Prognostics – being able to predict when a machine will stop being able to perform its intended function is a key element of effective predictive maintenance and probably the most exciting thing that the Industrial Internet of Things / Industry 4.0 will enable. We’ve put together a short guide to help you understand this technology as applied to the manufacturing sector:
It is accessible in ways that other sectors struggle with
Manufacturing has some great advantages in the use of prognostics over safety critical sectors such as aerospace. The main thing holding back prognostics in aerospace is that over-maintenance and overhauling of assets is driven by strict safety and airworthiness requirements. This prevents assets from being allowed to run long enough to confirm failure conditions with sensor data.
Within non-safety critical sectors such as manufacturing, assets are run for greater lengths of time and in some cases to failure. Combining this with high levels of utilisation results in a far richer data source to base accurate prognostics on. With the ability to run assets into the early stages of the failure mode there is a maximisation of availability of the asset.
In the aerospace sector there is always the dilemma of knowing when to take action. If confidence can be gained in the use of prognostics to extend maintenance schedule periods the result will be that prognostics can notify of the early stages of failure as much as 200 operation hours before functional failure, but what action should be taken? If the aircraft is to be sent off on a 4 hour flight, would an engineer have the confidence to sign off the aircraft for flight knowing a component is degrading, even if the prognostics reports a remaining useful life of 200 hours?
Using prognostics for manufacturing allows predictive maintenance regimes to take advantage of a technology that up until now has been stifled due to stringent aerospace safety requirements.
Large numbers of similar assets
The manufacturing sector typically has a high number of machines with high utilisation and these machines tend to be the same or similar across geographies and sub-sectors. This presents the benefit of greater statistical significance in the analysis of the data, plus the ability to share analytical techniques.
Having a proven failure identification method for a machine provides a primary benefit for that machine, but the method can be used on all of the similar machines. This provides economies of scale in dramatically reducing the costs of prognostic systems.
Assets are in fixed locations and tend to have good connectivity
Having assets that a) don’t move thousands of miles every time they operate and b) are generally not in hostile, remote environments, makes the collection of data more straightforward. In addition to the ease of data collection, changes in data from external factors is minimised providing greater consistency, which makes the task of normalisation of the data simpler.
Impact on operations and business performance
Prognostics is more than a means for engineers to better understand the assets they support. Used correctly it can have direct business impact. Being able to take the results of a prognostics system and drive operational activity will reduce the need for reactive maintenance activity, focus effort on the most critical assets, and ensure increased availability and performance.
Sensor data has value
Your machines and systems will most likely already be generating data. In many cases this is used for short term needs such as monitoring production processes. If you are in the fortunate situation of having a data historian then you are potentially sitting on a wealth of information. In our experience you probably realised this, but are unsure as to how to extract more value from the data in a way that is economical and will provide a return on the effort that is measureable as a business benefit.
Contextualise with events
In addition to sensor data, organisations are rich in data related to the maintenance and other operational events that will give context to machine generated data. Knowing the changes to machine states due to production demands, and batch processing, aids in identifying the difference between an emergent failure and a change of operation. Knowing when maintenance is carried out and being able to contextualise it with condition monitoring information is of great value when combined with historic sensors data to provide greater confidence on the identification of failure models.
Tell me more!
Want to learn more about prognostics and our new, easy-to-use product for predictive maintenance, PROGNOSYS? Download our free flyer here: