INSPECTION DATA

Everybody says the same thing – we can’t get the right data, and what we get we cannot rely on.  Given that a large proportion of our data comes from inspections, let’s see what we can do about it. 

What is an Inspection Anyway?

First let’s agree that an Inspection is just that – a task that requires us to inspect a piece of equipment, or a system or the quality of a job done.  It’s not a PM – although of  course often if we inspect, there will be an adjustment or a PM done at the same time.   Three major reasons for an Inspection:

1. to verify a trend or an OK status
2. to comply with regulatory requirements
3. overwhelmingly - to test whether a potential failure has occurred.

For those who are not well-versed in RCM, a potential failure recognizes degradation which – if nothing is done – will predictably lead to a functional failure.  The potential failure is therefore used to trigger a PM to prevent, avoid or delay a functional failure.

To be effective, an Inspection needs to measure a condition against a standard that is set for the potential failure – a critical missing link in most RCM applications.  If the actual measurement violates the standard for the potential failure, then the logical next step is to prepare a PM to execute an adjustment, a repair or a replacement.  This is the core of condition-based maintenance (CBM).

What Happens to the Data

So what happens to the data that has been collected on the Inspection?  If it is to verify the OK status or for regulatory purposes, then it is recorded in the equipment history files – end of story.  But if it is for testing against a potential failure, then it can then be used:

1. as part of a simple trend analysis (where there is a one to one relationship between the condition and the functional failure)
2. for further analytical processing where multiple inter-related conditions are used to predict the failure (– as with EXAKT for example)
3. for failure analysis or reliability analysis

In these cases, we can see that the data should include condition readings such as temperature, pressure, vibration etc;  “as found” and “as left” data before and after an install, a repair or replacement; work remaining to be done etc.  This data should be integrated with the standard work order data relating to work hours, materials consumed, tools used etc to provide a full picture of the events relating to the life of the equipment and thus be the basis for reliability analysis. 

Improving the Quality of Inspection Data

Let’s now turn to how we can improve the quality of the data.  Data collection needs to be consistent, timely and accurate.  Any one of these invalidates subsequent analysis – but who like collecting data?  Some steps:

1. Only collect what will actually be used – how frequently have you heard the comment – “there’s no point in collecting the data, no-one uses it”.  All too frequently we start with the data we have and try to make sense of it.  We have to work backwards from the knowledge that we need to acquire (often prompted by an analytical tool) to the data that we need to collect to build that knowledge.

2. Automate the data collection wherever technically and financially reasonable.  If not, make the data collection process as easy as possible.
3. If the automatic data collection is not feasible, make sure the data collection task (and the data analysis task) is specified on the work order.  That way, time and resources will be allocated.  So will a priority.  That means that it will be part of the backlog management process.
4. Document the data collection process and train the technicians; they need to understand why, what the collection process is, what will be done with the results, the effect of errors, the availability of retraining if needed.
5. Set the responsibility and accountability with the technicians and their supervisors.  Catch and correct errors fast and do not accept suspect data.
6. Publish the results – and more importantly, show what action has been taken as a result of the data collection.  If no action was taken, then ask why the data is being collected in the first place.

Increasingly, maintenance managers are being required to justify budgets and expenditures based on need and returns to the organization; without reliable data collection, this is not possible.

Comments and questions welcomed - Ben Stevens – ben@omdec.com

19 November 2009