In-situ stress variability within a reservoir is a primary parameter that controls hydraulic fracture initiation, growth, connectivity, and ultimately drainage and well spacing. This paper highlights the importance of characterizing the variability of in-situ stress and demonstrates the risk of underestimating stimulation treatment size when a treatment design is applied in a “copy-paste” fashion without any modifications to account for variation in pore pressure and in-situ stress across the basin.
Thermal maturity and hydrocarbon generation from unconventional shales has a direct effect on pore pressure and the in-situ stress distribution in reservoir and barrier rocks. An examination of the Bakken Petroleum System (BPS) identifies regions of thermal maturity and higher pore pressure due to hydrocarbon expulsion. Thus, the elevated pore pressure and the resulting in-situ stress vary vertically and laterally within the basin.
Multiple pore pressure profiles and corresponding stress profiles across the BPS were considered to quantify the impact of in-situ stress variability on hydraulic fracture geometry. These profiles include effects of either normal pore pressure regime or over-pressure regime or pressure ramps transitioning from over pressure to normal pressure regimes. For a given stress profile, hydraulic fracture geometries are estimated using a fracture simulator, with multiple calibration points. The hydraulic fracture system and reservoir interactions are simulated in a subsequent production modeling phase which estimates drainage area characteristics, recovery forecasts and optimum well spacing for developing an asset.
Results from stress profile sensitivity emphasize the need to address variability in in-situ stress as it directly impacts well spacing at asset development level. For example, stress profile with a normal pore pressure regime results in longer hydraulic fracture lengths in Middle Bakken (MB) thus requiring three wells per section to infill the asset. Conversely, stress profile with over-pressure regime in MB results in much shorter hydraulic fracture lengths thus requiring more than three wells per section to develop the asset. Assuming over-pressure when its normal pressure could mean an operator over-engineering wells and spending more money than would be needed for optimum economic production, whereas the other way round could lead to a sub-optimal completion and a production shortfall.
Ma, Y. Zee (Schlumberger) | Moore, William (Schlumberger) | Kaufman, Peter (Schlumberger) | Wang, Yating (Schlumberger) | Gurpinar, Omer (Schlumberger) | Luneau, Barbara (Schlumberger) | Gomez, Ernest (Schlumberger)
This article is concerned with the identification of hydrocarbon zones in unconventional plays through lithofacies classification using wireline logs. Identification of hydrocarbon zones in unconventional formations is often different than in conventional formations. For example, in conventional reservoirs, the fractional volume of clay (Vclay) is generally correlated strongly and positively with gamma ray (GR); high GR values commonly indicate low hydrocarbon potential. In unconventional reservoirs, however, Vclay is often correlated weakly with GR, and sometime even negatively. There are several possible explanations for this change of correlation. In source-rock reservoirs, high GR often indicates high total organic carbon (TOC), kerogen, and gas or oil concentrations. In some tight-gas sandstone reservoirs, hydrocarbon-bearing sandstones have abnormally high GR. The change of the correlation between commonly used wireline logs in conventional formations to a different correlation between those same logs in unconventional formations is, in fact, a manifestation of the Simpson's paradox--a counterintuitive phenomenon in probability and statistical data analysis. Here we show how to classify lithofacies and identify high-potential hydrocarbon zones in unconventional resources by discerning the Simpson's paradox when analyzing relationships between wireline logs.
Gomez, Ernest (Schlumberger) | Al-Faresi, Fahad A. Rahman (Kuwait Oil Company) | Belobraydic, Matthew Louis (Schlumberger) | Yaser, Muhammad (Schlumberger) | Gurpinar, Omer M. (Schlumberger) | Wang, James Tak Ming (Schlumberger) | Husain, Riyasat (Kuwait Oil Company) | Clark, William (Schlumberger) | Al-Sahlan, Ghaida Abdullah (Kuwait Oil Company) | Datta, Kalyanbrata (KOC) | Mudavakkat, Anandan (KOC) | Bond, Deryck John (Kuwait Oil Company) | Crittenden, Stephen J. (KOC) | Iwere, Fabian Oritsebemigho (Schlumberger) | Hayat, Laila (KOC) | Prakash, Anand (KOC)
The Burgan Minagish reservoir in the Greater Burgan Field is one of several reservoirs producing from the Minagish formation in Kuwait and the Divided Zone. The reservoir has been produced intermittently since the 1960s under natural depletion. A powered water-flood is currently being planned. The pressure performance of the reservoir has proved hard to explain without invoking communication with other reservoirs. Such communication could be either with other reservoirs through the regional aquifer of through faults to other reservoirs in the Greater Burgan field. Recent pressures are close to the bubble point.
