Log in Article Discussion Edit History Go to the site toolbox

HTM ComDoc 6.

From HTMcommunityDB.org


This page has been updated. Go to HTM ComDoc 6

Choosing the right PM interval

(This document was last revised on 3-4-15)

6.1 Introduction

As we will show below (in paragraphs 6.3 and 6.4) there is a well-established RCM theory that, in effect, states that when the MTBF of the hidden failure is greater than 10 times the interval at which the PVST testing is performed, then the level of patient safety with respect to hidden failures will not be affected very much by a relatively modest change in the testing interval. By relatively modest change in the testing interval, we mean changes such as from, say, 6 months to 12 months or from 12 months to 24 months.

Because the level of safety from hidden failures is not very sensitive to the length of the PVST interval it is permissible to allow the need for the rejuvenation of any non durable parts to be the primary determinant of the length of the PM interval. And since the TPM (part rejuvenation) interval is also not usually very critical, this leaves some latitude for giving consideration to practicality and convenience. While on-time PM completion may have become an important survey-driven discipline, documented findings from current PM work will probably show that “accidental” but "delinquent" PM intervals of 7 or 8 months rather than the prescribed 6 months will have very little, if any, measurable impact on the levels of maintenance-related patient safety.

If the device has one or more non-durable parts (NDPs) that require periodic attention then the estimated useful life of the NDP with the shortest useful life should be used to set the initial PM interval. If there is a manufacturer recommendation on what this should be, then this provides a logical starting place. If there is no manufacturer recommendation then a generally recognized industry consensus interval such as those published by the American Society for Healthcare Engineering HTM ComRef 10. should be used as the starting point. Once this initial interval has been set, the technique known as “interval exploration” should be used to determine whether or not a longer or a shorter interval would provide any better outcome with respect to patient safety. If there is very little difference between the outcomes at several different intervals then it is usually better to choose the longer option. Longer intervals are always more cost-effective than short intervals.

6.2 Interval exploration

In order to explore whether or not one interval is better than another it is necessary to systematically collect some kind of PM effectiveness metric at the time that each PM is performed. The Task Force has chosen to use two relatively simple PM effectiveness metrics (see HTM ComDoc 5.) The first measure is a three-state letter indicator (A, B or F) corresponding to the presence or absence of any hidden performance or safety degradation(s). The second measure is a similar four-state numerical indicator (1, 5, 9, or 0) corresponding to the presence or absence of any premature deterioration in the physical condition of any non-durable parts (NDPs) that were rejuvenated during the PM.

The second indicator will provide us with documented evidence as to whether the PM interval is too long, too short, or just right.

  • If the interval is too long for one particular kind of device (at the manufacturer-model level), there will be a higher rate of PM findings that the rejuvenated part(s) were found to be in worse physical condition than was expected (PM Finding Code 9)
  • If the interval is too short, there will be a higher rate of PM findings that the rejuvenated part(s) were found to be in better physical condition than expected (PM Finding Code 1)
  • If the interval is about right, there will be a predominance of PM findings that the rejuvenated part(s) were found to be about as expected (PM Finding Code 5)

6.3 Impact of the testing interval on the detection of hidden failures

At first thought, since hidden failures are most often caused by some random occurrence, it may seem to be prudent to test for possible hidden failures as often as possible, particularly if the outcome of the hidden failure could be serious or life threatening. However, the following, more careful consideration provides us with a greater insight into the relative risk levels at different intervals, allowing us to make a more informed judgment.

The first things to consider are the MTBFs of the most common and also the most serious hidden failures for the device in question. At this time, before we have much in the way of hard evidence as to what the typical MTBFs really are, we will have to work with tentative, best-guess numbers. Hopefully we will be able to show that the actual MTBF values are greater than, say, 50 years. The soft evidence that we do have at the moment from the apparent level of incidents pursuant to this type of failure seems to indicate typical MTBFs of at least 50 years and perhaps as high as several hundred years. Now let’s look at the impact of changing the testing interval from 6 months to 12 months on hidden failures with MTBFs of 50 and 250 years.

