Using Detection Technology to Enhance a Predictive Maintenance Program, Part I

By Robert Martin posted 02-13-2019 11:57 AM


We live in challenging times. This is particularly true in the predictive maintenance world, and acutely true in the area of vibration analysis. Today, most manufacturing sites have active reliability programs in place with the means of measuring plant performance and processes to find and correct repetitive problems that effect plant uptime. While this is a positive development, it is also true that vibration analysis programs, which help provide the data necessary to achieve these results, find themselves understaffed as a result of attrition and retirements. Much to the surprise of plant management, these positions are not easily filled and yet the routes need to be done, the data from these routes needs to be analyzed, and the report needs to be written.  The result is that reliability engineers are faced with the question: “How do I get the data?”

One popular answer is to contract the vibration program out to a third party. While this is an attractive option, the expense can be an issue, as well as the fact that these service companies face the same problems as the rest of us. Good qualified people are hard to find. Staffing changes are common and quality varies even with the most well-meaning service companies. An important fact to remember is that it takes around two years of experience and training to become a proficient vibration analyst.

A second option is to try to downscale the vibration program to meet your capabilities. This is much harder than it sounds. First, some assets are shifted from monthly to quarterly data collection. When this is done, these assets are really not in the predictive maintenance program any longer and are now in a reactive maintenance program that uses data collectors. This may sound harsh, but consider the fact that you are assuming that these assets will show signs of impending failure three to six months before action needs to be taken! A second option is to simply remove some assets from the program all together. This is the most uncomfortable option of all for the reliability manager. He or she now has the burden of knowing that someday one of these orphaned assets will come back to haunt them with an unexpected breakdown, usually at the most inopportune time. It’s not question of ‘if’, it’s a question of ‘when.’

Detection Technology

A third option that has gained popularity recently, is to employ detection technology to these assets as a way of monitoring their condition while eliminating the need to run frequent data collection routes on them. This does not necessarily mean that these assets are always removed from the vibration program. They may be sampled less frequently because you have them being monitored automatically with detection technology.

Detection technology comes in all sorts of configurations from wireless to electromechanical devices, but is best defined as sensors that can detect a change in vibration from an asset in a way that can enable prompt data collection and analysis when the condition of the asset indicates the need to do so. In practice, these sensors can be used to help eliminate assets from periodic route data collection and assign them to a pool of equipment that is visited when a change in vibration level indicates potential trouble. Simply put, when a change is detected, it’s time to check out that particular asset. 

This approach is similar to how we use our doctors and hospitals. We monitor our condition either subconsciously or consciously all of the time, not on a scheduled basis. Besides our yearly physical examination, we tend to wait until we detect something is wrong with us to visit our doctor to get examined. The doctor then performs tests and prescribes a treatment or sends us to a specialist for further evaluation. We have all been there and know the process. The important point here is that we don’t visit the doctor once a month, unless this is prescribed or the symptoms are not resolved. If the problem is significant, we are sent to the hospital for further treatment or repair. As your personal operating hours increase, these visits tend to become more frequent.

Types of Detection Technology

As mentioned above, detection technology comes in all sorts of configurations. Each of these have their own particular benefits and deficiencies. As mentioned, these range from vibration switches to wireless sensors. The most basic of methods employs the humble vibrations switch.

The Electromechanical Vibration Switch

The vibration switch has been around for a long time. They come in two basic styles, electromechanical and electronic. The electromechanical type employs an adjustable magnet that is attracted to a metal target. To set the unit you adjust the magnets exposure to the metal target in such a way that the two will become disengaged when vibrated past a particular level. This disengagement will mechanically open or close a set of electrical contacts. Originally, the electromechanical vibration switch would be wired to interrupt power to the monitored device once the contact had tripped due to a vibration level being exceeded. Obviously, this is more of a safety system rather than a detection technology.

Because of advancements in technology, these devices are of limited value anymore. Their main purpose is to prevent a failure from getting worse. The only prediction involved is that you will predictably have to visit the machine when the machine stops operating because of this switch operating.


