top of page
rreale5

SnifferDRONE™ for Penetration Pre-Screening


On October 23rd, Sniffer Robotics proudly submitted a request to the U.S. Environmental Protection Agency (USEPA) for an Other Test Method (OTM) approval, seeking to utilize the SnifferDRONE for quarterly compliance monitoring of landfill penetrations. If approved, this will substantially decrease the manual effort of walking across landfills to monitor penetrations, resulting in significant safety benefits without added environmental risk.


But how does it work, how effective is it, and why is it valuable?


SnifferDRONE™ conducting surface emissions monitoring at a well head penetration

"Our SnifferDRONE has been streaming landfill methane data into our geospatial data servers for some time now," Stated David Barron, CTO. "We've observed a consistent phenomenon: our GIS team can predict which penetrations are likely to emit methane above exceedance criteria and which are not."


Starting in mid-2023, Sniffer Robotics began building a specific database of landfills that included both SnifferDRONE flights and manual inspections. From this data, we developed an algorithm to replicate the intuition of our GIS team. This algorithm analyzes the methane signal (and other data) collected by the SnifferDRONE and classifies the risk of emissions at each penetration. The algorithm outputs a binary metric indicating whether a penetration is at low risk and does not require further inspection or if it does require manual inspection.


This algorithm is effective for two main reasons:


  • Landfill gas is heavy and tends to pool near the base of the penetration.

  • The SnifferDRONE, sampling per OTM-51, collects air samples directly at ground level, where it is highly sensitive to heavy landfill gases.


In developing the algorithm, we quantized SnifferDRONE methane data by the distance to the penetration of interest. If enough data was collected, the algorithm could assess the risk/reward of further manual inspection. We analyzed 12 sites with 2600 penetrations and millions of SnifferDRONE methane data points. After more than a billion calculations, we determined that sufficient data was collected at 357 penetrations ("addressed penetrations"), including 55 with emissions above 500 ppm.


Image of penetrations with predictive red and green markings.
Penetrations requiring manual inspection, as determined by Sniffer's algorithm, are shown in red.

Applying the algorithm to these 357 penetrations showed that 48% were at very low risk of emitting methane above the exceedance definition. This means that on a 100-acre landfill, miles of walking to penetrations could be eliminated by analyzing SnifferDRONE data, enabling better-informed risk decisions. Half of the penetrations still require manual inspection, as the algorithm is designed to minimize the “false negative rate”—the rate at which the pre-screening algorithm exempts a penetration from manual inspection when it is actually above the exceedance definition. The algorithm, as submitted to the EPA, has a false positive rate of just 0.56% relative to the total number of addressed penetrations.



The Bar for an Other Test Method (OTM) is High...


High Pole Vault Bar

The alternative must not affect the stringency of the applicable regulation. The question then becomes: How effective is manual inspection of penetrations today relative to the penetration pre-screening algorithm's false negative rate? Sniffer Robotics analyzed penetrations where manual inspection resulted in concentrations above 500 ppm. Our technicians then took extended duration measurements at the location of the maximum emission and recorded the output. It is well known that landfill emissions are highly affected by environmental factors such as wind and gusts. Plots of methane concentration at these exceedance penetrations clearly showed gusts knocking the concentration down, followed by a gradual rise as methane accumulated. These data points demonstrate a stochastic element to penetration inspections created by the sampling time domain relative to unpredictable environmental factors like wind gusts. From the long-duration measurements at exceedance penetrations, we quantified the probability that these environmental factors would result in a false negative for manual inspection. This study proved that the penetration pre-screening algorithm has equivalent performance to manual inspection regarding our cardinal metric, the false negative rate.


Given this equivalent performance, Sniffer Robotics is excited to share that we have submitted a formal request with a thorough technical data package to the USEPA, requesting approval to use the penetration pre-screening algorithm in conjunction with manual inspection of at-risk penetrations for compliance with Surface Emission Monitoring. If approved, this will substantially decrease the manual effort of walking across landfills to monitor penetrations, resulting in significant safety benefits without added environmental risk. In the meantime, we are operationalizing the penetration pre-screening algorithm to provide our clients with a better understanding of their landfill's overall health and offer another tool to optimize their CapEx for the greatest emissions reductions.

Comments


bottom of page