Predictive Asset Management – why and how?
Predictive Asset Management – getting from where you are to where you want to be.
Historically preventative and scheduled asset management has used the general case, based only on a limited number of criteria, to define replacement and maintenance schedules. This means that an asset of a certain type usually gets a fixed, standard maintenance interval such as every 6 months, every 10000 km and so on. And usually this is based solely on obvious characteristics such as machine type. When investigating individual assets this then leads to one of two types of inefficiencies: either we intervene too early and incur unnecessary costs or too late which results in unplanned equipment failure.
We’ve always known that planned maintenance is better than unplanned. One of the key metrics in managing a good maintenance organisation is the ratio of planned to unplanned maintenance. To help maintenance practitioners focus and improve this area many tools have been developed over the years. These include Reliability Centered Maintenance (RCM) tools that have been used to document and record accepted performance and to manage failure to meet this performance requirement. But behind those activities I’ve always thought we could do better. On individual assets are we sure we’re not maintaining too early, could we have been more proactive to prevent a failure? Are we analysing enough data to make the right decision, backed-up by robust statistical analysis?
In projects that Deloitte has done recently it turns out that we can do better. In fact, just looking at cost savings, customers are realising reductions in both operating and capital costs. We can do better because of advances in the tools and methods used to understand data and tools to query, visualise and analyse data better. With these tools we can now analyse the performance of all individual assets within an organisation and with clear statistically valid confidence levels estimate the life of the asset.
So how does an organisation go about moving to a predictive asset management regime? There are 5 steps:
- Collect, integrate and validate data (ERP, GIS, SCADA, publicly available info)
- Analyse cause and effect relations
- Refine predictive models and predict probability of failure
- Visualise outcome of the analysis
- Act on insight and update individual asset maintenance plans
During recent project work we’ve seen – and this is not surprising – that organisations have various levels of readiness to perform predictive asset management. A clear view of the benefits that can be realised from predictive asset management helps define current gaps and frames a roadmap to reduce both risk and cost.