. Vp/Vs ratio). Finally, once the log facies classes are defined at the well log scale, we use a dedicated technique to scale-up the petro-elastic log facies to the seismic domain, together with the petrophysical and the elastic curves. While the log facies are the hard data for geological modelling and support for petrophysical properties distribution, the corresponding upscaled log facies curves are used to populate the template for seismic elastic attributes classification to facies probabilities. The successful application to a green field in ultra-deep water environment, characterised by few development wells, proves the robustness of integrating all available data from thin sections to seismic scale in reservoir characterisation.

In order to better understand reservoir behavior, reservoir engineers make sure that the model fits the data appropriately. The question of how well a model fits the data is described by a match quality function carrying assumptions about data. From a statistical perspective, improper assumptions about the underlying model may lead to misleading belief about the future response of reservoir models. For instance, a simple linear regression model may have a fair fit to available data, yet fail to predict well. On the contrary, a model may perfectly match the data but make poor prediction (i.e. overfitting). In both cases, the regression model mean response will be far from the true response of the reservoir variables and will cause poor decision making. Therefore, a suitable model has to provide balance between the goodness of the fitted model and the model complexity. In the model selection problem, realistic assumptions concerning the details of model specification are the key elements in learning from data. With regard to conventional history match scheme, the data fitting is usually performed by linear least-squares regression model (LSQ) which makes simple, yet often unrealistic, assumptions about the discrepancy between the model output and the measured values. The linear LSQ model ignores any likely correlation structure in discrepancy, changes in mean and pattern similarities reflecting on poor prediction. In this work, we interpret the model selection problem in data-driven settings that enables us to first interpolate the error in history period, and second propagate it towards unseen data (i.e. error generalization). The error models constructed by inferring parameters of selected models can predict the response variable (e.g. oil rate) at any point in input space (e.g. time) with corresponding generalization uncertainty. These models are inferred through training/validation data set and further compared in terms of average generalization error on test set. Our results demonstrate how incorporating different correlation structures of errors improves predictive performance of the model for the deterministic aspect of the reservoir modelling. In addition, our findings based on different inference of selected error models highlight an enormous failure in prediction by improper models.

A workflow applied to achieve a multi-scale characterisation of a carbonate reservoir is presented. Carbonate rocks are strongly heterogeneous due either to complexity of the primary fabric or to diagenetic over-printing. The combination of these features leads to complicated pore systems, thus a proper definition of pore types using either pore size or pore throat size distributions, is important to indirectly capture diagenetic modifications and to get a link to dynamic properties. A new approach was developed in order to define a Rock Type classification (RRT) each time the approaches based on Winland's and Hydraulic Flow Unit methods do not give a reliable core facies characterisation when moving to the log scale. Moreover, the proposed workflow accounts for stratigraphy and seismic since RRT are linked to the elastic properties. In the new MICP-based Rock Typing workflow, RRT are identified by describing dominant pore types using mercury injection (MICP) curves parameterisation and routine core data (RCA). Clustering and subsequent extrapolation of MICP derived RRT to RCA samples, are the first two stages to achieve a predictable classification into the log domain. Log RRT are then defined at the log scale using curves of elastic properties, like Poisson's Ratio (PR), Frame Stiffness ( f k ) and Flexibility ( γ k ) Factors. These elastic parameters (calculated with the Extended Biot Theory), can capture the effects of pore structure on the petrophysical properties and link RRT prediction at well position to seismic attributes. Since the RRT are characterised in the elastic space, the facies model – properly upscaled – represents the basis to classify elastic attributes from seismic inversion in a Bayesian framework. The seismic classification can then be used as a driver for RRT distribution in the inter-well space into the 3D model. A further benefit is the direct relationship to the original RRT porosity/permeability distributions, when modelling petrophysical properties. This new workflow was a successful solution to define homogeneous reservoir intervals in a carbonate environment characterised by the lack of a significant relationship between depositional facies and petrophysical properties.

