Improved, flexible and faster access to critical data is vital for fully informed decision making in today’s multidisciplinary asset development teams. It is also imperative to include the highest fidelity quantitative temporal data when building reservoir models to improve development plans and maximize returns on investment. A key area is the integration of dynamic reservoir simulation models with 4D seismic data. Current solutions are often ad-hoc and resort to static time points, making data integration painstaking and inflexible. Furthermore, procedures for data conversion and data management can also be time-consuming and lacking in generality, and may only lead to visual, qualitative solutions rather than fully integrated quantitative results.
The CoViz® 4D software platform, developed by Dynamic Graphics,® Inc., provides a fully geo-referenced quantitative 4D solution. CoViz 4D lowers the barriers-to-entry for data from diverse applications and minimizes time spent on data management. This platform is capable of simultaneous fusion of a very wide variety of static and temporal data, critical to asset management. When all these data are made rapidly and seamlessly available to all members of a multi-disciplinary team, the full value of all data can be extracted, resulting in improved understanding and decision making. Furthermore, the software goes far beyond mere visual integration of the diverse data; it includes robust, quantitative, and customizable tools and workflows which are an integral part of the CoViz 4D environment.
The following asset development case history from the UK North Sea shows how the CoViz 4D software, using multi-disciplinary data, revealed previously overlooked inconsistencies between the predicted and actual fluid flow in the reservoir. When the various data were integrated, animated, and analyzed in CoViz 4D, the discrepancies between the datasets were revealed and quantified. These discrepancies were verified through the CoViz Sim2Seis forward modeling workflow, which led to critical remedial action resulting in improved production outcomes.
Closing the loop using the Sim2Seis workflow in CoViz 4D. Data used with permission of owner.
Seismic Forward Modeling in a North Sea Reservoir
The UK development team compared reservoir simulation predictions against the measured 4D seismic response to monitor the water flood in the North Sea reservoir. Specifically, they considered two of the many reservoir sands, an upper sand and a stratigraphically lower sand. Figure 1 (UPPER SAND, TIME 2003) shows a qualitative, temporal analysis of the upper sand. The blue cells from the reservoir simulation grid show where water saturation was predicted to be greater than 45%. The blue areas in the seismic attribute, positioned on the interpreted upper sand horizon, indicated an acoustic impedance increase (interpreted as water replacing oil, or decreasing pressure). Whereas red areas in the seismic attribute suggested an acoustic impedance decrease (interpreted as gas replacing oil, or an increasing pressure). The team noted how the injector wells (blue) matched up with the blue (hardening) seismic response, and this in turn matched the waterfront as predicted in the reservoir simulation; overall there was a broad correlation between these diverse datasets at the 2003 time point.
Figure 1. UPPER SAND, TIME 2003.
However, as the team advanced forward in the lifetime of the field (to 2006; Figure 2) they observed an increasing discrepancy between the seismic response and the predicted fluid flow model. Notice how in the upper sand, the 4D seismic anomaly extends significantly up-dip beyond the waterfront indicated by the simulation grid. Also, this observation did not fit with the known volumes of water that were being injected into the support well.
Figure 2. Tracer flowback data shown as green and blue (cumulative) towers. Cumulative ellipse symbols highlight optimal fluid pathways. Monthly gas, oil and water production towers also shown.
The lower sand (Figure 3) showed a different story. Here the team also looked at the seismic attribute difference since 1993 and the predicted water saturation greater than 0.45%. By 2006, while the reservoir simulation grid still showed a significant water flooded volume, the seismic difference extraction showed some strong negatives (reds). The team was immediately attentive to this discrepancy between the two sands, with the upper sand seemingly taking more water than was predicted, and the lower sand less.
Figure 3. LOWER SAND, TIME 2006.
Figure 4. Animation in CoViz 4D showing comparison of 2003 and 2006 time points for both the upper and lower sands in the reservoir.
The team did further research in the CoViz 4D software and uncovered additional clues by looking carefully at the water injection into the two sands. Both sands were using the same well as the down-dip water injector and the team noticed long-term decline in injected water volume starting in 2001. Combining all these observations, the team concluded that the perforations into the lower sand were getting progressively blocked. Consequently, water was not being injected into the lower sand at the same rate as the upper sand. The water saturation in the upper sand therefore increased dramatically, more than the simulation model predicted. Whereas the pressure in the lower sand dropped. This led to the discrepancy in the 4D seismic response when compared with the reservoir simulation prediction in both sands.
These observations were qualitative, and the team wanted to support these findings with a more quantitative approach. Therefore, they used the Sim2Seis module within the CoViz 4D package. Sim2Seis offers forward modeling of the expected 4D seismic response from a dynamic reservoir model. By using Gassmann fluid substitution modeling based on fundamental rock physics, and a convolutional seismic approach, the team was able to quickly model various reservoir fluid saturations scenarios and match the synthetic seismic response with the 4D seismic observations. What the team found reinforced the qualitative observations: the water predicted to be injected into the upper sand was insufficient to explain the seismic observations, and the lower sand also mismatched the predictive model. By modifying the reservoir simulation model to reflect the influence of a progressively blocked perforation into the lower sand the team established a much better fit between the model and the seismic observations.
After the findings, the well was worked over to clean out the perforations into the lower sand, water injection was reestablished, and production was restored. The 4D capabilities in CoViz 4D software enabled the necessary temporal data sets to be simultaneously visualized and analyzed, revealing the discrepancies and leading to rapid action and improved production outcomes.
See the CoViz 4D Sim2Seis page for more information.
Data Sources/Credits:
Data used with permission of owner.
