CCL Anomaly Detection in Practice: Lessons from Corva & YPF’s Field Applications

In unconventional completions, casing deformation is often discovered only after it creates an operational challenge. A restriction appears, a plug hangs up, or a tool becomes stuck, forcing teams to react in real time and absorb the resulting nonproductive time.

But what if operators could identify those risks before they disrupt execution?

That question is driving new approaches to casing deformation detection. Through a collaborative effort between Corva and YPF, machine learning, CCL data, and operational expertise are being combined to help completion teams identify potential deformation earlier, reduce uncertainty, and make more informed decisions throughout the completion lifecycle.

Historically, casing deformation has been difficult to manage because visibility often arrives too late. Restrictions are frequently identified only after a plug encounters resistance or a completion operation is interrupted.

YPF’s approach focuses on identifying potential deformation before those events occur.

Using machine learning models trained on CCL data, the team can detect anomalies that may indicate developing casing deformation and provide engineers with additional context for evaluating risk. Rather than replacing engineering judgment, the technology serves as an early warning capability that helps teams investigate potential issues sooner.

The purpose of this technology is not simply to detect deformation. It is to provide operators with the insight needed to make proactive decisions before restrictions become operational constraints.

High-quality data is the foundation of every successful analytics and machine learning initiative. As completion teams increasingly rely on data-driven technologies to guide decisions, the quality of the information being collected becomes just as important as the technology itself.

The ability to identify anomalies, uncover hidden risks, and generate reliable predictions depends on consistent, accurate inputs. Factors such as wireline speed, tool maintenance, remagnetization practices, and data filtering all influence signal quality and can have a significant impact on the insights that follow.

The takeaway is clear: better data produces better outcomes. When operators invest in strong data acquisition practices, they create the conditions for advanced analytics and machine learning to deliver meaningful value. The result is a stronger foundation for confident decision-making, earlier risk identification, and safer, more efficient completion operations.

Detecting deformation is only the first step. The real value comes from what operators do with that information.

When restrictions and casing anomalies are identified early, completion teams gain the opportunity to evaluate mitigation strategies before operations are affected. Whether that means adjusting stage spacing, modifying perforation designs, selecting alternative plug configurations, or revising operational plans, earlier awareness creates more options and reduces the likelihood of costly surprises downhole.

As completion programs continue to grow in scale and complexity, the ability to move from reactive troubleshooting to proactive decision-making becomes increasingly important. Identifying potential issues sooner allows teams to address risks before they escalate, helping reduce nonproductive time, protect lateral footage, and improve operational efficiency.

Ultimately, the goal is not simply to detect deformation. It is to provide actionable intelligence that enables better decisions throughout the completion process.

While machine learning has proven effective at identifying casing anomalies and potential deformation events, Corva and YPF continue to work together to deepen their understanding of how these signals translate into operational outcomes.

Today, anomaly detection serves as an early warning capability, helping teams identify areas that may warrant closer attention. The next phase of development focuses on strengthening the relationship between model outputs, restriction severity, and the operational risks associated with casing deformation. As additional data is collected and analyzed, Corva and YPF are working to better quantify those relationships and develop more precise decision-making frameworks.

The teams are also exploring how deformation correlates with other indicators, including tension signatures and operational performance metrics, to build a more complete understanding of how restrictions develop and evolve over time.

This ongoing collaboration reflects a shared commitment to moving beyond detection and toward prediction. By continuously refining models, validating results, and expanding the dataset, Corva and YPF are laying the groundwork for more actionable completion intelligence that helps operators anticipate risks, prioritize interventions, and optimize execution before issues become operational constraints.

Perhaps the most important takeaway is that casing deformation detection is not about replacing engineering expertise. It is about equipping teams with earlier awareness of potential risks so they can make more informed decisions with greater confidence.

When machine learning is combined with operational knowledge, historical performance, and field experience, it becomes a powerful tool for turning data into action. Rather than reacting to problems after they occur, teams can identify emerging risks sooner, evaluate mitigation options, and make adjustments before those risks impact execution.

For operators focused on reducing completion risk, minimizing nonproductive time, and improving operational consistency, that shift from reactive troubleshooting to proactive risk management can deliver meaningful value across the completion lifecycle.