When evaluating the reserves and generating the scenario for the development plan, it is necessary to select the epresentative values, i.e. area, thickness, porosity etc., for the specific reserves such as P10, P50 and P90 reserves.
The easiest way to determine these values is to calculate the common percentile values that can generate the specific reserves. For instance, generating the reserves distribution multiplied by three (3) lognormal distributions of area, thickness and recovery factor, the P10 reserves can be generated by multiplying the common percentile values around P23 for each parameter distribution. However, this method does not account for the difference in the size of each variance. The percentile values are selected according to the specific percentile such as P23 in each parameter distribution, even though some variances of these distributions are small and others are large. Also, if the distributions are different from the lognormal and their types are mixed, in general case, it is difficult to derive the analytic solution of common percentile values, then an alternative search algorithm and tool is required.
The approach presented here is the alternative way to select the representative values at the specific percentile of probabilistic reserves by direct usage of the actual trials in the Monte Carlo simulation. Although, for example, there are infinite combinations of values generating the P10 reserves, the distribution and combination of the values that hit the P10 reserves can be derived and evaluated by actual simulation trials. Monte Carlo simulation can show each parameter's distribution that hits the specific point of reserves. For area and thickness that have large variances in the exploration stage, the realized distributions are varied largely at each point, and these means or medians are changed clearly according to the varied points such as P10, P50 and P 90. On the other hand, for porosity or oil saturation that has a relatively small variance, the simulation shows nearly same distributions, and the difference in mean or median is also small at the different point of reserves, even though the specific point is changed largely.
Adopting the proposed approach, the representative values can be easily derived accounting the variance and type of distribution for each parameter. This alternative method can provide more realistic values and scenarios through the process from the probabilistic reserves evaluation for the development planning.
抄録全体を表示