Simulation Modeling Challenges and New Concepts to Improve Dynamic Models History Match Quality

With existing of powerful HPC’s resources and advance simulation tool technology, new features and concepts can be easy implemented, which would participate in improving dynamic model’s history match quality in very reasonable short timeframe with no matter how much complexity in the reservoirs. In the following study project, we had introduce several new concepts that worked together to boost dynamic model representation by better honoring physics while replicating static water saturation into dynamic model at high resolution.

The main new concepts that were introduced into this study project, and they are not existing in the previous model, are as following:

1. Fault corridors, by assigning different properties to the grid cells that are connected to the fault surfaces, to allow Water/Gas flow parallel to the fault networks.
2. Poro-Perm linear regression relationship.
3. Water saturation “ADNOC Fast-Track” approach.
4. Design of “Well Region” bodies, which allows modification of properties around wells at each simulation model layer to enhance fluid distribution and pressure match.

Its worth to mentioning that the study project has been completed in 4 months timeframe, which is showing very reasonable history match quality. The main findings are:

• This project has demonstrated the value of applying best practices and using advanced tools to history match a high-resolution model of a giant reservoir. The simulation performance results show that it is feasible to use a multi-million cell grid for practical studies, including compositional modeling for EOR prediction models.
• The subject dynamic model was utilized to investigate and benchmark several technical aspects that would drive the future dynamic modeling long-term strategy in ADNOC, with respect to upgrading plans for hardware’s resources, understanding the capabilities of the software tools, and establishing best practice guidelines in reservoir engineering technology.
• The optimum size of any dynamic model depends on (A) the ultimate target objective of the study, such as optimization of field development, testing EOR applications, pilot design, etc., (B) the density and quality of the measured field data linked with the historical production period, (C) the available hardware and software resources to accommodate and finish the runs in due time.
• ADNOC Fast-Track approach (Concept showed great added value to replicate water saturation log profiles in the dynamic model at high resolution, which also resulted in a very stable dynamic model, better history match, and improved run time performance.
• Permeability related parameters, including absolute cell permeability, relative permeability curves, transmissibility, connectivity and Kv/Kh ratio are the most effective/sensitive parameters influencing history match quality. Nevertheless, it is a challenge to evaluate all parameters within reasonable time, and the well region bodies model (Concept D) and poro-perm linear-regression relationship approach (Concept B) have really helped to obtain a good history match using AHM tools.

Following same subject, development of a fully matched compositional model is work-in-progress. The tests performed during this project have shown that there could be benefits in coarsening the grid for better performance when simulating multiple realizations for uncertainty and optimization. However, we have also shown that the resolution of the grid can make a significant difference to the results of the forecast model, particularly when studying gas injection scenarios. These topics will be presented and discussed in more detail in a future study which will cover Part-2 of the project.