Two fundamental building blocks support the concept of exploration for disorder. The first building block is a sequential chain of petroleum system events resulting in a lognormal field size distribution. Output from the first building block becomes input to the second. The second building block is the intersection of multiple petroleum systems (prospects) in one exploration well or by multi-well exploration campaign. This second building block utilises the Central Limit Theorem whereby an aggregation of lognormal prospect distributions tends toward a normal field size distribution. Exploration for disorder in practice is characterised by: Targeting multiple stacked reservoirs or multiple fluid phases in one exploration well Targeting a field being the product of more than one sequence of petroleum system events Both 1 and 2 above operating collaboratively. The benefit of exploring for disorder is that the geological earth delivers, in the limit, a normal field size distribution and an exploration probability of success of 50% respectively supported by the Central Limit Theorem and Binomial Probability Theory. That is, the Central Limit Theorem dictates that the exploration probability of success cannot exceed 50%. These attributes contrast with exploration for ordered events which are allocated both a probability of success less than 50% and a lognormal field size distribution [according to Binomial Probability Theory]. Exploration in tune with the disordered geological earth delivers the highest probability of success. Operating in parallel, gross rock volume (GRV) exists in a state of maximal entropy, hypothetically characterised by a normal distribution. Hypothetically, smaller GRV structures have a higher probability of being filled-to-spill resulting in a normal field size distribution for smaller fields. That is, smaller GRV structures require a lower volume of migratory hydrocarbons to fill-to-spill. Conversely, larger GRV structures may experience a wide range of fluid-fill variation thereby forming a lognormal field size distribution within the framework of a hypothetical, normal GRV distribution. This hypothesis was tested by Statistical Experiment. Exploration for disorder is manifest in disruption to the monotonicity of creaming curves. It is expected that creaming curve disruption is caused by: An unplanned event (risk exposure): in a positive sense the geological probability of success (GPOS) and/or pre-drill GRV estimates fail to act as a proxy for field size and a larger-than-expected field is discovered; pre-drill field size estimates typically have a plus or minus one order of magnitude error band Design: the monotonically-diminishing lognormal field size creaming curve is disrupted by exploring for disorder; discovered field size jumps from a lognormal to a normal creaming curve. A conventional lognormal field size creaming curve may encounter a "false summit" at the point where the discovery of small to medium field sizes coincides with: Low GPOS values (<50%), and/or A field size sample-variance minimum. The term "false summit" implies bypassed exploration potential. Exploration for disorder may then be utilised to fulfil exploration potential.

Polymer injection might lead to incremental oil recovery and increase the value of an asset. Several steps have to be taken to mature a polymer injection project. The field needs to be screened for applicability of polymer injection, laboratory experiments have to be performed, and a pilot project might be required prior to field implementation. The decision to perform a pilot project can be based on a Value of Information (VoI) calculation. The VoI can be derived by performing a workflow capturing the impact of the range of geological scenarios as well as dynamic and polymer parameters on incremental Net Present Value (NPV). The result of the workflow is a Cumulative Distribution Function (CDF) of NPV linked to prior distributions of model parameters and potential observables from the polymer injection pilot. The impact of various parameters on the CDF of the field-wide NPV can be analyzed and in turn used to decide on what measurements from the pilot have a strong sensitivity on the NPV CDF and are thus informative. In the case shown here, the water cut reduction in the pilot area has a strong impact on the NPV CDF of the polymer injection field implementation. To extract maximum information, the response of the pilot for water cut reduction needs to be optimized under uncertainty. To calculate the VoI, the Expected Monetary Value (EMV) difference of a decision tree with and without the pilot can be used if the Decision Maker (DM) is risk neutral. However, if the DM requires hurdle values through a Probability of Economic Success (PES), Value Functions (VF) and Decision Weights according to the Prospect Theory should be used. Applying risk hurdles requires a consistent use of VFs and Decision Weights for calculating VoI and the Probability of Maturation (POM) of projects.

Natural gas produced from shale formations has increased rapidly in the past decade altering the oil and gas industry markets remarkably. The use of horizontal drilling and hydraulic fracturing has allowed previously unrecoverable shale gas to be extracted. Shale gas development is more expensive compared to conventional developments and as such, understanding the economic feasibility is of greater importance in successfully developing the resource. Using the estimated ultimate recoverable expressed in terms of P10, P50, and P90 from 2751 horizontal well production data (all starting production from 2008) from the Barnett shale, a discounted cash flow economic model (MS- Excel based) was used to quantify the effect of finding and development costs (F&DC) and gas prices on the economic viability of horizontal wells within four out of five basins (Strawn Basin, Ouachita Folded Belt, Forth-worth Syncline and Bendarch Basin) in the Barnett shale. The investment hurdle in the economic model was a rate of return of 20% and a payback period of 5 years or less. This paper helps determine the percentage of wells within basins in the Barnett shale that would be economically viable at various F&DC and gas prices subject to satisfying the prescribed investment hurdle.

