Abstract

History-matching is commonly used to calibrate numerical simulation models to validate geological and reservoir engineering inputs and ensure the reliability of model predictions. While history-matching can yield a powerful forecasting tool, thermal SAGD history-matching presents a host of challenges, particularly in terms of computational time and numerical tuning difficulties associated with large thermal simulation models. Typically, single well or small multi-well models are developed and calibrated to generate meaningful production forecasts. However, these predictions are only valid prior to steam chamber coalescence and the models are insufficient for life-of-well forecasting. Hence, it is necessary to build and calibrate large multi-well simulation models within a reasonable timeframe.

The oldest SAGD wells at Suncor's Mackay River (MR) property are part of the initial Phase 1 development. This phase consists of 25 well pairs that commenced steam injection in September 2002. This paper presents a calibration process for Phase 1 simulation model, generated through Suncor's geostatistical modeling process, incorporates more than 9 years of production history and contains approximately 1.4 million active cells. The process described has several components: 1) rigorous data quality control, 2) establishing appropriate boundary conditions, 3) numerical tuning, and 4) a focus on global rather than detailed local changes.

Using the process described herein, a Phase 1 history match was achieved which honours actual field rate and pressure history. The history match results are presented on well/pattern/phase basis. This process has subsequently been applied to MR Phase 2 and 3 with significant efficiency improvements.

A process is presented that addresses many of the issues that arise in validating and history-matching large, complex simulation models. It is shown that history-matching multi-wells SAGD performance with a long operation history is feasible, practical, and useful. The calibrated modeling processes gives confidence in understanding the reservoir and reduces the uncertainty of future forecasting results.

You can access this article if you purchase or spend a download.