6.2 Communication, communication, communication...
What asset team doesn’t joke about geoscientists not understanding what engineers need and/or about engineers not getting what geoscientists do? These jokes are as much a way to exorcise any possible communication issue to come as they are a way to vent out the frustration of on-going problems caused by miscommunication. And beyond that, these jokes are simply funny! Our teams are under a lot of pressure. A good joke is always a nice way to lift some of the tension we face and we should enjoy them for that!
Every geomodeler should be vigilant about this potential problem though. Too many geomodeling projects don’t reach their full potential, because of miscommunication between geoscientists and engineers. It’s unfortunate, but luckily it can be largely avoided. The remainder of this paper provides some ways to do so.
Nothing gives a reason for a good laugh (or a fair amount of frustration) more than a geomodeling project already in progress for a few weeks (months…) and everything has to be redone because the team suddenly realizes that the model doesn’t take into account a few wells needed later for flow simulation. The question is not who shall have given the information to start with – the team, the geomodeler, the geoscientists or the engineers. The point is that it is a problem that a proactive geomodeler can easily fix, at the beginning of a project, by agreeing on the list of wells to be used.
Firstly, we must validate the list of wells with our geoscientists. On their side, it will be linked to which wells have geological/petrophysical/geophysical data that must be taken into account in the model. Secondly, we must crosscheck this initial list with the list of wells the engineers are looking at. Many wells will be on both lists. But engineers will also consider wells with some production history, even if these wells have no data useful for modeling the geology of the reservoir. Horizontal wells tend to fall in this category. They have some production attached to them, or they will in the future and so they must be taken into account for predicting future production. Yet, they might have no data usable for geomodeling per se. Too often geomodelers forget to make sure that these wells fall in the correct geological units. That is, wells known to have been drilled in a sand layer might end up crossing into some shale units located above or below the targeted sand. It happens when the geomodeler interpolated the horizons incorrectly between vertical wells, not realizing that it placed the horizontal wells in the wrong place. Geomodeling packages have options in their workflows to take into account the complete geometry of the horizontal wells.
At the same time, it is wise to check with the engineers what the lateral and the vertical extent of the volume of rock they need modeled is. Figure 2 in the section 1.4 of the first chapter and its associated paragraphs give an example of such problems.
Once the well list and the volume to model is approved by the whole team, the project can start. During the project, the geomodeler will communicate about his processes and his results to the geoscientists. Among other things, he will explain why he picked some specific geostatistical workflows and he will show that his model is indeed respecting the ideas the geoscientists have about the reservoir. It is wise to include engineers in these discussions. Firstly, it will give them more confidence in the project. Secondly, it will emulate discussions about the model inside the whole team. Geoscientists tend to focus their review on how the geomodel respects their ideas about 3D facies distribution. This is crucial, of course, but it can sometimes overshadow some mistakes a geomodel might have in term of respecting the laws of physics in general and the laws of flow dynamics in particular. Engineers will often spot such mistakes. During the presentation of his model to his team, the geomodeler should go as far as stating that he needs the geoscientists’ feedback on the facies and the porosity as well as the engineers’ feedback on the water saturation and the permeability models. In so doing, everyone knows what your expectations are for her/him.
Figure 1 gives an example of a project in which the engineer’s feedback on water saturation was crucial. It is based on an anecdote that happened to one of the authors a few years ago. The reservoir was a simple sandy geological unit. There was no facies modeling per se as the whole unit was considered made of sand. The porosity modeling didn’t cause any issue either. Water saturation proved more challenging (Figure 1A). The water saturation log showed low values everywhere with the exception of a zone close to the top of the unit, and only in the South-West corner of the reservoir. There, the water saturation was getting close to 100%. Due to a large number of input wells, this information was overlooked by the geomodeler. Water saturation was modeled using geostatistical techniques, in the same way it had been done on many other projects before. The 3D water saturation model was showing, locally, a zone of high values around the wells. Everything was consistent as far as geostatistics was concerned; the hard data were respected as well as the global saturation distribution and the global variogram. The geoscientists and the geomodeler reviewed the project. Satisfied by their model, they gave it to their engineers and they moved on to other tasks. Months later, the geomodeler and the geoscientists discovered that the engineers were struggling with the geomodel; water was literally “raining” in their model from the zone of high saturation. To them it was, in fact, impossible that such a zone of water saturation existed there. It did not make any sense in terms of flow dynamics. Gravity would have made this water drop to the bottom of the reservoir (water being denser than the oil in this reservoir). They decided to manually edit the saturation in the problematic area to get some good flow simulation results. Naturally, they were frustrated by this situation. Reviewing the geomodel and the input data, the geoscientists discovered the source of the problem - the water saturation logs were valid, but not the facies description. It had been missed that the reservoir was showing a local continuous shale in that zone. The water saturation model was correct, but the permeability model was not. High permeability values, believed to be in a pure sand unit,had been distributed in the whole sand. Instead, it should have been set to zero in the shale unit. In that case, the water would not have “rained” in the sand below. The geomodel was rebuilt. A zone of shale was added to the facies model. The water saturation was now modeled by facies – very low in the whole sand and close to 100% everywhere in the shale. At last, permeability was computed by facies as well - high in the sand, null in the shale.
