Leak Source Isolation™
Proprietary algorithms enabling highly accurate localization and quantification of gas emissions to proactively guide LDAR teams and operators with measurement-based emissions intelligence.
Where LSI™ Fits in the MethaneTrack™ Technology Stack
MethaneTrack™ is designed for continuous monitoring of methane emissions and broader hydrocarbon emissions environments.
Leak Source Isolation™ (LSI™) is the intelligence layer of MethaneTrack™. It consists of real-time cloud-based algorithms that continuously analyze field measurements collected by MethaneTrack™ hardware and generate emissions validation, localization, and quantification outputs.
How does Leak Source Isolation™ Work?
The LSI™ algorithms use proprietary plume modeling to analyze MPS™ sensor data from wireless IIoT Endpoints, wind data from anemometers, and Endpoint location data to determine:
- Leak validation (presence and start time)
- Leak localization (source position)
- Leak quantification (current rate and total emitted volume)
- Leak lifecycle tracking (start time, end time, total duration)
The Role of Close-Proximity Continuous Monitoring™ (CPCM™)
MethaneTrack™ uses Close-Proximity Continuous Monitoring™ where Endpoints are placed on or near likely emission sources rather than at distant boundaries. This approach captures gas signals before significant dilution or dispersion occurs.
Key Outcomes of Close-Proximity Continuous Monitoring™
- Earlier detection of leaks
- Higher signal-to-noise ratio
- More reliable inputs for localization and quantification algorithms
- Improved performance in complex site geometries
Plume Modeling in LSI™
LSI™, now in its 5th generation, uses empirically driven plume modeling tailored specifically to decades of experience with the NevadaNano’s MPS™ sensing technology and has been further refined by learnings from thousands of devices deployed in the field at customer sites globally.
The system uses the following inputs:
- Time-aligned gas concentration data from 675 readings per hour per Endpoint
- Time-aligned wind speed and wind direction data
- Time-aligned environmental data such as temperature and humidity.
- Known sensor placement geometry
What is Plume Modeling?
Plume modeling is a computer-based simulation technique used to predict the movement, concentration, and dispersion of airborne contaminants—such as gas, smoke, or hazardous particles—in the atmosphere.
- The accuracy of plume modeling depends on several factors including:
- The quality and resolution of the input data
The suitability of the model to the specific sensing technology and site environment
LSI™ Outputs and Benefits
MethaneTrack™ LSI™ provides:
Leak Validation: To reject false positives
To prevent unnecessary site repair visits and provide accurate historical emissions records.
The ULP MPS™ Emissions sensor enables the MethaneTrack™ system to have a nearly ideal Receiver Operator Characteristic (ROC) curve. The ideal ROC curve demonstrates a 100% probability of true leak detection and a probability of false leak detection of 0%.
MethaneTrack’s gas sensor uses onboard sensor algorithms to implement a smart, variable Lower Limit of Detection (LLOD). The LLOD is updated on every sensor cycle and is based on real-time environmental measurements made by the sensor. The algorithms adjust the LLOD to maintain a false positive rate (at the sensor level) of less than 5 occurrences in 1 million sensor cycles. Further false positive rejection techniques are then employed at the LSI™ level.
Leak Localization: To enable efficient repair
Localization determines the physical origin of an emission within a monitored site.
LSI™ performs localization by evaluating how gas concentrations vary across multiple endpoints under time-aligned environmental conditions. By reconciling distributed measurements with plume modeling outputs and known sensor geometry, the system infers the most probable source position.
Once LSI™ determines a leak location, that location is evaluated within the MethaneTrack™ data architecture. Endpoint positions and site geometry are mapped to assets within the system database, allowing the localized emission to be associated with specific equipment or infrastructure.
This mapping enables:
- Identification of the emitting asset or asset group
- Classification of emission events by operational context
- Development of measurement-informed activity datasets
This asset-level attribution supports the replacement or refinement of assumed activity factors traditionally used in emissions inventories.
Leak Quantification: To help prioritize repairs efforts
Real-time emissions rate and total emissions volume over the duration of the leak.
Quantification determines the instantaneous emissions rate as well as the total emissions volume over the duration of a leak.
Because Endpoints are placed near likely emission sources, meandering flow physics dominate gas dispersion behavior near the leak source. Conventional far-field Gaussian assumptions do not apply in this regime. Therefore, MethaneTrack™ calculates leak rates via a proprietary empirical model trained with, and optimized for, the MPS™ sensing technology used in the Endpoints and the anemometers deployed with MethaneTrack™. The model’s variables are wind speed, wind direction, the Euclidean distance determined by the model between the leak location and the Endpoint(s) detecting the leak, and the concentration(s) of methane detected at the Endpoint(s).
Once LSI™ determines the leak rate and leak location, those measurements are evaluated within the MethaneTrack™ system against customer-configurable emission rate and cumulative volume thresholds defined at the site or regional level.
This enables:
Severity-based alert notifications routed to designated personnel
Dashboard filtering and prioritization by leak severity
Emission volume reporting by asset, site, region, or time period
Historical analysis of asset performance based on total emissions, emission rate, and severity classification
Summary
MethaneTrack™ Leak Source Isolation™ combines high-resolution MPS™ sensor data, empirically derived plume models, and generational algorithm refinement informed by extensive field deployments and third-party validation. The result is a continuously improving system for emissions validation, localization, and quantification in real-world operating environments.