|
|
Case Study: TyumenNIIgiprogas Obtains Positive Results from Computer-Assisted History Matching of Dynamic Gas Reservoir
"The combined power of ECLIPSE software, MEPO, and SIS expertise really accelerated our history-matching processes and produced more accurate results."
Nesterenko Alexander
Deputy General Manager
Development of Gas, Condensate, and Oil Deposits
TNGG
|
|
ECLIPSE and MEPO pilot project demonstrates a higher-quality matched model than trial-and-error history matching
Challenge
Produce history-matched model of large oil/gas/condensate field with 20-year production history and complex characteristics (i.e., reservoir scale, geological structure, and lack of separate data from wells with commingled production).
Solution
Replace trial-and-error history-matching processes with computer-assisted history-matching technology and create an automated workflow using
- ECLIPSE reservoir simulation software
- MEPO interactive historymatching software
- HP high-performance computing (HPC) clusters.
Results
Generated several viable history-matched models in just 4 weeks (instead of 7 to 9 months using traditional approach); decreased uncertainty of predicted field performance and remaining reserves; established a common workflow practice.
Quantify reservoir uncertainties
The Urengoyskoe oil/gas/condensate field, one of Russia’s largest, is being developed by Gazprom dobycha Urengoy, a subsidiary of Gazprom. The Urengoyskoe asset is located in the Yamalo-Nenets Autonomous District of Siberia, and during its 20 years of production has produced about 40% of the country’s gas resources.
TyumenNIIgiprogas (TNGG) was asked to produce a history-matched model in support of a much-needed prediction development strategy for this important field, taking into consideration the following specific complexities:
- multiple layers, with different fluid properties in each layer
- wells with commingled production
- faults with unknown conductivity
- variable porosity and permeability through the reservoir.
Initially, TNGG used a conventional trial-and-error approach to history-match the reservoir model. This procedure included manually changing simulation input parameters, comparing simulation and historical data, and then analyzing error distribution. Such manual methods take a long time (7 to 9 months) to produce a single match option, and any changes to the model require the whole process to be repeated.
In addition to improving speed and accuracy, TNNG needed to find all possible combinations of input parameters that matched historic data and be able to rerun the process quickly with the availability of new data.
Launch pilot project: computer-assisted history matching
Schlumberger Information Solutions (SIS) and TNGG launched a pilot project to accomplish the history-matching task within a limited time frame. This project relied on the implementation of an advanced hardware-software solution—computer-assisted history matching (CAHM)—to significantly reduce the manual work of previous history-matching processes.
Multivendor solution
The project team selected Schlumberger ECLIPSE software to run simulations and MEPO (a product of SPT Group) to study a wide range of uncertain reservoir and fluid parameters during history matching. The following cluster resource management software and Web interface, purchased by TNGG and implemented by SIS, created a user-friendly simulation environment:
- LSF—queuing and resource management software for HPC (from Platform Computing Corporation)
- EnginFrame—job submission, control, and monitoring system (from NICE srl).
Cluster-based optimization
In conjunction with the above software, and using an SUSE Linux operating system and an InfiniBand/high-speed network, the following HPC cluster technology was run:
- 10 compute nodes—20 AMD Opteron 285 CPUs, clock speed 2.6 GHz, 40 cores in total, 8 GB RAM per node
- one management node—two AMD Opteron 280 CPUs, clock speed 2.4 GHz, 16 GB RAM
- storage solution—real capacity 4 TB, type SCSI.
Obtained positive CAHM results
The CAHM pilot project helped TNGG produce a history-matched model within a very short time frame. The model’s quality allowed reevaluation of remaining hydrocarbon reserves in the studied field after 20 years of intensive production. Not only did the project’s automated method provide TNGG with positive results for history matching of this particular dynamic gas reservoir, it provided a new workflow for use in modeling other dynamic gas reservoirs and reduced uncertainty of predicted field performance and remaining reserves.
|
|
|
