Abstract Although papers comparing some standard functions with saturation models have been published, no consistent review exists comparing the performance of most of the universal saturation-height function quantitatively. The universal SHF is fast and straightforward, but robust enough to account for limited data and while another full data acquisition and advanced analysis are in progress (partially obtained). The method can help the subsurface team in understanding the water saturation nature in quick turnaround time before the completion of ongoing volumetrics estimation. Two best practices of this workflow are rapid and robust. The paper reviews three of the universal saturation-height methods, namely those proposed by Choo, Kyi-Ramli, and K-Function. The comparisons between modelled and measured capillary pressure measurements over the most common functions and through different reservoirs are discussed. The advantages and drawbacks of each method are highlighted. Each technique is compared by investigating how accurately they model the saturation-height profiles of several wells from Offshore Malaysia. The work was carried out to independently assess which equations should be tested first during saturation-height studies. The differences for each capillary pressure between the water saturations estimated using the equations and those measured on the samples are examined in both graphic and quantitative terms. The results of this study show that Choo (2010) model is one of the better performing saturation-height functions. However, the best results are achieved using this function, but this method is also the most challenging to execute in petrophysical and static modelling software. Of the conventional equation-based approaches, the K-Function model appears to have the most utility and are recommended as first choice saturation-height models to test. It only has two inputs for the modelling comprising of RQI and HAFWL. This study continues the extended concepts of Adams (2016) and Harrison (2001) to describe quantitative comparisons between modelled and measured capillary pressure measurements over the functions and through different reservoirs. The review presented could not include all possible equations, but shows which of the most frequently cited functions, is likely to be of utility. Areas for future improvement are also highlighted.

Abstract One of the reservoirs in a giant field in onshore Abu Dhabi has been producing for six decades. The reservoir was already saturated at the time of production commencement, with a large oil rim and a gas cap. Both water injection and lean gas injection have been relied upon to sustain production, and will play an even more prominent role for the future development of oil rim and gas cap. Due to the stakeholders’ different entitlements / equity interests in the hydrocarbons originally existed in oil rim area versus gas cap area, it is important to be able to allocate liquid hydrocarbon production and injection gas utilization among the stakeholders, based on a systematic framework. This paper presents a comprehensive comparison of two modeling-based approaches of fluid tracking for condensate allocation and gas utilization – a tracer modeling option in a commercial reservoir simulator, and a full component fluid tracking approach implemented for this reservoir. The component tracking approach is based on the idea that if individual components represented in a fully compositional reservoir model are tracked separately starting from model initialization, one can trace back the source of hydrocarbon production from both gas cap and oil rim. This approach is implemented through the doubling of the number of components in the equation of state fluid characterization – one set of components for the gas cap, and another set for the oil rim. In order to track the net utilization of the injected lean gas, additional components are needed – in this case one more component representing the lean gas, as the injected gas is a dry gas. The results of the comprehensive comparison demonstrate very clearly that these two approaches yield consistent condensate allocation and gas utilization results over the entire life of field (including history match and prediction). For condensate allocation, the hydrocarbon liquid production split depends on how the injected lean gas is tracked. For gas utilization, the injected lean gas must be tracked as a distinct component separate from both oil rim and gas cap components. The comparison also shows that although the tracer-based approach is numerically more efficient with less runtime, the full component tracking approach is simulator agnostic, and therefore can be implemented in any reservoir simulator. In addition, the full component tracking method can be used for cases where injection gas is a known mixture of oil rim and gas cap gas – something the tracer-based option cannot handle. In summary, this paper presents a first comprehensive comparison of the two (2) different fluid tracking modeling approaches, with practical recommendations on modeling-based hydrocarbon liquid production and injection gas utilization allocation in cases where the commercial framework makes such allocation necessary.

Abstract Forecasting oil and gas production for a well or reservoir is one of the most valuable tasks of a reservoir engineer. This paper elaborates on the assessment of production targets deliverability using a dynamic and integrated approach to perform short term production forecasting. The case also studies the seamless integration of sub-surface with well and facility network models providing options to examine the feasibility of production plans. The principal approach employed in the methodology comprises an automated workflow, which includes reservoir simulation data, wells, and network models enclosed in a dynamic loop, where workflow iteration takes place until the production target is achieved. Within this implementation, the process allows the estimation of short-term production forecasts mainly used for optimizing production operations and business planning, among other tasks. Some of the main steps followed in order to assess the feasibility of the production targets are: Well, Network and Reservoir data QA/QC and further alignment Narrowing down of gaps between the surface and sub-surface system Integration among the several data-driven sources Iteration of the overall process allowing minimal human intervention Throughout this implementation, it was clearly appreciated that production forecasting represents a highly complex task due to the number of different components included in an integrated system and their intrinsic interconnection, where essentially every piece of the calculation influences others. The case study highlighted how performing a dynamic reservoir integration run in an integrated digital production system can help engineers to provide a way to check the feasibility of short-term production targets while considering full surface system configuration. Moreover, the integrated production system provided flexibility in terms of setting up forecast scenarios in an efficient manner, thereby minimizing users' time and efforts in data handling and driving maximum user focus on results and analysis. A dedicated forecast server helped in achieving run performance, thereby enabling the user to carry out various what-if scenarios in a short amount of time. The case studies also discuss a few key challenges encountered during the process that represented a difficulty in overcoming unless addressed in an integrated collaborative system: Data size and complexity Lack of data and/or data inconsistency Surface and Sub-surface model configuration for dynamic integration Gaps between surface and sub-surface performances at initial time step. The application of this integrated and automated workflow approach improved confidence in the reservoir target deliverables by providing robust data management and better predictions resulting from evaluating the entire system (including the performance of wells and reservoirs at the same time). This helped in saving user analysis time significantly by avoiding the process of analyzing all the sections of the system in isolated silos, which is usually the approach followed by many operators with large amounts of wells.