A coarse simulation model of the nearby fields and the regional aquifer was constructed based on data from the fields and regional geological understanding. This model could be history matched to allow all regional pressure data to be broadly matched, a result which supports the view that communication is through the regional aquifer. Using this model to predict future pressure performance suggested that injecting at rates that exceeded voidage replacement by about 50 Mbd could keep reservoir pressure above bubble point. It was recognized that the process of history matching performance was non-unique. This is a particular concern in the context of this study because the model inputs that were varied in the history matching process included aquifer data that was very poorly constrained. To address this problem multiple history matched models were created using an assisted history matching tool. Using prediction results from the range of models has increased our confidence that a modest degree of over-injection can help maintain reservoir pressure.
This paper demonstrates the utility of computer assisted history match tools in allowing an assessment of uncertainty in a case where non-uniqueness was a particular problem. It also emphasizes the importance of understanding aquifer communication when relatively closely spaced fields are being developed.
The Mississippian-Devonian Bakken Formation is a relatively tight mixed carbonate - clastic sequence in the Williston Basin of North Dakota. Although production can exceed 1000 BOPD, hydraulic fracturing is necessary to induce economic production. In late 2007 seven (7) operators along with Schlumberger formed a consortium for the purpose of applying the best available technologies to examine the geologic factors and the drilling and completion principles that affect production. As part of this, three (3) horizontal wells (each 4000 feet in length) were drilled 1500 feet apart into the Middle Bakken Member. Schlumberger positioned an array of 16 triaxial geophones, spaced 100 feet apart, in the middle lateral (Nesson State 42X-36) to monitor the microseismic activity during the hydraulic fracturing of the two (2) outside wells. Different hydraulic fracturing methods were applied in each stage; ranging from a single treatment in the northern lateral (Nesson State 41X-36) to a six (6) staged treatment with swell packers in the southern lateral (Nesson State 44X-36). The microseismic events were studied in the context of our geologic understanding of the area, comparison to the distributions of radioactive and chemical tracers and a reservoir simulation to develop a robust interpretation of effectiveness of the hydraulic fracture treatments. In addition to the microseismic monitoring conducted by Schlumberger, three other microseismic monitoring arrays were installed and monitored. The results of the various monitoring efforts varied dramatically and these differences are worthy of examination.
Stimulation Results and Completion Implications from the Consortium Multi-well Project in the North Dakota Bakken Shale
After a great deal of success with Bakken open hole horizontal completions in Richland County, Montana between 2002 and 2005, operators began to move to the more aerially extensive, yet heterogeneous North Dakota side of the Bakken play in the Williston Basin. However, the early results in the North Dakota Bakken were much
more variable than the Richland County Bakken wells. With the huge resource at stake in North Dakota, it was realized that better understanding of the nature of the
formation's transmissibility, completion, and stimulation efficiency could shorten the learning curve on the economic exploitation of this important oil resource. For this reason a consortium was formed with seven operators, a major service company, the state of North Dakota, and the DOE to drill and complete three science wells. Extensive vertical and horizontal log and core data were collected in these wells. Two wells were completed and one well was utilized as a monitoring well. During the stimulation several seismic arrays were deployed to map out the micro seismic events, including the largest seismic array ever deployed in a horizontal well. In addition, multiple radioactive isotope (RA) and fluid chemical tagging were employed.
This paper presents some of the results of this project and the completion analysis that was done with this data as well as other released data in the Williston Basin. In addition, some conclusions are drawn on the nature of fracture initiation along the wellbore and an attempt is made to provide some insight to the completion optimization.
Conceptual models are used to solve specific problems in selected sectors of reservoirs; study production mechanisms; understand behavior of a particular process in a reservoir system, and assess impacts of changing input parameters during reservoir modeling. They are tools of choice for assessing risks, evaluating "worst-case" scenarios, validating analyst's intuition, and to support informed decision making. Our objective is to demonstrate via two case studies how conceptual numerical models were used to shorten the time required to make reservoir management decisions. The first case study involves making a decision, either to develop or sell an oil property. Target formation is sandstone saturated with heavy oil (12°API gravity) which is overlain by a gas cap. Conceptual numerical simulation models provided answers to two questions:
• What is the impact of gas production from the gas cap on the underlying heavy oil zone?
