An IoT-based Multi-Gas Sensor to Identify Precursors Used to Predict Flameout Events

The oil and natural gas industry is the largest source of methane emissions in the United States, responsible for about 30% of total emissions. The problem is that numerous peer-reviewed publications are being published regularly indicating that emissions are either higher than reported or higher than expected. Within oil and gas, refineries are one of the larger sources of methane. This is because methane that is supposed to be burned off (e.g., flaring) doesn’t burn efficiently, or a flameout condition occurs. Since methane’s 100-year Global Warming Potential (GWP) is ~28 times more dangerous than carbon dioxide, any system that can provide an early-warning alert to signal a flameout should be incorporated into the mitigation strategy. The question to be answered in the proposed work is whether a fundamental emissions signature can be identified (a precursor) that occurs prior to a flameout. In other words, are there specific species of gases that are emitted either right before -- or immediately after -- a flameout event (and what minimum thresholds were required to sense the precursors)? It is expected that if a multi-gas platform can be developed with the required sensitivity and selectivity, flameouts can be predicted -- resulting in quicker response times -- or even prevention. The sensor platform to be developed is based on an internet-of-things (IoT) multi-gas sensor platform that has been developed in partnership with the University of New Mexico’s Center for Micro-Engineered Materials, where additive manufacturing is leveraged, together with edge and cloud-based data analytics, to create a low-cost mixed-potential electrochemical (MPE) multi-gas sensor platform. This system is capable of identifying and quantifying emissions. The key to the system is that it can differentiate various sources of methane (e.g., anthropogenic, wetlands and agriculture) through machine learning algorithms. This capability can be applied directly to determine the precursors of a flameout event and will provide the input to the early-warning alert system proposed here. Prototype/pre-production NOx sensors (with more than 1000 hours of real-world testing) based on the proposed technology have been tested in vehicles. Additively manufactured methane sensors have been field tested at the Methane Emissions Technology Evaluation Center at Colorado State University. Finally, preliminary work has been performed on detecting hydrogen for Hythane applications in the laboratory using the same technology. The biggest challenge in the proposed system is the sensitivity and selectivity of the MPE sensors. Improved sensitivity and selectivity will leverage techniques such as (1) adding signals using multiple sensors to improve signal to noise and (2) pulsed biasing techniques. The latter technique was used to bring NO sensitivity in zirconia sensors down to sub-ppm level (<0.5 ppm). To commercialize these sensors, one focus must be on manufacturability. The manufacturing of these sensors using traditional screen-printing technologies is possible. Packaging will be similar to the lambda oxygen sensors (used widely in vehicles today) demonstrating there is (already) a reasonable path toward low-cost, mass production of these devices for widespread deployment. Engine dynamometer testing found that MPE sensors withstood exposure to exhaust gases -- as well as demonstrated the ability to resolve changes in engine conditions with resolution on the order of seconds (sensors are fast). The ability to create a sensor platform that can “predict” flameouts has significant impact on refinery emissions. Once the sensor platform has been developed, the system can be coupled to a secure (modular) communication system that is capable of remote monitoring.