Machine Learning Emissions Estimation Using LiDAR Remote Sensing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for accurately and timely reporting of gas emissions at hydrocarbon operation locations are often inaccurate and untimely, leading to regulatory fines, as they lack the precision and real-time data necessary for effective emissions tracking and management.
Innovation Solution
A machine-learning model is trained using high-resolution, real-time data and granular equipment-level data, including LiDAR remote sensing and historical leak data, to predict emissions factors, enabling precise emissions estimation and facilitating smart decision-making for asset managers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional emissions reporting methods are used, then reporting process is simple, but measurement precision and timeliness are insufficient leading to inaccurate emissions data
Solution Approach 1:
The patent replaces traditional mechanical/manual emissions measurement systems with optical sensing technology (LiDAR) and machine learning algorithms. The LiDAR system uses laser light to detect and measure emissions remotely, while ML models process the data to provide accurate emissions estimates, eliminating the need for complex physical sampling equipment and manual analysis procedures
Solution Approach 2:
The patent introduces machine learning models as an intermediary between raw LiDAR sensor data and emissions reporting. The ML models process and interpret the complex optical sensor data, translating it into accurate emissions estimates that can be directly used for regulatory reporting, thereby simplifying the overall system while improving precision
2Reliability
If traditional emissions monitoring is implemented, then implementation cost is low, but real-time data collection and accuracy are insufficient
Solution Approach 1:
The patent extracts only the most relevant features from the large volume of LiDAR sensor data using machine learning models. Instead of processing all raw data points, the system identifies and extracts key emission-related features, reducing computational burden while maintaining high reliability in emissions estimates
Solution Approach 2:
The patent uses LiDAR technology to create remote optical copies of emissions data without requiring physical contact with the emission sources. This allows for non-intrusive, continuous monitoring of emissions from distant locations, providing reliable data without the complexity of on-site sampling equipment
3Productivity
If manual emissions reporting is used, then reporting process is simple, but timeliness is poor leading to delayed reports and potential fines
Solution Approach 1:
The patent implements continuous automated emissions monitoring using LiDAR sensors that continuously scan and measure emissions in real-time. The machine learning models continuously process the incoming data stream, providing ongoing emissions estimates that are immediately available for reporting, eliminating the discontinuous nature of manual sampling and analysis
Solution Approach 2:
The patent establishes a feedback loop where LiDAR sensor data is continuously fed into machine learning models that generate real-time emissions estimates. These estimates can be immediately reported to regulatory authorities, and the system can also provide feedback to operational personnel for immediate corrective actions if emissions exceed thresholds
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The solution provides reduced uncertainty in emissions predictions, helping asset managers meet yearly emissions commitments and enabling timely, accurate reporting, thus avoiding regulatory penalties.
Implementation Method 1
LiDAR remote sensing
Data Source
AI summary
A system can collect a first set of equipment data and emissions data from a first hydrocarbon operation location. The system can train at least one machine-learning model to estimate an emission factor of at least one equipment component of the first hydrocarbon operation location using the first set of equipment data and the emissions data of the first hydrocarbon operation location. The system can then apply the at least one machine-learning model to a second set of equipment data to estimate total emissions over a predetermined amount of time at a second hydrocarbon operation location.