Abstract Acquiring reservoir fluid samples through formation testers is critical to asset evaluation in most oil and gas drilling operations. From the time this technology was introduced to the industry, the key challenges have been in planning the job, estimating contamination during operation, and obtaining clean fluid samples in the shortest time possible. The objective of this paper is to create a new data-driven model to proactively simulate the cleaning process in order to provide a practical job-planning tool that optimizes fluid sampling. After detailed analysis of formation pump-out cleaning behavior and oil-well sampling, a parametric study with nearly a hundred thousand scenarios was designed to model fluid behavior during sampling. The simulation scenario is a multi-component model with radial geometry, capable of handling complex reservoir rock, fluid composition, probe geometry, and sampling conditions. Compositional simulation output is then used to generate the comprehensive database of the fluid sampling and cleaning processes. The study is used to determine the sensitive parameters related to sampling and contamination. Full factorial experimental design was used to build nearly one hundred thousand scenarios with more than 10 relevant parameters. Outputs were analyzed through a variety of visualization and statistical techniques to understand cleaning behavior in different initial and operating conditions. One-factorial analysis and statistical tests, including analysis of variance (ANOVA), were used to determine the significance of the different parameters. The most influential parameters have been selected and used as input to the representative model in order to predict pumpout volume and corresponding contamination. In this work, multiple data-driven models such as Neural network, Random Forest, and Gradient Boosting are presented. Furthermore, multiple mathematical equations have been compared to fit the contamination trend, and methods of estimating their best fit parameters are presented. Blind testing has been performed to evaluate performance of the developed models, showing promising results. The workflow, database, and the developed models can be used to perform forward modeling of sampling jobs in different reservoirs, drilling muds, and operating conditions for both wireline and logging while drilling (LWD). This enables an effective and practical job-planning tool implementation, whereby a tool string can be optimized to reduce sampling time while improving the quality. The state-of-the-art workflow deployed in a commercial reservoir simulator combines physics, programming, statistical analysis, and machine learning techniques to tackle the challenging problem of sampling. The workflow and data can be used during operations with various wireline formation testing (WFT) and LWD testing tools to optimize cleanup and sampling of formation fluids. Simulations of different realizations of reservoir properties, drilling mud invasion profiles, and cleanup operations also helped develop a useful and diverse pumpout database.

Abstract This paper presents a novel approach of continuously measuring drilling fluid rheology and density by use of sound signals. A unique apparatus is built with a series of pipe sections designed to exact pre-calculated dimensions to achieve equivalent standard shear rates as stipulated in the American Petroleum Institute (API) Recommended Practice 13D for measuring the rheology of oil-well drilling fluids (from 3 to 600 RPM). Acoustics waves are passed through the fluids of interest and their interaction is recorded and analyzed to deduce the density and rheological properties of the fluids. The concept of resonance as demonstrated by the Barton's pendulums are the basis of the methodology. Sound signals are known to exhibit a damping effect when passing through various media. Pairs of sensors are employed in this set-up and their signal response are first characterized and calibrated with fluids of known properties. Electric current is converted into acoustic signals by piezoelectric sensors mounted of the flowline which are then emitted through the fluids desired to be measured without interrupting the flow. A matching sensor receives these damped signed and reconverts them back to electromotive potentials for recording by a data acquisition unit. The signals are then analyzed by applying statistical techniques to interpret and obtain the fluids physical properties. Owing to the nature of the task, the goal of accurately achieving simultaneous measurement of density and viscosity is attained by applying an ensemble machine learning algorithm, known as Multivariate Random Forest. Pure chemicals and fluids of known properties form the training group on which the predictive model is built for subsequent testing on new mud samples flowing through each section. The pipe sections generate shear rates covering the standard range adopted in oilfield reports. Results from each pair of sensors are analyzed and compared with dial readings from rotational viscometers; these have shown to be within a narrow band of error. As a result of this work, the voltage outputs are sent continuously and in real-time to a processing computer that converts the values to dial readings at standard shear rates, while not disrupting the flow. This can aid in the better monitoring and surveillance of the entire fluid system of the well, which is highly beneficial to well control. The system can also be arranged to acquire gel strengths or how the fluid behaves after a fixed period of rest. Improvements can be made on the current procedures for fluid characterization which have remained relatively static for many years. This work engages the disciplines of rheology, acoustics and machine learning, creating a mechanism for continuous and real-time drilling fluid surveillance critical to the enhancement of safe development of petroleum resources.

