Rotating Machinery Predictive Maintenance Testing

Mechanical Solutions, Inc. can provide testing or technology to meet your predictive maintenance needs. Avoid unplanned downtime by catching system failures before they happen. No one can afford unplanned downtime.

Development of "Active" Electric Current Monitoring for Plant Machinery Maintenance

The Maintenance Problem Being Addressed: Condition monitoring has made good progress in recent years in identifying many types of deterioration in plant machinery, so that pro-active maintenance can be performed, improving overall plant productivity. However, predictive maintenance is not yet a mature technology. Existing condition monitoring methods are not sensitive enough to certain lubrication, wear, and fatigue problems in their early stages, when remedial action would have the most benefit in terms of the avoidance of unplanned downtime, the elimination of parts which are out-of-tolerance due to tool degradation, and the limitation of repair costs to the fewest number of machine components. This presents an opportunity for the development and application of a new technology which is superior to existing methods in the early detection and identification of certain common and costly problems. Possible fruitful areas for such developments are the monitoring of vibration, oil, temperature, and electric current.

The Novel Approach Being Suggested: The proposed project is a new means (called "active", as explained below) of monitoring the electric current of electric motors in order to detect problems in the motors, and more importantly in the tools or equipment being powered by these motors. The method is very different from existing methods of current monitoring because it does not rely on the current naturally produced by the motor. Instead, the concept is to use an external "clamp-on" to induce a very small "blip" in a machine’s electrical system, but while it is operating, so that a second "clamp-on" can then detect the system’s very small operational reaction. This information is transmitted back to a "walk-around" PC or a central computer, for either manual or automated interpretation. During interpretation, the "bounce-back" signal is evaluated like radar, to determine whether all the components in a process "power train" affected by the "blip" (including lubricant films, tool/ work interface cooling, bearings, supports, and driven tools) are in good working order.

The new method is patterned after research being performed by Mechanical Solutions, Inc. (MSI) on behalf of the electrical power industry, where it is being applied to nuclear reactor coolant and safety pumps. For either the power plant or the plant machinery application, the proposed system’s accurate "pulse-echo" method of monitoring power train integrity and performance would find problems early, while avoiding false alarms. In addition, it would allow a plant to track and fix efficiency problems in specific machinery as a function of operating condition and time, which could potentially save large sums of money each year in electrical operating costs. The system, which requires no hard-wiring or expensive sensors, would be inexpensive enough to be installed on most machinery, and would have an option for wireless information transmission that would minimize optional permanent installation costs.

Background:

Existing off-the-shelf condition monitoring and predictive maintenance methods are not reliable enough in indicating certain important problems in manufacturing equipment, such as cracking of power transmission components or their structural support members, inadequacy of local lubrication, and excessive cutting edge wear of a tool or die. These situations lead to unnecessary scrap that could be avoided by a timely component repair, or a change of lubricants, bearings, or tools. In addition, the occasional unsuspected catastrophic failure of plant machinery can result in costly repairs and down-time.

There has been considerable interest over the past ten years concerning the use of electrical current monitoring in order to detect problems with electric motors and the equipment that they supply shaft horsepower to. Existing current monitoring techniques have proven useful in detecting certain types of problems, and are much more sensitive than other measurement parameters (such as vibration, bearing temperature, or lubricating oil condition) to certain common types of electric motor faults. Broken rotor bars, phase-to-phase shorts, and degraded electrical wiring or contacts (such as brush interfaces) are detected earlier and with greater reliability by observing their effects in the motor current versus frequency spectral "signature", as opposed to monitoring more commonly observed on-line parameters or steady state RMS current, resistance, and voltage. In particular, the work of Don Casada, John Kueck, and co-workers Kryter and Haynes (Ref. 1), Smith and Castleberry (Ref. 2), and Linehan, Bunch & Lyster (Ref. 3) at Oak Ridge National Laboratories is notable for its pioneering investigations, and its progress in this area of research.

