Accurate Small Leak Detection for Liquid Pipelines with LeakGeek
Protect your assets and the environment. LeakGeek offers the AI-driven accuracy needed for reliable small leak detection that standard systems often overlook. This UTSI software enhances your monitoring system by using intelligent analysis to find small leaks precisely. It cuts false alarms by more than 90%.
Data Quality Method (DQM)
The foundation of accurate leak detection is reliable data. UTSI’s Data Quality Method (DQM) is the first step LeakGeek uses to ensure the data feeding your models is always correct, precise, and useful.
Cleaning the Signal: We remove the “noise” or bad signals like sensor glitches or temporary disruptions that often mislead conventional systems and trigger false alarms.
Ensuring Reliability: DQM checks data quality at every stage. This solid foundation allows the algorithm to work with unmatched stability. As a result, your leak detection models begin with the best possible information. This proactive data preparation step is essential for achieving the high accuracy needed for effective small leak detection.
Automated Data Labeling
The next important step in achieving LeakGeek’s superior performance is smart data preparation through automation. Our AI-driven tool performs automatic data labeling. This process provides the necessary context for the raw information coming from your pipeline’s SCADA system.
Context for Confidence: Auto data labeling instantly and accurately tags every normal pipeline event, such as planned pump changes, valve movements, or temperature shifts, in real-time.
Smarter AI: This context helps the algorithm quickly point out what is normal behavior and what indicates a subtle, abnormal leak signature. By training itself on clearly labeled events, the AI gains the necessary confidence to avoid mistaking standard operations for emergencies.
Actionable Insights: This process eliminates data overload, providing operators with clear, actionable alerts instead of confusing data streams.
Small Leak Detection
LeakGeek is specifically engineered to overcome the industry’s biggest challenge: finding leaks that are less than 5% of the total flow.
High Sensitivity, Low False Alarms: By using DQM and automated data labeling, the detection model achieves high sensitivity without the common issue of many false alarms. It consistently verifies and checks for the real signs of a leak.
Faster Response: The model has been tested in real-world settings. It can confirm a small leak detection event hours faster than many leading conventional systems. This speed is key to reducing environmental damage and decreasing cleanup costs.
Protecting Throughput: By identifying genuine leaks and distinguishing them from routine operational events, you maintain continuous operation. This prevents unnecessary and costly emergency shutdowns.
Meet the Leak Geek Team
Gustavo Sanchez
VP Data Science
Gustavo is the Founder of Pandata Tech (acquired by UTSI) and a 2X entrepreneur with over 15 years of Data Science, Machine Learning and Business Development experience in the energy, national defense, national intelligence, geothermal and geospatial industries.
Education: BA – Lafayette College, MBA – Hult International Business School,
John Kozicz
Senior Data Scientist
John has over 35 years of experience in Offshore Drilling, Technology Management for Drilling Operations and Innovation with Drilling Data Acquisition, Analysis and Modeling System.
Education: Petroleum Engineering at SAIT in Calgary Alberta
Brandon Lovatt
SCADA Lab and R&D Lead
Brandon is experienced in SCADA Systems Design, Implementation and Support, Database Design and Management and Systems Integration and Automation.
Certifications: OASyS Systems Integrator, Nozomi Networks Certified Engineer
Education: BS – Computer Engineering, U of H Clear Lake
UTSI HEADQUARTERS
Toll-Free: +1.800.324.8874
Telephone: +1.281.480.8786
Fax: +1.281.480.8008
1301 Fannin St Ste 775 Houston, TX 77002
Need additional information
or want to speak with us about a specific project or need?