Static Considerations
For a typical oilfield, the areal grid spacing is between 50 and 100m. For large fields and in exploration larger grid spacings may be used. The grid is usually oriented parallel with the major fault trend to minimise the number of jumps in grid rows along faults.
What is the well spacing?
If occasional wells are very close together and penetrate the same cell, they can be separated by using an arbitrary trend line.
Unless well data is very dense, there is no need to use fine grids of 25m as this will have little benefit and may be detrimental if there is a lot of seismic noise.
What level of detail of faulting needs to be captured?
Models used for well planning may have more faults than those used for volumetrics
The 3-D modelling grid mid-point should be close to the elevation of interval of interest to reduce distorted cells
This can be difficult when field boundary faults have large throws.
May need to clip top or base of fault planes closer to reservoir limits
Dynamic Considerations
This will be driven largely by the time taken for a reservoir
simulation run. Apart from the number of active cells in the model, the
simulator will be slower if there are a lot of low permeability cells.
Consider starting with a coarse model and then it can be refined.
How many cells can be handled by the simulator?
How much of the aquifer needs to be included?
The aquifer sometimes needs to be modelled – it
is not just a case of having an aquifer or not
How thick are the beds in terms of significant
changes in porosity and permeability?Is upscaling going to be required?Is
any particular cell orientation needed?Cells may be oriented parallel to the main faults to improve flow behaviour
Reservoir top and base
Seismic Horizons
Unless there is very dense well data, the seismic interpretation will be the key to the reservoir depth structure (See Gallery, Image 2).
What is the best quality seismic horizon?
It is much simpler
to generate a model using a single seismic surface
At what position in the well does the top / base
reservoir seismic event occur – synthetic seismograms.
Does this horizon tie to the actual top reservoir, or is an isochore needed?
Is there a base seismic horizon and does it
follow the top reservoir horizon?
If there are significant changes
in thickness away from well data, then two seismic horizons are recommended
Is the top reservoir an unconformity?
Does the reservoir thin below seismic resolution in any areas? e.g. pinch-out, truncation, onlap
Are there any good quality intra reservoir seismic events that can be used to assist the model?
Does the seismic mapping fit with any spill points or other features?
If more than one seismic horizon is
used, the resulting isochore must be checked and edited to remove artefacts – rapid
changes that could not occur in the depositional environment.
Depth Conversion
The depth conversion can be carried out in many ways, ether in the seismic system or in the modelling package. In any case, a simple time-depth well plot for top and base reservoir is very useful as a first pass and to highlight any unusual features. It is not intended to cover this in this article, but a few relevant points are listed below.
How is the initial depth surface going to be finally adjusted to the wells?
Never use a method that automatically fits the well data without checking with another method.
Make sure the seismic surface is not spikey near the wells, otherwise inconsistent shifts may occur during well adjustment.
Faulting
Major faults may bound the reservoir and will need to be modelled. Major internal faults usually need to be incorporated, but
minor faults may have lesser effect. For a particular seismic horizon, it is important to avoid gridding artefacts around faults often caused
by noise and unwanted extrapolation of the seismic events. The reservoir offset across faults mapped on seismic will also vary depending on the
degree of rollover assumed.
The actual angle of major faults should be checked on seismic with no vertical exaggeration. Depending on the fault hade and the purpose of the model, the faults can be built either vertically, or sloping at a realistic angle and this will complicate the 3-D grid. If two faults converge with depth and meet in or just below the reservoir, this causes difficulties for most gridding software and special treatment or editing of the faults may be attempted.
- Reservoir boundary faults greatly affect GRV, are they sloping at a realistic angle?
- What is the net to gross or the reservoir?
High net to gross reservoirs will have good juxtaposition across faults - Are there any particularly high permeability layers that may be juxtaposed and cause flow past shale layers?
- Is there any core or log evidence of cementing at faults or other deterioration in properties?
- Are any of the seismic faults seen in wells?
Tying the faults to well fault cuts can be difficult, but can be important for horizontal wells.