Dynamic current monitoring in real-time can provide a significant improvement in electric motor fault detection. It can also provide information on changes in the "current draw" because of anomalies in other components in the power train. Certain potentially critical faults, however, require more information than monitoring of operating current frequency content is able to provide, when such monitoring is done on a purely "passive" basis. For example, shaft or spindle cracks, marginal lubrication, incipient bearing degradation, mechanical looseness, imbalance, and misalignment are not detected well by passive monitoring of electrical current. To address this issue, MSI is developing an approach whereby the electric motor system would be "actively queried" as well as "passively monitored" in order to broaden the range of problems detectable by current monitoring, and to provide a cross-reference that would make false alarms less likely. By combining the proposed active monitoring approach with passive current monitoring, a Condition-Based Maintenance system for manufacturing machinery would be able to seek out and detect a wide variety of problems within electric motors themselves, as well as within the associated mechanical components of an electrically driven tool or process in a manufacturing plant.

Discussion of Past Work:

Some research organizations have been investigating passively monitoring motor electrical circuits, through observing the value of the electrical current over a given frequency range, in order to detect operating characteristics and anomalies that could be useful in evaluating the efficiency of the motor and/ or the machine that it drives or operates, and to perform diagnostics and prognostics on such systems. This passive monitoring method is explained in various articles and technical papers that researchers at Oak Ridge National Laboratories have written on this subject (e.g. references 1 through 3). It’s approach and intent is analogous to vibration signature monitoring, which is illustrated in Figure 1.

An "active" monitoring method, in which some kind of a known signal is sent into the system of interest, to see whether or not it reacts in a healthy manner, can substantially increase the capability of such an approach. One type of "active" technique is a frequency response function "impulse" or "modal" test, which is familiar to vibration analysts. Typical results of such a vibration test are shown in Figure 2, with particular frequencies of interest (the "poles" and "zeros") pointed out.

The approach proposed in the research proposed here is an important step forward in the state-of-the-art in that it involves active current testing while the motor is on-line, i.e. actually in-service and operating. The basic concept is a form of the experimental modal analysis (EMA) described in the preceding paragraph, which in that case was applied during vibration testing to determine mechanical natural frequencies. The procedure follows the techniques previously published by the author and others (e.g. references 4 through 6) for such mechanical systems, in which a mechanical means is used for the "active" excitation. EMA is used by the author’s company and other organizations in order to determine the so-called poles and zeroes of a structural system. The poles represent the mechanical natural frequencies (e.g. "critical speeds" in a shaft), and the zeros reflect locations where the vibration cancels to zero, due the particular vibrating motion flexural "mode shape", as illustrated in Figure 2.

In the case of passively monitored amplitude versus frequency, current amplitude rather than vibration can be plotted versus the frequency. Depending upon the data reduction technique chosen, changes in the carrier frequency (i.e. frequency modulation) can be separately tracked, as a plot of peak values at modulation frequencies versus frequency, or as current versus frequency where the plot has been "demodulated" to remove from the spectrum the domination of the peak at the carrier frequency (i.e. line frequency). This is the essence of the Oak Ridge technique, and is illustrated in Figure 3. Figure 4 shows how, similar to the mechanical vibration case, a frequency response function (FRF) of the current can be created and evaluated, excited by a current impulse instead of a mechanical one, as discussed in Reference 7.

Trending the poles and zeros alone suggests whether or not parts are still tightly connected, whether cracks are developing in loaded components, and whether friction has increased due to lubrication break-down or excessive heat at the tool/ workpiece interface. Furthermore, the coupled mechanical/ electromagnetic system natural frequency shifts could be used to deduce flaws in the motor wiring or coils, and in the driven systems in the "power train", including spindle mechanical assembly, or the tooling and tool/ work interface itself.