Abstract Closed loop reservoir management is challenged with building reliable and fast predictive reservoir models to make field decisions. Traditional numerical simulation models can be difficult to characterize, tedious to build and calibrate, and at times computationally prohibitive for short term decision cycles in field applications. On the other hand, pure data-driven methods often lack physical insights and have limited range of applicability. For operational scenarios such as short-term production forecasting, waterflood optimization, production control and understanding major reservoir mechanisms, it is desirable to use a reservoir modeling methodology that is easy to build, history match, compute and interpret. In this work, we propose to use a hybrid and efficient reservoir graph network (RGNet) modeling approach based on time of flight concept that can be built using routinely measured field measurements (such as pressure and rates) and can be used for real-time forecasts, scenario modeling, production optimization and control. We propose a gridding method based on discretized time of flight for multi-well scenario with interference. It simplifies the 3D reservoir flow problem into a graph network representation that can be solved with any commercial reservoir simulator, which enables the RGNet model to be readily applied for various types of fluid physics. The parameters in RGNet model are obtained through assimilating observed data. The RGNet model has a very compact model representation that requires significantly less complexity compared with full-physics 3D models, which leads to very fast simulation. The efficiency of RGNet makes it appealing for applications where many simulation runs are needed. We applied the proposed approach on SPE benchmark reservoir simulation models for single well, multi-well with interference and injector-producer pairs. The calibrated models are used to quantify uncertainty for production forecasting. In all cases, the range of uncertainty is reduced effectively and efficiently with data assimilation. The posterior RGNet models are also shown to provide reasonable estimates of reservoir and well drainage volumes. By virtue of the reduced complexity, the modeling methodology is highly scalable while still retains physical interpretability (in terms of pore volume and transmissibility). We also discuss the potential applications of the method such as reservoir connectivity analysis and well control optimization. The proposed reservoir graph network (RGNet) modeling approach provides a unique and sustainable way to combine advanced analytics and physics to develop an explainable dynamic reservoir model that can be effectively used to understand reservoir behavior and optimize performance. The lightweight model lends itself naturally to fast computation that are required for scenario analysis and optimization.

Abstract This paper elaborates on the concept of successfully applying one combined platform that includes gas condensate dynamic simulation models, surface network, and individual well models interacting and running sequentially within a closed loop. The study also highlights the value created by integrating dynamic modelling, simulation data, history matching (covering gas condensate reservoirs consisting of gas producers and injectors under the recycle mode) with continuously calibrated well and network models, thereby allowing end-users to make the best use of an integrated system for their dynamic production forecasting. The dynamic reservoir integration methodology incorporates as a first step the data coming from the reservoir simulator model as the main source of reservoir parameters to build a comprehensive system for enhancing production forecasting profiles. In an automatic routine, the simulation data provides the Inflow Performance Relationship, which gets transferred to the well's models, so a well performance curve (WPC) can be generated automatically. Once the latter is generated, it gets transferred to a recycle production-injection network model where a user-configured surface network scenario optimizes in an IAOM (Integrated Asset Operation Model) environment to calculate the rates corresponding to each well taking into consideration distinct constraints. The rates generated are transferred back to the reservoir simulator as well control parameters to initialize the next step of the loop and begin the process under updated conditions. The number of steps, termed as the schedule of the run, are determined by the user based on the forecasting objectives. From the practical point of view, this dynamic reservoir integration mainly targets at getting the best possible assessment from the available data, assumptions, and constraints. The value generated by having a dynamic integration, including all main components of the field/reservoir production, initially relies on the accurate understanding of the dynamic behavior of the hydrocarbon reservoir in order to predict future performance under different development and production approaches. There are several reasons why an integrated approach proved to have strong value creation: Reliable evaluation of the entire production system from reservoir to processing facilities. Continuous assessment of well and network performance. Verifying consistency of data reducing uncertainties. Minimizing underlying assumptions and constraints. It is worth mentioning that during this implementation, the entire system employed compositional models where a high number of components and pseudo components were part of the system, and the thermodynamic behavior added further rigor to the overall calculations. This advanced methodology of carrying out dynamic integration of surface to sub-surface in a production platform framework enhances various key factors of numerical simulation, such as run time estimation, optimal incorporation of surface parameters, identifying gaps between the surface and sub-surface system and enabling the user to perform key business scenarios in an efficient and flexible workflow-based production platform system.

