A Quantum Approach to Studying Intrinsic Uncertainties in Basin Modeling

Øyvind Sylta¹, Are Tømmerås¹, and Christian Zwach^2
1Migris AS, Trondheim, Norway
²StatoilHydro ASA, Oslo, Norway

The study of uncertainties is a key issue in basin modelling and when applied to real exploration cases, a quantification of the uncertainties for a prospect may be of significant economic importance. The modelling of hydrocarbon generation, expulsion and migration from source rocks to traps through geologic time allows us to address the uncertainties in estimates of trapped volumes of oil and gas in prospects by means of many different numerical techniques. In many exploration scenarios, the objectives may have to be reduced to the task of estimating the probability of finding hydrocarbons in undrilled prospects. This may be due either to a lack of data or a lack of tools to make reliable volumetric estimates. When volumetric estimates of uncertainties for prospects can be made, however, a natural next step is to try to reduce the uncertainties in the estimates. We cannot expect that these uncertainties can be reduced to zero anytime soon. Important questions to address are therefore how small the uncertainties can become, and which of the input properties contribute the most to the uncertainties.

We use a novel method to describe the intrinsic uncertainties of undrilled trapped oil and gas volumes and/or columns and compute their probability distributions by applying a Quantum approach to Monte Carlo simulations. Hydrocarbon generation and migration software can be used to simulate the transfer of oil and gas within the sedimentary basin from sources to traps. All important input variables to the simulator must be described with uncertainties to use this methodology. As an example, we use the average volumetric content of shale (Vsh) in each layer of the model to demonstrate the formulation of the method. Setting Vsh = 0.7±0.16 means that the volumetric shale content is defined by a Gaussian statistical distribution with a mean value of 0.7 and a standard deviation of 0.16. This distribution is used as input to a full basin scale Monte Carlo simulation including the processes of burial, thermal heating, generation and migration of hydrocarbons.

In the Quantum approach to running Monte Carlo simulations we input values from the statistical distributions with fixed distances from the mean value of each parameter, and we use only these exact values. In the case of the Vsh parameter, this means that we only use the values of 0.7-0.02 (0.68) and 0.7+0.02 (0.72). The exact value is controlled by a Quantum distance parameter, which is the probability distance from the mean, e.g. 5%. For the Vsh value, each simulation run would then use one of the 3 values: 0.68, 0.7 and 0.72. We use the same approach for all the input variables to the simulations. In a typical case we may simulate with, say, 15 layers, using at least 15 input variables for each layer, of which 4 are source rock parameters (TOC, HI, thickness, kinetics) and 11 are migration parameters (permeabilities, entry pressures, saturations), making the total number of required simulation runs to span the entire parameter space 315*15 which is more than 10107 simulation runs. Random sampling of the input variable space with a smaller but significant number of simulation runs can, however, provide volumetric results that can be used to compile the intrinsic uncertainties of each trap and may relate these uncertainties to source, carrier and seal properties.

The method may be used to estimate the intrinsic uncertainty of each exploration prospect. If the intrinsic uncertainty in the predicted oil volume of a prospect is, say, 10 million barrels, then it will not be possible to predict the volumes of hydrocarbons in that prospect with less uncertainty than 10 million barrels by applying forward basin modelling techniques. If, on the other hand, the intrinsic uncertainty is modelled to be, say, 0.1 million barrels, then further constraining the input parameters for that sedimentary basin can result in significant reductions in the exploration uncertainty, and help mature the prospect.

The intrinsic uncertainties of a prospect are inherent properties of each prospect, and the estimation of these properties can help us focus our modelling efforts towards the more critical items.

This paper will explain the novel technique and its implication for petroleum system modeling within the exploration work process. The method is demonstrated with results from a demonstration case. We will also compare the presented approach with other uncertainty methods previously applied in basin modeling.

AAPG Search and Discovery Article #90091©2009 AAPG Hedberg Research Conference, May 3-7, 2009 - Napa, California, U.S.A.