Performance criteria is one of the most—if not the most—important aspects of a successful preventive maintenance (PM) and condition-based maintenance (CBM) program. The asset owner must refine and accept all performance criteria for it to be meaningful, and the criteria must be documented to ensure consistent application. Specifically, when referencing national and international consensus standards as a basis for a facility's performance criteria, personnel must evaluate the criteria to ensure it is representative of the normal operating characteristics of the asset.
In all of the CBM (and predictive maintenance [PdM]) courses I have developed over the years, I have always shared the following: "It is the facility owner's responsibility to establish performance standards (objective criteria). Any defined performance standards should provide alarm values and shutdown values."
If consensus standards or even original equipment manufacturer (OEM) recommendations are used as a basis for specific asset performance standards, those recommendations must be evaluated and, in most cases, adjusted to fit a specific set of asset operating conditions. The facility owner's engineer is typically responsible for that decision.
Documenting the criteria and providing training are also necessary to ensure that the defined performance criteria are globally understood and that consistent and coherent data is collected for analysis.
Ask yourself: Has your facility made this level of investment in its PM and CBM programs?
Low-Frequency Vibration Monitoring
The following discussion focuses on broadband, filter-out, low-frequency vibration monitoring (less than 2 kilohertz [KHz]). The discussion also deals with International Organization for Standardization (ISO) consensus standards associated with general equipment vibration. The notions and concepts presented in this article are also applicable to those consensus standards that directly apply to in-situ pumps and fans, Hydraulic Institute and Air Movement and Control Association (AMCA) International.
In the days before ISO 10816, Mechanical Vibration – Evaluation of machine vibration by measurements on non-rotating parts, Parts 1 through 6, there was ISO 2372, Mechanical vibration of machines with operating speeds 10 to 200 rev/s – Basis for specifying evaluation standards. ISO 2372 was replaced by ISO 10816 in 1995.
One example of performance criteria that must be evaluated and adjusted to fit a specific set of asset operating conditions is the old\'97from 1974—ISO 2372 Vibration Severity Chart. A version of this chart is packaged with just about every vibration meter, vibration analyzer and vibration data collector on the market today. It is still referenced 20 years after ISO 2372 was withdrawn.
After 40 years of presentation, you would think that most would understand that this chart is not an absolute—it provides guidance only. It offers a place to start when developing a CBM process.
This chart was also referenced in the original ISO 10816 (ISO 10816-3, to be specific) as an appendix. In more recent revisions of ISO 10816, the appendix has been removed and the severity chart has been rewritten. However, no matter what color you paint it, the severity chart is still there. ISO 10816 (the original and the latest iteration) also contains the following statement: "The ISO does not provide pass or fail criteria. It provides reasonable guidance ensuring that gross deficiencies (read as unsafe) or unrealistic requirements (read as too limiting) are avoided."
For whatever reason, few people notice this statement. Many technicians and engineers see the charts and tables and jump to the conclusion that these are absolutes. Figure 1 (page 72) shows a variation of the old 2372 severity chart. Several things are important to note:
- The severity chart is no longer specifically referenced in a consensus standard.
- The chart is not used as pass or fail criteria but as general guidance.
- It is the facility owner's responsibility—not a third-party committee's—to establish performance standards (objective criteria).
As an example, a global manufacturer with a fleet of facilities in the U.S., Taiwan, China, Japan and Korea that has been in business for more than 100 years drank the ISO 2372 Kool-Aid. During initial visits to each facility when the technicians walked through their vibration monitoring program(s) for small-frame pumps, they showed their latest manual data collection and on-line monitoring schemes.
Not surprisingly, none of the programs was consistent or even similar in terms of data collection and analysis. That was the first fix: to establish a consistent and coherent program. However, one thing was consistent: a devout following of the ISO 2372 severity chart. The alarm points in all of the monitoring schemes were always in line with the severity chart. A red line at 7 millimeters per second (mm/s) root mean squared (RMS) was drawn across all of the trending data for their small-frame pumps—a solid D according to the ISO severity chart.
Despite this diligent effort to establish a limit, the mean value of all vibration data collected on the facility's small-frame pumps was less than 2 mm/s RMS. The alarm value was established at 300 percent of any actual monitored condition. In effect, the equipment could fail—and actually had a history of failure—before the condition that would cause a failure was identified. Based on the alarm value, the program could not provide a return on investment. This was the No. 2 fix: to establish meaningful performance criteria.
Common explanations for how the alarm value was derived included:
- "The vendor that sold us the monitoring equipment set the alarm values."
- "The commercial vibration training course I attended referenced these alarm values."
- "These are ISO alarm values."
When asked to show a copy of the ISO, they pulled out the old ISO 2372 severity chart that came in the box with the monitoring equipment or device—no standard, just the severity chart. When asked why they allowed a vendor (a third party) to define what was or was not acceptable criteria for their equipment, the management looked astonished. They invested heavily in a program (training, equipment and software) and had not seen any real return.
Did this facility's maintenance team change course? Yes, they did. ISO 10816 includes a caveat that states that changes in vibration amplitude greater than 25 percent warrant investigation. So with a mean value of 2 mm/s RMS for all readings, they adjusted the alarm value down to 2.5 mm/s RMS. For a shutdown value, the facility chose 3 mm/s RMS. Once the facility set these limits, assets that performed in a range greater than the alarm value but had not reached the shutdown value were put on a watch list. Assets with readings in the shutdown range were, of course, shut down.
Shutdown assets became a priority, prompting personnel to correct deficiencies and at least get vibration performance below the shutdown values in order to place the equipment back in service. Watch-list equipment was considered a secondary priority, and personnel aimed to bring performance in line with similar assets with better performance.
Most performance issues associated with the company's small-frame pumps at all of its facilities were coupling and alignment issues. This challenges the idea that balance is the most common cause of vibration issues.
This is what CBM is all about: finding indications of potential failure, and prioritizing, planning and correcting the cause before actual failure occurs. Facilities will see return on investment and asset performance improvement.
In the discussion above, you probably noted that I used a value listed in a consensus standard (in this case, ISO 10816, Part 1) to establish an alarm value. However, past experience and how I incorporate that experience into training on vibration acceptance is a bit different. For common equipment types, I have the end user collect data on all bearings in all plains to establish a baseline. Then I have the individual use a spreadsheet to list the equipment on the rows and the data collected at each bearing position in the columns. We look at the mean value of data from each column and the mean value of all of the means. The first standard deviation from the mean of the means is the alarm, and the second standard deviation is the shutdown. Then I round the values up to the next whole number (of half value). It is a fairly simple process. What is particularly interesting is that the alarm value generally comes in at 25 percent.
We are establishing performance criteria based on the equipment's operating characteristics in a specific set of circumstances. We are using the consensus standard as guidance and reference to ensure the decision is reasonable—not too high and not too low. Using this method, facility owners are taking responsibility for establishing performance criteria to evaluate the condition of their equipment.