Ensuring The Inspections Will Catch the Defect Before A Functional Failure Occurs

Ever wonder how some organizations make their vibration or thermographic program work, and not only work but deliver huge results to their organization?  They use a systematic approach to establishing the correct frequencies of inspection.   Establishing the correct frequencies of maintenance activities is critical to the success of any maintenance program.   Too infrequently and the organization is subjected to failures, resulting in poor operational performance.  Too frequently, and the organization is subjected to excess planned downtime and an increased probability of maintenance induced failures.

This article will continue the discussion on establishing the correct frequency in a maintenance program.  There are three different approached to use, based on the type of maintenance being performed;

• Time-Based Maintenance
• On-Condition Maintenance
• Failure Finding Maintenance

This article will focus on On-Condition Maintenance.  While establishing the frequency for Fixed Time Maintenance activities is complex and is more of science, establishing the frequency for Condition Based Maintenance inspections (or On-Condition) is a mix of science and art.

Construct the P-F Curve & Establish the P-F Interval

The first step to determining the inspection frequency for on-condition tasks is to construct the P-F curve and P-F interval. Constructing a P-F curve requires recording the results of the inspection and plotting the result versus the elapsed time.  If enough measurements are taken, a fairly consistent curve can be developed for each failure mode. Making sure that the data is gathered carefully and consistently will aid in increasing the quality of the P-F curve.   Lets use an example from RCM2;

• The tread depth on a tire is directly related to the linear distance traveled.  Based on the data collected, it is safe to say that for every 3000 miles the tire wears 1mm.  So for a tire with 12mm tread when new, a potential failure point of 3 mm and a failure point of 2mm, the P-F interval is 3,000 miles.

Now this works quite well for linear P-F curves because it is predictable.  So how do you construct a P-F curve for a non-linear failure mode?  It is a bit more complex, and a bit more of art.  Let’s use another example;

• A bearing will operate with minimal vibration under normal operations.  As a defect materializes, the vibration will increase exponentially as the defect gets worse.   While the P-F Interval will be the time (or operating cycles) from the point the defect can be detected (potential failure point) to the point it becomes a functional failure, its rate of deterioration will increase dramatically towards the end of its life.  This can be quantified just as the tire in the above example, with the right data.

With P-F curve and P-F Interval (PFI) established, the frequency can be determined.

Select the Right Frequency for Inspection

Once the P-F Interval (PFI) is established, the inspection frequency can be determined.  Thankfully it is not as complicated as establishing Fixed Time Maintenance frequencies.  To determine the inspection frequency, the formula is either PFI/3 or PFI/5.

• Standard Inspection – the frequency of inspection for most equipment should be approximately 1/3 of the P-F interval (Formula = PFI/3).  For example, a failure mode with a P-F interval of 3000 miles should be inspected every 1000 miles.
• Critical Equipment Inspection – the frequency of inspection for critical equipment should be approximately 1/5 of the P-F Interval (Formuala = PFI/5).  For example, a failure mode on a critical piece of equipment with a P-F interval of 3000 miles should be inspected every 600 miles.

Now the above works well for linear P-F curves, so how do you establish the frequency for the non-linear curves?  You use the same approach as above for the initial inspection frequency.

However, once a potential failure is detected, additional readings should be taken at progressively shorter intervals until a point is reached that a repair action must be taken. For example; the initial inspection frequency is every four weeks.  Once a defect is detected, the next inspection will be at three weeks, then two weeks and then ever week.

This is only guidelines and should be adjusted based on the method used to track and trend data, the lead time of the repair parts (if not kept on site), and how quickly the data will be analyzed, and the repair work planned.  If your planning process is poor, the frequency should be more frequent, to allow for a high chance of detection sooner.

How much thought was put into your Condition Based Maintenance inspection frequencies?  Have you broken down each failure mode trended the data and established the frequency using a systematic approach?   As with the Fixed Time Maintenance activities, you may be over or under inspecting, costing your organization reliability or money.

