In today’s fast-paced competitive business world, manufacturers are seeking every competitive advantage they possibly can to increase their production and minimize costs while maintaining product quality. The identification of defects within a machine, reducing equipment failures and unscheduled downtime is increasingly demanded of condition monitoring technologies.
Vibration analysis has proven to be one of the most effective tools for identifying mechanical and electrical faults within machinery. Most vibration programs use a combination of on-line monitoring and offline (walk around) monitoring. Off line programs require the resources of a trained technician to walk from machine to machine to collect the vibration data.
The primary goal of vibration analysis is to identify faults within a machine and then alert personnel that some type of action needs to occur. Problems start to occur when the needed frequency of the data collection is not aligned with the maintenance strategy. A machine’s criticality, its risk priority and its failure modes establish the frequency required; however, far too often frequency is determined by the availability of a local contractor, internal staffing or, even worse, on how much money will be saved if the frequency is changed from monthly to quarterly?

Today’s on-line systems have the ability to provide continuous monitoring and can send alarm notifications which can be incorporated into a site’s process control system so operators are alerted of a problem.  Some systems can be configured to distribute emails or even send text messages to specific individuals based on an alarm state.
Most vibration analysis systems today also have the ability to monitor and alarm off-process data such as temperature, pressure, voltage, current, flow or speed and can provide alarming if a process measurement goes outside of a predefined range.
Some of today’s on-line systems can incorporate on-board logic and decision making and some vendors offer machine diagnostics so that data is analyzed and screened for alarm violations automatically.  Data storage can be accomplished by the end user locally or the data can be stored and accessed via the cloud.  Utilizing a cloud server allows Reliability Engineers, Vibration Analysts or Condition Monitoring Contractors the ability to analyze and view data, alarms, trends and reports from anywhere in the world.
The “Industrial Internet of Things” (IIOT) is changing the way vibration data is viewed and managed.  Developments in Artificial Intelligence, Smart Machines, Embedded Intelligence, Machine Learning and Data Analytics are changing and significantly affecting how condition monitoring data is collected, processed and presented to users.

by Dave Leach CRL CMRT CMRP

How many facilities only collect vibration data when it doesn’t interfere with other activities? So often collecting and analyzing data is only one part of a given person’s responsibilities and workloads dictate that the collection and/or analysis take a back seat. When this happens, machine problems are not detected and therefore not reported for corrective action to be taken. If a machine then fails management has all the right to ask why the problem was not found and reported, even if management itself is the reason the data was not collected or analyzed! Vibration data collected should also be analyzed in a timely manner (within two business days of collection) to allow for proper scheduling of any needed repairs; of course, if problems are detected while collecting data that are believed to be severe enough to merit immediate attention, then they should be reported immediately to the facility. Many analysts do not know how long it will take to approve, plan, order parts, kit out, and schedule the resources to execute the repair work. Therefore, one must collect, analyze, and report the data as soon as possible. Generally, you may find several problems in most facilities; however, if you hand in 20 or 30 reports to the Reliability contact, they can quickly be overloaded. I would collate and deliver all the necessary reports but would focus on the top 5 priority problems first, based on safety, criticality, severity, and production demand.

by Trent Phillips CRL CMRP - Novelis

Grease,  excess paint,  etc. can affect the quality of collected vibration data. The below 3 steps will ensure that the best possible data quality is always collected:

  1. Always carry a rag, scraper, brush, etc. to clean each measurement location before acquiring vibration data.
  2. Make sure that the measurement surface is flat and the sensor is not rocking during data collection.
  3. Hold the cable during data acquisition to prevent movement that may induce electrical noise and affect measurement quality.

Learn about the Triboelectric effect on vibration accelerometers

by Trent Phillips CRL CMRP - Novelis

Problem: In a number of real world applications the speed of a machine cannot be held to a fixed RPM while vibration data is collected.
Solution: The order tracking capability of the VIBXPERT allows accurate data collection on machines that experience constant speed changes during data acquisition.Applications such as a winder that spools paper coming off the end of a paper machine, by its design continuously decrease in RPM as the diameter of the paper accumulated on the winder increases.  Other applications such as a pump that feeds a production process may be required to continuously change speed as the process demand dynamically changes.  A spectrum collected during an RPM change will result in smeared and skewed data.  During the spectrum measurement process a digitized time waveform of the vibration is collected.  The time waveform is collected for enough time to gather repeated cycles of vibration.  The FFT process converts the waveform into a spectrum which displays vibration amplitude versus frequency.  If the speed of the machine changes during the waveform collection process the peaks generated in the waveform will not be evenly spaced.  The resulting spectrum will have vibration peaks widened or smeared and frequency information from low frequency to high frequency that is frequency shifted or skewed. The resulting spectrum data taken from a machine whose speed is changing is most often unusable.  In this changing speed scenario if frequency peaks are displayed in the spectrum both the frequency and amplitude will be inaccurate at best. Order Tracking Spectrums allow one to successfully collect spectrum data from a machine that is continuously changing speed. The resulting spectrum amplitudes will be accurate and the frequencies will be displayed in orders of running speed.The collection process requires a tachometer to track the machine speed as the time waveform data is acquired. As the equipment speed changes during the measurement process, the start of the actual measurement has to be synchronized from average to average.  The number of averages is user selectable. Otherwise, the 1× on one average will not be the 1× on the next average, etc., for the measurement.  This results in the same peak showing up in multiple places or being merged with other peaks. This can create analysis havoc.  Order tracking results in an accurate spectrum with no data skewing or smearing.
Setting Up Order Tracking SpectrumsOrder Tracking Spectrum setups in OMNITREND
There are 5 preset Order Tracking Spectrum setups in OMNITREND.  To view these setups in OMNITREND simply select the menu item <Tools> then <SetupManager> from the drop down menu.  Measurement setups 260 through 264 are examples of Order Tracking Spectrums. The preset Order Tracking Spectrums are available in 5 different units of measure.  Like all preset measurements in OMNITREND these are fixed and cannot be changed.  If the preset measurements are not suitable for your application, they may be modified per the following instructions:Order Tracking Spectrum setups in OMNITREND1) From the <Setup Manager> window, highlight and right mouse click on the Order Tracking Spectrum you are interested in modifying.2) Click on <Duplicate Setup>3) An identical measurement will be created and show up at the bottom of the setup manager list.4) The spectrum parameters of this newly created spectrum may now be customized to better fit your needs.Once you have finalized your Order Tracking Spectrum setups we simply add them to a Measurement Location on a Machine in the equipment database or in the Template Editor.The Machine can then be downloaded to your VIBXPERT data collector as part of a route or as a machine template. At the present time Order Tracking Spectrums are not available via the Multi-Mode area of the VIBXPERT.
Collecting Order Tracking Spectrums 
When collecting Order Tracking Spectrums the appropriate accelerometer and tachometer are required per the measurement setup in OMNITREND.  The tachometer must be aimed at a speed reference such as reflective tape mounted on the turning shaft.  When the machine is running, the RPM may change while you are taking the measurement.

Order Tracking SpectrumsTo the left is a spectrum taken as the speed constantly changed between 1775 RPM and 910 RPM.  Notice the waveform time units are recorded as number of rotations.  The spectrum frequency units are orders of running speed. The lower half of the display can be changed to list the maximum 10 amplitudes in the spectrum. Toggle to the lower screen by pressing the F key and then select the peak you would like to see.  In the spectrum the cursor will jump to the corresponding frequency.In OMNITREND the data is equally simple to view and analyze.Order Tracking Spectrums Software

by Ana Maria Delgado, CRL