Abstract ADNOC has completed the second phase of its ambitious integrated capacity model (ICM) with the overall aim to optimize its fluid production portfolio from the well level to the processing facilities. The business drivers are to establish capabilities to optimize high-value products and proactively react to market demand changes effectively. Such capabilities required a robust thermodynamics engine with component-wise tracking based on a country-wide capacity model network comprised of a myriad of wells, pipelines, and separators. Fluid samples are not available for all the wells in a field. An innovative workflow was created to assign appropriate composition at the well level based on the data set available for a subset of wells. The captured compositions were then passed to the ICM's hydraulic calculation engine to track the fluid compositions at the required nodes across the network. The existing data model was expanded and user interfaces were created to capture the complexities within the network and visualize the changes in fluid properties, particularly composition, density, and flow rates, at the defined nodes. This digital transformation initiative had to overcome the following complexities to improve accuracy and enable faster decision-making: Incorporation of data from more than 20 fields, 150+ reservoirs, 5000+ wells Optimizing the country-wide network model comprised of wells, pipelines, and separators Performing multiple pre-conceived daily scenarios with 60-month forecasts for production and injection rates Accounting for lateral and vertical composition variation within a reservoir Mixing of fluids at different points in the network at different pressures Implementation of a unified equation of state (EOS) to enable component tracking The network model successfully captured this complexity and predicted capacities for all custody-transfer points between upstream and downstream networks demonstrating a good match (>90%) between the actual laboratory-based measurements and the ICM results. The tool also offered the capability to maximize production of desired components at the source level to meet the dynamic energy demands of the country, allowing a 1-3% profit improvement in the base operating plans. Alternate scenarios offer additional views on how to obtain the same upstream liquid production targets while maximizing downstream gas revenues, hence overall country profitability. The ICM recommended suitable targets during crisis conditions to react accurately to unexpected market fluctuations. Implementation of the unified EOS along with component tracking creates new avenues for digital transformation by allowing the operator to optimize high value products and answer to demand changes quickly. Multiple scenarios can be analysed and visualised to support decision makers to increase profitability in a highly competitive hydrocarbon market from rock to stock.

Abstract 14 reservoir fluids from the Middle East and Gulf of Mexico were used to investigate whether cubic equations (SRK EoS) are capable of modeling published asphaltene data with and without gas injection or if it requires more complex equations of state. Each fluid had published data for: 1) fluid composition, 2) routine PVT/EOR data, 3) amount of asphaltene (wt% in STO), and 4) asphaltene onset pressure data with and without gas injection. The starting point for the model developments was a reservoir fluid composition. The models were first tuned to match routine and EOR PVT data. For Asphaltene modeling, the heavy components were subsequently split into an asphaltene and a non-asphaltene part. EoS parameters were assigned to the asphaltene components to accurately simulate measured asphaltene onset pressures. The development of an asphaltene model for a selected fluid was also compared between SRK, CPA, and PC-SAFT EoS. This study shows that it is possible to obtain a good match of both PVT data and asphaltene onset pressure data for all 14 fluids with the SRK EoS. Using optimized EoS parameters obtained from this study, a model simulating the asphaltene behavior of a fluid can in most cases be developed by adjusting only the T c of the asphaltene component to match an asphaltene data point. For fluids with a high asphaltene content, further tuning may still be required, but the optimized parameters provide a very good starting point. A step-by-step guide is provided which shows how to develop a model that gives a good match of asphaltene onset pressure data with the SRK equation. An equally good match could be achieved by the CPA and PC-SAFT equations, but these two more complex equations provided no advantage over the simpler SRK equation and the tuning was found to be more cumbersome.

Abstract Prior to field-scale development of chemical EOR processes, pilot tests are widely accepted in the oil industry as a standard method to determine the efficiency of the formulated chemicals. During such tests there can be significant differences in temperature between the injected and reservoir fluids. This results in a cool-down of the wellbore, near-wellbore and inter-well regions which can be aggravated in high temperature reservoirs. Key features of surfactant flooding, such as interfacial tension (IFT) reduction between the oil and water phases, depend strongly on temperature. As a result it is necessary to estimate the strength of this cool-down effect upon designing pilot tests. This is the topic of this paper which addresses several scales ranging from near-wellbore to pilot pattern. This work assesses the impact of temperature gradients during a pilot test on the efficiency of surfactant injection using advanced reservoir simulation. We first determine the temperature window seen by an injected surfactant solution with the aim of understanding how it may drive surfactant formulation. We then apply our findings on a pilot design study, with a model including a temperature dependent IFT. We analyse the sensitivity of given injection sequence and operational constraints to specific properties of the injected surfactant solution (low-IFT temperature windows) and then propose a methodology to determine the most efficient injection sequence for a specific surfactant formulation. We show that the temperature window encountered by the surfactant is very sensitive to thermal history of the reservoir and injection temperature. The analysis of chemicals slug thermal and compositional mixing with in-situ fluids is found to be a game changer for reliable pilot design and production forecasts. Obtaining the lowest IFT between oil and water phases is the key in surfactant flooding efficiency: as such the in-situ temperature profiles obtained by simulation and the formulation design at the laboratory should be closely linked. We demonstrate that the process is considerably sensitive to temperature and suggest as a result the following workflow for the design of injection sequences during a pilot test: 1) assessing the temperature window that will be seen by the surfactant using simulation, 2) designing an adequate surfactant formulation, 3) estimating an optimal and robust surfactant injection sequence using simulation, 4) iterating between the three previous steps until an optimal recovery is achieved with a laboratory-formulated, cost-effective surfactant. The impact of temperature on surfactant pilot tests is a specific, not so well documented subject, although it is a capital step in the feasibility assessment of a field scale deployment of surfactant EOR technology. Our workflow yields a reliable assessment of temperature landscapes seen by the injected fluids, which may then be used to test surfactant formulations from near-wellbore to interwell/reservoir scale (e.g. for designing and performing single well chemical tracer tests). As such it should be of interest to petroleum engineers, production engineers and chemists working on the design of chemical EOR processes.

