Shaya Ecuador, an oil production services company operating a mature oil field in the Amazon region, partnered with SLB and Cognite to address significant operational challenges and unlock operational excellence. By implementing a comprehensive digital ecosystem, Shaya successfully moved "from the data to decisions" using Industrial AI and advanced Machine Learning to enable high-value, proactive operational improvements.
Problem: Operating a Complex, Remote, and Mature Field
Shaya manages a mature field in the Ecuadorian Amazon that relies on a heavy water flood injection program. The field presented four major obstacles to efficient operation:
- Waterflooding Uncertainty: The waterflooding program involved a complex, heterogeneous reservoir spanning more than five different reservoir areas and multiple layers. This complexity made it difficult to predict how much oil would be recovered when a certain amount of water was injected.
- Operational Losses: Significant production losses, ranging from 2% to 3%, were primarily attributed to electrical supply issues and failures of the downhole systems, particularly Electrical Submersible Pumps (ESPs). A core objective was to maximize production and to minimize these losses.
- Data Heterogeneity and Management: The data originated from highly diverse sources, including high-frequency downhole sensor data acquired via satellite, SCADA data from surface facilities, Excel sheets, and well diagrams. This mix of high-frequency, low-frequency, Excel, and SQL data made it very challenging not only to homogenize the frequency and precision of the data but also to store the data effectively.
- Challenging Logistics & Remote Access: The field is located in the middle of the jungle, with narrow roads and bridges, resulting in an average travel time of around three hours from north to south. Operational constraints, sometimes due to natural events or community restrictions, made physical movement difficult, necessitating ways to minimize physical inspections.
Solution: A Digital Ecosystem of AI-Driven Workflows
Shaya's digital transformation began by addressing its fundamental data problem: scattered, diverse, and siloed data. They partnered with SLB and Cognite to build a digital ecosystem centered around SLB’s Operation Data Foundation, powered by the Cognite AI and Data Platform.
The linchpin of this architecture is the Industrial Knowledge Graph. This technology is critical because it performs the essential step of data contextualization by integrating high-frequency sensor data, SCADA data, and unstructured historical data from diverse sources (e.g., Excel, SQL, diagrams) and mapping the relationships between them. This process transforms siloed data into a unified, reliable source of truth.
This reliability is crucial for providing accurate, contextualized data for AI/ML solutions and enabling the deployment of interconnected solutions, where the output of one tool (e.g., reservoir pressure estimation) can be reliably fed into another (e.g., smart optimization).
Reservoir Management
- Workover Candidate Selection: A customized Optiflow ML model integrated reservoir data (pressure, permeability, rock type) and production history to generate a preliminary list of wells for workover opportunities. The value lay in the speed, shrinking the analysis time for 200 wells from an engineer's three months to a couple of days.
- Automated Reservoir Pressure Estimation: This critical enabler analyzed intake pressure transients following well stops (due to power outage or downtime) to detect updated reservoir pressure. This tool allowed for pressure information coverage for more than 80% of the wells, dramatically surpassing the 5% to 10% coverage that engineers could achieve manually.
- AI Waterflooding Insight Tool: This tool uses physics-based models to run 'what if' scenarios, advising engineers on optimal water injection volumes and liquid production to maximize short-term recovery.
Production Engineering
- Machine Learning ESP Health Checker: This data-driven ML system compiled sensor data, production history, and historical failures to predict downhole system failures. The goal was to provide up to three months of anticipation. It operates with a 72% recall rate.
- Production Twin Models: These pipe-flow analysis models, embedded in Optiflow, are automatically recalibrated daily with new production data. This provides an immediate estimate of the flow rate when pressure or frequency changes, enabling staff to identify the loss in real-time.
Operations and Facilities
- Autonomous Operations for Scale Inhibition: A virtual flow meter provides hourly flow rate values, enabling a closed-loop system to automatically calculate and inject the correct, constant dosage of scale inhibitors required without human intervention.
- Operational Planner: This tool runs scenarios for workover and stimulation jobs by changing the well's position in the execution timeline. This optimization process is designed to maximize the Net Present Value (NPV) or the cumulative volume recovered
Impact: The Value to the Customer
By building a digital ecosystem of interconnected digital solutions, Shaya was able to magnify the value of each tool, achieving substantial, quantifiable results over a rolling year:
- Saved $4.3 million in Operating Expenses (OPEX)
- Increased production by 100,000 barrels of oil during the rolling year.
- Reduced 5,200 man-hours of work in general workflows
- Reduced 290 tons of CO2 emissions
The partnership between Shaya, SLB, and Cognite successfully navigated the complexities of a remote, mature oil field by establishing a unified digital foundation and deploying targeted Industrial AI solutions. By shifting the engineer's focus from time-consuming data gathering to rapid, proactive decision-making, Shaya realized significant, measurable gains in production, cost savings, and operational reliability, setting the stage for a fully integrated, end-to-end data environment.