Vibrations systems Testing & Evaluation
-
Vendor relations and evaluating their products
-
Fluke vs SymphonyAI vs Banner
-
DUS family can be up to 4 hours away from maintenance
-
Current solution: parent site is incharge of maintenance
-
Solution: remotely monitor the machines and generate work orders when needed
-
Result: work orders automatically generated, reducing the need for travel and oversight from parent site
Vibration Systems Project Scope
The DUS Family of sorters is what the post office puts in rural post offices which can be 4 hours away from maintenance capable sites. There are about 500 nationwide. If an issue arises with one you're looking at a day's pay for a maintenance worker just to go inspect it and figure out what's wrong. This doesn’t fix the actual problem with the machine.
To that end the post office wanted to put monitoring systems on those machines so that they could reduce maintenance cost and downtime. I initially started with a competitive demo between Fluke and Symphony AI. They both use vibration systems to monitor motors and issues with the backbone that is the part that sorts the packages into their designated bins. I set up a design of experiment to test different failure modes of the backbone to see if they could detect the issues through vibration. Fluke came out on top however their solution would have cost $20,000 per machine and these are $50,000 machines which is not cost feasible.
I had a working relationship with Banners. They had recently developed something they were calling an AMG and suggested that they might be able to do a similar job but for much cheaper. I arranged for them to have some lab time on one of the machines to install their systems. I worked with them to create a way for us to pull their data from their system and into our database using modbus protocols through node-red.
Some hurdles I had to work through was that their documentation was not 100% accurate for modbus registries. There was a shadow registry in the twenties causing the registries I was trying to pull to be off by one and giving incorrect numbers. Once the problem was solved I was able to pull correct numbers and get them into our database.
Their solution only cost $4,000 per machine but required a little more work on my end. The Post Office customers were so impressed that they rolled it into a DEMO to the C-suite with PTC on one of the new machines being deployed as part of the ten year plan.
