Recording the instant an rotating event happens

[Ralph2]

Hello all

As part of our service we overhaul and adjust the speed an overspeed trip mechanism will activate at. This "overspeed mechanism" relies on centrifugal force on some imbalanced part to overcome the pressure of a spring, moving it to a new position, which hits a lever and causes a sequence of events. We mount this in temporary bearings and rotate it to ~4000 RPM. The trip rings work independently, must eventually trip at close to the same speed and during the trip phase (they will reset once the speed drops) fling out ~3/8 inch. Picture included of a typical overspeed jackshaft.

Up to now we have relied on the ear to note when the trip rings "trip" and the eye to capture the speed on a hand held tachometer. This is neither accurate or modern and our customers who witness this would like us to upgrade with some new technology

Other than hand held tachometers I am not familiar with what is available. What we need is a tachometer that will (output to a computer possibly) somehow record the speed trip event occurs. A tachometer that will record the maximum speed of a run will not work as the speed of the rotor can increase after the trip event.

Anyone with any thoughts on how we might accomplish this?

Thank you for your time.. and any ideas

 

[IRstuff]

I think you're mistaken in some concepts. A tachometer measures speed, so using your switch to trigger or tag some sort of recording of the tach output would result in a positive record of speed at trip. A tachometer could be any number of off-the-shelf tachometers with serial or USB output. Someone would need to log the streaming tach data, timetag it, and timetag the switch transition. There are dataloggers that have this sort of functionality already built-in and programmed.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[Ralph2]

Thanks IRstuff.
There is no switch when we are testing / setting this. The ring rotates with the rotor and on "tripping flings out. Catching the speed (RPM) this happens at is the issue. In service there are levers that are struck that dump oil pressure and start a shutdown process. Actually tripping is an extremely rare event and the design is not one for multiple uses. On our test stand there is nothing.. and making something mechanical for the ring to hit (multiple times while setting the speed) would subject the unbalance ring to wear / damage that would be unacceptable.

 

[IRstuff]

Nonetheless, one can design an optical sensor that can detect the "fling" and have it recorded and timetagged. I would suggest that you hire a knowledgeable EE to do this, but theoretically, you should be able to get almost everything to do this from, say,

http://www.omega.com/index.html

or any number of similar suppliers.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[Ralph2]

Thanks for the Omega link IRstuff.. have contacted them to see if they have any ideas. The EE idea may be the way to go.. I was hoping for some clever ideas from this forum, I have some myself but wanted a fresh look at the problem.

 

[FACS]

We use encoders (resolvers) to monitor speed and position on our precision conveyors and packaging machines. I don't see why you can't simply install an encoder to monitor the speed with a PLC, and have that do anything you want. Seems awfully simple to me so what am I missing?

Charlie

http://www.facsco.com

 

[makesparks]

The output of an approriately sized and mounted photo sensor could signal when the ring flings out. Add to a plc system with an encoder as FACS sugessted, and it should work great to capture the speed at trip.

 

[TheTick]

Accelerometer

 

[FACS]

Yes, a photo eye or proximity sensor to catch the rings... I forgot to mention that... thanks makesparks. Even have it break a fiberoptic/laser

Charlie

http://www.facsco.com

 

[Ralph2]

Thanks all
You are all pointing towards a similar idea that I have been toying with as well. Supposing I place the reflective tape for my pickup in such a position that when the ring trips it is no longer where the pickup can see it. That would be the trip RPM.. but is there a lag? How instantaneous are the readings.. All variables likely dependent on the quality of our tachometer. I will do some experimenting the next time we have a overspeed to service. I have also considered FACS's idea of breaking a fiberoptic/laser but have dismissed it as being too complicated considering all the guarding etc that needs to be in place before we run this thing.

 

[byrdj]

Ralph Your customer that wants something modern wouldn't just happen to have the initial BNP?
I say this because I was a third party witness to verification of such a test a couple years ago. the vibration that occurs when the ring goes eccentric generates enough noise like you say is very easy to hear.

My arguments that the vendors test setup was "accurate" enough was;
what method did the customer use to perform the OST test?
for the official value, an operator used a laser tach and recorded the value when he heard the unit trip. the accelration rate for the test was 1 rpm/sec. the laser tach had an update rate of 1 sec. so ther could +/- 1 rpm update error. the cal shope calibration standard was of 0.05% of full scale range. even saying it was +/-0.05% of range used (say 4000 rpms) that gives a tolarance of +/- 2 rpms. I used the customer laser tach to verify the vendors tach when holding rated speed just prior to the trip. i could have used it to recorded trip speed also, but I chose to witness the vendors display and verify what their tech recorded

next question was, what was there acceptable short term deviation?
there was a blank stare. I like the old ASME PTC20.2 standard of +/- 0.5% which is the range of the multiple OST for this test. so for a 3600 rpm unit that means I would expect all trips to be with +/-18 (excluding the first trip) to say the emergency governor is repeatable.

