Hydraulic fracturing execution has historically relied on surface treating pressure as the primary information source for evaluating a stage. However, treating pressure is a composite signal that blends pipe friction and perforation friction, limiting its usefulness for diagnosing the underlying drivers of performance changes.
As a result, operators can see that a change has occurred, but not why it occurred or how fluid is distributed across clusters. This lack of subsurface visibility means that inefficiencies in cluster utilization are often recognized only after the treatment (i.e., through production data or post-job diagnostics) when the opportunity to take corrective action has already passed.
To address this limitation, high-frequency pressure data is used to deconvolute downhole effects in real time, enabling a closed-loop, measurement-driven fracturing workflow (Fig. 1).
Acoustic friction analysis (AFA) uses high-frequency pressure during pumping to quantify key metrics, including perforation efficiency and uniformity index (UI) (URTeC 4044703). By separating and quantifying the friction components driving stage behavior, the workflow avoids relying on aggregate surface pressure trends as a proxy for subsurface response.
These quantified outputs support a physics-based artificial intelligence (AI) approach that combines acoustic modeling with predefined decision logic to translate friction response into actionable stage-level recommendations, enabling operators to move beyond surface-based interpretation and make decisions based on measured subsurface behavior.
In this context, physics-based AI is anchored to measurement rather than used as a substitute for it. High-frequency pressure response is used to quantify perforation friction, perforation efficiency, and flow distribution, while AI supports interpretation and decision logic. This distinction is important because purely data-driven models rely on historical patterns, which may not represent outlier stages or changing downhole conditions.
Using Direct Measurements
Because these outputs are available while pumping, decision criteria can be established to guide execution (Fig. 2). Perforation efficiency and UI thresholds are used to flag performance degradation or emerging issues in real time. These alerts trigger targeted adjustments, such as modifying rate, proppant concentration, or fluid system, deploying diverters, or reallocating volumes, which can be implemented immediately. This shifts fracturing operations from a reactive process to one where execution is continuously evaluated and optimized as each stage progresses (SPE 230647).
Applying this in the field has shown that real-time insights can be used to identify optimal flow-velocity ranges, refine proppant ramp strategies, and dynamically redistribute volumes away from ineffective stages.
Importantly, these improvements can be achieved without increasing total fluid or proppant volumes. Instead, resources can be deployed more effectively, based on immediate feedback. This represents a shift toward volume-neutral gains in stimulation efficacy and more consistent cluster utilization across stages.
This case study describes the first field deployment of this workflow, where live AFA outputs were used to evaluate perforation efficiency and trigger logic-driven interventions based on predefined evaluation points and performance thresholds.
The objective was to demonstrate how threshold-based responses can limit efficiency degradation within a stage and improve overall execution consistency, providing a scalable framework for measurement-driven fracturing.
Evaluation Criteria and Intervention Strategy
To operationalize the closed-loop process, evaluation points and intervention criteria were established prior to pumping. Stages were assessed using perforation efficiency derived from AFA at key points during execution, enabling both early-stage diagnosis and continuous validation of treatment response (URTeC 4495082).
The initial evaluation of perforation efficiency was conducted at the start of the extended 1.3-ppg proppant hold. A threshold of 70% was established to trigger intervention. When efficiency fell below this threshold, a corrective response was initiated immediately during pumping (Fig. 3).
Following any adjustment, additional measurements were taken to evaluate treatment response and confirm effectiveness. At least two subsequent assessments were performed: one at the end of the proppant concentration hold and one at the end of the stage. In some cases, additional intermediate evaluations were conducted to monitor trends and ensure stability following intervention.
Interventions were primarily focused on modifying hydraulic conditions to improve flow distribution across clusters, specifically, a 5-bbl/min rate increase designed to shift flow velocity toward a more favorable regime for cluster activation. The intervention leveraged the relationship between flow velocity, perforation friction, and perforation flow area. Insufficient velocity can limit cluster participation and create imbalance if conditions are not corrected.
This structure—early evaluation, threshold-based triggers, and repeated validation—ensured that perforation-efficiency loss was identified and addressed within the same stage. By embedding these criteria directly into execution, the workflow enabled consistent, repeatable decision-making across stages and pads while maintaining flexibility to respond to actual downhole conditions. The structure outlined also ensured that corrective action was taken early enough to influence the remainder of the stage rather than simply diagnose performance changes after they had already occurred.
