From the Midland Reporter-Telegram
University of Texas of the Permian Basin Professor Ahmed Alzahabi has been studying shale rock for the past four years.
As Permian Basin oil and gas operators seek to optimize production in their wells by better determining the “sweet spots,” the University of Texas of the Permian Basin is looking to help.
Ahmed Alzahabi, a member of the petroleum engineering faculty at UTPB’s College of Engineering, has been studying shale rocks for the last four years and is the co-author of two new books on the subject: “PVT Property Correlation: Selection and Estimation” and “Optimization of Hydraulic Fracture Stages and Sequencing in Unconventional Formations.”
A better way
He’s working on a new model that would make hydraulic fracturing more efficient and could save companies time and money as they drill for more oil. Alzahabi wants to eliminate the process of crews drilling multiple wells at one site and sending sensors down the wellbore in the hope each well will strike oil. This process does not always result in optimal production, he said.
Alzahabi believes a special algorithm and screening criteria on the surface of a drilling site can more accurately determine the “sweet spot.”
“We believe companies will be able to cancel unnecessary wells. They can cancel unnecessary frac stages before even drilling it,” he said. “We will save companies billions of dollars.”
His students are assisting him with his research, which is being sponsored by the College of Engineering. He said he hopes his students can start testing the model immediately.
Alzahabi took some time from the busy first week of UTPB’s fall semester to discuss the research.
Exactly how will this model work? Is it a computer program crews can use?
We are building a set of correlations to work as a predictive model for the Wolfcamp data. These correlations can be used by anyone with easy available data on the surface to predict the number of fracture stages, clusters and perforations in horizontal wells.
This work targets a predictive model to evaluate the number of stages, clusters and perforations as one of the completion strategies and treatment for the Wolfcamp shale play and to guide future selective optimum completions. Many important parameters that control producing well behaviors such as number of days on production, depth, fluids in each barrel, horizontal well completion configurations, stages per well, fracture type, average water requirement, proppant type, fluid type, hydraulic horsepower per stage, pounds per square foot of proppants per stage, number of stages, and lateral length and completed interval of the horizontal wells, have been analyzed.
We analyzed the performance of thousands of horizontal wells right from the Wolfcamp formations using available data. The analysis of the data identified key parameters — depth, barrels of fluids, type and amount of proppant, fluid type, and 30-day initial potential in defining the number of stages, clusters and perforations.
The procedure used in exploring the data can be used as a decision criterion for similar cases in deciding the number of fracture stages and the main important factors that lead to the optimum way of developing these resources. A multivariate linear regression predictive model for number of fracture stages, clusters and perforations were advised. The models show relatively good in-sample predictions.
Would you say this is part of the “Big Data” that is increasingly a part of oil and gas operations?
The answer is simply, yes, it is. The concept is to make sure that we frack the optimum number of stages in our horizontal wells here in the Permian Basin that would provide us with the optimum hydrocarbon production.
How will you test this model? Will you test it with a small number of wells? How long do you think the testing period will last?
We will test the model using publicly available data.
Do you find – as others have – that operators are willing to provide data to help develop such algorithms?
We have an interested local company that kindly supports us with the data for building these set of new correlations.
Are your students helping to develop this model?
Yes, I am guiding our undergraduate students to build the models and test them. I am collaborating with a professor from Texas Tech as a part of my research team.
