The popularity of these new data gathering boxes is such that one is more likely to find one of these compact devices in the maintenance department of a given plant than to find a fully equipped dual channel spectrum analyzer in the engineering department.
Thus, a strange thing has happened over the years: the Maintenance Department has become better equipped to handle some engineering problems than the Engineering Department. While the predictive maintenance technician may well gloat over his superior capabilities in the area of vibration measurement, he should recognize a higher responsibility to his employer by learning some of the non-P/M applications of his FFT-based box. With some additional training, the P/M technician can assist Engineering in solving some of the engineering problems which may cause his company the loss of future sales or the damaging of their reputation for quality, causing harm to the Company Profit and Loss statement.
This article will discuss some of the areas in which the user of a hand-held FFT monitoring device may be pressed into service to help solve engineering problems for the Quality Control, Manufacturing Methods, Engineering, and the Service departments of a manufacturing plant. Be warned at the outset, however, that a hand-held device is no match for a fully equipped dual channel spectrum analyzer. We are talking about making do with what you have, not about what could be done with the proper equipment.
Quality Control:Demonstrating compliance to a specification
As the popularity of predictive maintenance has grown, it has become more common for the buyer of new machinery to insist that the supplied machinery meet certain vibration and/or noise specifications before accepting delivery. It has long been the practice of governmental agencies, such as the Navy, to require that equipment meet certain sound and vibration specifications. When one supplies critical machinery to a nuclear facility such as a power plant, it is often required that the equipment manufacturer submit proof that his machinery will perform satisfactorily during earthquake conditions.
Since the P/M technician is capable of obtaining a narrow band FFT data on manufactured equipment, he is in a position to help the quality control department verify the ability of the plant's products to meet these various customer specifications.
Customer Sound and Vibration Specifications:
Customer vibration limits are often included in the original purchase order for the equipment. Sometimes, due to an unfamiliarity with vibration analysis, the added cost of verifying compliance to a vibration spec. is not added to the price of the machinery. A vibration test is never thought about until the customer either notices the lack of the test amongst the papers transmitted from the vendor or, worse, when the machine is mounted in place and shaking violently.
The P/M technician is well qualified to deal with the problem of verification of acceptable vibration limits during the operational tests of the unit. All that is necessary is that the machine be rigidly mounted in an area of low background vibration. The test should be run at normal operating conditions, after the machine has reached thermal equilibrium. The P/M data gathering box should the be used to gather FFT vibration data at each bearing of the machine in each of three mutually perpendicular directions.
During the above impromptu shop acceptance tests, care must be made to ensure proper alignment of the motor, proper operating conditions, and adequate mounting. If the test is run using a dynamometer rather than the motor to be shipped with the machine, it is important to determine the forcing frequencies of the Dynamometer. If these frequencies have sufficient amplitude to exceed the vibration limits of the specification, some negotiations with the purchaser will have to be made before shipment. It is always best to be honest before shipment than to face back charges and/or litigation when all remedies must be taken at the job site.
Government Acceptance Tests:
Governmental agencies such as the Navy typically require that machinery meet certain Octave Band sound levels and 1/3 Octave Band vibration limits as spelled out in MIL STD 740. Although it is possible to simulate Octave and 1/3 Octave Band data by taking two or three sets of 400 line FFT data and combining the energy in each of the 400 filters in certain ways via computer, don't bother.
Firstly, the MIL STD requires certain levels of accuracy which your hand-held FFT is unlikely to meet. Further, all of the test equipment used must be calibrated to certain Government standards every six months. As this is quite a costly matter, it is probably less expensive to perform the final tests at the facilities of an outside vendor who specializes in Governmental testing.
All is not lost, however. The Government's Octave and 1/3 Octave Band tests are throw-backs to the days before narrow band FFT analyzers. It is quite difficult to find the cause of a problem in these wide band filtered sets of data. Therefore, the machinery to be tested should undergo a narrow band analysis of its vibration and sound characteristics to avoid the embarrassment and cost of flunking the official test. Note that a flunk will occur before the very eyes of the government DCAS officer who will witness all of your future qualifying tests.
The preliminary vibration tests can be run in the same manner as any predictive maintenance test. The sound tests require replacing the accelerometer with a microphone and power supply designed to fit the particular FFT instrument. Calibration for sound can be approximated by following the instructions of the microphone vendor. By going to the trouble of performing the suggested tests, your company can proceed to the official Government tests with confidence of passing on the first try.
The nuclear regulatory agencies of various countries (including our own) insist that various pieces of critical machinery be capable of withstanding operation during an earthquake. One of the "standard" earthquakes is shown in Figure 1. The verification of this capability can be achieved either through a long series of analytical calculations (whose underlying assumptions can be thrown out at any time by the inspector assigned to review) or via an actual test on a shaker.
A given piece of well designed machinery can usually pass the static g loading of the earthquake. The reason for failure is more likely due to the existence of a natural frequency in the region of frequencies where the earthquake has high energy. This has the effect of amplifying the earthquake at that frequency by a large factor. Thus, a machine without a natural frequency at, say 1.0 Hz. will see a velocity excitation of 100 in/sec. (see Figure 1). A machine with a lightly damped natural frequency at 1.0 Hz. might think that the same test is exciting it at 2,000 in/sec. at 100 Hz. This machine will probably fail.