Electromechanical Vibration Switch example

The Electronic Vibration Switch

The electronic version of the vibration switch uses either an internally or externally mounted accelerometer to produce the vibration signal that is sensed by the on-board electronics. The externally mounted accelerometer is preferred. Electronic vibration switches can be set to alarm in acceleration, velocity, or displacement. These can also be configured for two alarm levels as well as providing a 4-20Ma signal to the controls system.  As with the electromechanical vibration switch, when an alert or fault vibration level is reached, a set of contacts will change state. The important feature with these switches is the 4-20Ma output that is available and the addition of an external accelerometer for vibration detection. 


Electronic Vibration Switch 

There are several disadvantages in the electromechanical vibration switch that are eliminated by the modern electronic vibration switch alternative. First, by having the ability to use an external accelerometer to gather the signal, the necessity of mounting the entire device to the machine is eliminated. This makes for a more practical installation and a far better signal source. Secondly, the electronic switch can alarm in units that make sense for the application rather than just displacement. Finally, the optional 4-20Ma output gives a real time signal that can be monitored remotely giving the ability to trend and shut down within the same device.

Smart Sensor Vibration electronic vibration switch

A more recent development is the electronic vibration switch which incorporates all of the electronics into the accelerometer body creating the first generation of smart switches. The sensor is a programmable device that will open its two wire circuit when the programmed vibration level is exceeded. The program is modified within the accelerometer via a cable that connects to the sensor and a standard computer with a USB port. This sensor can be used to shut down the monitored device or to send a signal to the controller indicating a problem. As an added feature, the alarm level can be set with an external magnet.

Image_5.pngThe smart vibration switch 

The smart vibration switch's programming accessories

The new generation of 4 – 20 Ma smart sensors 

The most significant disadvantage of a simple 4-20Ma vibration sensor is that it is providing data with a particular amplitude unit and frequency range. Many times this is configured by the manufacturer. The challenge is whether this is the correct configuration for the machine being monitored.

Recently, I visited a customer who had just had a major gearbox failure. The output shaft bearing had failed causing the rolling elements to travel into the gear set causing the expected result. Although the gearbox had several standard 4-20 Ma sensors monitoring it, this disaster occurred without warning. The sensors were simply not looking in the frequency region where this problem could be detected and were looking in the wrong amplitude units. I am sure that the sensors were reading Hi levels of vibration once the gear set came apart, but then again, anyone standing nearby could have probably detected the failure as well.  The problem was that High frequency acceleration was needed, but low frequency velocity was used.

Smart sensors now have the ability to be programed to the correct amplitude units and proper frequency range. This can now provide a continuous stream of useful data that can detect a significant change in machine behavior. As is the case with the smart switch above, the accelerometer housing holds the CPU and the programming needed to have this level of flexibility.



Smart 4-20Ma sensor 

Smart 4-20Ma programming screen. A special cable connects the sensor to the computer

Second Generation Smart 4-20Ma sensors

While still keeping with the concept of a sensor with it’s CPU enclosed within the accelerometer case, the next generation of smart sensor provides for much greater flexibility in amplitude units. Because of these improvements, these sensors can be considered bearing fault detectors rather than simply vibration monitors. The end result is that the sensor is now tailored to the application focusing on the potential failures rather than overall vibration.

These sensors can monitor the vibration in one of five amplitude units. The sensor is programmed as described above via a special cable to a computer.

  • RMS Acceleration
  • True Peak Acceleration
  • Peak Acceleration with Correction
  • Crest Factor
  • Crest Factor Plus


 Image_5.png Smart bearing fault detector


 Smart bearing fault detector programming screen

Using external electronics with a standard accelerometer

Sometimes due to environmental concerns, a standard style accelerometer needs to be used. When this is the case, external electronics can provide similar flexibility for your 4-20Ma detection technology. In this case a standard sensor is powered and its data manipulated by a din rail mounted electronics package mounted in a standard enclosure, providing two 4-20 Ma outputs. One output is selectable for acceleration, velocity, or displacement while the second output provides a 4-20Ma signal in True Peak Acceleration. These selections are done thru dip switches inside of the electronics module. An added bonus of this design is that raw vibration is available via either a BNC connector or through screw terminal. The advantage is that you can view the vibration from the sensor in real time with a data collector or oscilloscope.


