Probabilistically Mapping Well Performance in Unconventional Reservoirs With a Physics-Based Decline-Curve Model

The description below outlines a series of problem statements tackled in the paper SPE-195596-PA and shares related supplemental material.

Firstly, there was a need to process a large amount of publicly available production data from unconventional reservoirs with the aim to get valuable insights.

Secondly, it is well-known that the Arps equations are unreliable and not originally designed to forecast production in unconventionals due to the extended transient period. There are other models available, but we needed to incorporate more physics into the models while still keeping them simple and adjustable to field data.

Thirdly, the extended transient period leads to multiple outcomes because the onset of boundary dominated flow is frequently not observed. Therefore, a single model is not sufficient; we need to tackle the problem from a probabilistic perspective and have criteria to calibrate our estimates.

Finally, mapping the results of the probabilistic physics-based decline curve analysis allows the visualization of spatial patterns dominating the decline trends across the field. This can lead to uncertainty reduction in forecasts and better screening of potential infill drilling locations.

Our new paper SPE-195596-PA "Probabilistically Mapping Well Performance in Unconventional Reservoirs With a Physics-Based Decline-Curve Model" proposes a robust framework to address the problems listed above.

We used production data from 814 gas wells in the Barnett shale, the sequence that the wells were drilled are shown in the figure below.

The figure below shows the best-fit history matched models to 814 gas wells from the Barnett shale.

The details are available in the paper. Hopefully, you have some time to enjoy reading it. Here is a link to the publication.

I would like to acknowledge the co-authors Dr. Eduardo Gildin and Dr. Peter P. Valkó from Texas A&M University.