The Value of Information (VoI) analysis typically assesses information opportunities to manage uncertainty. The VoI is traditionally estimated using the expected monetary value (EMV), which overlooks the decision maker's (DM) attitude towards upsides and downsides. In this study, we assess the VoI for DMs with different attitudes, under different levels of information reliability. We define attitude to be as: neutrality to upsides and downsides, aversion to downside risk, and willingness to exploit upside potential. We applied a simple and flexible formula, which incorporates EMV with lower and upper semi-deviations from a benchmark, to quantify downside risk and upside potential. We determined VoI using many uncertain scenarios to maintain interactions between parameters instead of deterministically isolating the uncertainty under analysis. To accelerate analysis and reduce computational costs, we used a set of candidate production strategies optimized for extremely different scenarios. Our case study was the UNISIM-I-D, a benchmark reservoir model with a key structural uncertainty affecting production strategy selection. We used an appraisal well as an information source, and four hypothetical DMs with different attitudes. Our results showed that these DMs value information differently, that one DM may decide to acquire information while another may not, for the same situation. In our case study, information reduced downside risk but did not increase upside potential, meaning that information was more valuable to risk-averse DMs (which was up to 20 times higher), and less valuable to DMs exclusively focused on maximizing upsides.

This paper is intended to highlight some of the misconceptions that enter decision making whether as a consequence of human bias or as statistical inaccuracies and misuse of mathematically robust equations. There are three basic questions in the oil and gas industry: 1) How big is it? i.e.: Resources, 2) How quickly can we get it out? i.e.: Rate, and 3) How much can we make from it? i.e.: Value. All three contain risk and uncertainty. Several methods are used by oil and gas professionals to manage these variables, ranging from the simple toss of a coin, to complex mathematical algorithms and statistics. If only it were so simple to use a formula for the decision and have the confidence to trust it. However, business decisions are often more than just complex logic and the worlds of subsurface geoscience and reservoir engineering add yet more variables and uncertainty. The paper uses real world examples in the oil and gas industry and highlights the difference between Risk Management and simple Gambling. The paper draws upon the following three themes: Human Factors: No matter how precise the mathematics of risk analysis, ultimately decision making relies upon individuals who are subjective by definition. A group of similar but independent professionals will have a different perception of risk and their own biases depending upon the number of parameters used to calculate for example, chance of success and different biases on marking at extremes of a range. Statistics: One solution to deal with the human variable is to attempt to take individuals out of the decision making process completely and trust mathematics instead. Unfortunately, many decision makers have forgotten what assumptions exist in Portfolio Theory and often use the methods blindly, placing too much confidence in the predictions and unaware of other equally likely outcomes. Popular Decision tools: Expected Monetary Value – ‘Expected’ is a subjective term and numerically tends towards the Mode of the distribution. The authors question whether EMV really has any true meaning and instils a false sense of confidence in the outcome. Hurdle Rates - The Theory of Inevitable Disappointment highlights how a decision maker who fails to understand uncertainty can often fall short on forecasts. Conclusion: There are no simple rules when decisions involve information which contains uncertainty and interpretation. The responsibility for decision making should rest with individuals with ‘skin in the game’. A decision maker should be aware of all assumptions and always question those assumptions. In the end, if a decision maker does not use a spectrum of tools to examine all possible outcomes, then they are simply relying on luck. Whatever the process, review, feedback and calibration are critically important.

At any stage of the Exploration and Production cycle decisions must be made with limited information. This situation can always be improved by investing more money or time in improving the quality and variety of information available. However, this in itself presents a decision as to whether the investment to get new information such as a new seismic survey is worth it. This paper outlines a practical and structured method to estimate the value of new information in monetary terms, which can be used to make an informed business decision on whether the new information is worth it. The method is comprised of two primary elements. Firstly the problem is defined using a decision tree analysis approach that considers what decisions are to be made and their possible outcomes. Secondly, a collaborative workshop is used to estimate the probabilities required to populate the decision tree. This then enables the Expected Monetary Value to be calculated both with and without the technology that acquires the new information. The difference between the two is approximated as the value of the information. This paper uses a case study approach to explain this method that considers an operators dilemma in deciding whether or not to drill an appraisal well. Obtaining a meaningful result from this process requires time and commitment from management and technical staff. However, at stake are the considerable sums of money in acquiring the new information and the subsequent investment decisions that draw on the new information. The method described as a minimum provides fiscal information to make investment decisions and at best, provides a framework in which the benefits and risks of investing in new technology can be fully understood. There has already been much published on the use of decision trees to estimate the value of information. However, in practice it can be difficult to estimate the probabilities required for a decision tree analysis to work as they are not intuitive to describing the situation. This approach simplifies this by arranging the mathematics to make the task of the workshops clearer and more achievable. The result is a practical method that can be easily understood and implemented.