Abstract Carbonate acidization is the process of creating wormholes in carbonate rocks by injecting acid to increase reservoir permeability and consequent oil production. Nevertheless, some reservoir oils might contain certain amounts of asphaltenes that affect the wormholing process. During acidization in the presence of these asphaltenes, the injected acid might not effectively react with rock. This study illustrates the processes of interactions between acid, rock, and asphaltene. It is found that a cumbersome and intricate compound known as acid-asphaltene sludge is formed when the acid reacts with the deposited asphaltene. The formation of this sludge would reduce oil productivity and acid injectivity. In this study, a numerical model is developed to predict the amount of asphaltene deposition, acid-asphaltene reaction, and formation of acid-asphaltene sludge. A 1D numerical model was developed to predict the deposition of asphaltene and acid-asphaltene reaction in radial geometry. Then, the developed model was utilized to predict sludge formation and its related effect on reservoir permeability. Furthermore, the developed asphaltene deposition model was validated against two different sets of experimental data where various factors affecting sludge formation were determined. The detailed analysis showed that although sludge formation is few percent, its concentration increases with time and acid injection. The results also indicate that acid-asphaltene sludge causes an irreversible and permanent formation damage from 40 percent blockage to total pore blockage, depending upon the location of sludge deposition in pore body or pore throat, respectively. Hence, reduction in hydrocarbon productivity is expected at a much lower rate as opposed to that before the acidization job. In addition, the results showed that the amount of acid-asphaltene sludge formation depends on asphaltene percentage, acid concentration, acid flow rate, and flow period. To the best of our knowledge, there is no published work focusing on modeling acid-asphaltene sludge formation and its subsequent wormhole development. This paper provides more insight into the effects of asphaltene deposition and acid-asphaltene sludge formation on wormhole creation and propagation.

Abstract The present paper regards a high-resolution 3d fracture DMX modelling exercise applied to Triassic dolomites of the Ghail Fm, outcropping along Wadi Bih in the territory of Ras Al-Khaimah (UAE). The outcropping rocks show a low primary porosity, are well bedded and highly fractured (jointed) up to cm scale. They are an equivalent of the hydrocarbon producing Khuff Fm in the subsurface, and as such the exersice shows the relevance of applying new technologies to outcrop observations to confront with subsurface situations. Secondly, the exersize shows several new unique elements and related technologies not presented previously. The workflow presented which allows to arrive from observations to a 3d data driven fracture model, applying new technologies and algorithms using specific modules of the DMX protocol. Data collected comprise direct hand measurements of fractures and bedding planes, and 3d calibrated HR photographs. The scale of the model ranges from cm to meter scale fractures (core to borehole image scale). The parameters that have been analysed are size (length, height), orientation (set recognition), truncation relationships (BIXTV) between fractures and bedding planes, and spatial distribution (spacing) of the single fracture sets. The analyses and confrontation between the bedding tilt and the orientation of the 5 recognised fracture sets demonstrates that the fracture network is passively tilted with the bedding plane. A methodology is presented to arrive from in principle 1d (lineament) observations upon 2d sections (bedding planes surfaces and vertical sections) to statistical models and the usage in 3d fracture modeling. The resulting 3d model is generated as a Data Driven Model using the DMX protocol, which is directly connected to the observed features which are themselves present in the model, and comprises 5 sets with the following parameters reflecting the observed mathematical models: Specific spatial spacing relationships, size distributions (lengths and heights), relationships with modelled bedding plane surfaces, undulation of fractures, and 3d spatial truncations (XTV) between fractures and between fractures and bedding planes. This model is interrogated for 3d connectivity in different directions, P32 density (related to fracture porosity) and can be confronted with subsurface datasets from (preferably oriented) cores. The example shows for the first time how relatively simple and straightforward observations, which can be collected by hand, or with 3d stereofotogrammetrical techniques or drone technologies, can be used to build 3d analogue models that accurately mimic geological relationships.

