Innovative High Permeability Streak Characterization and Modeling Utilizing Static and Dynamic Data in a Complex Giant Mature Oil Field in the Middle East

This paper proposes an integrated approach to model High Permeability Streaks (HPS) using the case study of heterogeneous carbonate Reservoir B, utilizing static and dynamic data. Modelling the HPS is critical as they play an important role in fluid dynamics within the reservoir. The impact is observed from 60 years of development, where flood front movement is captured by rich density of Pulsed Neutron and recently drilled open hole logs. Injection water is overriding from tighter lower subzones (injected zones) to permeable upper subzones of the reservoir, thereby leaving the tighter lower subzones unswept. Gas cusping down to the oil zone occurs through the HPS resulting in non-uniform gas cap expansion, which leads to early gas breakthrough in producers near the gas cap. The problem with characterizing HPS is associated with their thickness- in Reservoir B it ranges from 0.5 to 2.5ft and occur in multiple subzones in the upper part of the reservoir. The standard triple combo suite of logs does not have the resolution to detect these thin HPS. In addition, the cored interval of the HPS is mainly disintegrated which is attributed majorly to well sorted grain-supported lithofacies. Therefore, sampling for porosity & permeability via Routine Core Analysis (RCA) and Capillary pressure as well as pore throat distribution using Mercury Injection Capillary Pressure (MICP) method is extremely difficult. This results in a gap in the input dataset for the static models, where the higher permeability samples are not captured in logs or cores and are therefore under-represented. Current approach to unify this gap is to use permeability multipliers, which does not honor geological trends. The HPS in Reservoir B has added complexities when compared to other regional HPS. Not only are they multiple and distributed across subzones, there is also preferential movement of water through the HPS within the same area. Of the 3 upper subzones that have HPS, in some areas, water injected in lower subzone will override the HPS in the middle and move right to the HPS in the top subzone, thereby ignoring the hierarchical flood front movement from bottom to the top. A robust workflow was developed in order to address and resolve the above mentioned uncertainties related to High Permeability Streaks. The proposed integrated workflow consisted of five stages: Developing a robust geological conceptual model Mapping spatial distribution & continuity Capturing the vertical presence in cored & uncored wells (depth & thickness) Permeability Quantification of HPS using Well Test Measurements Modelling High Permeability Streaks The paper highlights the utilization of a range of static (core, Routine Core Analysis (RCA), image logs, OH logs) and dynamic data (Pulse Neutron Logs (PNL's), later drilled Open Hole Logs, Production Logging Tools (PLTs) and well test data). Quantitative (HPS depth indicated by water saturation profile indicated by waterflood movement) and Qualitative (Flooding observed but HPS depth is uncertain) depth indicators/flags were generated from the data set and became the foundation of the modelling the HPS. The first step in the workflow is to establish a robust geological conceptual model. For Reservoir B, certain facies contribute to HPS, which are mainly leached Rudist Rudstones and Coated grain Algal Floatstones as well as well sorted Skeletal Grainstones. Based on core observations, they have confirmed vertical stratigraphic presence in each subzone (top, mid, base) which is attributed to storm events. These were consequently mapped using average thickness from core descriptions and revised using contributing facies trend maps and qualitative dynamic observations. These maps served as basis for probability trend distribution for static rock type models. The vertical presence of HPS was increased from 10% to 30% by re-introducing them in the missing core intervals using quantitative dynamic flags and thickness from isochores. Consequently, permeability were assigned in the missing section using the proposed permeability enhancement technique that honors the verified well test measurements. Based on the above improvements, the HPS intervals were mapped to the static rock type with best reservoir quality (SRT 1), which is also linked to certain geological attributes (i.e. lithofacies, diagenetic overprint & depositional environment). The enhanced permeability in the identified HPS intervals is also reflected as upgraded SRT (from lower SRT 2 to best SRT 1). The overall impact is observed by improvement of poro-perm cloud, with added control points for HPS SRT (1), which is vital for permeability modelling. The updated permeability model, captures high perm streaks in terms of vertical presence and magnitude. By introducing higher permeability in the upper subzones of the reservoir, the water overriding/gas cusping phenomena could then be mimicked in the dynamic model. The proposed methodology is an integrated workflow that maximizes the input from each disciplines (G&G, Petrophysics and Reservoir Engineering) to create a robust static model through incorporation of high permeability streaks. The use static and dynamic data, has helped to establish HPS existence/preference, which then could be used to upgrade the permeability/SRT. This will in turn lead to a better static model and a better history match in the dynamic model. It will also led to better remaining in place prediction and enable accurate prediction for future field development, especially where EOR is involved.

