A New Methodology to Forecast Solution Gas Production in Tight Oil Reservoirs

摘要

Due to favorable economics, more and more oil companies are now drilling multi-stage fractured horizontal wells to develop tight oil reservoirs. While oil production forecasting in these wells has historically been the primary focus of most operators, gas (solution gas) production forecasting has been largely neglected. There are two distinct reasons for this oversight: First, when compared to oil production, solution gas rates typically have a much lower impact on overall economics. Second, solution gas production is often very difficult to forecast due to abnormal GOR's which are typically caused by incorrect gas rate measurements. This paper presents a simple methodology to predict solution gas production based on forecasted oil production. This methodology introduces a new specialized plot which can be used to determine various parameters necessary to forecast solution gas production without any costly PVT and pressure history data. Additionally, this paper presents a step-by-step procedure used to analyze the available history to obtain an accurate representation of GOR performance. This data in turn can be used to predict solution gas production. Missing data and/or incorrect gas measurements are common issues that can directly affect a well's GOR history. The methodology presented in this paper will outline the steps required to repair this kind of bad data and ultimately generate a reasonable GOR forecast that is representative of a well's true performance. This methodology was initially validated using synthetic data (generated by a commercial simulator) and has been subsequently tested on over one thousand oil wells that are producing from tight formations such as the Bakken, Niobrara and Wolfcamp in the USA and the Cardium in Canada. Each test was carried out by hindcasting - using the early part of production data to history-match the later part of history. This methodology has consistently resulted in good agreement between the forecast and the real field data. A number of practical examples from different reservoirs have been presented in this paper to illustrate/validate this new methodology.