Bearing fault detector module wires to a standard accelerometer. Configuration done within module with dip switches.


Signal processing done in the external module. Notice how output signal (3), True Peak, is basically digital demodulation.

Wireless Vibration Monitoring

Wireless vibration monitoring has been a hot topic for several years now. Everyone seems to have an offering that does something special. For those who are not electronics engineers, RF engineers, or computer science engineers the question should be asked “What do I need this system to do?” and “What makes wireless vibration monitoring better than other methods?”  

The first misconception is that these systems offer continuous monitoring.  Few of these systems provide continuous streams of data in any way similar to what a wired on-line system would provide.     In practice these systems are limited by battery life and band-width. At their best these are periodic data collection systems that provide value in that they are collecting data from your asset four or more times a day. This is significantly more than you once a month program.

A second misconception in that these systems can work well on any asset. The best asset choice for a system that is collecting data four times a day, without consideration of machine state, is an asset that is running during all of these data collections. A perfect candidate would be an asset that is running in a steady state for long periods of time. This would rule out cyclic equipment although, much to my amazement, I have seen these systems applied to machine tool spindles on machines that are always changing state. 

The chief advantage of these systems is that little wiring is required. Another advantage is that most of these systems send their data in an easy to read format directly to your laptop, but this is where the real challenges come up. “What kind of data do I need from this system and how do I want to get this data?”

This question brings up another challenge with wireless systems. Most of these systems want to send data across your plant network. For some companies this requirement begins a journey to the most frightful of destinations, the IT department. With the grave threats that are out there to our IT security, getting a monitoring system, with its data, onto your network is a challenge. The good news is that it can be done. If you take the time to explain what the system truly does and that it in no way will affect system operation, most IT people have no problem with this data on their network.

Above everything else, the main challenge with wireless vibration systems is that they typically use batteries for power. All batteries have one thing in common, the more energy (work) that you do, the faster the battery runs out of energy. Battery life is always an issue and changing batteries is always a nuisance. This is one of the main reasons that these systems are set up to collect data only a few times a day. If you want more data, buy more batteries. 

A final challenge for battery powered wireless transmitters is that they have an exclusive connection to the accelerometers they use to gather data. This means that you cannot use these accelerometers for anything else. A recent development that helps resolve this issue is the advent of the wireless switch box. This device acts as a conventional switch box for route data collection while also transmitting data to the wireless receiver on a periodic basis. Of course, this system is not battery powered and must have DC power provided.  

Wireless Vibration Data
As we have seen already, there are a wide variety of data types available. The challenge is to match the data available to the capability of the one looking at the data to understand it and react to it in an efficient way. In most cases this is the reliability engineer. Most of us know, from experience, that showing spectral and waveform data to the untrained is seldom productive. For this reason, trendable values with alarm values are the most useful.

 Even as a trained and experienced vibration analyst I would ask you again “What do I need this system to do?” The answers may be:

  1. Tell me when an asset has changed state in a negative way.
  2. Tell me how severe the problem is.
  3. Tell me how this problem has developed.
  4. Tell me what type of problem it is.

Over the years I have asked the same question to experienced and inexperienced vibration analysis staff. “What is the first thing that you would do if this system showed that an asset has gone into alarm and was trending upward?” The answer has always been “I’d get my analyzer out and go check out the machine”, or words to that effect. I have never gotten the answer that they would check out the last spectra and waveform taken or that they would request spectral data be taken by the system. That’s just not how people think. Basically they want to go out and kick the tires.

Wireless vibration collection systems can be very useful if applied correctly. They are easy to install and will provide good data quickly. These systems collect data in places where we don’t want to go at times we don’t want to work to do it, without complaint. When used as problem detectors with sophisticated trend and alarm values, these systems can provide valuable information far more frequently than any route based program. The thing to remember is that you will probably have to visit the machine at some point to finalize the diagnosis.