Chapter 1
Introduction

The purpose of this practical guide is to summarize the procedures and workflow towards building a 3D model: the principles are applicable to any modelling project regardless of the software; in other words, this is an attempt at a practical approach to a complex and varied workflow (Figure 1.1). What we are not trying to do in this book is to build detailed geological models of depositional environments but to capture the heterogeneity due to structure, stratigraphy and sedimentology that has an impact on flow in the reservoir.

Figure 1.1 Reservoir modelling workflow elements presented as a traditional linear process showing the links and stages of the steps as outlined in the following chapters.

The key to building a reservoir model is not the software; it is the thought process that the reservoir modeller has to go through to represent the hydrocarbon reservoir they are working on. This starts with a conceptual model of the geology and a diagram of the 'plumbing' model to represent how fluids might flow in the reservoir. Modern integrated modelling software starts with seismic input in terms of both interpreted horizons and faults and seismic attribute data that characterizes reservoir from non-reservoir and ends by linking to dynamic simulation; the so-called seismic-to-simulation solution. I have always been concerned that geophysicists and reservoir engineers might forget the geology that actually creates their oil or gas accumulation.

Wikipedia defines reservoir modelling as 'the construction of a computer model of a petroleum reservoir, for the purposes of reserves estimation, field development planning, predicting future production, well placement and evaluating alternative reservoir management.' The model comprises an array of discrete cells arranged as a 3D grid populated with various attributes such as porosity, permeability and water saturation. Geological models are static representations of the reservoir or field, whereas dynamic models use finite difference methods to simulate the flow of fluids during production. You could of course construct a reservoir model using paper and coloured pencils, but analysis of that model is challenging!

Geo-modelling is 'the applied science of creating computerized representations of the Earth's crust based on geophysical and geological observations.' Another definition is 'the spatial representation of reservoir properties in an inter-well volume that captures key heterogeneities affecting fluid flow and performance.' However you define it, geo-modelling requires a balance between hard data, conceptual models and statistical representation. Whether you are working on a clastic or carbonate reservoir, the workflow is the same, although the challenges are different: in carbonate reservoirs, characterizing the petrophysical properties properly is paramount because diagenesis will usually destroy any primary deposition controls on reservoir quality. We will look at carbonate reservoir characterization separately.

A few key statements should be made at the outset:

• Every field is unique and therefore has different challenges
• Every challenge will have a unique solution
• Every solution is only valid for the given situation and therefore .
• KEEP IT SIMPLE .. at least to begin with.

1.1 Reservoir Modelling Challenges

Building a model of an oil and gas reservoir is complex and challenging as much because of the variety of data types involved as the many different steps required. The process is made easier if you can establish why you are building the model; what is the objective of the model? Today, we generally build 3D geocellular models for volumetric estimation, dynamic simulation, well planning and production optimization or to understand the uncertainty inherent in any hydrocarbon reservoir. Above all, a successful 3D model aids in the communication of concepts and the interpretation of data used to characterize a potential or producing oil or gas field.

We model reservoirs in 3D because nature is three dimensional and because the reservoir is heterogeneous and we have restricted opportunities for sampling. Additionally, to understand flow in the reservoir, we need to consider connectivity in three dimensions, rather than simple well-to-well correlation. Having built a 3D representation of the reservoir, it can be used to store, edit, retrieve and display all the information used to build the model; in effect, a model is a means to integrate data from all the subsurface disciplines, so the data are not just stored in the minds of geologists.

Reservoir modelling is also a challenge because we are dealing with a mix of geological and spatial properties and also the complex fluids present in the reservoir. The data available to build a representative model are generally either sparse, well data or poorly resolved, seismic data. The resulting model is dependent on the structural complexity, the depositional model, the available data and the objectives of the project. Building a usable reservoir model is always a compromise: we are trying to represent the reservoir not replicate it.

The advances in computer processing power and graphics over the past 20 years has meant that geoscientists can build representative models of a reservoir to capture the variability present at all the appropriate scales from the microscopic to the field scale. However, as reservoirs are complex, we need to be highly subjective about the scale at which we model and the level of detail we incorporate: a gas reservoir may well be a tank of sand but faults may compartmentalize that tank into a number of separate accumulations.

1.2 Exploration to Production Uncertainty

Even before the first exploration well is drilled on a prospect, a geologist will have estimated the likely volume of oil or gas contained in the structure; by comparing one field with another, a reservoir engineer may even have estimated a recovery factor. The volume estimated will have an upside and a downside to provide a range of values. At this stage, the volume range may have been calculated using deterministic or probabilistic methods, or a mixture of both, and will generally be quite a large spread. At each stage in the life cycle of the field, the median value and the range should ideally change in a predictable way as uncertainty is reduced through appraisal drilling and data acquisition (Figure 1.2). When there is sufficient confidence in the estimates, a development decision is made and a project can begin to spend real money! In reality, the evidence from many different field developments is that the ranges in volume are always being revised to account for new data, new ideas or new technology. This often has the effect of delaying the timing of decision-making, especially in smaller fields, where the risk of getting it wrong can have a bigger impact on value (Figure 1.3).

Figure 1.2 Idealized evolution of resources with time over the oilfield life cycle showing the reduction in uncertainty after each stage.

Figure 1.3 Actual example of resource change during appraisal and development of an oil or gas field; appraisal often continues after project sanction as too many wells may erode project economics.

In 1997, the UK government commissioned a survey to review the track record of field development in the UK sector of the North Sea with the aim of determining how reserves estimation changed from the time of project sanction to field maturity. Fields that were sanctioned between 1986 and 1996 and containing greater than 10 MMBOE were reviewed. This was done to establish the confidence the operator had in the estimated ultimate recovery they reported, the methods adopted to make the estimation and the major uncertainties in their estimation.

Until 1996, 65% of respondents said that they used deterministic methods to provide a single value with explicit upside and downside ranges; 53% reported using Monte Carlo/parametric methods, and 30% said that they adopted probabilistic estimation using multiple geological or simulation models: several companies use a mix of all the methods, which is why the percentages add up to more than 100% (Thomas, 1998). In the same survey, 30% of the respondents said that gross geological structure accounted for much of the uncertainty in the estimation of ultimate recoverable reserves; the remainder believed that the reservoir description accounted for the uncertainty. For fields under appraisal, the level of uncertainty was greater than those in production and that, in general, the estimates tended to be pessimistic rather than optimistic.

While analysing the results of the survey, it became apparent that reserves estimates have varied by plus or minus 50% in more than 40% of the fields after project sanction; this was particularly true for fields where the estimates were based on deterministic models (Figure 1.4). The economic impact on field development cannot be ignored; 60-80% more wells were required to achieve the reported reserves estimates, together with very expensive retrofitting of equipment on offshore platforms. Many of the fields surveyed were compartmentalized, fluvio-deltaic reservoirs of the Brent Province that required significant additional investment over their lifetimes. With the increased use of 3D geocellular models over the past twenty years, one would...