# Why do we need to measure the vibration of rotating machinery (e.g., acceleration, velocity, and angular velocity) in three axis?

Why do we need to measure the vibration of rotating machinery (e.g., acceleration, velocity, and angular velocity) on all three-axis? Why aren't single-axis vibration signals (e.g., only X or only Y, or only Z) enough?

Is there any way to integrate/combine the 3-axis signals into a representative single-axis measures/metric? For example, combining the acceleration signals on X, Y, and Z into one representative signal.

• I won't write an answer because I'm an amature and the engineers here can probably come up with better or other reasons, but one obvious big one is that the direction of vibration is relevant to engineering and construction of anything that must withstand it. Looking at any one axis you also can't know the vibration's total magnitude either. If you only measure vertical, you could completely miss vibration in the horizontal. Rotating machines themselves vibrate more radially than axially and radial vibration can reveal imbalances in the core or magnetic field caused by faults or design.
– K H
Jan 29, 2021 at 0:12
• So the axial axis of the motor would be easiest to disregard, but there is still a possibility it would be relevant, especially if the motor is attached to something. Bearings must withstand normal radial forces plus vibration, so excessive vibration will cause bearing wear or the necessity of upgrade, and excessive vibration on the axial axis may necessitate different or extra bearings, or in the case of an already designed system could be an indicator of reason for premature wear.
– K H
Jan 29, 2021 at 0:22
• There is no universal best way to combine 3 phase into 1 as the applications are too diverse. There are many ways to analyze the results from phase, FFT harmonic multiple, peak vector direction, rectified envelope analysis, selective filtering etc. The question is too naive.and lacks focus Jan 29, 2021 at 0:23
• When you need to combine forces on different axes, you use vector addition, which means you use 3D trigonometry to connect the vectors tip to tail and use trig to calculate the resulting angle and length. Note that the measurements must be synchronized for this to work and if you're adding formulas this way instead of raw numbers, the formulas must take synchronization into account or you won't get the true vector.
– K H
Jan 29, 2021 at 0:24
• @TonyStewartSunnyskyguyEE75 could be a tradesman trying to edify himself. Sometimes we ask questions above our skill level to get an idea of the justification for "black box" orders. Could be one of those things where you send an electrician or EIT to take the measurements and you only call in the appropriate engineer if measurements come back too bad.
– K H
Jan 29, 2021 at 0:29

I am wondering Why we need to measure the vibration of rotating machinery (e.g., acceleration, velocity, and angular velocity) on the three-axis? why single-axis vibration signals (e.g., only X or only Y, or only Z) are not enough?

To research what the machinery does initially, you need to collect all the information you can. You don't know what it's going to be doing, maybe the key signal you want is going to be on the X, Y or Z axis, so look at them all.

Once you've characterised what the machinery does, you might find that just two or even one axes are sufficient to extract the information you want.

Also, please, is there any way to integrate/combine the 3-axis signals into a representative single-axis measures/metric. For example, combining the acceleration signals on X, Y, and Z into one representative signal!

There are several methods of reducing the number of axes you have available. For instance, the rms of the three vectors would give you something like the total energy in the vibration. However, this might not be what you want, if the machinery has a normally operating strong X axis vibration, and the fault-indication vibration is a tiny Y axis acceleration. You would need to try out several methods in your initial research phase, and see which one worked best for your particular device.

By knowing the geometry of a design and the rotating part sizes, one can estimate the fundamental frequencies of each part.

Historically, a Spectrum Analyzer by sweep or FFT tells you the dominant frequencies which are more likely to be harmonics with some mixing of frequencies from non-linear surfaces.

I suggest you use a microphone or accelerometer and record on Audacity then use the FFT function to obtain the best orientation for select different ball bearing frequencies, Axial or radial

Look for attenuating harmonics to indicate a bump on a smooth round surface and a strong fundamental to indicate an unbalanced rotating mass. the rotating phase and amplitude tells you for 2 axes the amount of balance error, from which you can create a process by trial and error to balance the rotating mass.

A Drop into a soft foam can be captured on a scope to capture -1g for calibration rough checks. The ratio of stop/drop=k height with an elastic surface will respond with +k value for g's.