Kuva Systems: Harnessing infrared imaging tech for monitoring methane emissions [Gas in Transition]
Q: Can you give us a brief overview of what Kuva Systems offers for oil and gas companies looking to lower their carbon footprint?
A: Kuva enables the oil and gas industry to address methane emissions. We do this by having an end-to-end turnkey solution for continuous monitoring of methane emissions consisting of a proprietary cost-effective infrared camera which acts as an edge device, a cloud solution that continuously takes the data coming from the camera, stores and further refines that data, and two options to communicate with the customer: a web portal and also API/MQTT for customers looking for integration between their systems and Kuva’s solution.
The National Gas Company of Trinidad and Tobago Limited (NGC) NGC’s HSSE strategy is reflective and supportive of the organisational vision to become a leader in the global energy business.
Q: Can you explain how your Gas Cloud Imaging technology works? What makes it so effective at identifying gas leaks?
A: The way our technology works is that we use sunlight. Sunlight comes down to earth and it gets reflected and scattered into different directions, and we use what is called spectroscopy to find out if there is a certain amount of infrared light missing which happens when there is methane or VOC in between the surface reflecting the sunlight and the camera. This technique is known as open path detection. Kuva’s camera works like an open path detector except that we take an image, which means that we have to follow the above-described process for every pixel. Open path detectors have an infrared light source at one end. We use the sun as the source of light.
When you see an image from a Kuva camera, every pixel on that image is like the measurement done by an open path detector which is spectroscopically detecting the emission, and it is the sunlight hitting the ground and spreading in every direction including the camera, and we just look at what is missing in terms of infrared content, which we can determine as the content for the sunlight is known, and then using our algorithms we can say per pixel this is the amount of methane and VOCs present, and that’s how we colour every pixel in accordance with that amount.
We work the same way as a traditional gas safety open path detector: it is infrared, but not thermal infrared. We work at 2.3 um. That has several advantages: one is that it is not confused by water vapour or steam. That’s the reason open path gas safety is done at 2.3 um. Otherwise, if you have fog or if it is offshore it would alarm. We can detect underground leaks because we are not looking at temperature differential, in fact we are not measuring temperature, it is a spectroscopic device. It is a very established technology.
The camera is also self-calibrating: we do a dark calibration with every picture we take.
Thermal cameras work by the expansion of the gas: when it cools down relative to the background, you have to look at the difference. Now you have to deal with the incoming light, thermal emissions, and reflected thermal emissions, and with whatever the gas does, and it is hard to keep those apart because you have different backgrounds with different emissivity, something was in the sun, now it is in a shade, then it changes again. All this makes it very hard to automate thermal imaging-based detection. It is possible, and some companies have done it, but it is a harder physical problem.
For us, we don’t measure the temperature, so we don’t get confused by the temperature profile of the background. We still have to deal with the variable reflectivity of the background, but in our case we don’t look at the temperature profile of the gas plume we look at the absorption.
The key resulting benefit of our Short Wave Infrared (SWIR) approach is that it produces less false positives, or false alarm detections, which can be costly for operators. Kuva Systems employs a proprietary method where false positives are rejected. Soon machine learning algorithms will be part of the solution. Kuva strives to deliver only true positive, confirmed detections.
So, the first reason we have less false positives is that our problem is not as hard.
The second reason is that we have a proprietary technology analysing whatever remains of false positives.
Q: How are you harnessing machine learning to both automate the process and improve leak detection accuracy through false positive elimination in continuous monitoring?
A: Machine Learning (ML) is a very active area of investigation for Kuva. Over the last year, Kuva engineers have been collaborating with a team at the Alberta Machine Intelligence Institute (AMII) to design algorithms to advance our automated gas detection capabilities. Together, Kuva and AMII have developed ML models trained on tens of thousands of images, and we can report that the results are promising. The most mature of these ML algorithms significantly improves our detection accuracy. We are now working on incorporating the candidate ML algorithm into our commercial offering by the end of this year. ML research will continue with refinements on this model while also exploring different approaches and targeting different aspects of our product beyond the gas detection problem.
Q: There is getting to be more and more companies that are offering continuous monitoring of methane emissions. What makes Kuva Systems different?
A: Our camera is a proprietary camera. Our founders worked with the US Military where they worked for decades on infrared imaging for defence and national security purposes. They quit their jobs and funded Kuva as their way to apply all their knowledge and learnings to the cause of having a better future for our planet. We have 7 patents on our imaging technology and another patent for the tower we use to install the camera. All this means our take on the methane problem is unique.
Now, are we the only ones that can do a good job with visual detection of methane and VOCs? Certainly no, but no other company has the same performance to cost ratio.
What makes infrared cameras expensive is the semiconductor. Our technology base is the same as open path gas safety detection. That technology is very mature, and it has been around for decades. This lowers the cost of the technology.
There are good cameras out there that are like a Ferrari. We just like to think of ourselves as a Ford F150 pickup truck, just that we are lower cost… much lower than a pickup truck.
In addition to that is our no false positives approach: it is expensive for customers to send a crew to verify an alarm, and methane detection is well known for the large amount of false alarms produced by most detection methods. Not only does our technology make the elimination of false alarms easier (see answer to “Can you explain how your Gas Cloud Imaging technology works?”), but we have fine tuned our algorithms to help us in our no-false positives goal. Last, but not least, we have a team inside Kuva tasked with checking gas emission candidate images to determine if they are really emissions or not. The result, as independently tested and confirmed by the Methane Emissions Technology Evaluation Center at Colorado State University (METEC), one of the world’s leading authorities on methane monitoring technologies, is 0% of false alarms during their testing.
Q: Can you give us a brief overview of your company history and your growth trajectory?
A: Kuva was founded in 2016. We started with a breakthrough idea and have been refining our product ever since. Currently we have 40+ people, 25 of them are based in Calgary, 15 in the Boston area, a sales team in the US, and some team members in British Columbia and Ontario.
Q: Kuva Systems has been very successful in winning grants from Emissions Reduction Alberta (ERA), the Clean Resources Innovation Network (CRIN), and Natural Gas Innovation Fund (NGIF). How important have these funding and grant programmes been in accelerating your technology development?
A: Grants have been a critical component in accelerating our technology. One benefit that these grants have provided us is that they’ve helped us start and grow our Canadian operations since 2017 to more than 20 strong and establish working relationships among many of the top producers here.
The grants have also helped expedite our hardware and software development and validate it by facilitating multiple deployments in the harsh Canadian climate. Several key learnings have come out of these grants and our product would not be as robust today if it were not for these early stage projects.
Q: Geographically, what region are your customers based in and do you have aspirations to expand to other regions?
A: We are currently in Canada, USA and Australia. We are in conversations with entities in a dozen or more countries which are interested in our technology, but we are taking our time to ensure that the quality of our products and services is not compromised by expanding too quickly. It is a good problem to have!