Transforming to Predictive Maintenance in Six Steps

By Bob Gill

The traditional fix-it-when-it-breaks, reactive approach to maintenance of plants and equipment has been so deeply embedded in most industrial companies for such a long time that those plants that do make the move to implement new digital-based proactive maintenance methods, can gain significant business benefits – especially lower costs and higher productivity from the same asset base. This was the key message from Karl Watson, Global Asset Performance Management and Safety Digital Leader at ABB in his presentation entitled Six Steps to Predictive Maintenance, given at this year’s ARC Industry Forum Asia.

As companies push the reliability curve upwards by adopting techniques, like predictive maintenance and in-depth root cause analysis (RCA), he explained, the emphasis shifts to eliminating failures rather than reactively managing them. This gradually means being able to tackle even the most difficult maintenance problems by cultivating a culture of continuous improvement.

Increasingly, said Mr. Watson, ABB customers that are looking for improvements in their maintenance strategies are also embracing digital technologies. They realize that digital solutions offer greatly expanded ways to facilitate predictive, prescriptive, and other advanced maintenance approaches by enabling and enhancing connectivity, visibility, transparency, predictability, and adaptability. And this combination, in turn, gives deeper insight into the equipment and processes, leading to greater maintenance efficiency and fewer breakdowns.

Begin the Journey with Small Steps

Step 1

Journeys always start with a single step. And the first step in eventually getting to a situation where you can predict impending failures in any plant is to get visibility of the asset data. That’s easier said than done, since industrial companies currently only use a small fraction of the data they presently generate to deliver any actionable insights. And sorting out which data is the most relevant in your specific maintenance context is no simple task.

To help solve this “data dilemma”, as Mr. Watson calls it, ABB has developed easy-to-implement solutions to gather all the key data from disparate plant devices and equipment and collect it in the cloud. This is Step 1, and from there the data can be contextualized and dash-boarded, to give operations teams both faster and deeper understanding.

As a concrete example, in a project done with the Norwegian oil producer OKEA, ABB was able to extend asset lifetimes, improve operational capacity and reduce costs by streaming 5,000 tags to the cloud and contextualizing the data to deliver better asset-specific visibility and better production overview information.

Close Data Gaps

Steps 2 & 3

Once you have enabled ready visibility and access to plant data in Step 1, it is likely that you will notice there are gaps in the data. And this leads to Step 2 – understanding those gaps and closing them with additional sensors.

The ABB Smart Sensor is a good example of a device that can plug such gaps. It can easily be physically attached to rotating equipment, like pumps, fans and compressors, to measure troublesome vibrations, which are very often an early indication of upcoming equipment malfunction. Its easy wireless connectivity, battery life up to 15 years, and certification for hazardous areas make the Smart Sensor a simple solution for expanding the sensing envelope and closing data gaps in industrial environments. And it seamlessly enables Step 3, which involves using the targeted data you now have available and linking it through analytical KPIs to see possible correlations with potential equipment faults.

Transforming to Predictive Maintenance


Get Early Warning to Predict Faults

Step 4

Now the plant can move to Step 4, which is the key stage of truly being in a position to predict equipment faults well in advance of their likely occurrence. This is an enormous benefit, of course, as it can help avoid very costly breakdowns and unplanned downtime.

To achieve this, ABB applies a novel approach that combines data science with reliability engineering. It involves generating training data for a machine learning model that knows what “good” looks like when the equipment is operating correctly. This is then fed into a fault model, in which the equipment faults are well understood through reliability modeling.

The fault model enables the generation of both a Key Severity Index (KSI) and a Key Probability Index (KPI) for all faults related to anomalies in the equipment’s operating behavior. These fault indices are fed into standard predictive algorithm, which provides insights into when the fault is most likely to occur.

In a predictive maintenance project with Italy’s Enel Hydropower, ABB predictive analytics software was deployed across 33 industrial sites. Using this hybrid of Data Science and Reliability Engineering, Enel now gets early warnings to help avoid critical failures and, at the same time, allows it to monitor health diagnostics on all critical assets at the 33 sites. The project aims to deliver significant benefit, giving an increased power generation capability of 10%, as well as maintenance cost savings of around 2%.

Crucial Final Steps

Step 5

With the key enabling technology now implemented, Step 5 is about learning and adjusting operations and processes. An example of this is a 30-year-old refinery in Asia for which the owner was looking to increase the turnaround intervals from one year to three. ABB was selected to deliver a digital transformation project for the refinery upgrade.

As well as deploying an Asset Performance Management (APM) solution across the refinery, ABB expertise was used to perform a “constraints and criticality” analysis to help the company change longstanding maintenance and inspection routines. The result: improved asset integrity; higher uptime, quality and production efficiency; annual savings of $7 million; and achievement of the turnaround extension target.

Step 6

Finally, Step 6 is centered on culture, specifically about establishing a reliability culture. This means getting to the point where everyone in the organization is continuously involved and committed to the identification, understanding, and application of targeted, proactive reliability behaviors and practices throughout an asset’s lifecycle.

Realistically, cautioned Karl Watson, a company’s reliability culture is not going to change overnight from reactive to proactive, to reliability focused on world class. It will certainly be a gradual process, and the speed depends on a number of factors, including the company’s existing culture and history; a strong driver such as company survival; and the long-term commitment and investment by senior management.

So, in summary, transforming companies’ maintenance strategies is a major undertaking, but the potential benefits are huge. Adoption of digital can help smooth the transformation. And the step-by-step roadmap outlined by ABB at the ARC Asia Forum can certainly help speed up the effort.

Engage with ARC Advisory Group