Abstract

Operated by Thai Shell, the Sirikit field lies onshore Thailand, about 350kms north of Bangkok. The geology of the shallower reservoirs (study zones) comprises sediments in an alluvial environment setting which have been highly faulted. Oil coming from these several stacked/amalgamated channel reservoirs contributes a significant chunk of Thai Shell's daily production.

The field carries significant uncertainties regarding reservoir characterisation. One of the most important of these is sand distribution/connectivity, hence lithology differentiation between the sands and shales is of key concern. Furthermore the sand units are very thin (often beyond the resolution of seismic reflection data), averaging 3-4 metres in thickness. This makes them difficult to follow on conventional seismic sections. Reservoir connectivity across faults is a key issue for successful field development.

In mid 2000, Jason Geosystems was commissioned by Thai Shell to carry out an advanced reservoir characterisation study as part of a larger initiative to help to address several development issues. The project was undertaken in two phases. First conventional Acoustic Impedance (AI) inversion was used to generate a more detailed interpretation of the seismic data. This re-interpretation was then used for the Phase II High Resolution AI Inversion (also known as Stochastic or Geostatistical Inversion). Rather than producing a single answer, this approach generates a range of possible models. As well as fitting all the available data, each model is also equally probable. The range of models can then be used to create best-case and worst-case scenarios of sand development (e.g., P10, P90 etc) for quantitative interpretation and uncertainty analysis.

The Phase II results are currently being used to address water flooding and a number of other key issues for field development. Sand probability models were built showing the probability of encountering sands at any location in the 3D volume. "Blind well tests" were performed at locations where new well data became available (not part of the initial study), in order to compare the predicted results with actual well results (figure 1). Reasonable matches between the predicted and actual lithology emerged, thus adding confidence to the inversion's results. In order to assess the all important aspects of reservoir connectivity, the lithology models were analysed in 3D voxel-space to determine the distribution and connectivity of the sand units (figure 2). This allows a much better understanding of the character of the reservoir to enable well location planning with reduced risk and greater confidence. Since completion of the project certain lithology predictions have been backed up by an oil discovery at the sand locations as predicted by the study.

Figure 1. Blind well test: Predicted lithology section (left) shows a good match with the actual lithologies from the well path (No figure available).

Figure 2. 3D perspective view showing only those oil sand bodies which are internally connected (No figure available).

Presented at: 2003 South East Asia Petroleum Exploration Society (SEAPEX) Conference, Singapore, 2003