Coiled Tubing Drilling (CTD) has been growing and developed rapidly through the last two decades. There have been numerous highly successful applications of CTD technology in Alaska, Canada, Oman and the United Arab Emirates (Sharjah Sajaa and Dubai Murgham fields), among other places. Currently, Saudi Arabia has undertaken a campaign for the last seven years that has shown successful results in gas reservoirs. ADNOC initiated a trial Coiled Tubing Underbalanced Drilling (CTUBD) project in the onshore tight gas reservoirs in Abu Dhabi, United Arab Emirates beginning operations 1-December-2019. The initial trial will consist of three (3) wells. The purpose of the trial is to assess the suitability of CTUBD for drilling the reservoir sections of wells in these fields, and further application in others. The reason for choosing coiled tubing for drilling the reservoir sections is based upon the high H2S content of the reservoir fluids and the premise that HSE can be enhanced by using a closed drilling system rather than an open conventional system. The three wells will be newly drilled, cased and cemented down to top reservoir by a conventional rig. The rig will run the completion and Christmas tree before moving off and allowing the coiled tubing rig to move onto the well. The coiled tubing BOPs will be rigged up on top of the Christmas tree and a drilling BHA will be deployed through the completion to drill the reservoir lateral. The wells will be drilled underbalanced to aid reservoir performance and to allow hole cleaning with returns being taken up the coiled tubing / tubing annulus. The returns will be routed to a closed separation system with produced gas and condensate being primarily exported to the field plant via the production line, solids sparge to a closed tank or pit and the drilling fluid re-circulated. The primary drilling fluid will be treated water; however, nitrogen may be required for drilling future wells in the field and will be required regardless for purging gas from the surface equipment during operations. A flare will also be required for emergency use and for start-up of drilling. If the trial proves a success, a continuous drilling plan will be put in place.

The molecular-based soft-SAFT equation of state has been applied to describe the chemisorption of CO 2 in non-aqueous hybrid solvents of mixtures of AMP and glycols. The reactive nature of the CO 2 absorption process in non-aqueous amines was implicitly modelled by the formation of CO 2 -amine physical aggregates bounded by strong and localised intermolecular interactions. This modelling framework only required VLE data on the absorption of CO 2 in amine solvents, without the need for additional information such as speciation reactions or equilibrium constants. Subsequently, the developed models were used to examine the CO 2 capture performance of these hybrid solvents in terms of absorption cyclic capacity and heat of regeneration as key performance indicators using a simple and short-cut estimation method. Results show that for the same total amine mass concentration, non-aqueous AMP solvents possess a 30-40% decrease in total heat of regeneration compared to their aqueous counterparts at the expense of a 20-50% reduction in cyclic capacity. These results validate the reliability of the molecular modelling approach as an attractive and valuable tool for the screening of chemical solvents and process modelling.

Meeting the gas production target while maximizing the total gas condensate and hence the revenue from condensate reservoirs is one of the key business drivers for an operating company. This paper describes a comprehensive simulation process to strategize production optimization, which helps in meeting the target molar fraction at the delivery point from the overall asset producing from complex reservoirs with varying fluid properties and achieving the overall target of lean gas production. This process uses an integrated approach encompassing the various nodes in a production system, starting from reservoir to the export system to assure a representative and accurate prediction. In the first step, using the representative fluid model and the desired target production from the field, the well capacity-based rates are allocated to the individual production strings. In the second step, a component level optimizer is used to estimate the contribution of each well based on the well production stream composition. In the third and final step, this contributed production figure is fed back to the surface network hydraulic simulator to assess the back-pressure impact on the overall production and the achievable field-target. The objective of maximizing the condensate production was fulfilled considering the provided constraints of operating guidelines, reservoir, wells, and surface facility capacities. Two different scenario runs were put together where the target gas production was achieved while increasing the condensate production by maximizing the condensate specific molar components and minimizing the heavy molar components. The expected condensate production was forecasted to increase by 5% in the scenario. As the predictive hydraulic model is seamlessly integrated with the true field operating conditions, the outlined optimization process ensured that various business-scenarios accurately forecast the system behavior under various operating conditions. In the scenario, the forecast was able to maximize the condensate while meeting the production target within 1% tolerance limit. The outlined approach provides a clear step by step standardized process approach which can be expanded to cater to the other business needs, such as minimizing the H2S for reducing the corrosion problem, minimizing the C7+ components to produce a high-quality condensate. Such a framework based standardized methodology incorporates a seamless integration between the molar composition optimization and the associated hydraulic calculations-based optimization. Solving the both objective optimization functions simultaneously, this approach provides a novel way to address a vital industry business objective.

