Predictive Analytics for Rail
4Atmos Technologies goes “Beyond the Lab” using AI based solutions to predict and prevent catastrophic failures for rail transportation companies. 4Atmos is short for 4Atmospheres in diving (below 99 feet) where only master divers can take their customers safely down and back. We dive deep into the data to find the treasures or the risks that are beyond view of the human eye.
Risk Identification & Assesment
4Atmos has deep domain experience in rail. For more than 14 years, we have helped large class 1 freight and passenger rail (global) companies identify at-risk critical assets (locomotive prime mover engines, traction motors, Head End Power- HEP engines, coolant systems, compressors, transformers) using proprietary solutions that are measured 6 digits below the decimal point (parts per million – PPM). These methods and processes break down lab data into known failure modes (e.g. fuel, crankcase overpressure, turbo, exhaust, bearing wear, electrical and contamination) while ranking the fleet from “worst to best”. Beyond identification, the 4Atmos model includes prescriptive recommendations (step-by-step) for inspection and repair ahead of catastrophic failures.
Critical Failure Intervention
In the US, our work with a large class 1 freight railroad has resulted in more than 2,000 “good catches” where critical failures were avoided by early intervention, saving millions of dollars and increasing availability, life cycle management and safety. The scope of services include: warranty overwatch, third-party contract maintainer support, compliance with oil/fluid sampling and new technology experiments (e.g. renewable diesel, alternative fuels, new concepts/methods).
The 4Atmos model provides real-time access via cloud based web-dashboards and detailed inspection plans that are available remotely on desktop computers, laptops, tablets and mobile devices. Our work in passenger rail has been featured Forbes.com.
Non-Invasive Inspections
What makes 4Atmos different is that our team of experts stand side-by-side with the crafts and help direct non-invasive inspections targeted to known failure modes. More than 80% of our “good catches” occur at element limits that are often times 40%-50% below lab thresholds.
Using cluster models we are able to rank at-risk locomotives based on their performance against the fleet, the class of power, number of times / duration that failure mode signatures have been present…even predict the date of failures accurately. More than data scientists, we are experts in engines / components and have helped our clients recognize their ROI in under 30 days. The insights to action provided to our customers provide a comprehensive view of their risk and the value of prevention… long before the onboard alarm systems ever alert. Outside of each client’s fleet, we compare performances against our database of all our customers. This provides us with a critical advantage in terms of transfer learning and refined failure signatures.
There are weekly teleconference calls with our customers to review highest risk / action plans as well as daily notifications via text or email for units that need immediate attention.
During the global pandemic of 2020-2022, 4Atmos never stopped working and there were no interruptions with any of our customers.
In the world of AI / Internet of Things there are a lot of companies claiming to have solutions but 4Atmos has a proven track record that is affirmed daily with real-world results and proven / quantifiable savings.
We can deploy a new program in 5-30 days using a railroads existing data.