This anecdote illustrates several important points. Firstly, the engineers might indeed spot issues with fluids and permeability models that geoscientists and the geomodeler himself might miss. Secondly, if the engineers are not involved in the review and they just received the geomodel as a package thrown over a fence, there is a greater chance that they will try to correct such problems themselves rather than reporting them to the team. The team might then end up with two geomodels; the original, and the one edited behind closed doors by the engineers. And who is to blame for such situations – the engineers for not communicating about what they saw or the geoscientists and the geomodeler for not properly involving them in the review process? To avoid having to argue about such questions at a later stage, we believe it is in any geomodeler’s own interests to include engineers in their project at the same level than geoscientists are.
Involving engineers in their projects will also help the geomodelers to address two of their most common questions. Many engineers wonder why we are spending so much time building a facies model while they need only porosity, water saturation and permeability. Many also wonder why the 3D geological grid we work on has a complex mesh and millions of cells while they specifically asked for a “sugar box”, simple 3D grid.
Flow simulation engineers need 3D grids which are aligned with the main direction of flows in the reservoir. In a simple, layer-cake reservoir with no fault and no folding, it means that the K axis of the 3D-grid should be indeed perfectly vertical. The horizontal mesh will be, for example, parallel and perpendicular to the horizontal wells around which the flow simulation is run. If the reservoir is fractured, the horizontal mesh will likely be built parallel and perpendicular to the main direction of the fractures. As a last example, if the reservoir is faulted, the horizontal mesh will likely be built parallel and perpendicular to the fault surfaces. In addition to this, the mesh of the flow simulation grid should be made of cells of constant size, with no truncated or eroded cells. These constraints ensure that the computations in the simulation software run faster and are more stable numerically.
The geological 3D grids are built to populate petrophysical properties in 3D. As these properties are primarily controlled by the facies distribution, we have to model facies in detail as well. Geostatistics are our main toolbox to do this. In chapter 2, we explained that the orientation of the mesh of the geological 3D grid is the primary control on how the facies (and the petrophysics) are interpolated around the wells. Use a mesh that doesn’t reflect the directions of sedimentations and you are likely to get an incorrect 3D facies distribution.
In fact, building a geological 3D grid and building a flow simulation 3D grid follows the same problem. In both cases, we need a 3D grid that is aligned with the main directions of the physical phenomena we are modeling. In flow simulation, it means solving the equations of flow dynamics and the mesh must follow the directions of flow. In geomodeling, it means mimicking with geostatistics the results of physical phenomena like erosion and sedimentation and we need a 3D grid with a mesh parallel and perpendicular with the directions of deposition. Those directions are usually different from the directions of flow simulation. That’s why we can’t use the flow simulation grid to model facies, and in reverse that’s why it is unwise to run flow simulation in a 3D grid fit for facies modeling. We need two 3D grids with specific cell size, with specific orientations for the mesh and with or without eroded cells. We need a geological 3D grid and a flow simulation 3D grid. It implies that we will have to transfer rock properties from the geological grid to the flow simulation grid (it will be covered in the next part).
Explaining the need for a specific geological grid by building an analogy with flow simulation constraints have proven efficient to the authors on several occasions.
Figure 2 and Figure 3 illustrate these points. Let’s assume we have three vertical wells, each showing a succession of shale and sand. An engineer might convince a geomodeler to use a “sugar-box” 3D grid for modeling facies – that is a grid with constant cell size and horizontal and vertical mesh. The type of mesh flow simulation would be run into. If we do this, the facies model would look like a succession of horizontal sand and shale layers (Figure 2A). Let’s assume that dipmeter data shows that the sands and the shales are in fact dipping. It makes sense to build a 3D geological grid with an inclined mesh (Figure 2B). The new 3D facies model is now very different from the original one. This second approach is better than the first one as it not only respects the facies at the wells, but also the information of the dipmeter data and, from there, the geological concept developed for the reservoir– the facies are dipping. Figure 3 shows how the water saturation model would look like. This is the type of property (with porosity and permeability) that reservoir engineers need. Shall we give them our geological, inclined 3D grid for their flow simulation? Maybe, if our engineers confirm such a 3D grid is good for their work; but very likely, they will ask for the 3D petrophysical models to be transferred into a sugar‑box grid.
Without proper understanding, by the geomodeler, of what is needed for facies modeling and later for flow simulation, either the facies model would be wrong (Figure 2A) or the 3D-grid sent to the engineers could potentially be inadequate for flow simulation (Figure 3).