What areas are you guys using triaxial acclelerometers? I'm in the chemical industry (pumps, holding tanks, centrifuge, small to medium fans etc.) and looking for pros and cons for its usage.

Thanks in advance!

Scott,

Are you talking about using them with portable data acquisitionor with on-line systems?

Skip Hartman

Good question... I'd like to know both...

I figured for sure on an "on-line" system...

Not so sure I've heard of anyone using them on standard route data collection.

Scott,

I can't rally see a big advantage to using triaxial for normal route data collection if you do not have a data collector that is designed to take advantage of the triaxial sensor. The DLI data collection scheme, for example, is designed around this method.

There have been lots of discussions about whether or not you get better data using a triaxial sensor than with proper placement of a single axis accelerometer. On that, I come down squarely "in the middle" of the issue. That is, I know from experience you can have an excellent, equally successful program both ways. I think that, if you could place a single axis accelerometer perfectly in all applications, you would have superior data, but it is often the case where at least one of the three axes of measurement is compromised by the inability to find a suitable mounting for a magnet mounted transducer. Therefore a single good mounting location for a triaxial sensor, that is mounted using a through-bolt to a properly prepared mounting pad, will often yield more consistent and repeatable data that is better for trending.

It must also be remembered that the triaxial should not be used in a Pdm program with magnetic mounting and that the maintenance of the triaxial mounting pads will require some effort.

For on-line systems it is usually more a matter of economics. The cost of the system is largely determined by "channel count". The higher the number of channels (i.e. the more sensors there are) the higher the cost of the system. So, if money is no object (not often the case), use triaxial sensors and you will be able to do a more thorough job of diagnosing the machine's problem from the on-line system's automatically acquired data. More often a compromise must be made to prevent the channel counts from getting unreasonable. This compromise is often found in a routine of using a single radial sensor per bearing and one axial per shaft. Many people try to locate the single radial sensor at an angle of 45 degrees from the vertical (on horizontal shaft applications) in order that the sensor will detect some changes in energy that may be directional, to either the vertical axis or the horizontal axis.

Sometimes the best location of a single sensor per bearing is determined by application. For rolls in web process machines, for example, where the only real need is to detect bearing problems, the preferred axis of measurement is the axial direction. This is because there is a good body of evidence that indicates that this is the direction of measurement which usually gives the earliest indication of a bearing defect for these (usually spherical roller type) bearings. If you were monitoring fans where process material builds up on the blades and creates an imbalance the radial direction would be the preferred single sensor location.

Sometimes those who are accustomed to triaxial data in their portable route collection are concerned about not having triaxial when they install an on-line system. Actually, the on-line system usually performs very well at detecting defects simply because it is usually working with better, more repeatable data, because the sensor is permanently and rigidly attached and because the on-line system acquires so much more data. Trends are much more well-defined with dozens of data points per day compared to one per month. The variability of the vibration data due to process fluctuations is often much better understood in a system has had an on-line system installed for a few days than on a system that has been monitored monthly with portable data collectors for years.

I hope this helps.

Skip Hartman

I agree with Skip that triaxial measurments are typically not required for online monitoring. If one was interested in triaxial readings in an online system it is usually more cost effective to purchase 3 separate single axis sensors and either mount them to a "block" in one machine position, or directly on the machine in each axis. Our recommendation for machines such as pumps, blowers etc. Is 1 radial point per bearing (unless the distance between the bearings is less than about 30") and 1 axial per component (i.e. one on the motor and one on the pump). Again it comes down to the economics though.
Regarding stud mounts vs. magnets, besides repeatability issues, magnets are only good up to about 3 kHz (on a clean flat unpainted surface) whereas stud mounts are good up to about 5-7 kHz. So consider what you need to measure and then choose.
One more benefit of triaxial sensors (assuming the use of mounting pads) is data collection speed. If you have a 4 channel analyzer, you can get all axis simultaneously with one button press instead of: move sensor, take test, move sensor take test, move sensor, take test.

The top 3 documents on this page discuss sensor mounting locations as well as single vs. triaxial issues:

http://www.dliengineering.com/section.asp?nID=16

"Triaxial Vibration Spectral Data - An Important Ingredient for Proper Diagnosis" can be downloaded from this page:

http://www.dliengineering.com/section.asp?nID=6

Hope this helps.

Thank you Skip and Alan...

This is good info for sure.

Skip is right on the money with the NO MAGNETS for triaxial advice. A magnetic mount in the shear direction will not provide reliable results. When using a magnet with a single-axis accelerometer, it should be attached to a smooth flat paint-free surface. With a good mounting surface, I believe magnets do a better job than they are given credit for. When used an a rough surface with a heavy layer of paint....the response is extremely bad and unrepeatable.

For permanent installations, the use of tri-axial acceleromters boils down to economics. A single sensor and cable are easier to install than 3 sensors, but the cost of the hardware is far more.

CTC's new low-cost triaxial starts at less than $600. This compares to their low-cost single axis accelerometer at $69, or $207 for 3. A cost difference of $400 (or more if less than 3 directions are required) may be difficult to justify.

For routine measurements with a portable data collector, I'm entirely in favour of tri-axial measurements. But for continuous monitoring, bear in the the problems likely to come up. A large airhandler fan can be well protected with a radial and axial accelerometer on one end and a radial accelerometer on the other end. Three accelerometers cover all the problems you're likely to see on a unit of this type (plus sensors on the motor of course).

Jon
Spintelligent Labs

I have never used tri-axials but the problem I'd have with them is, as Skip points out, the best data is usually not taken at a single location. I like to get my accel as close to the bearing as possible and I always take a horizontal and vertical measurement unless I am physically prevented from doing so. I have found that the machine itself determines whether or not a defect shows up best in the horizontal or vertical direction -- not the type defect. So I don't agree that a single radial point per bearing is sufficient.