The ideal reservoir model is now possible using a fully integrated Seismic to Simulation (STS) workflow

By Graeme Eastwood, Fugro-Jason

The ideal reservoir model is the repository for all the information you have about the reservoir – petrophysical, geological, geophysical, and production information. In the ideal world, this model will, without major modification, replicate the production history of the field in a dynamic simulation, predict drilling and production results years into the future, and enable you to compare development scenarios, among other things. But this is seldom the case, so how do you construct such an accurate and reliable model?

Description of the conventional reservoir modeling workflow
A typical reservoir model is built using the horizons and faults from a seismic interpretation, along with the petrophysical and lithology information from the wells. Information on the geological setting – net to gross, reservoir geometry and aspect ratios, Lithology proportions, etc. – is also included to help guide the interpolation of the well data as the model is populated. The methods of model population are varied, from simple interpolation to full use of geostatistics in methods such as Sequential Gaussian Simulation. The resulting model is then usually upscaled and input to a dynamic simulation workflow, and the results compared to the production history.

 

Usually the results are somewhat unsatisfying. Often, objects, designed to resemble the presumed geology, are included in the model in an attempt to model the inter-well complexity and, realizing that the seismic data has important inter-well information, seismic attribute maps or volumes are frequently used to guide the interpolation of well properties in the model. However, this does not generally yield dramatic improvements in the quality of the model. Other products attempt to vary the static model automatically to try to get a better history match, but then all connection to the input data has been lost, making the model nearly impossible to update. What is needed is a static model which is dramatically more accurate to start with.

Key Advantages of Fugro-Jason’s STS workflow
Getting the various Geoscience disciplines to talk to each other is difficult, because each is working with their own data in their own way. Data that doesn’t fit into their paradigm is not included in their reservoir model. For example, geophysicists don’t use corner point grids because their seismic data is in regularly sampled cubes. Geological modelers don’t use seismic because it is in the wrong grid, and how do you put “amplitude” into a simulator anyway?

As it turns out, seismic data does have huge amounts of useful lithologic information that greatly enhances simulation model results. But in order to unlock that potential, the seismic has to appear in a form the geological modeler and the simulation engineer recognize and can use. The JasonSTS process accomplishes this by:

• Transforming the seismic to highly-detailed elastic property and Lithology models
• Building a petro-elastic model from the wells to relate petrophysical properties to elastic properties
• Stratigraphically linking the seismic grid and corner point grid to facilitate two-way transfer of petrophysical property models between them.

Geostatistical inversion is far more than simply constraining simulations to seismic data. It is about using geostatistical information in addition to a convolutional model and a wavelet and truly inverting from the measured 3D seismic data to the rock properties of interest in high detail.

Geostatistical Inversion enables asset teams to simultaneously incorporate all field data – including well logs, cores, geological beliefs, and seismic (pre- or post-stack) – into an analysis that results in multiple highly-detailed and realistic models, each of which honors all of the data known about the reservoir. These models are accurate near and away from wells, have realistic detail beyond the seismic bandwidth, and together provide more accurate estimates of uncertainty and bias than other methods.

This is the first key difference with this new approach. These models are high enough detail and have enough heterogeneity for use directly in a reservoir simulator, but they are elastic properties, not the petrophysical and engineering properties needed in such an application. In order to link the petrophysical world to the elastic world the petrophysical information from the wells is used to perform an integrated rock physics study, which yields a petro-elastic model relating the properties of impedance and velocity to the petrophysical properties in the wells – lithology, porosity, Sw, etc.

 

This is the second key difference. The petro-elastic model is fully integrated with all the same parameters and settings fully synchronized between the petrophysical and rock physics equations. The petrophysical model can now be used to create highly detailed models of petrophysical properties from the Geostatistical Inversion results. The different realizations can be ranked, and a selected set upscaled and put through dynamic simulation. Alternatively, if time and compute resources allow, all models can be dynamically tested.

The third key difference is in how the resulting volumes of data, which occupy a seismic-based, orthogonal grid system, are populated into the geological, corner point grid. An intermediate grid is constructed that connects to both the layers defined by the horizons and faults in the geophysical grid and the layers in the corner point grid of the geological model. Therefore we can accurately transfer the properties into the geologic model, preserving thin permeability barriers and other features and ensuring there is no “property leakage” between layers.

A final difference to this method is that the flow simulation results can be used to validate the compartmentalization derived from the seismic interpretation. Often, additional faults are be added, but these would come from the seismic interpretation and would cascade all the way through to the revised dynamic model. With the new STS workflow, the simulation model remains tied back to its original input information. This makes the model both more accurate and easier to maintain. The realization(s) with the best history match are then used to predict future production and to evaluate alternative management scenarios.

About Fugro-Jason
Fugro-Jason is dedicated to delivering innovative products and services to help our clients identify and produce hydrocarbon deposits by integrating information from the various geoscience disciplines. In 1993 Fugro-Jason introduced the Jason Geoscience Workbench, making it possible to integrate geological, geophysical, geostatistical, petrophysical and rock physics information into a single consistent model of the earth. In 2001 Fugro-Jason added the PowerLog family of petrophysical applications, and in 2003 added the FastTracker family of geological modeling applications.

The application of Fugro-Jason’s technology, either through use of the software or through our consultancy services, substantially improves returns on E&P investments by adding invaluable reservoir model information that reduces the risks, costs and cycle-times associated with field development.

Fugro-Jason is a unit of Fugro NV. Fugro was founded in 1962, and is listed on Euronext NV, Amsterdam, the Netherlands. Fugro has more than 13,000 staff in 250 offices and a permanent presence in over 50 countries.

Contact details:
Graeme Eastwood, Region Manager, Fugro-Jason
c/o Fugro Survey (Middle East) Ltd
PO Box 43088
Abu Dhabi
United Arab Emirates
T: +971 2 554 1011
F: +971 2 554 7811
E: geastwood@fugro-jason.com