Acquiring reservoir fluid samples through formation testers is critical to asset evaluation in most oil and gas drilling operations. From the time this technology was introduced to the industry, the key challenges have been in planning the job, estimating contamination during operation, and obtaining clean fluid samples in the shortest time possible. The objective of this paper is to create a new data-driven model to proactively simulate the cleaning process in order to provide a practical job-planning tool that optimizes fluid sampling. After detailed analysis of formation pump-out cleaning behavior and oil-well sampling, a parametric study with nearly a hundred thousand scenarios was designed to model fluid behavior during sampling. The simulation scenario is a multi-component model with radial geometry, capable of handling complex reservoir rock, fluid composition, probe geometry, and sampling conditions. Compositional simulation output is then used to generate the comprehensive database of the fluid sampling and cleaning processes. The study is used to determine the sensitive parameters related to sampling and contamination. Full factorial experimental design was used to build nearly one hundred thousand scenarios with more than 10 relevant parameters. Outputs were analyzed through a variety of visualization and statistical techniques to understand cleaning behavior in different initial and operating conditions. One-factorial analysis and statistical tests, including analysis of variance (ANOVA), were used to determine the significance of the different parameters. The most influential parameters have been selected and used as input to the representative model in order to predict pumpout volume and corresponding contamination. In this work, multiple data-driven models such as Neural network, Random Forest, and Gradient Boosting are presented. Furthermore, multiple mathematical equations have been compared to fit the contamination trend, and methods of estimating their best fit parameters are presented. Blind testing has been performed to evaluate performance of the developed models, showing promising results. The workflow, database, and the developed models can be used to perform forward modeling of sampling jobs in different reservoirs, drilling muds, and operating conditions for both wireline and logging while drilling (LWD). This enables an effective and practical job-planning tool implementation, whereby a tool string can be optimized to reduce sampling time while improving the quality. The state-of-the-art workflow deployed in a commercial reservoir simulator combines physics, programming, statistical analysis, and machine learning techniques to tackle the challenging problem of sampling. The workflow and data can be used during operations with various wireline formation testing (WFT) and LWD testing tools to optimize cleanup and sampling of formation fluids. Simulations of different realizations of reservoir properties, drilling mud invasion profiles, and cleanup operations also helped develop a useful and diverse pumpout database.

This paper presents a novel approach of continuously measuring drilling fluid rheology and density by use of sound signals. A unique apparatus is built with a series of pipe sections designed to exact pre-calculated dimensions to achieve equivalent standard shear rates as stipulated in the American Petroleum Institute (API) Recommended Practice 13D for measuring the rheology of oil-well drilling fluids (from 3 to 600 RPM). Acoustics waves are passed through the fluids of interest and their interaction is recorded and analyzed to deduce the density and rheological properties of the fluids. The concept of resonance as demonstrated by the Barton's pendulums are the basis of the methodology. Sound signals are known to exhibit a damping effect when passing through various media. Pairs of sensors are employed in this set-up and their signal response are first characterized and calibrated with fluids of known properties. Electric current is converted into acoustic signals by piezoelectric sensors mounted of the flowline which are then emitted through the fluids desired to be measured without interrupting the flow. A matching sensor receives these damped signed and reconverts them back to electromotive potentials for recording by a data acquisition unit. The signals are then analyzed by applying statistical techniques to interpret and obtain the fluids physical properties. Owing to the nature of the task, the goal of accurately achieving simultaneous measurement of density and viscosity is attained by applying an ensemble machine learning algorithm, known as Multivariate Random Forest. Pure chemicals and fluids of known properties form the training group on which the predictive model is built for subsequent testing on new mud samples flowing through each section. The pipe sections generate shear rates covering the standard range adopted in oilfield reports. Results from each pair of sensors are analyzed and compared with dial readings from rotational viscometers; these have shown to be within a narrow band of error. As a result of this work, the voltage outputs are sent continuously and in real-time to a processing computer that converts the values to dial readings at standard shear rates, while not disrupting the flow. This can aid in the better monitoring and surveillance of the entire fluid system of the well, which is highly beneficial to well control. The system can also be arranged to acquire gel strengths or how the fluid behaves after a fixed period of rest. Improvements can be made on the current procedures for fluid characterization which have remained relatively static for many years. This work engages the disciplines of rheology, acoustics and machine learning, creating a mechanism for continuous and real-time drilling fluid surveillance critical to the enhancement of safe development of petroleum resources.

Closed loop reservoir management is challenged with building reliable and fast predictive reservoir models to make field decisions. Traditional numerical simulation models can be difficult to characterize, tedious to build and calibrate, and at times computationally prohibitive for short term decision cycles in field applications. On the other hand, pure data-driven methods often lack physical insights and have limited range of applicability. For operational scenarios such as short-term production forecasting, waterflood optimization, production control and understanding major reservoir mechanisms, it is desirable to use a reservoir modeling methodology that is easy to build, history match, compute and interpret. In this work, we propose to use a hybrid and efficient reservoir graph network (RGNet) modeling approach based on time of flight concept that can be built using routinely measured field measurements (such as pressure and rates) and can be used for real-time forecasts, scenario modeling, production optimization and control. We propose a gridding method based on discretized time of flight for multi-well scenario with interference. It simplifies the 3D reservoir flow problem into a graph network representation that can be solved with any commercial reservoir simulator, which enables the RGNet model to be readily applied for various types of fluid physics. The parameters in RGNet model are obtained through assimilating observed data. The RGNet model has a very compact model representation that requires significantly less complexity compared with full-physics 3D models, which leads to very fast simulation. The efficiency of RGNet makes it appealing for applications where many simulation runs are needed. We applied the proposed approach on SPE benchmark reservoir simulation models for single well, multi-well with interference and injector-producer pairs. The calibrated models are used to quantify uncertainty for production forecasting. In all cases, the range of uncertainty is reduced effectively and efficiently with data assimilation. The posterior RGNet models are also shown to provide reasonable estimates of reservoir and well drainage volumes. By virtue of the reduced complexity, the modeling methodology is highly scalable while still retains physical interpretability (in terms of pore volume and transmissibility). We also discuss the potential applications of the method such as reservoir connectivity analysis and well control optimization. The proposed reservoir graph network (RGNet) modeling approach provides a unique and sustainable way to combine advanced analytics and physics to develop an explainable dynamic reservoir model that can be effectively used to understand reservoir behavior and optimize performance. The lightweight model lends itself naturally to fast computation that are required for scenario analysis and optimization.