Remember, to find success; you must first solve the problem, then achieve the implementation of the solution, and finally sustain winning results.
I’m James Kovacevic
Eruditio, LLC
Where Education Meets Application
Follow @EruditioLLC

References;

• RCM2 by John Moubray
• Establishing Fixed Time Maintenance Frequencies

How to Select The Optimum Fixed Time Maintenance Intervals

Think about your maintenance program. How often are your PMs scheduled?  How were those frequencies established?   If you are in the majority, the chances are that the frequencies were either established from the OEM manual, or by someone in the department without data.

Establishing the correct frequency of maintenance activities is critical to the success of any maintenance program.   Too infrequently and the organization is subjected to failures, resulting in poor operational performance.  Too frequently, and the organization is subjected to excess planned downtime and an increased probability of maintenance induced failures.  So how do you establish the correct maintenance frequencies for your organization?   There are three different approached to use, based on the type of maintenance being performed;

• Time-Based Maintenance
• On-Condition Maintenance
• Failure Finding Maintenance

This article will focus on Time Based Maintenance Tasks.

Time-Based Maintenance Tasks

“The frequency of a scheduled task is governed by the age at which the item of or component shows a rapid increase in the conditional probability of failure” (RCM2).  When establishing frequencies for Time Based Maintenance, it is required that the life be identified for the component based on data.

With time-based failures, a safe life and useful life exists.  The safe life is when no failures occur before that date or time.  Unless the failure consequence is environmental, or safety related, the safe life would not normally be used.   The useful life (economic life limit), is when the cost of consequences of a failure starts to exceed the cost of the time-based maintenance activity.   There is a trade-off at this point between the potential lost production and the cost of planned downtime, labour, and materials.

So how is the safe life or useful life established?  It is established using failure data and history.  This history can be reviewed using a Weibull Analysis, Mean Cumulative Failure Analysis or even a Crow-AMSAA Analysis to statistically determine the life of the component.   Once that life is determined using a statistical analysis, the optimum cost effective frequency must be established.

Establishing the Optimum Economic Frequency

This formula is used to establish the economic life of the component, balancing the cost of the downtime vs. the cost of the replacement.

Where;

• C_T= The total cost per unit of time
• C_f= The cost of a failure
• C_P= The cost of the PM
• T = The time between PM activities

The formula will provide the total cost based on the maintenance frequency. Since the calculation can be time-consuming, Dodson developed a table which can be used if;

• The time to fail follows a Weibull Distribution
• PM is performed on an item at time T, at the cost of C_P
• If the item fails before time = T, a failure cost of C_f is incurred
• Each time a PM is performed, the item is returned to its initial state “as good as new”

Therefore when using the table, use formula; T=mѲ+δ.  Where;

• m is a function of the ratio of the failure cost to PM cost and the value of the shape
• Ѳ is the scale parameter of the Weibull distribution
• δ is the location parameter of the Weibull distribution

In the example below, you can see how the table can be used with the formula;

The cost for a PM activity $60.  The cost of a failure for the same item is $1800.  Given the Weibull parameter of B=3.0, O=120 days, and δ =3 how often should the PM be performed?

• C_f/ C_P = x
• 1800/60 = 30

The table value of m given a shape parameter B of 3.0 is 0.258.  Therefore;

• T=mѲ+δ
• T = (0.258)(120)+3 = 33.96
• T = 34 days for each PM

As you can see, determining the frequency of Fixed Time Maintenance tasks is not as simple as picking a number out of a manual or based on intuition.  Armed with this information, a cost effective PM frequency based on data can be developed for your Fixed Time Maintenance tasks.   This will ensure the right maintenance is done at the right time, driving your plant performance further.

Does you Fixed Time Maintenance Tasks have this level of rigor behind them?  Why, not?  After all, your plant performance (operational and financial) depends on it.   Stay tuned for next week’s post on establishing frequencies for On-Condition tasks.

Remember, to find success; you must first solve the problem, then achieve the implementation of the solution, and finally sustain winning results.

I’m James Kovacevic
Eruditio, LLC
Where Education Meets Application
Follow @EruditioLLC

References;

• RCM2 by John Moubray
• CRE Primer – Quality Council of Indiana