Three year R&D-project yields important innovations

Our technology has been further improved through a three year R&D project; Migrisk. The main objective of the project has been to implement, test and verify new methods for evaluations of uncertainties for complex geology using our migration simulator, Migri. Tullow Oil Norway and Statoil were partner companies and has contributed with testing and verifying the methods in the Migrisk project, which was completed in December 2013. Our software and methods are now more effective and our analysis more accurate.

The project was an IFU project, partly funded by Innovation Norway (Innovasjon Norges Industrielle Forsknings- og utviklingskontrakter).

Key research challenges for the project included:

  1. Improved description of primary migration as a geologic process.
  2. Analysis of leakage from oil and gas traps coupled to seismic.
  3. Improved handling of complex geometry around faults.
  4. Improved description of migration rock properties at basin scale.
  5. Improved optimisation and uncertainty analysis.
  6. Use of state of the art graphics and visualisation for improved decision support.
  7. Multi-core simulation and visualisation.

Several of the advancements made in the project have already been implemented into our technology, and contributed to the results seen in our recent Barents Sea Charge Study.

“Riskchart” efficient for uncertainties in complex geology
One of the results from the Migrisk project includes a “Riskchart” component, which can be an extremely efficient method for addressing the uncertainties for complex geology. Typical for complex geology in charge (hydrocarbon migration) modelling is that finding a single solution that fits all data may be very difficult and often impossible. One reason for this is that as the complexity increases the possibilities for input data errors and risks of inaccuracies also increase. Our Riskchart technology was therefore developed to plot both errors with respect to observations data (oil and gas columns heights). The uncertainties of input parameters can also be evaluated.

Riskchart analysis of a series of 5000 Monte Carlo simulation
Riskchart analysis of a series of 5000 Monte Carlo simulation runs. Each row in the plot shows the deviation between modelled and observed column height (in meters) for one observation (well) compiled from all simulation runs. The deviation values are sorted in increasing order. Blue colours represent simulations where the model predicts smaller column than the observed value while red colours shows simulations where the model predicts a larger column than what is observed. Perfect matches are shown in white colour. This gives an quick overview of which observations that can be matched by the model and which that are always either underfilled or overfilled.

Beyond present day technology
The Riskchart method can be extended with simulation runs where only the intrinsic uncertainties are modelled. These are the modelling uncertainties that we cannot resolve with the present day technology. Using the risk offset plots of the Riskchart tool it will become fairly straightforward to find which parameters have the largest effect on the total intrinsic uncertainties and which prospects have low intrinsic uncertainties. Low intrinsic uncertainties means that estimates of trapped resources can be made less uncertain by reducing the uncertainties of the input distributions.

Validating and ensuring stability
Extensive work on the capillary pressure leakage modelling has been done in the Migrisk project to validate the methods and ensuring stability of the approach. The leakage modelling approach used here is different from existing techniques in several ways, the most important ones being accounting for hysteresis effects and that it does not depend on the lateral model resolution. The effective area of leakage is calculated by the simulator in an iterative solution where the permeability of the most effective seal above the trap is important together with the hydrocarbon column heights within the trap.

Multi-core platform
The use of a multi-core platform is critical to the success of Migri in the future. A single simulation can execute (parts of) the code in parallel (1) and/or multiple simulation runs can be run in parallel (2). A small Red Hat Linux compute cluster with a total of 50 cores has been used in the project with addition of a Xeon PHI multi-core processor. We have concluded that type (2) solutions, with multiple simulation runs in parallel, are the most important for uncertainty modelling at the present time. This allows for the efficient testing of multiple cases / scenarios on workstations, small clusters or using a cloud solution. In Migrisk we have therefore implemented a solution where Migri can launch separate processes on (multiple) cores. We call the processes MigriBee and the idea is that each MigriBee is asked to run up to e.g. 1000 simulation runs. At the time of completing the Migrisk project, the MigriBee system and the associated Riskchart analysis tool have already been used with success in case studies and we believe it will be an important element in the continued success of our simulator technologies.

Important results – tools already in use
The Migrisk project has significantly improved our ability for doing improved optimisation and uncertainty analysis. In fact, this is probably the single most important result of the project. The tools are already in practical use within Migri and the two partner companies both have access to the full technology. This new technology has opened several new avenues for improvements and Migris will seek to exploit these avenues while making sure that the new technology is moved into the market.

Great potential
A number of technologies have been prototyped in the Migrisk project and several of them have the potential to become key components in reducing exploration uncertainties in the future. Validation and testing of these new technologies will be done in cooperation with oil company clients over the next few years and using case data from the oil companies.