Optimal number of Monte-Carlo simulations for hydrocarbon resources probabilistic estimation

Raman Sadykov

2023

Full text Open access: Gold Metadata (CSL-JSON)
DOI 10.54859/kjogi108590
Data confidence 70 medium
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

Background: Probabilistic resource estimation allows to assess the potential of oil and gas prospect, which is used for decision-making in the industry. The main result of the estimation is resource potential, expressed in probability distribution function of geological or recoverable hydrocarbon resources. Monte Carlo simulation commonly used for these purposes. It simulates the numerical value of the objective function. Aim: The aim of the work is to find the optimal number of Monte-Carlo simulations in the probabilistic resource estimation. Methods: Monte Carlo simulation is based on repeated calculation of the function describing the process using a random number generator. The function variable defined by statistical distributions is calculated using random number. Further, mathematical operations are performed on all variables of the function according to the mathematical model. Summarizing the obtained set of results, the statistical distribution describing the desired function is approximately estimated. Results: The accuracy of distribution function estimation increases with an increase of the number of simulations. However, this tend to increase the computation time. Thus, there is a choice between the speed and accuracy of the solution. The Latin hypercube reduces the influence of the random number generator imperfection, but intermediate calculations with Latin hypercube attenuate the decrease in the number of simulations compared to random sampling. Conclusion: Based on the results of the work, a logarithmic dependence of the calculation accuracy on the number of Monte Carlo simulations was obtained. In most cases, 10000 simulations with accuracy about 1% will be sufficient.