Question: How many of you have other personnel such as operators or maintenance collect PdM data such as vibration,oil samples and Infrared surveys? Part B to the question is how is that working out? I realize that there are some precautions to look for in letting other persons collect data, but I wanted to know if there has been any successes in this kind of format.
Any suggestions will be greatly appreciated.

Rafael

We currently have a millwright collect the vibration data, an electrican for infrared and motor current and area mechanics testing the oil. Outside of the millwright, there is no interpetation by the data collector. I gave them some basic rules of thumb. When they return to the office, I evaluate the data. Now, with vibration, the millwright who has been doing it for 6+ years, gives me a heads up on what is good and bad and some analysis. That helps a lot. It is my goal here to have them collect and due first pass analysis. Anything they cannot determine, then I can step in and help.

I work for a vibration products and services company and we have clients worldwide (some since the 1970's) that collect their own data and send it to us electronically for analysis. (approx 5,000 machines / month). Results automatically appear at their sites or can be accessed via secure web browser.

There are a number of ways that this can be accomplished that limit data collection errors and invalid data. This is especially important with some of our clients as they constantly change personnel and therefore cannot rely on having to train them extensively. That said, these contracts have produced excellent results year in and year out for over 20 years (and are renewed year in and year out because of this).

On the other side of this, we have developed very efficient analysis procedures to allow us to analyze vast quantities of data and provide accurate results with limited manpower. Results include specific mechanical faults and severities as well as repair recommendations.

We have had excellent results with this. If you are interested, I would be happy to describe some of the major things that need to be considered, both on the data collection side and the analysis side to limit errors, save money and make the programs extremely efficient. (let me know and I'll post this info)

Information can also be found here:

www.dliengineering.com

Alan, I would like to hear your thoughts on this subject. Any information on how to be more efficient and reduce errrors is always wanted.

Jim:
Sounds like it's working and your goals are very respectful. Is there any field balancing done and if so, by whom.

Alan:
Yes, I'm interested to read more details about the considerations you have to offer.

Thanks

Rafael

I'll give you a bit at a time as it will turn into a book (in fact it may eventually - especially if there is interest - each item here could easily take up a chapter in explaining the long term benefits and in how to actually accomplish these things) - Alot of the info on our website ( www.dliengineering.com )actually points to these solutions though not necessarily in this context.

Data collection error reduction:

1. Define standard test conditions for all machines (this may be the toughest but most important step - we can discuss this further if you like)

2. Use stud sensor mounts (insures repeatibility of test location and gives better freq. response) - more bang for the buck if you use a triaxial sensor (complete data and less time spent moving the sensors to different axis)More up front cost but big long term savings in time and better results.

3. Create a pocket test book for data collectors (personnel) that includes: diagram of machine with test locations (they can find them if they are not obvious and also replace studs if necessary), instructions for testing (conditions, special test setups or configurations etc.) and nameplate info (they can make sure the machine wasn't moved or replaced with a different model) - once / year or something have them check the whole plant against the books and then check the books against what is in the software to make sure your info is up to date. (note, the book has many other benefits in maintaining the system in the log term - also applicable for the analyst - more on this later - if you manage the books well you'll see a big increase in efficiency on a number of fronts)

4. Use barcode driven data collectors or some other way to make sure person is testing the point their collector thinks they are testing. (barcode numbers can be in the book and typed in also) Our first data collector was a yellow box with a sensor, a barcode wand and a little led display that said "OK, ready, done" etc.) - 5 minutes of training to learn it. Newest data collector has the same capability but you can also punch in the number with your finger.

5. Create a procedure for adding new machines to the system or updating information about them (tell them all the info you need about the machine in order to set test conditions and enter the info. in the software). There are forms you can use to facilitate this.

6. Educate them briefly on the importance of test conditions and the value of what they are doing - best way of doing this is to produce good results and make sure they know about them. Also always inform them when you receive bad or suspect data. Teach them how to replace test studs and equip them with extra studs and glue kits.

7. If possible, have them re-test machines before overhauling to insure the data the analysis was based on was good.

8. Use "note codes" in the data collector - a quick way for them to tell you if there was excessive noise, oil leaks or other obvious indications that something is not right.

when I get a chance I'll add the next steps (i.e. what to do on the analysts' end to manage this - and how to move information around if your client is on the other side of the world) -

PS. I respond well to encouragement ;-) and would be happy to clarify things.

Alan:
Thanks for your detailed answer, it is very informative. Is there ever a time when the persons collecting data, whether it be vibration, oil samples or thermography, get bored of doing just that, collecting data? I would think that over time (6 months or so)that individuals would feel that were just data collectors and maybe attitudes would start changing, which could possibly affect the accuracy of the data collected. I'm assuming that there has to be some kind of understanding and education involved to maybe give them a sense of owernership?

Jim, I noticed that a millwright has been collecting and is doing some analysis for 6 years. Has his/their attitude(s) changed any over the years?

Please correct me if I'm seeing this wrong.

Rafael,
In my experience, alot of the people collecting data also have 100 other jobs to do as well. They often see data collection as a break from their other - often more dirty and difficult work. Screwing a sensor onto a machine, checking some gauges and pressing a button on the data collector is often preferable to the other tasks they are responsible for. We always encourage people to learn as much as they can and we certainly try to get people to feel ownership for their work. Oftentimes however, the reality is that the employees have so many other duties that this technology doesn't necessarily become the center of their focus or their interest. We encourage interest when we see it but that's about all we can do.
On the other side of the coin, we can tell when someone is not collecting good data and do our best to convince them to not waste their time or ours. We had a case once where data on every machine on a route looked identical, and it was pretty obvious that the data collector person attached the sensor to one machine and then just kept collecting data....At the very minimum we try to educate technicians on the use of the technology and the reason they are doing what they are doing, and how good data leads to good results and bad data to bad results....

www.dliengineering.com

When the program was started here at the plant, I was scare that as a millwright making trouble shooting I was to downgrade to a datas collector, a button presser. But I have been involved in the program since the beginning and now I am responsible of it. I collect datas, analyse them and make the recommandations. I also do or supervise balancing jobs. That's interesting and I would not have accept to only collect datas and let someone do the interesting part of the job. On the other hand, some millwrights just can't sit in front of a computer for a full day of analysis. They need to work with tools with their both hands and as Alan said, collect datas can be a break for them. If you have someone who has an analist mind, make him or her collect the datas but involve him in the analysis phase. It will be good empowerment and you'll make him a good collaborator.
J-Marc

At our plant we currently have 4 operators that take oil samples and 2 mechanics that take vibration routes. I would like to see the mechanics get to a level 1 analyst level, but basically they are just data collectors. The vibration data is analyzed by the technical personnel. Every training class that I've ever attended for vibration analysis implies that, the person taking the vibration data should be the person analyzing the data. This is not taking place at our plant. Any suggestions on how to remedy this problem would be appreciated

Food for thought (and just my opinion): 80% of the diagnostic "information" available to an experienced analyst as he collects data on a machine, does not show up in the vibration data collected by the "box".....

As a result, much of the data is misinterpreted. Either bad machines are missed, or -- more often -- machines are called "bad" because the data seems to say so, even though there are no other symptoms present (but the analyst sitting at the computer doesn't have the missing symptom info available to him).

That said, if I had a choice between "suspect" data collected by inexperienced personel, or no data at all.... I'd take the suspect data!