Insights on Water Sources for Injection Development of Large Oilfields in Southeast Iraq and Appraisal of Paleogene Shallow Water Source Formation

As many large oilfields in southeast Iraq entered the final stage of depletion development, water injection appears to be the most economical and technically feasible method to enhance oil recovery. Considering the shortage of freshwater and huge investment in seawater supply project, it is very important to appraise and optimize the favorable shallow water source formation to ensure sufficient water injection supply. Based on regional seismic, well data, core analysis and production test data, Paleogene sequence stratigraphy was determined by integrating well and seismic interpretation. Under framework of sequence stratigraphy, the sedimentary evolution of main water source formations was characterized. Subsequently, combined with core analysis and special logging data, the petrophysical characteristics of the formations were evaluated, and the volume of the regional water source was estimated. The research shows that: 1) Dammam limestone and Ghar sandstone are the two main Paleogene shallow water source formations; 2) Dammama developed carbonate shelf, from southwest to northeast, the formation thickness decrease with the sedimentary evolved from inner slope to out slope. Expose and dissolution increased the porosity which is favorable for water storage; 3) Ghar developed alluvial and delta, from southwest to northeast, the formation thickness increase with the sedimentary evolved from alluvial fan, alluvial river to delta. Delta developed abundant and unconsolidated sandstone with high porosity and permeability; 4) The water sample analysis showed the water belong to Cacl2 type with total dissolved solids greater than 250,000 ppm which indicated well sealing condition. Production tests have shown that both Dammam and Ghar have a water supply capacity of 8,000-10,000 barrels per day. The preliminary evaluation of the water volume in the Ghar area can up to 1 trillion barrels. Paleogene shallow water formation is currently the most realistic and economic water source choice for water injection to enhance oil recovery in large oil fields in southeastern Iraq. Dammam formation and Ghar formation of Paleogene had the characteristics of shallow buried, good water quality and sufficient reserves. Thus, they are the preferred target water source formations for injection development of large oilfields in southeastern Iraq.

Development of HFU-based permeability prediction models using core data for characterisation of a heterogeneous Oligocene sand in the Nam Con Son basin

Core data by both routine and special core analysis are required to understand and predict reservoir petrophysical characteristics. In this research, a total number of 50 core plugs taken from an Oligocene sand (T30) in the Nam Con Son basin, offshore southern Vietnam, were tested in the core laboratory of the Vietnam Petroleum Institute (VPI). The results of routine core analysis (RCA) including porosity and permeability measurements were employed to divide the study reservoir into hydraulic flow units (HFUs) using the global hydraulic elements (GHEs) approach. Based on five classified HFUs, 16 samples were selected for special core analysis, i.e., mercury injection capillary pressure (MICP) and grain size analyses for establishing non-linear porosity-permeability model of each HFU based on Kozeny-Carman equation, which provides an improved prediction of permeability (R2 = 0.846) comparing to that by the empirical poro-perm relationship (R2 = 0.633). In addition, another permeability model, namely the Winland R35 method, was applied and gave very satisfactory results (R2 = 0.919). Finally, by integrating the results from MICP and grain size analyses, a good trendline of pore size distribution index (λ) and grain sorting was successfully obtained to help characterise the study reservoir. High λ came with poor sorting, and vice versa, the low λ corresponded to good sorting of grain size.

Application of Micro-resistivity Scanning Imaging Logging Data in Reservoir Thickness Division

Reservoir parameter interpretation is one of the main contents of reservoir description, which affects the whole process of oilfield development. According to the characteristics of micro-resistivity scanning imaging logging, which can directly reflect the changes of lithology and physical properties of reservoirs, this paper compares the thickness and interbed division of reservoirs with conventional logging data, this paper finds out the shortcomings of the conventional logging data in the interpretation of thickness and the division of interlayers, and combines the core analysis data to examine the differences in the correlation on the coring wells, and obtains good results, it has laid the foundation for the establishment of new interpretation procedure.

