Incorporating Geophysics into Geologic Models

New approach makes geophysical results available to engineers in a form they can use. History matching, volumetric calculations and reserve estimates may all be improved.
This article appeared in Vol. 5, No. 5 - 2008


Most oil and gas exploration and production companies have a substantial investment in seismic data and interpretation of that data. This data contains a wealth of information about the reservoir properties between the well locations, and this information would be of great interest to geologic modelers if only it were available in a form they could make use of in their geologic models.  

The use of seismic data

Geologic models are built from well data, making the model very detailed and reliable at each well location. But what is going on between the wells? A basic step is to connect the stratigraphy from well to well, but this does not account for faults or other anomalies. Seismic data is an obvious source for inter-well surface data, but for reservoir property information it is inherently incompatible in its measuring system-time instead of depth and samples instead of grid cells and layers. Converting these two systems into a compatible grid has been a serious challenge for many years.  

Well data and seismic data are in a way quite opposite. Well data is very detailed in a small area. Seismic data covers a large area but is much less detailed. Well data gives rich detail in specific locations but the rest of the field remains unknown at that level of detail. Seismic data provides a coarse view of everything, but sharp focus is not possible with seismic alone.  

An interesting change occurs when well data and seismic are combined. Seismic sheds light on the unknowns between wells, while well data helps sharpen detail. Overall, a model combining well and seismic data thus becomes more useful to geologists, geophysicists and engineers alike. The only question remaining is how to put the two together.  

Combining seismic and well data

Sometimes seismic is used as a backdrop to well data in cross-section view. This provides a bit of context to the well analysis but contributes nothing quantitative. The geologist may believe the formation will behave in a certain way because of what he sees, but there is little hard data to back up that belief.  

Another approach is to add the seismic into the corner point grid (CPG) using geometric resampling. Unfortunately, the difference in measurement systems and grids makes this a process highly susceptible to error. Assigning the seismic time values to cells in the CPG often results in data that is simply wrong.  

Figure 1 compares the seismic grid to a corner point grid. The smearing of the reservoir property values outside the stratigraphic layering in the corner point grid is unavoidable without a common grid linking the seismic and corner point worlds.  

Yet another approach collapses the seismic data into maps using an average for volume. This, unfortunately, means that none of the maps are correct for use as a guide to propagating the well data.  

The uncertainty caused by these errors is quite severe. Volumetric and reserve calculations can be inaccurate, and obtaining a history match can require substantial and often unrealistic adjustments in the reservoir model parameters. As a result, even when seismic information is brought into the geologic model, it is used as a very loose constraint and adds minimal value to the model.

Correlating well data to seismic datas

What is needed is an approach that reconstitutes seismic-derived property data in terms of the stratigraphy of the geomodel. The well-based model and seismic data must be brought together into a neutral grid that enables the correlation of well and seismic data with far greater accuracy. Geostatistical modeling can then be performed using the correlated data as a strong trend in 3D. This has the effect of constraining the geologic model to agree with the seismic information between the wells, leading to dramatically reduced variance among the geostatistical realizations. This reduced variance in turn leads to better P10, P50, and P90 estimates of recoverable reserves, reducing overall uncertainty. Because the model more accurately matches both well and seismic data, history matching can usually be achieved more rapidly.  

Once a model is generated in this way, individual realizations can be used in a flow simulator and compared to actual production data to determine which of the equi-probable reservoir models is the most likely model.

Introducing RockScale

A module of FastTracker, RockScale correlates the corner point grid and the seismic grid in a neutral stratigraphic model grid (SMG). Using Zonal AdjustmentTM, Rockscale manages the transform of properties from the seismic grid into the SMG, and ensures that they are structurally and stratigraphically correct when transformed from the SMG into the corner point grid. This includes the management of detailed structural models with many faults and sub para-sequences.  

Figure 2 illustrates the difference in models generated using geometric resampling and a neutral grid such as the stratigraphic model grid supported by RockScale. This example shows the original seismic-derived porosity model (Figure 2a), and simple geometric resampling (Figure 2b), and a RockScale Zonal Adjustment case (Figure 2c).   

The simple geometric resampling shows obvious grid artifacts and coherent regions of high and low values are broken up by ‘noise'. This comes from small lateral and vertical shifts, which make the property values incorrect. When Zonal Adjustment is applied, flow units and associated net to gross are properly preserved.  

RockScale provides three dramatic benefits to geologists and engineers:

  • More accurate volumetrics-providing a better match to real subsurface geology.

  • Faster and better history match - the simulated model matches all available data and behaves like the reservoir itself does

  • Improved field management program-fewer wells placed more accurately. 

RockScale features a click-and-drag interface that is easy to use and speeds the overall modeling process. Its 3D view of the model before and after the addition of seismic detail provides an interesting view of the change in seismic data from its original form to its new shape in the stratigraphic model grid.

Improving history matching

3D seismic-derived porosity cube after import to the corner point grid. This porosity model can be used directly or as a 3D trend model for geostatistical modeling. RockScale provides a neutral grid for geologic and seismic data. This neutral grid-the SMG-enables the accurate transformation of seismic-derived reservoir property information into the corner point gridded geologic model in 3D. Using this information as a 3D trend in geologic modeling improves accuracy and reduces uncertainty, enhancing the field management process.  

Seismic data is tremendously helpful in understanding reservoirs when it is presented to each geoscience discipline in a form that it can use. For engineers, including seismic properties in the reservoir model makes history matching faster and improves volumetric calculations. Using a model that accurately integrates well and seismic data also ensures better accuracy in reserve estimations.


Related Articles

GEO Science Explained Europe

The Snorre Field Pilot

In 2008 the Optowave fibre-optic seismic sensor cable was trenched at the StatoilHydro-operated Snorre field with the aim of achieving insight into installation, data acquisition and monitoring procedures. The Snorre pilot demonstrated ‘real time' data transfer and quality control - the fundamental requirement for reservoir monitoring ‘on demand'.
Slide3 thumb