In addition to trending electro-magnetic and mechanical poles and zeros, amplification factor at the poles can be observed, and from this the system damping and energy absorption of the driving element (motor) and driven element (such as a stamping die) is able to be determined. In addition, from such measurements, the machinery system electrical efficiencies can be determined, and steps can be taken to optimize the efficiency and overall operational costs of the driven equipment. In fact, direct variations in torque in given frequency band can be calculated from the combined measurement of passive current and the impulse excited pole/ zero FRF. An example of how the torque is the product of the current and the FRF, on a frequency-by-frequency basis, is shown in Figure 5.

Correlations of torque at specific frequencies that occur in the machining process would be directly related to the cutting or stamping forces required, and would therefore be a function of tool wear and lubrication adequacy. By deducing the torque instead of merely measuring passive electrical current peaks, machine-to-machine differences and workpiece dynamics could be properly taken into account.

Anticipated Problems and Solutions, in Practical Application:

Unfortunately, it is very difficult to do sufficiently accurate pole/ zero tracking (either mechanical or electrical) on equipment while it is operating. At first glance, in order to clearly identify poles and zeros in competition with the naturally occurring operational signature of the tested system, the actively applied exciting signal must generally be applied so strongly that the exciter must be impracticably large, and the excitation force might become large enough to itself damage the system being observed, or at least to trip the unit. Therefore, the MSI TAP™ method is used, as discussed in MSI references 4 through 7. The MSI method is very efficient at eliminating unwanted naturally occurring signals, and producing a clear frequency response plot in spite of a small level of excitation force. The resulting information is clear and able to be correlated with machinery condition with respect to a wide variety of potential problems, and with respect to optimization of process performance and efficiency.

Closure:

The novel current condition monitoring approach as described could be applied to any type of electric motor, including but not limited to DC brush and AC induction motors of both large and small sizes. Such motors are used, for example, to drive machine tools, stamping or forging presses, air conditioning compressors, ventilation fans, and motor operated valves, various types of pumps, and even plant elevators. Motors driven by variable frequency drives (VFD’s) would also be able to take advantage of the proposed technique.

Information from the proposed method could provide early indication of power train cracking or tribological/lubrication problems, as well as indicating a variety of electrical problems in the motor. "Success" of the new system initially would be indicated by an increase in the justified alarms, combined with fewer false alarms than existing systems. Such success would result in electrical motors and driven machinery systems that would be of greater reliability, lower maintenance cost, and less expensive to operate than the systems currently in existence today. This would lead to the final goal of decreased production costs, and better control over production schedules.

References:

1. Kryter, R.C., and Haynes, H.D., Condition Monitoring of Machinery Using Motor Current Signature Analysis, Sound & Vibration Magazine, Sept. 1989, p. 14-21

2. Smith, S.F., and Castleberry, K.N., Advanced Techniques in Current Signature Analysis, Proc. 46th Meeting of the Mechanical Failure Prevention Technology Society of the Vibration Institute, April 1992

3. Linehan, D.J., Bunch, S.L., and Lyster, C.T., Method for Detecting Periodic Abnormalities on a Periodic Carrier Wave, Invention Disclosure under Dept. of Energy contract DE-AC05-84OR21400

4. Marscher, W.D., Determination of Rotor Critical Speeds During Operation through Use of Modal Analysis, Proc ASME 1986 WAM Symposium on Troubleshooting Methods and Technology, Anaheim Cal, Dec 1986

5. Marscher, W.D., How to Use Impact Testing to Solve Pump Vibration Problems, Proc. EPRI Power Plant Pumps Symposium, Tampa, June 1991

6. Marscher, W.D., and Jen, C.-W., Using Impact Testing on Live Machinery to Perform Accurate Diagnostics, Proc. 51st Meeting of the Machinery Failure Prevention Technology Society of the Vibration Institute, April 1997

7. Marscher, W.D., A New Method for Detecting Shaft Cracks and Tribological Problems in Electrical Rotating Machinery, Proc. 52nd Meeting of the Machinery Failure Prevention Technology Society of the Vibration Institute, March 1998