Comparison of Porosity Change Due to Geochemical Reaction between Samples from High CO2 Field and Depleted Field

When developing a high CO2 field, oil and gas companies must consider the best and most economical carbon capture and storage (CCS) plan. After considering the distance of the storage site and storage capacity, PETRONAS has identified 2 carbonate fields, known as X Field and N Field in East Malaysia as the potential CO2 storage site. Interestingly, both fields are different, as X field is a high CO2 green field, while N field is a depleted gas field. The research team’s initial hypothesis is that N Field would have more severe geochemical reaction between CO2, brine and carbonates compared to X Field, since X field is already saturated with CO2. In order to test the hypothesis, samples from these two fields were selected to undergo static batch reaction analysis, and changes in porosity were determined using Digital Core Analysis (DCA). Both X and N fields are carbonate gas fields, with aquifer zone located below gas zones. The aquifer zones are the preferable CO2 injection zone because the deeper the zone, the longer it will take for the plume migration to happen. For static batch reaction analysis, samples each field were selected from the aquifer zone. After Routine Core Analysis (RCA) and Quality Control (QC), the samples were scanned under the high resolution microCT scan, before they were saturated into the respective synthetic brine. After saturation is completed, both brine and samples were placed inside a batch reactor, where the reactor’s pressure and temperature are set according to the field’s pressure and temperature. Once stabilized, the supercritical CO2 is injected into the brine, and was left for 45 days with constant observation. After aging with supercritical CO2, the samples were then scanned under microCT scan once again, using the same resolution, before being analysed via image processing software. Using registration algorithm software, both pre and post CO2 aging images were overlapped and subtracted digitally. The difference images were analyzed to determine the change in porosity. Samples from X Field has around 1% p.u. increase in porosity, while samples from N field shows increment of 2% p.u. porosity. While N field (depleted field) has higher reaction compared to X field (high CO2) field as per hypothesis, the difference is very minimal, which is much less than expected. The usage of DCA in the analysis enabled the team to determine minute changes that were happening during CO2 batch reaction. Without DCA, the 1% changes usually regarded as equipment’s error margin. The next step would be modelling, where the lab results will be upscaling into field scale, for modelled longer period of time. Hence, although the porosity changes between X and N field are very small under laboratory condition, it can have greater impact in field scale.

Machine Learning for Multiple Petrophysical Properties Regression Based on Core Images and Well Logs in a Heterogenous Reservoir

Reservoir characterization requires accurate prediction of multiple petrophysical properties such as bulk density (or acoustic impedance), porosity, and permeability. However, it remains a big challenge in heterogeneous reservoirs due to significant diagenetic impacts including dissolution, dolomitization, cementation, and fracturing. Most well logs lack the resolution to obtain rock properties in detail in a heterogenous formation. Therefore, it is pertinent to integrate core images into the prediction workflow. This study presents a new approach to solve the problem of obtaining the high-resolution multiple petrophysical properties, by combining machine learning (ML) algorithms and computer vision (CV) techniques. The methodology can be used to automate the process of core data analysis with a minimum number of plugs, thus reducing human effort and cost and improving accuracy. The workflow consists of conditioning and extracting features from core images, correlating well logs and core analysis with those features to build ML models, and applying the models on new cores for petrophysical properties predictions. The core images are preprocessed and analyzed using color models and texture recognition, to extract image characteristics and core textures. The image features are then aggregated into a profile in depth, resampled and aligned with well logs and core analysis. The ML regression models, including classification and regression trees (CART) and deep neural network (DNN), are trained and validated from the filtered training samples of relevant features and target petrophysical properties. The models are then tested on a blind test dataset to evaluate the prediction performance, to predict target petrophysical properties of grain density, porosity and permeability. The profile of histograms of each target property are computed to analyze the data distribution. The feature vectors are extracted from CV analysis of core images and gamma ray logs. The importance of each feature is generated by CART model to individual target, which may be used to reduce model complexity of future model building. The model performances are evaluated and compared on each target. We achieved reasonably good correlation and accuracy on the models, for example, porosity R2=49.7% and RMSE=2.4 p.u., and logarithmic permeability R2=57.8% and RMSE=0.53. The field case demonstrates that inclusion of core image attributes can improve petrophysical regression in heterogenous reservoirs. It can be extended to a multi-well setting to generate vertical distribution of petrophysical properties which can be integrated into reservoir modeling and characterization. Machine leaning algorithms can help automate the workflow and be flexible to be adjusted to take various inputs for prediction.