This article deals with the difficulties of modeling subsurface volumes, (such as oil and gas accumulations). It discusses delivery realistic scenarios leading to a scientifically acceptable range of results.

Introduction
Most oil/gas companies are using probabilistic mathematics (Monte-Carlo simulations etc) to constrain uncertainty and risk, and to estimate volumes and value of undiscovered accumulations and ‘reserves.’
In a probabilistic approach, all prospect parameters are combined in all possible ways. It also relies heavily on input values displayed in a statistical distribution. Here we face the first serious flaw of this methodology: rarely we dispose of many hard data points (in excess of a population of 100 points).
The Monte-Carlo approach also delivers a large number of outcomes that do not comply with ‘reality’ (in geo-scientific terms: the cumulative knowledge of how oil and gas fields can be described, and how they originated). Although ranges of potential outcomes may be encompassed, ‘truthful’ outcomes remain scattered in a cloud of fake results.
Many of my professional colleagues share my concerns - probabilistic modeling tends to yield complex, and equally unreliable results, also subject to ‘creative’ tweaking. As so often an increased complexity does not further accuracy and truthfulness. My personal credo being: the simpler, the better.

Talking nature
Let’s talk some real examples, and consider the impact of aligning play- or success parameters.
In several shallow oil fields in Asia, there is a correlation between the length of the oil columns, and the API degree of the oil within the reservoir. Short columns commonly mean very poor API, and highly viscous oils. There is also an obvious correlation between gross reservoir thickness, and column length. Can we ignore such correlations? Obviously not. Can volumetric evaluation programs cope with interconnected parameters? Some programs may do a better job than others, but building correlations into software is a tricky, if not dangerous task- a procedure, that can affect the regional consistency of evaluation.
Conversely, an incorrect alignment of success parameters can create exploration prospect portfolios with a high probability of success, whilst a killer parameter is overlooked. As an example, I would like to cite the razor-sharp boundary of charge derived from the Iabe (late Cretaceous) source rock in the Congo-Angola Basin. In this area, the scope for structures, reservoir and seal is generally benign, which can paint a picture of ‘high probability of success’ or ‘low risk.’ Far from that. Source rock maturity and access to charge, however, have been found to be absolutely critical. A slight decrease in the Tertiary overburden might dictate a difference of finding billions of barrels, or literally nothing.
Obviously, the real challenge is to see, extract and compute the true ‘scientific’ components, or alignment/correlation of components, that form building blocks to a unique reality. How can we ever dream of describing geo-scientific realities, if these remain hidden within a cloud of artificial probabilistic outcomes?
Unfortunately this won’t remain our only problem. We have to consider the practical side, too. A reservoir engineers wants to see one, or two scenarios for planning a well path, a completion. He or she cannot do reservoir calculations on an unlimited number of probabilistic outcomes. The development planner also requires concrete scenarios. He or she does need discrete scenarios to calculate the costs of a platform, and of topside equipment. Nobody can make a probabilistic purchase order.
My conclusion, succinctly worded, is: Probabilistic mathematics might appear fundamentally unsuitable for the calculation of oil and gas volumes.
How can we approach the issue of volumetric in a creative and correct way? Is there an alternative?
Let’s step back for a moment, and look how nature is organized. Obvious to anyone of a natural science background, nature doesn’t create an infinite number of (random) outcomes. There is a precise reason for whatever there is, and statistical nebulas demonstrate human ignorance, if anything. Nature isn’t quite organized along the lines of insurance or lottery mathematics.
· Crystals can be classified in a very few classes.
· There are only some 110 different atoms
· There is a finite number of chemical molecules

I would also like to cite a more mundane example. Go to any restaurant’s kitchen, and combine all the existing food ingredients under all possible cooking, frying and seasoning outcomes. Only a few dishes will deliver an acceptable taste, if we ignore the skills of finding the few unique appetizing combinations enticing our tongue. Look at a typical McDonalds or Burger King menu: though several hundreds of chemical components are used in the kitchen, these are combined in a small and very specific arrangement of outcomes. Needless to say that ice cream, cooked in palm oil and combined with coleslaw (as a statistical outcome of a fast food Monte Carlo run) might not appeal to everyone.
Alike we use cooking skills to obtain a dish that fulfills our wishes, we should equally drop lottery mathematics and approximate our target with proven tools we better understand: earth science.

The Periphery of Truth concept
Let me at this point introduce a concept, which I developed over the last years in search of a sound and pragmatic volumetric appraisal line of thought. It is called ‘Periphery of Truth.’
The basic line of thought is that we may not be able to find the only genuine, truthful subsurface realization (without drilling), but we might pinpoint the few discrete possibilities that Mother Nature may have realized. The task here is to develop credible scenarios, with a list of supportive arguments. Unlikely scenarios (“let’s give it the benefit of doubt”) should be eliminated if we are unable to support these by strong geological or geophysical evidence. This approach requires a thorough look at the physical realities of a hydrocarbon basin, combined with a lot of creative thinking, mental power, and, yes, honesty.
I would like to compare this to a dartboard: The red circle in the middle is a target that is very difficult to hit; an experienced thrower will, however, be able to hit the periphery of the inner circle quite often.

Basin and prospect evaluation
How does this work in practical terms for prospects, or basins?
A small number of ‘hard candidates’ is evaluated, and assessed in terms of cost, and potential reward. These scenarios are used as cornerstones of the cost/reward reality characterizing an oil/gas prospect, or a sedimentary basin. It may be important here to count (and to compute/ economically evaluate) all viable scenarios, without pushing (as human nature tends to do) the positive outcomes only.

Knowing that none of these outcomes will ultimately represent the one and only genuine subsurface realization, they form an envelope of (ultimately wrong, with some truthful element) results that could point toward the genuine, but hidden value.

Now, what about risk? Using a ‘Periphery of Truth’ approach, we can quote the balance of economically favorable scenarios against unfavorable outcomes. No mathematics of whatever kind will protect against making the wrong decision. Let’s assume there are four possible scenarios selected. If only one of the four scenarios is economically viable, the risk of failure is 75%.

The bottom line
Where does this lead us? The ‘Periphery of Truth’ approach certainly requires a lot of intellectual strength, combined with a readiness to accept challenge. Concluding, I might argue that improving exploration success in the oil and gas industry does not depend on fancy black boxes or computer programs, but instead on a sharp, solid and waterproof exploration logic combined with a lot of experience, creativity, and honesty.

©2005 by Franz L Kessler