Field Performance and Response to Intervention
In this study, stage execution was assessed at each predefined evaluation point, and corrective actions were applied when the perforation-efficiency threshold was not met.
Stages in which intervention was applied experienced consistently smaller declines in perforation efficiency throughout the remainder of the stage. By the mid-stage evaluation, the average efficiency decrease was approximately 5% for stages where action was taken, compared with 9% for comparable stages without intervention. By the end of the stage, this divergence widened, with decreases of about 12% vs. 18%, respectively (Fig. 4).
In a subsequent pad, the same decision criteria and intervention logic were applied to evaluate repeatability under similar operating conditions. The overall trends remained consistent. At mid-stage, stages with intervention saw only an approximate 1% decrease in perforation efficiency compared to approximately 7% without intervention. By the end of the stage, this divergence persisted, with declines of approximately 7% vs. about 14%, respectively, confirming that early corrective action consistently reduced efficiency decline (Fig. 5).
In both pads, performance degradation was detected early enough to enable timely response, reinforcing the robustness of the threshold-based approach without increasing total treatment volume or cost.
Economic Implications of Real-Time Intervention
Improving cluster utilization during stimulation has a direct impact on well outcomes. Prior studies have shown that a 0.1 increase in UI correlates with approximately 6 to 10% higher production in the Permian Basin (URTeC 4494923) and 7 to 12% higher production in the Williston Basin (SPE 230663) (Fig. 6).
Real-time intervention consistently reduced perforation-efficiency loss relative to untreated stages. Because this metric has been tied to production outcomes, the ability to evaluate and correct degradation during pumping makes it actionable rather than purely diagnostic.
As closed-loop fracturing is deployed at scale, small improvements in cluster utilization across a pad or development program can translate into meaningful production uplift and incremental economic value without increasing total treatment volume or cost.
Conclusion
Field deployment demonstrates how integrating live AFA with predefined decision logic enables consistent, in-stage optimization of fracture execution. By moving beyond surface pressure interpretation, the operators gained direct visibility into cluster distribution and responded to degradation as it occurred rather than after the treatment.
Across multiple pads, the application of consistent thresholds and intervention strategies resulted in measurable improvement in perforation efficiency, with adjusted stages consistently outperforming those without intervention. While the magnitude of improvement varied between wells, underperformance was reliably identified, enabling timely response and confirming the repeatability of the approach under varying field conditions.
These results demonstrate that real-time assessment paired with immediate, logic-driven response can materially limit efficiency decline as a stage progresses. This approach actively manages execution while pumping, representing a significant advancement in scalable, measurement-driven fracturing optimization.
For Further Reading
SPE 230647 Employing Live Subsurface Measurements To Support Supervised and Autonomous Intelligent Decision-Making While Fracturing by K. Fatheree, S. Aghdam, and J. Conaway.
SPE 230663 The Effect of Cluster Uniformity Index on Production: A Look at 200 Wells in the Williston Basin by K. Fatheree and S. Aghdam.
URTeC 4044703 Case Study—Using an Acoustically Derived Surface Metric To Approximate Fluid Distribution Uniformity During Stimulation by S. Aghdam, C. Cipolla, M.M Khan, J. Lassek; J. Kroschel, M. McKimmy, and K. Fatheree.
URTeC 4494923 Correlation of Acoustically Derived Perforation Efficiency to Early-Time Oil Production and Relevance to Real-Time Completion Optimization by M. Schult, S. Gabel, M. Aghababa, A. Majer, M.M. Khan, M. Mullett, and J. Klostermann.
URTeC 4495082 Closed-Loop Hydraulic Fracturing Optimization Using Real-Time Surface Measurements and Automated Control Systems by J. Conaway, C. Parra, C. Skinner, M.M. Khan, and K. Fatheree.
Kinleigh Fatheree, SPE, is a completions manager at Seismos, where she supports operators in applying noninvasive, real-time acoustic measurements to improve hydraulic fracturing execution and optimize stimulation. Her work focuses on using live subsurface measurements to evaluate perforation efficiency, cluster distribution, and treatment response during pumping. She previously worked in advanced fracture-treatment design and onsite implementation across multiple North American basins and has provided direct technical support for refracturing, well rescues, and fracture diagnostic services. She holds a BSc in petroleum engineering from The University of Oklahoma.