If hidden failures are the result of some randomly occurring event (as is usually the case), and the testing for hidden failures is performed on some regular schedule, then there will always be some period of time during which the patient is exposed to the device in its (hidden) failed state. Ideally this period of exposure should be zero, and we would, of course, like it as close to zero as is reasonably possible. According to standard RCM theory (see paragraph 6.4 below) the relationship between: (a) the average fraction (or percentage) of the time between PVST tests during which the patient will be exposed to the hidden failure (what we will call the “average exposure to hidden failures” – or AEHF - but which in RCM terminology is called the “device unavailability” or DU), (b) the MTBF of this particular failure mode, and (c) the testing interval is:

AEHF (or DU) = 0.5 x PVST testing interval / MTBF

So, for an MTBF of 50 yrs and a testing interval of 6 months, the AEHF is:

(0.5 x 0.5 / 50 =) 0.005 or 0.5%

And for an MTBF of 50 yrs and a testing interval of 12 months, the AEHF is:

(0.5 x 1.0 / 50 =) 0.01 or 1%

With greater MTBFs the AEHF becomes proportionately lower (better). For an MTBF of 250 years (which we consider a quite likely value) the AEHFs for testing intervals of 6 and 12 months are 0.1% and 0.2% respectively. We believe that the difference between the indicated levels of patient exposure to a possible hidden failure (the AEHF percentage) when doubling the testing interval from 6 to 12 months – whether it is from 0.5% to 1% and particularly when it is from 0.1% to 0.2% - is not really a significant overall increase. The impact on the AEHF (which is an excellent measure of patient safety with respect to hidden failures) of using the longer testing interval is quite small, which makes it very reasonable (and much more convenient) to conduct the PVST testing at the same interval as is required for the TPM rejuvenation tasks. If the device has no non-durable parts and consequently no designated interval for TPM rejuvenation, then it seems reasonable to choose a traditional (and entirely practical) interval in the range of 6 months to 4 -5 years. The choice might be influenced by the level of severity of the particular hidden failure and the actual MTBF of that particular failure.

6.4 The relationship between the AEHF and the SV interval/ MTBF ratio

The following is an intuitive derivation of the relationship between the AHEF and the ratio of the SV testing interval to the MTBF of the failure mode. We have adapted this particular illustration from one of the standard RCM textbooks (HTM ComRef 1.) See pages 175-6.

Ten motorcycles have been in service for 4 years. This means that the total service life (what we call the device experience base) of the motorcycle fleet is (10 x 4 =) 40 years. The brake light on each motorcycle has been checked once a year for 4 years. (This example assumes that no attempts were made to check the lights between the annual checks – and remember that this example is supposed to be illustrative of hidden failures).

Over the 4 years, the annual testing found a brake light in a failed state on 4 occasions. So the MTBF of the brake light failures is (40/4 =) 10 years. In this case the failure-finding interval (what we call the PVST testing interval) is equal to 10% of the MTBF (1/10 = 0.1 or 10%). However, we don’t know exactly when each light failed. One may have failed one day after the check, another the day before it was checked, and the other two somewhere in between. All we know for sure is that each of the 4 brake light failures occurred sometime during the 12 months preceding the check.

So, in the absence of any better information, we assume that, on average, each failed brake light failed halfway through the testing interval. In other words, on average, each failed light was out of service (in our world we would say, exposing patients to the potentially harmful hidden failure) for half of the one-year test interval. This means that over the 4 year period, the 4 failed lights were in a failed state for a total of:

4 failed lights x 0.5 years in a failed state = 2 years

So, on the basis of the above information, it seems that we can expect an average “device unavailability” (period in which patients were exposed to a possibly hazardous failed device) of:

2 years in a failed state / 40 years of service = 0.05 or 5%

This example suggests that there is a linear relationship between the DU (or AEHF) of 5%, the failure-finding (PVST) interval (of 1 year) and the reliability of the brake light, as given by its MTBF (of 10 years) which is as follows:

DU (or AEHF) = 0.5 x failure-finding interval / MTBF of the failure mode

According to the text book (HTM ComRef 1.) it can be shown (2 specific references are provided in the book to other peer-reviewed papers) that this linear relationship is valid for all DUs (AEHFs) of less than 5%, provided that the failures conform to a true random failure pattern.

Site Toolbox:

Personal tools
This page was last modified 22:32, 2 October 2018. - This page has been accessed 11,329 times. - Disclaimers - About HTMcommunityDB.org