Within the European context, CO 2 storage operations shall address the potential impacts of large scale CO 2 storage through risk assessment. The key risks identified for this onshore CO 2 storage site were the migration through faults and ground deformation. To quantify the CO 2 migration along a fault, flow modelling and uncertainties management codes are coupled to compute the failure probability i.e. the probability of CO 2 migration towards a control aquifer. Such probability of failure is characterized by low to very low probability of occurrence which requires a large number of simulations to enable its evaluation. Each failure scenario models the CO 2 migration from a storage aquifer to a control aquifer when altering the flow properties of the fault zone. Fault failure analyses are performed on the surrogate models. They show that limited CO 2 migration is occurring along the fault but no breakthrough in the control aquifer. The injection induces some pressure disturbance in the control aquifer in about 30% of the cases which lead to effective stress changes. To quantify effective stress changes due to CO 2 injection and the subsequent ground deformation, the mechanical responses of the different sediment layers are modeled coupling flow and geomechanics. The impact of the stress changes on porosity and permeability of the storage reservoir is modeled along with the impact of uncertainties of the mechanical parameters. For this onshore CO 2 storage site study case, the expected ground displacement is negligible (below the limit of the measurement capabilities).

Reservoir monitoring has constantly used 4D seismic data and has become an important tool for reservoir management. However, from the economic perspective, the information is only valuable if it influences decisions. It is necessary to decide if the acquisition of new information will be useful for field management. Thus, a methodology was developed to estimate the chance of success of a 4D seismic project. The methodology is an iterative process; it quantifies the increase of the project's value through the optimization of the production strategy for several reservoir models. As the number of reservoir scenarios representing the problem can be high, the optimization process can be time consuming. Therefore, the selection of representative models is an important step of the methodology. The present study discusses the impact of the number of representative models on the expected value of information (EVOI) and chance of success results. The chance of success methodology was applied to a synthetic model to validate the results and show its benefits. The selection of the representative models was based on production and economic results from the reservoir scenarios. Five evaluations were performed; each one considers a different number of representative models. The expected value of information and chance of success varies depending on the number of representative models. However, the EVOI and chance of success results obtained stabilize as the number of representative models increases. The methodology estimates the variation of the expected benefits due to the acquisition of 4D seismic data. The results obtained support the decision-making process, making the methodology an important tool for reservoir management. The decision-making process must consider the current uncertain scenario, the engineering and geophysical studies and also the economic aspect. Measuring and, especially, predicting the economic impact of new information is complex. Thus, the chance of success methodology is a tool to assist the decision maker in the evaluation of new 4D seismic projects.

4D seismic is important information to optimize well planning, interventions and drainage strategy. It provides unique information regarding the properties of the reservoir between and beyond the wells; it can be used to locate bypassed oil, to identify not drained compartments or to mitigate risks such as an unexpectedly early breakthrough of injected fluids. The integration of geophysics and engineering information increase the predictive capability of reservoir simulations, and hence has positive impacts on field management. The estimation of the best time to acquire seismic data has a major impact on reservoir management since, depending on the acquisition period, seismic may not be able to identify important variations in the reservoir dynamic properties, important changes may no longer be detected, well history matching can be able to identify properties, and it can be too late for effective action in field management. 4D seismic reservoir monitoring has no value unless it impacts operations and its utility is related to the period of seismic data acquisition. The challenge is to use effectively the available information; the integration of geophysics and engineering information must be accomplished by performing the history matching process that incorporates both production and 4D seismic data. Changes in saturation and pressure, derived from 4D seismic, can be used to improve the quality of the reservoir model allowing a better understanding of the reservoir model and fluid flow. Given the importance in defining the time for seismic data acquisition, this paper evaluates the quality improvement of a synthetic reservoir model by incorporating seismic data over different production periods and identifies characteristics that impact on the best time estimation. It also, estimates the best time to the case studied considering a practical condition. The study evaluated two cases: a posteriori (Case 1) and a priori (Case 2) analysis. In Case 1 the true model is known; a history match process was performed to 2, 4 and 6 years of production using two methods considering: (1) only production data; (2) production and seismic data. By analyzing the reservoir model quality improvement it was possible to identify the main characteristics that determine the moment to acquire seismic data. As, in practice, the earth true model is unknown, in Case 2, a process flow considering a risk analysis associated with reservoir simulation was applied to estimate the best time to acquire 4D seismic information. Analyzing the reservoir model quality improvement it was possible to identify the main characteristics that determine the best moment to acquire seismic data, to a non-mature oil reservoir with water injection production mechanism. These characteristics include: saturation and pressure error maps, production data error and breakthrough distribution within production wells.