Now the real problems that result in our test deviating form the actual when back in the unit would be the operating envoroment of an oil spray/mist at about 120 to 140F. the other biggy is establishing the vibration in magnitude and phase angle of the attached turbine rotor.

I talked with the vendor about testing acceleration drift. this would require a high speed data logger for speed since the desired accel rate for the testing would be 10%/sec. they had considered, but the increase in drive HP was more of a headache than they wanted to tackle.


I haven't seen a double emergency governor in a long time

 

[Ralph2]

Thanks byrdj
No the initials are not BNP These are for coal fired generating station(s) in central Alberta. Your process was much more refined than ours. Our acceleration rate is a lot of "as slow as we can adjust our hydraulic drive motor". Another area I have been asked to look at.

Our "system" leaves a lot to be desired and rightfully our customers would like a bit more precision. They too have had repeatability issues when tested in the real world and I have often wondered about the effect of oil and induced vibration.

My job at work is dynamic balancing and the customer recently wanted to know how much unbalance there is in the rotor. Turns out a lot.. which got their engineers all concerned and I suspect generated a lot of emails to England. No resolution on that so far except we do a base line of each rotor now.. just to record what unbalance there is. The whole rotor would need to be redesigned from the ground up to have it balanced in an untripped condition. However, in determining where the unbalance is coming from it is easy to see where pockets of oil could change the unbalance mass and thus change the trip speed.

In my opinion.. we (the company) are trying to set and demand repeatability on antiquated technology. Springs and shims do not lend themselves to that very well. But we do our best to please our customers..

 

[IRstuff]

If you want accuracy, then some sort of relatively high speed, timetagged, video is required. This would be compared against the timetagged tach data will get you your result. 4000 rpm is ~66 Hz. If your speed increases by 1 Hz per second, then with 30 Hz video, you'd get plenty of samples, and rotational accuracy to something like 2%. A conventional optical encoder has thousands of cycles per revolution, so given a sufficiently slow velocity ranp, you can get extremely good accuracy. Even with a ton of averaging, you'd only be off by 1 Hz or so, which is a couple of percent. So, all told, you could get something like 3% rms error.

Theoretically, you could use the sound instead of video, but sound will have lots of interference from noise sources.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

 

[ScottyUK]

Are you - or perhaps is your customer - really interested in the point at which the trip ring deploys, or the time at which the OST oil is dumped to drain? The latter is probably more significant in terms of the machine behaviour, but I appreciate that you possibly don't have that equipment available at the works test. In field tests we would normally rely on the speed probes and the OST oil pressure switches which were scanned at 100ms intervals by the turbine controller. 100ms of uncertainty was trivial in relation to the acceleration of the turbine.

 

[byrdj]

During my visit to the vendor's shop, I witness the test of smaller 3600 RPM EG (about your size) and testing at room temperure and then heating to 140F had a 2 to 3% change in trip speed (I don't recall the direction). the EG I was witnessing was a much larger 1800 rpm and they had no provisions to heat it.

My methodology was I gathered the customers last 10 years of testing and determined what I expected the trip speed to be that year (customer did not perform a down coming OST). I then compared that value with the shop trip speeds before diassembly and cleaning and thus had an expected deviation from actual to shop. this customer only performed the OST twice during startup every two years so there was not enough information to determine the repeatblity

another observation I made during there test prodical was that a +/- 0.001" change in TIR in the setup would change trip speed about 1%. thus I had them try to duplicated the angle of TIR noted during initial shop trip

My customer was upset over the fact that actual trip speed 30 rpms higher than the venders report (and was 3 rpms higher than the predicted trip speed using my observations corrections). I informed my customer they were unreasonable in thier expectations given the design acceptable repeatablity was +/- 0.5% and the in abilty to duplicate actual running conditions. How would you duplicate the #1 bearing vibration?

I heard they approached the vendor to "Modernize" their prodical. I got paid, but no "at a boy"

 

[Ralph2]

Thanks ScottyUK
All we are interested in is setting the speed the rings trip at and how repeatable the process is. We also do one ring at a time, locking the other one as we have found that the tripping of one will invariably trip the other due to vibration. There are LOTS of variables that cause this to be less than 100% accurate but our current method, relying on reflexes.. hopefully can be improved upon.

 

[byrdj]

Do you have vibration trend recorder that you use for run up/coast down data?

If it has a speed channel use it and then a siesmic probe to be the event marker (the nominal vibration going high will be the trip). could you cut a key phasor notch in your drive coupling and use what ever transducer your vibration equipment likes.

what maker uses the dual rings, I think GE went to single ring in the 50's