• Can gas production from up-structure wells meet field deliverability requirements?
Second case study uses conceptual models to optimize well placement and support infill drilling. Infill well placement posed a challenge because thickness of target formation is not well known, and oil zone is bounded on top by a massive impermeable shale boundary, and by oil-water contact (OWC) located about 20-40 feet below.
Conceptual models answered the following questions:
• What type of well to drill--vertical or horizontal?
• What is the impact of horizontal well's vertical placement (offset distance from OWC) on oil recovery and water breakthrough times?
• What is the optimum horizontal well lateral length and its impact on oil recovery?
This paper describes modeling methodology, major observations and conclusions. We discuss the benefits and lessons learned from the case studies and demonstrate that successful application of conceptual models requires identifying key well/reservoir performance drivers and assessing their impacts on the reservoir management decisions.
As a branch of spatial statistics, geostatistics is commonly used to model geologic facies and petrophysical properties. The spatial characteristics of geostatistical methods in variogram, kriging and stochastic simulation have made them the tools of choice for reservoir modeling. Such techniques are especially useful to characterize the reservoir connectivity and sweep efficiency. However, geostatistical modeling methods do not always make an accurate inference of reservoir properties from well-logs to a reservoir model because of the stationarity and ergodicity assumptions and the multiscale of subsurface heterogeneities. This often causes incorrect frequency statistics of reservoir properties, which typically exhibit a non-Gaussian distribution. As a result, estimation of the hydrocarbon in-place and recoverable reserves can be grossly inaccurate and
hundreds of millions of barrels of hydrocarbons can be lost or fictitiously added in a reservoir model. An accurate reservoir characterization should include not only the realistic description of the spatial continuity but also the sound inference of the reservoir properties from fine-scale well-logs to coarse-scale reservoir models. The latter can be achieved through appropriate inference of frequency statistics coupled with spatial statistics.
An application of using both spatial and frequency statistics in a real reservoir modeling example is presented. Facies probability maps were derived from geologic propensity analysis coupled with well-log data, and used for constraining the model to honor the facies spatial association of the reef complex. Careful examination of frequency statistics helped to detect the estimation biases in the initial geostatistical model. An enhanced reservoir modeling workflow with inference coupling spatial and frequency characteristics of the geologic propensity and well-log data was developed, which resulted in a more realistic model. The history match and production performance forecast using the new model was straightforward. More importantly, the new model shows more than 100 million barrels of additional oil in-place compared to the previous geostatistical models and uncertainty evaluation based on the facies scenarios and other geologic and petrophysical variables confirms the new result. The realization of more subsurface resource has implications for future field development to target unswept oils in the reservoir.
Modeling naturally fractured reservoirs is difficult because of the need to characterize the fractures, matrix and the matrix-fracture interaction. It becomes more challenging if the naturally fractured reservoir produces wet gas, condensates and water. Three different three dimensional, compositional models--single-porosity (SP), single-porosity with alpha factor (SPWAF), and dual-porosity single permeability (DPSP) of the study area, were studied.
The models were calibrated against measured pressure, historical oil production, layer contributions and gas-oil ratios. The calibrated models were then used to forecast the performance of wells in the study area. The impacts of the methodology of describing the natural fractures on fluid flow behavior and recovery mechanisms, as well as on the ultimate hydrocarbon recovery were evaluated. The results also were used to ascertain the risks of selecting the optimum methodology for the field development plan.
The forecasted results show little variations in oil recovery, pressure and oil saturation distributions under identical operating strategy for the three models. This is attributed to the absence of some critical properties required to model the oil recovery mechanisms in dual porosity system. For example, imbibition capillary and relative permeability functions were not input in the dual porosity (DPSP) model. However, the DPSP model is considered more efficient than the single porosity (SP and SPWAF) models because it took less time and modifications to obtain reasonable history match of the field performance. It was also more difficult to obtain a reasonable and acceptable history match using the SP and SPWAF models compared to the DPSP model, and the reservoir properties in the single porosity models had to be modified extensively and unrealistically to obtain history match.