Dynamic Balancing

It is quite bit complicated and specialized field, and equipment and experience from, say major power generators to airplane propellers is not safely transferable.

First of all you need a shaft encoder of some kind to know the angle of the shaft at any moment. Then you need at least 3 accelerometer (usually more) or strain gauges. The two are not interchangeable, as there is a variable phase lag between them as the rotating shaft is driven thru resonance. Then you take into account the electronic phase shifts occurring in the electronic amplifiers, delays in the microprocessor execution loops. Some of them can be gained only with calibration setups.

As the result, a fairly simple case of the balancing measurement of the propeller of a single engine airplane will cost you up to $2000. If your certified mechanic whittles metal safe and fast, it can be done in less than a day.

But outside theory, anything else is not directly transferable. Amplitudes, phase shifts, calculations and interpretations do differ.

Read these and if you have specific questions let me know. I made a good living balancing turbines for a time...

By the way you don't have to use a DBM, I field balanced equipment all the time, with the rotor in the machine.

http://faculty.uml.edu/pavitabile/22.403/web_downloads/Single_Plane_Balance_091401.PDF

Hi,

we did build our own dynamic balancing measurement setup.

No encoder but a black or white marking on the shaft to measure by reflected light the phase of the rotor, arbitrarily set to zero where the marking is put.

Two accelerometers (most commercial machines use two voice-coil systems) to catch the vibration signals. These accelerometers to be mounted with some suitable axial distance on the to be measured item. If not complete with housing a suitable bearing assembly and drive has to be added. As this is introducing additional errors we measure preferably in completed spindles. Both possibilities (incomplete with auxiliary bearings and drives or complete spindles with drives) have to be suspended in a flexible support that allows operation of the to be balanced rotor well above the first resonance (this support with the rotor). Isotropic flexibility is best and sometimes mandatory but primitive supports (soft rubber foam) sometimes works well.

Bandwidth of the accelerometers has to be considerably higher (factor of 5 or to be corrected for phase and amplitude) than the maximum speed at which unbalance should be measured.

Accelerometer to be fed to a preamplifier of switchable gain to adopt the various vibrations that are often not synchronous to speed: all signals have to pass the amplifier else there are severe errors! The preamplifier and first analog filter has a subsequent (digital) filter to adopt to the rotor speed.

From the switching signal derived from the marking on the shaft - this is assumed to be derived from a shaft of non changing location and constant speed - a synthetic (in the computer or analog circuitry residing) sine and cosine is generated phase locked to the rotor and the marking on the rotor.

This synthetic sine and cosine is multiplied with the amplified signal from the accelerometers. Then integrated.

This is giving as result the zero time-lag cross-correlation of the accelerometer signal with the synthetic sine and cosine thus filtering out the (often very high) disturbing signals that are not synchronous to speed. This filtering is much better than any ordinary bandpass filter.

The magnitude is giving the amount of unbalance.The relation of sine and cosine derived measurements the phase in relation to the rotor marking.

The amount of unbalance has to be calibrated with a known unbalance as the mass and the inertias of the total system is often not known. And we are doing balancing in the assembled spindles so there is the rotor and the housing contributing to the mass and inertias.

The accelerometers have to be located at some axial distance to measure in two planes (perpendicular to the rotor-axis) as unbalance in a rigid rotor is invariably composed of a radial plus a tilting component - this can be thought as being composed of two radial components in the arbitrarily chosen planes. Calculation is by simple vectorial representation of the unbalance in the respective plane.

Adding or removing mass is used to balance according to these measurements. This can be done in the planes of measurement or in any other two planes.

Caution is recommendable if the two planes of measurement or the two planes of mass removal are on one side of the center of mass.

Sometimes pre-balancing at low speed is necessary for a safe transition through the resonances.

Complete systems available on request.

More detailed information available if questions remain.

RHABE