We are looking after a plant having 200 gearboxes of defferent sizes ranging from 2 hp drive to 300 hp drive. It has mixed population of worm , helical, bevel and geared motors. The drive is for Belt conveyors, Feeders, cenrtifuges and similar equipment.

At present there is no predictive maintenance schedule, but we would like to introduce it immediately. Since we are entering into first stage we dont wish to procure many aparatus for monitoring.
What technology sould be adopted by us for predictive maintenance( Ecconomical and reliable),
The Ultrasound analysis
Vibration analysis or
Oil analysis ?

Please help us in deciding about the initial procurement and adoption of programm,

RP

Doesn't AGMA have any RPs (Recommended Practices) for gear maintenance practices? This is to start with I guess.

Dear RP, you're gonna open a can of worms here - but all technologies can be used.
Usually it's safest to combine more than one for backup.
But - would first look at your equipment and classify it according to criticality.
Do the critical ones - leave the rest.
You're programme needs to be manageable.

Mike.

Seems to me it would be extremely risky and possibly costly in a variety of ways to try to do all the gearboxes the same.

Some are more or less critical to the process. I'm no gearbox expert, but I don't think the failure process is the same in each and their respective applications might have some bearing on how they fail. Plus, how soon do you need to know the failure? If some replacements parts or full pieces of gear are in short supply you need to know an impending failure much earlier than if the parts or gear is readily available.

All of these technologies will provide useful information and indications of impending failure, but probably at different points on the failure curve.

Most bang for your buck on a gearbox: oil and vibration analysis.

RP:

Mike, Joe and Dave have already provided sound advice. Where I work we use vibration analysis, Infrared surveys, motor current/circuit testing, and ultrasonic testing. Additionally, we have an on-site oil lab primarily utilized as a screening tool to determine what samples warrant further, detailed analysis, which in turn are sent to an outside resource.

Mike and Joe are both on-target with their advice. It is important to select those units to begin monitoring based upon a criticality basis. Attempting to do all at once would be too large a meal to consume.

I wholeheartedly agree with Dave, based upon the type equipment it is you wish to monitor: oil analysis & vibration. Given only one to choose, it would be oil analysis. I have seen here that oil analysis will pinpoint impending problems with gearboxes well before (in some cases up to two months) vibration data will reveal problems of concern.

The most important advice I would add is that whatever road you ultimately choose to travel, make sure you acquire trained, experienced resources to evaluate program output.

Gary Forsythe

RP,

You've excellent responses from all above. In order of priority, gearbox criticality classification, oil & vibration and finally trained resources (most important) in the program.
I've been in Plants where there is a full blown PdM team & gadget but miss out on the most obvious (visually) defects.
If you have to choose one tool, I will say start off with Vibration, as most of the time you can get the support of lubricant suppliers, for oil analysis; this depends on contractual agreement.

Good Luck....Rajan

In order of best predictive method:
1- Vibration analysis -2- Oil analysis and a remote -3- sound measurement

In order of cost from less:
1- Sound measurement -2- oil analysis -3- vibration analysis

In order of my recommendation: See the first order!

Sound analysis is not accurate and repeatable and prone to data contamination from outside.

Oil analysis produces results only after gears start wearing each other out whereas one would see a gearbox with misaligned shafts immediately with vibration analysis.

Vibration analysis produces results when the fault exists. However, some rightfully claim that vibration does not produce results until gears are worn out enough. One would see the wear particles in the oil sample they say and I think it is more correct than wrong.

Regards
Ibrahim

For Gearbox, Vibration Analysis and Oil Analysis is a most effective condition monitoring tool. You will monitor vibration readings regular basis (Biweekly, Weekly, Fortnightly, Monthly and once in 2 months on the basis of criticality of gearbox). You will monitor physical properties of oil (Viscosity, Water Content, TAN, TBN etc.) on regular basis and Wear Debris Analysis on once a 2 months or monthly on the basis of criticality of gearboxes.

Thanks K. K. GAJPAL

Thanks everyone. I have also attended a workshop of Ultrasound monitoring and they claim that initial defects can be detected by it.

