Uncertainty in Reservoir Fluid Fill Prediction and Its Implications on the Business – Case Studies from South East Asia

Abstract

LWD logs are becoming increasingly reliable for the primary (and often the only) source of downhole log data. The standard and advanced mud logs in conjunction with LWD logs are used to conclude on the reservoir fluid typing (gas, oil, water) in real time. However, there are examples where the mentioned logs can equally misguide the log interpreter. An incorrect fluid interpretation can result in huge financial implications. There could be decisions associated with laying new pipelines, upgrading surface facilities (separators, compressors) and/or installing new platforms. In addition, the fluid fill would have a direct impact on the company reserves and reporting. Historically, a number of exploration and development wells were drilled in the SE Asia region where minimal LWD log data was acquired. This data is often considered sufficient for evaluation of reservoir properties. However, for fluid fill, the data was insufficient for various valid reasons and the initial interpretation from the LWD logs could have led one astray. This paper highlights examples where an integrated effort from the subsurface disciplines provided insights into the misbehaviour of the LWD logs and mud log data. The examples shown in this paper demonstrate that similar looking reservoir log responses could lead to an uncertain fluid fill. In such circumstances, the requirement for additional log data needs to be highlighted at an early stage of the project and should also be reflected in the well and field economics. The uncertainties in the reservoir fluid fill needs to be reflected in the project risk register and in the field development plan (FDP). This paper provides alarming signals to the log interpreters that extra caution needs to be taken in log data acquisition, processing and interpretation stages. The paper also highlights the importance of integration between subsurface disciplines and the criticality for the log interpreters to be familiar with the depositional environments. It has also been demonstrated that a limited set of log data could equally mean having no information unless additional evidence is available to address the uncertainties.