The continuing dilema of data

Recently I heard an old chestnut that I haven't heard a reliability professional say for almost five years. He made the outright statement (attributed to a global consultancy that he had recently engaged) that "If we do not have any data for us to begin our reliability analysis with, then it is a waste of time starting!" He went on to inform the seminar group that I was teaching that they should not waste their time thinking about starting without data because it will be incorrect and they would risk creating problems rather than solving them. Where do these people keep coming from and who is feeding this garbage into their heads? I often wonder if they are victims of over-zealous E.R.P consultants, or if they have just lived in a cave for the past ten years. I have to admit to being confused about this issue myself a long time ago prior to meeting and working with John Moubray. But that was a long time ago and the entire discipline has moved on... So why is this such a mistaken view to be holding in 2007? A quick recap of tewenty years of reliability evolution... 1. Before we get to have a lot of failure data for reliability analyses we need (Guess what) failures! These cost a lot of money, they kill people and they damage the environment. So if the plan is to drive up the levels of failures prior to starting a reliability program then you will enjoy the discussions with the lawyers once it all goes pear shaped. 2. Some of the failures that you need to manage may not yet have occurred. 3. If any failure is serious, then we will never have a statistically significant quantity of data. Why? Because serious failures get attention, they get designed out, maintained out or some other activity. This does not even take into account the fact that many companies still have poor data capture technologies, they failure mode and equipment register taxonomies are often not the best designed for capturing data, and work processes are often not the best for capturing good failure data. Unfortunately this grates with the judgment of many engineers because without data they are unable to perform a high confidence analysis and they find themselves relying on qualitative information. Not good for people used to certainties and engineered solutions. Obviously this area is fraught with problems. Commercial databases provide some solutions. Particularly those that are able to validate the taxonomy that has been used in the collection of their data and its integrity. But there are issues here also. Differing operating contexts, differing equipment types and errors in data capture are just some of the potential issues. if companies wish to start out along the path to high confidence decisions regarding asset reliability then the following steps may be of use. 1. Do the analysis. 2. Use the resulting failure modes uncovered to populate the ERP coding systems 3. Implement the work processes and technologies to make sure that data captured against the equipment that has been analyzed will be of high integrity and quality. This will provide companies with a structure that will deliver continually improving asset data over time. Driving them forward along the road to higher confidence decisions regarding asset management.