This paper elaborates on the concept of successfully applying one combined platform that includes gas condensate dynamic simulation models, surface network, and individual well models interacting and running sequentially within a closed loop. The study also highlights the value created by integrating dynamic modelling, simulation data, history matching (covering gas condensate reservoirs consisting of gas producers and injectors under the recycle mode) with continuously calibrated well and network models, thereby allowing end-users to make the best use of an integrated system for their dynamic production forecasting. The dynamic reservoir integration methodology incorporates as a first step the data coming from the reservoir simulator model as the main source of reservoir parameters to build a comprehensive system for enhancing production forecasting profiles. In an automatic routine, the simulation data provides the Inflow Performance Relationship, which gets transferred to the well's models, so a well performance curve (WPC) can be generated automatically. Once the latter is generated, it gets transferred to a recycle production-injection network model where a user-configured surface network scenario optimizes in an IAOM (Integrated Asset Operation Model) environment to calculate the rates corresponding to each well taking into consideration distinct constraints. The rates generated are transferred back to the reservoir simulator as well control parameters to initialize the next step of the loop and begin the process under updated conditions. The number of steps, termed as the schedule of the run, are determined by the user based on the forecasting objectives. From the practical point of view, this dynamic reservoir integration mainly targets at getting the best possible assessment from the available data, assumptions, and constraints. The value generated by having a dynamic integration, including all main components of the field/reservoir production, initially relies on the accurate understanding of the dynamic behavior of the hydrocarbon reservoir in order to predict future performance under different development and production approaches. There are several reasons why an integrated approach proved to have strong value creation: Reliable evaluation of the entire production system from reservoir to processing facilities. Continuous assessment of well and network performance. Verifying consistency of data reducing uncertainties. Minimizing underlying assumptions and constraints. It is worth mentioning that during this implementation, the entire system employed compositional models where a high number of components and pseudo components were part of the system, and the thermodynamic behavior added further rigor to the overall calculations. This advanced methodology of carrying out dynamic integration of surface to sub-surface in a production platform framework enhances various key factors of numerical simulation, such as run time estimation, optimal incorporation of surface parameters, identifying gaps between the surface and sub-surface system and enabling the user to perform key business scenarios in an efficient and flexible workflow-based production platform system.

ADNOC has completed the second phase of its ambitious integrated capacity model (ICM) with the overall aim to optimize its fluid production portfolio from the well level to the processing facilities. The business drivers are to establish capabilities to optimize high-value products and proactively react to market demand changes effectively. Such capabilities required a robust thermodynamics engine with component-wise tracking based on a country-wide capacity model network comprised of a myriad of wells, pipelines, and separators. Fluid samples are not available for all the wells in a field. An innovative workflow was created to assign appropriate composition at the well level based on the data set available for a subset of wells. The captured compositions were then passed to the ICM's hydraulic calculation engine to track the fluid compositions at the required nodes across the network. The existing data model was expanded and user interfaces were created to capture the complexities within the network and visualize the changes in fluid properties, particularly composition, density, and flow rates, at the defined nodes. This digital transformation initiative had to overcome the following complexities to improve accuracy and enable faster decision-making: Incorporation of data from more than 20 fields, 150+ reservoirs, 5000+ wells Optimizing the country-wide network model comprised of wells, pipelines, and separators Performing multiple pre-conceived daily scenarios with 60-month forecasts for production and injection rates Accounting for lateral and vertical composition variation within a reservoir Mixing of fluids at different points in the network at different pressures Implementation of a unified equation of state (EOS) to enable component tracking The network model successfully captured this complexity and predicted capacities for all custody-transfer points between upstream and downstream networks demonstrating a good match (>90%) between the actual laboratory-based measurements and the ICM results. The tool also offered the capability to maximize production of desired components at the source level to meet the dynamic energy demands of the country, allowing a 1-3% profit improvement in the base operating plans. Alternate scenarios offer additional views on how to obtain the same upstream liquid production targets while maximizing downstream gas revenues, hence overall country profitability. The ICM recommended suitable targets during crisis conditions to react accurately to unexpected market fluctuations. Implementation of the unified EOS along with component tracking creates new avenues for digital transformation by allowing the operator to optimize high value products and answer to demand changes quickly. Multiple scenarios can be analysed and visualised to support decision makers to increase profitability in a highly competitive hydrocarbon market from rock to stock.