By Michelle Ferguson
An Edmonton entrepreneur is completely disrupting the field of maintenance engineering with his patent-pending vibration sensors.
An avid reader and consummate learner, Sunil Vedula has always been interested in the power of predictive modelling. Rather than get rich off stock market predictions, however, the mechanical engineer turned his attention to the field of material sciences — using the latest discoveries in artificial intelligence to help companies predict when and how their industrial equipment will fail.
“I’m not interested in stock market predictions; I don’t think you can do that reliably,” Vedula, CEO of Nanoprecise Sci Corp, says. “But I’m very confident that we can do material failure predictions, because materials behave in a certain fashion — we just need to understand the pattern behind that.”
Unplanned downtime can cost manufacturing, petrochemical and utilities companies between $500,000 to $2 million per day. Over half of these unscheduled events are due to equipment failure. Vedula hopes to eliminate a large portion of these disruptions with its series of AI-powered sensors.
The company’s flagship product, VibrationLF, collects vibration data, temperature, pressure and humidity readings every five minutes. The wireless sensor can be adhesive, stud or magnet mounted, and communicates securely with Nanoprecise’s cloud server. The raw data is analyzed, then published in real-time on a user-friendly dashboard that sends friendly reminders to order a part or schedule maintenance work.
While VibrationLF is designed for use on rotating equipment, such as pumps and compressors, Vedula envisions covering all four industrial asset types: rotating equipment; static and load-bearing equipment; piping and pipeline systems; and coatings and paints. The company has already developed a strain sensor, called FatigueLF, for use on static and load-bearing equipment, such as boilers and cranes.
Nanoprecise started off as a post-failure analysis company, but quickly pivoted when Vedula came across a powerful new algorithm, called Hibert-Huang Transform, that could be used to segregate different vibration frequencies.
“There are many moving components in a machine that vibrate at different frequencies,” Vedula explains. “Together they give rise to one vibration signal.”
Using Hilbert-Huang, Vedula created a patent-pending algorithm capable of predicting when and how a specific component of a larger asset will fail.
“That’s something that isn’t being done by anybody right now.”
Wireless sensors aren’t new to industrial maintenance, but none are as accurate as Nanoprecise’s. The company boasts 100 per cent fault detection accuracy, based on four case studies in which they saved companies close to a million dollars in losses.
“This specific [algorithm] was being used to detect brain tumours, earthquakes and solar flares. It’s a very sophisticated technique, so it was being used in very sophisticated fields,” Vedula says of the company’s competitive advantage. “And computationally, it’s very intensive … But thanks to low-cost cloud computing, using this technique for machine condition monitoring has become possible.”
To train its machine learning algorithm, Nanoprecise partnered with the University of Alberta’s Reliability Research Lab. Part of the mechanical engineering department, the lab has collected years of data on everything from bearing to gear failures.
The company also hired condition monitoring engineers (engineers whose job it is to take vibration, temperature, pressure and humidity readings once a month) to help refine the algorithm.
Nanoprecise has seen incredible growth since pivoting in December 2017. The company currently counts two offices — one in Edmonton and another in India — 20 employees and over 40 customers in Canada, the U.S., India and China.
In February, Nanoprecise was selected as one of the top 50 AI and IoT (Internet of Things) startups in the world by the Smart City Summit and Expo — Asia’s premiere IoT exhibition. Vedula was invited to present his technology in Taipei, Taiwan, on March 26-29, as part of the summit’s inaugural AI 50 program.
While the technology was developed with industry in mind, Vedula says the possibilities are endless: trains, wastewater treatment plants, airports.
Nanoprecise might even start supplementing or replacing your car’s mechanic.
“Imagine if instead of someone taking apart your car and performing a post-failure analysis, you had a sensor that was always tracking the aging of different components?” Vedula says.