Although a completely correct test for natural frequencies requires a dual channel spectrum analyzer, an approximation of what may occur on the shaker test stand can often be achieved using a hand held FFT analyzer. An accelerometer is mounted on the machine and the machine is struck with soft objects such as a 2X4 (to elicit a low frequency excitation spectra). A high amplitude response at some frequency, as seen by the FFT device, is an indication of a possible natural frequency. This test is not conclusive, but as stated at the outset of this paper, is better than nothing in terms of locating possible problems before the official shaker test.
Reducing Production Costs:
The labor costs in manufacturing components for large pieces of machinery is directly proportional to the speed at which the parts can be manufactured in existing lathes, boring mills, etc. There have been tremendous advances in tool bit cooling techniques in recent years which have allowed manufacturers to increase the speeds and feeds of many machining operations. At some point, however, the maximum allowable speeds and feeds are limited by tool bit chatter. The hand held FFT data gathering box can be used to investigate these problems, often helping arrive at a solution.
Tool chatter and excessive noise are caused by a phenomena called sticktion. The tool bit, which normally cuts through the metal being machined, adheres to the metal surface due to the force on the tip of the bit. The tool bit is pulled down with the motion of the rotating work piece until the force on the tip is enough to overcome static friction. Since the coefficient of static friction is greater than the coefficient of dynamic friction, the tool bit bounces back up until the force of cutting is sufficient to pull it down again, due to static friction. The resultant motion of the tip of the tool bit is similar to the motion of a spring mass system excited by a series of impacts. The frequency of the chatter is equal to the natural frequency of the tool bit/support system.
When sticktion develops, it is usually necessary to reduce the speed and/or the feed of the cut. This increases machining time and reduces profit.
Two other solutions to the problem exist: One can increase the natural frequency of the tool bit/support system by increasing the stiffness. This would move the chatter frequency to a point where it takes higher speeds and feeds to excite it. An accelerometer of sufficiently small mass to avoid mass loading the tool bit can be used with a hand held FFT box to investigate the problem in a manner similar to that discussed above for finding seismic natural frequencies for nuclear tests. Note that the impact device, in this case, must be hard in order to cause relatively high frequency excitation. These tests will reduce the cost of a trial and error solution.
A second solution to the problem is to ensure that the tool bit always remains in motion (so that it is seeing dynamic friction rather then a static/dynamic/static friction cycle). This can be done by mounting a small exciter on the tool bit/support system such that the tool is always moving at some high frequency. The frequency of excitation must be high enough that the displacement perturbations of the tool bit are small enough to allow for meeting the necessary surface finish specifications. Again, the FFT box can help in the development of the exciter system.
New Product Development:
The development of a new product is a costly venture involving design problems, the building of patterns for castings, the fabricating of parts, and many hours of prototype testing. By changing the input transducer of the hand held FFT data device from accelerometers to microphones to pressure transducers, the P/M technician can find any undesirable oscillatory characteristics of the new product while still in the prototype stage. This allows the engineer to make the necessary changes in design before the final designs and patterns have been finalized and imperfect machines have been shipped. Again, the use of a dual channel spectrum analyzer would be more help in the process than a hand held box, but the hand held box is better than ignoring potential problems altogether.
No matter how perfect a given piece of machinery is when it leaves the manufacturers door, the odds are good that the customer will misapply, misalign, or misuse it, causing high levels of vibration or noise. The finger of guilt invariably points to the machine manufacturer. It is in the best
interest of the machine manufacturer to have the ability to examine the operation of the machine under actual on site conditions to determine whether the problem is the fault of the customer or the vendor. This ability insures that the liability falls where it should. It is much easier to assign fault to the "other guy" if correct instrumentation is used to ascertain the facts. Often, the P/M technician has the only piece of instrumentation in the company able to do this. An actual example of a simple problem which could have cost the vendor tens of thousands of dollars was resolved with a simple filtered vibration measurement device.
A new power plant had two large pump packages mounted on a mezzanine as shown in Figure 2. Unit #2 ran well, but unit # 1 shook the entire mezzanine. The pump manufacturer was being blamed. The threat of multiple lawsuits filled the air. The power company was refusing to accept delivery of the plant until the problem was resolved. The engineering company and construction company were ready to sue each other as well as the pump manufacturer.
A set of tests were run by the author using a simple tunable filter vibration meter (Some of up are old enough to pre-date FFT equipment). Both pumps were shut down. A pneumatic impact hammer was fastened to the mezzanine near pump # 2. The speed of the hammer was slowly adjusted until everyone present agreed that their feet tickled most - tunable filter analyzers are too slow to run "real time" data as can the modern FFT data gathering box. Vibration data taken on the impact hammer showed that the predominant frequency excited by the hammer was equal to the blade frequency of the pump.
The conclusion was simple: The design of the mezzanine was such that it had a natural frequency equal to the blade frequency of the pumps. Pump # 2 was located near a node of the floor, making it almost impossible to excite the structure at that frequency from that location. Pump #1 was at an anti-node, making excitation at that location very simple. The structural design engineer was at fault. The pump manufacturer, through the use of simple filtered vibration equipment, was found to be innocent.
An FFT based vibration monitoring device, in the hands of a properly trained technician, can do far more for the good of the company than simply performing predictive maintenance functions. It can be used by quality control for compliance testing, to reduce manufacturing costs, in new product design, and in dealing with field service problems. The well trained P/M technician, then, has the ability to widen his horizons as far as is permissible by his employer.