Abstract

Unconventional developments generate a characteristic phenomenon known as well interference. They are signaled by a disturbance in a well given the influence of a neighboring well. We can classify three types of interferences: Completion to Drilling, Drilling to Production and Completion to Production. The last kind, also known as a frac-hit, is the subject of this study, being the most frequent and with the most productivity impact, caused by liquid loading or wells shut-in.

The goal of this project is to deliver an early frac-hit detection system for Vaca Muerta’s natural flowing wells integrated into YPF’s Unconventional Decision Support Center (DSC). Using this system, shutting certain wells for frac-hit protection could be avoided, shutting them in when there is a deviation from natural well-head pressure decline.

To achieve this objective, we use real-time pressure data from well head pressure gauges. The trend component of the head pressure time series is extracted, and a model is made using one-month data for each well. The model forecasts the next 3 days of well head pressure, predicting its's natural decay and it was tested with a set of 34 wells labeled as interfered.

By comparing the modelled pressure with the real data, we can detect the beginning of the interference when the pressure difference is above a defined threshold, and the difference persists a given number of hours. An alarm is raised when these conditions are met. The alarm is displayed in a dashboard accessible for DSC supervisors, and operational engineers, to take corrective actions, e.g.: decide which wells need to be shut-in.

Monitoring begins when the fracking equipment starts the new well stimulation process. If no alarms are registered, the model is automatically recalculated.

As a result of this work, a dashboard for monitoring the state of natural flowing wells on the DSC was developed. We can expect to have less production loss for wells shut-in, unless an early frac-hit detection alarm is raised.

This work is directly applied to the unconventional play of Vaca Muerta, decreasing the supervisory needs to detect frac-hits when several frac rigs are operating simultaneously with tens of wells under risk of being hit. The novelty is the application of time series analytics over well head pressure data to enhance DSC’s supervisor’s productivity and decrease production losses for the unconventional assets.

This project was developed working in a multidisciplinary team with agile methodologies, with people from the following teams: Analytics CoE, Upstream Data Science, Digital Oilfield, and Asset’s Reservoir Engineering, Production Engineering, and Production DSC.

This content is only available via PDF.
You can access this article if you purchase or spend a download.