With recent downturn in Gas & Oil prices, there is an increased focus on finding ways to optimize the hydrocarbon recovery cycle. In this new economic reality, technology and analytics will have to be counted upon to deliver cost reduction solutions. This paper outlines one such solution applicable to unconventional shale drilling operations.
A two-pronged approach has been taken to customize decision making processes for the two phases – planning and execution. Traditionally, some form of offset well information is used in planning phase of a well to assess risk, allocate and acquire capital, and logistical planning. The variables used in these calculations are stochastic in nature and, as such, necessitate a stochastic simulation of the well. To run an accurate simulation, we assign a customized probability distribution function to each variable based on offset information. Offset information comes from a meticulously maintained database of local wells. Reliability of simulation accuracy depends heavily on quality of offset wells in the database. A high quality database will include wells drilled across many different geographical locations, by rigs with varying technical limitations and crew experience levels and by different drilling operators. In the execution phase, we use a similar combination of historical data and stochastic simulation process to predict the result of a change in drilling parameter before making the change. This is done by using a real-time decision making dashboard.
The paper shows some real world examples of inaccurate results obtained by traditional methods. We also run some stochastic simulations on our model and compare results with observed results. We show the model to have reliable predictive capabilities.
This paper provides a new tool which combines established statistical methods with good quality drilling data to improve predictive capabilities. The tool enables risk to be accurately built into drilling budgets. It also allows for a more reliable prediction of "unexpected" events such as stuck pipe, tool failures, mud losses, etc. By building in the effect of such events, variability of outcomes is dramatically reduced.