Can you come up with a hand-held solution instead of using EARS?
This was the challenge presented to us during our first meeting with PTT Exploration and Production.
The current iteration of EARS is not yet ATEX-certified which is a strict requirement for any device brought into any of PTTEP’s facilities. Hence, leaving EARS in their facilities for long-term assessment was not an option. Fault detection, however, seemed to pique their interest. It was suggested that we provide them with a solution that would allow them to 1.) carry a hand-held device to record sounds, then 2.) return a result that would tell them of any faults detected. Cool idea? Sure. Could we implement it? We thought we’d find that out.
When people hear of our solution, they think: “Oh, you deal with hardware.” This couldn’t be further from the truth. While integral to our solution, EARS simply facilitates the collection and transmission of acoustic data.
What our clients are really paying for are the data we collect and the algorithms we’ve come up with to process all this data. We’re closer to an AI company than anything else.
Given that, we thought we’d give this experiment a shot. Not being able to use EARS was a limitation (the device is equipped with noise reduction and is tuned to listen to motor frequencies), but we wanted to see the level of accuracy we could achieve without it.
On an early morning in June we took a 2-hour long ride to Suphan Buri province in Thailand. Once we got to PTTEP’s drilling rig, we were led into a room where a safety seminar was conducted.
As mentioned, PTTEP takes protocol very seriously and since the current iteration of EARS is not yet ATEX-certified – we were not allowed to use it. We were, however, allowed to take recordings using our mobile phone.
We took a few samples of 2-minutes each per motor at specific settings to ensure quality collection of data. We began with a couple of water injection pumps on-site.
We then moved onto a few other motors attached to beam pumps. There was a lot of waiting involved having to record several 2-minute samples one after the other. Between recordings, we’d have a chat with some engineers from PTTEP.
After all was said and done, we headed back to Bangkok. We uploaded samples from a total of 6 industrial motors to our system. The results were sent to PTTEP which were then analyzed, compared with their own data, then sent back to us. We scored a mere 33%. Quite low, but not bad considering this was our first real-world test of a hand-held system and while using an uncalibrated device.
Our ATEX-certified device will roll out in 10 months. In the meantime, we’ll use what we learned here to improve this potential solution and to see if it’s a viable product.