The oil analysis is good but the equipment/aparatus quoted to us are costly.
The usual defect in the gearbox is not break down due to gear failure, it is mainly due to oil seal leakage - thus affecting the bearings which in turn affects the whole gearbox.
I would like to have some hand held system which can predict these problems. Since all the gearboxes are in one plant, collecting data once in a month for all the critical gearboxes willnot be a problem.
Can you suggest something cheaper in light of the new data provided
RP

RP,

There's no doubt in Ultrasound's ability in detecting component faults. The question lies in training someone to both use the equipment and attune/recognise fault conditions; in a timely manner to capture potential failures.
With regards to oil analysis, there are on-site portable test kits that can determine the oil condition (basic properties). Alternatively, if you have an Industrial laboratory close by, try getting their support/services.
The CSI system has portable packages for both Vibration and Oil analysis.
Remember, the cost of above systems are not cheap but, they are not exhorbitant either.
If you are serious in starting up a PdM program then, you got to be prepared to invest in test equipment and training your people. Otherwise whatever you buy will become "White Elephant" in no time.
My opinion, don't waste money on cheap overall level indicators.

Good Luck...Rajan

It seems to me that if you feel/know that most problems come from leaking seals, then you should be looking for a method/process to detect or replace leaking seals.

Waiting on vibration, oil analysis or ultrasonics to detect a problem likely means that damage and degradation of the machinery has already occured.

Am I missing something?

Thanks! every one specially Rajan and Joe.

RP,

Let me reassess your question again.

You are about to enter into predictive maintenance and want to know what technology is best suited for gear boxes.

First of all you should go through the definition of predictive maintenance. Tools used in it are expensive and needs an adept person to operate them.

You can buy an expensive vibration analyzer or an oil analysis lab but there is nobody qualified enough to achieve full out of it.

I think you should go for a DUAL CHANNEL Vibration Analyzer (eg. CSI 2130) and get ateast two people to get VA I certification. Out of these two, one must be a graduate experienced mechanical engineer.

Oil samples can be analyzed from outside and it is not expensive.

Trend all the machines and analyze them on priority basis (i.e., big gear boxes needs more attention than the smaller ones).

Regards

Zeeshan

I hate to twist the elephant's tail but why MUST one be "a graduate experienced mechanical engineer"? If you have a journeyman with good troubleshooting skills couldn't he do just as good as the mechanical engineer?

Aubrey

Lowly Hammer Slinger

I'm with Aubrey on this one. I've run into a statistically significant number of mechanical engineers who weren't worth the value of the iron ring on their pinkie.
The difference that I see between the _average_ mech.eng. and the _average_ journeyman millwright (like me) is that the mech.eng. will have more theory, and theory can be of great value when one is out of their element - outside the bounds of their past experience.
That having been said, I'm still with Aubrey... even if the average mech.eng. would be more useful than the average journeyman for this task there are above average journeymen out there... and if my experience is any indication, decidedly below average mechanical engineers.

I don't understand the engineer thing either. I think that it is clique thing.
You get your best journeyman mechanic who is eager to learn new things and would like to get off the wrnech pulling and you will get a better analyst than a graduate engineer I guarantee it.
If you are just starting out, I would suggest looking at Commtest vibration equipment. It is very user friendly and can be set up using proven alarms by Technical Associates. This would be a good place to start until you get someone with some experience. It would also be a good idea to get a consultant in there to help set up the program and mentor your people for a period of time.

The real issue seems to have ben sidelined with a new debatecoming up whether- graduate engineer is better!
I am confident with my 37 years of experience after my engineering degree in 1970, that people with good technical mind can do wonders- without having a degree and people with bad ideas can also cause desasters in plant even if they have excellent degrees.

RP, I see no need for an engineering degree either. Willingness to learn and a determination to solve the problem combined with on the job experience are the best mixture I'd say.

I also don't quite see the need for a 2 channel analyser right off the bat.

Rather than looking at a program as "too expensive", consider the cost of failure and determine a percentage of that you consider appropriate to spend on a maintenance program. Then your goal is to use your maintenance program to save at least that amount of money (preferably more of course).

My two cents...