Underbalanced operations are carried out to bypass drilling challenges that could be difficult to resolve using conventional drilling techniques. Steady state multiphase flow models are used to construct underbalanced drilling operational windows. These advanced software models are deterministically formulated. It is known that some of the model input parameters such as the multiphase flow parameters, friction factors, and reservoir productivity, are subject to uncertainties. Failures to capture these variabilities may introduce some error in the model prediction, resulting in poor well planning and implementation. The purpose of this work is to implement probabilistic modeling of underbalanced drilling using a simple steady state two-phase model. Both predefined uncertain and fixed factors serve as inputs to a pre-existing deterministic model. By applying Monte Carlo simulations, the model predicts outputs which follow a statistical distribution. A sensitivity analysis is conducted to determine the input factor mostly responsible for the uncertainty in the predicted bottomhole pressure. The results demonstrate that uncertainty modeling can improve underbalanced drilling design and operations. A more realistic operational window is obtained, ensuring that underbalanced condition is maintained throughout the target section. With a better understanding of uncertainties and the corresponding impacts, well planners can make better decisions regarding well design criteria and safe operational conditions, and avoid huge economic consequences.

The acquisition of unconventional assets has become viable under current technical and economical conditions. While producing the conventional resources, utilizing the reserve replacement potential offered by the unconventional reservoirs has brought up optimal extraction path considerations. Optimal extraction path analyses mainly focus on the timing of incorporating unconventional production within a company’s portfolio. This paper establishes a decision making framework that takes into account that different price scenarios can unfold with differing probabilities of occurrence and that switching from conventional to unconventional production comes at a cost. The model is based on an integrated discounted cash flow and optimization formulation to find the differing parameters necessary to populate a decision tree framework. A series of differing price paths are considered for a company with exogenously assigned probabilities of occurrence and an optimal path analysis is conducted under different assumptions regarding its risk propensity. A sensitivity analysis is also conducted on the probabilities of different price scenarios to better understand the risk the company faces in making production decisions. Using the framework introduced in this paper, a company should be able to make a more informed decision that considers price path uncertainty and risk preferences in the timing of the switch from conventional to unconventional production.

Many decisions in petroleum exploration and development have to be made with significant uncertainties. Stochastic simulations are frequently used to capture the consequences of a decision in relation to the uncertain variables within reservoirs, infrastructures and economics. The results of stochastic simulations are often used in the following risk management process to optimize decisions. A big challenge in this process is to frame the outputs of stochastic simulations in a manner which makes sense to the decision-maker and provides him with clear judgemental insights into the problem and its related uncertainties. Traditional decision-tree analysis provides a good way of formulating and structuring decision problems, but one of its disadvantages is that only the expected value is used at a decision node. Most of the decision issues involve both continuous and discrete uncertainties, which can only be fully evaluated using distribution functions. By combining the decision-tree approach with stochastic simulation the term of expected value can be extended to a distribution function. Distribution functions can provide more details than expected values for understanding the influence of relevant uncertainties in the context of a full risk spectrum. Depending on the decision-maker’s risk preference, decisions can be made based on not only the expected value but also upside or downside. A simple example on Value of Information is used to demonstrate the combination of decision-tree approach and stochastic simulation, where the value of information realized by a decision is quantified by a distribution function through Monte Carlo simulation rather than an expected value. A real decision situation concerning infill drilling in a North Sea field is presented to show the procedure of stochastic modeling of a decision tree in the framework of integrated project valuation. Important decisions along the time horizon include scheduling of drilling campaign, pilot or side-track drilling of adjacent prospects, and installing of a new template.