At Hitachi’s annual social innovation forum, which was held in Sydney for the first time, the company explored the application of innovative solutions in the era of the Internet of Things (IoT).
The forum addressed key trends and challenges in a range of industries including mining, public safety and healthcare, with the aim of accelerating social innovation through collaboration with leading businesses and decision makers.
Automation was one of the key focal points during discussion on the mining sector – how it helps increase productivity and maintain a company’s competitiveness while reducing labour costs.
When it comes to automation, one of the key factors Hitachi’s Andrew Walters emphasised was data management and how it improves processes at mining operations.
Walters discussed data analytics and the way it eliminates sudden disruptions to day-to-day operations.
“Almost all mining processes are inherently noisy, which means they have a level of unpredictability to them that inhibits automation,” he told Australian Mining.
“Data analytics is the enabler to removing some of that unpredictability and that makes processes more certain and therefore more automatable.”
Walters added that Australia was on track with implementing automated processes on site, although the level of automation varies between mines and companies. He referred to remote locations such as the Pilbara in Western Australia and how these can present high labour costs for companies, leading to the push towards automation.
“Automation is actually a necessity and we’re seeing more and more of that creep into the mining industry, which I think is a welcome development,” Walters explained.
One area Walters stressed was the potential for companies to move from completely manual processes to wholly automated on the maturity curve.
“In any given mine the level of automation ranges from completely manual to fully automated, with degrees of partial automation in between that may require some human intervention or intense monitoring because there just isn’t the repeatability in the process to enable full automation.”
And coupled within any level of automation maturity is the need to maintain equipment, with Walters comparing the different types of regimes used on site.
“When you have a condition-based maintenance regime, by definition you’re actively looking at the health of that asset and that really is a big improvement already compared with completely preventative methods.
“You’re able to maintain an asset when it needs to be maintained rather than when the log book says you should…The asset itself is under constant watch, so if it does diverge in its behaviour you can tell and you can trigger an alarm.”
Walters described predictive maintenance as another level of maturity higher on the analytics curve.
“What that does is it enables you to say ‘well a part that isn’t going to fail now but it might fail in two weeks’ time, so let’s arrange our maintenance service personnel to be able to do it at the time’,” he said.
“You’re looking at its conditions but you’re also predicting when you’re going to maintain that asset and that gives you a layer of control over your new maintenance regime.
“That actually takes the risk out of the change in your maintenance process.”
Walters said mining lags other sectors, particularly in the transition from predictive to proactive-based equipment maintenance.
“Mining, because of its remoteness – it’s a bit of a risky operation – people have been less inclined to change their maintenance practices,” he explained.
“This is a maturity level from completely preventative – which is schedule based and then condition based which is using sensors to improve your knowledge of the asset’s health, to predictive which is one level more of maturity. And then really proactive which is right at the top, which is really only maintaining machines when they need to be maintained.”
However, the biggest issue in developing automation maturity is correlating maintenance with the availability of parts.
“That’s one of the challenges in increasing that maturity, you need the parts and the service available at the time that you need to do the maintenance. That’s easy in preventative maintenance, not so easy in proactive maintenance.
“Mining is off the bottom in terms of preventative only and moving quite satisfactorily up the maturity curve but I think there’s a lot of opportunity there because of the vast amounts of data that are emitted from mining assets.”
Trapped in over-maintenance
While maintenance is routine for operators, there can be a danger of over maintenance, which increases downtime and operational costs.
“If you have an ultra conservative maintenance regime essentially over-maintaining your asset then it’s down for more of the time but it also costs more, because each maintenance request triggers a cost,” Walters said.
“So, by over-maintaining an asset you reduce its productivity and increase your operating costs, you also introduce the potential for human error into the maintenance of that asset.”
Walters added that equipment often broke down during errors in maintenance and therefore over maintenance provides added risks.
Analytics and business intelligence
Walters highlighted that while both are important, data analytics should not be confused with business intelligence, which focusses heavily on data visualisation.
“Data analytics should not be confused with business intelligence because it’s all about the data. It’s about advanced data analytics – you collect data, blend it, correlate it and enrich it and that’s what gives you the greater insight into your actual asset behaviour – the visualisation bit comes way later,” Walters said.
He urged companies to start with data rather than moving on to the visualisation aspect.
“I’m seeing a lot of projects that refer to data analytics projects that are a little bit too heavily skewed on the visualisation side of it rather than the data management side,” he said.
“You can have the most beautiful visualisation tools on the planet but they’re not going to tell you anything.
“The data preparation is 60-80 per cent of the time taken in normal analytics projects. It’s really important to blend data from disparate sources to clean it up and to enrich it using subject matter expertise or advanced mathematical algorithms before you present it.
“You can’t just have a data set in a data base but equally you can’t just have a visualisation of the data that doesn’t make any sense.
“We often find that automation maturity and data analytics maturity advance in parallel because you can’t really have one without the other.”