Using data analytics to quantify the impact of production test uncertainty on oil flow rate forecast

Abstract

In general, flow measurement systems in production units only report the daily total production rates. As there is no precise control of individual production of each well, the current well flow rates and their parameters are determined when production tests are conducted. Because production tests are performed periodically (e.g., once a month), information about the wells is limited and operational decisions are made using data that are not updated. Meanwhile, well properties and parameters from the production test are typically used in multiphase flow models to forecast the expected production. However, this is done deterministically without considering the different sources of uncertainties in the production tests. This study aims to introduce uncertainties in oil flow rate forecast. To do this, it is necessary to identify and quantify uncertainties from the data obtained in the production tests, consider them in production modeling, and propagate them by using multiphase flow simulation. This study comprises two main areas: data analytics and multiphase flow simulation. In data analytics, an algorithm is developed using R to analyze and treat the data from production tests. The most significant stochastic variables are identified and data deviation is adjusted to probability distributions with their respective parameters. Random values of the selected variables are then generated using Monte Carlo and Latin Hypercube Sampling (LHS) methods. In multiphase flow simulation, these possible values are used as input. By nodal analysis, the simulator output is a set of oil flow rate values, with their interval of occurrence probabilities. The methodology is applied, using a representative Brazilian offshore field as a case study. The results show the significance of the inclusion of uncertainties to achieve greater accuracy in the multiphase flow analysis of oil production.