07 June 2019
Downtime is bad for businesses - in fact, McKinsey estimates that "outages typically consume between a third and half of the overall maintenance budget and can reduce annual production volume by 5 to 10 percent." Can predictive maintenance turn the dial?
Recurring nightmares: A Short (Shutdown) Story
Your phone rings. You have a feeling it's not a call you want to take. No reason for it, just one of those feelings you get. It's the Operations Manager - the one that looks after the largest facility your company operates. You know before he even started talking - it happened again. The dreaded tripping incident. The cause of every 24-hour shut down you've had for the last 2 years. An almost weekly occurrence at this point - the longest stretch without a trip was a measly 14 days. Two great weeks, they were. This is the bane of your existence, the current curse of your professional life - the problem for which a solution is still evading an entire team of data scientists and engineers. There's just too much data (44bn data points and counting) coming from too many different sources for them to be able to point to a root cause with enough degrees of certainty. You don't really care how it gets solved you just know that every single time it happens it costs (literally) millions of dollars in downtime and lost revenue.
This might be a familiar story. Perhaps you've heard it before. Perhaps you've lived it before. It's a problem for your people’s safety, your productivity and your profits.
It's the kind of problem that needs a silver bullet. And it's usually hard to believe they exist - especially for something as complex and intricate as the machinery that runs a massive plant or factory.
But what if a solution did exist? What if technology had come far enough to give you the elusive silver bullet? What if big data, artificial intelligence, machine learning and predictive analytics could be combined to solve all your problems?
Shutting down shutdowns
Downtime is bad for businesses - there no denying, changing or avoiding that fact. McKinsey estimates that "for energy and materials players, outages typically consume between a third and half of the overall maintenance budget and can reduce annual production volume by 5 to 10 percent." This is an alarming amount for any company concerned with increasing productivity.
..for energy and materials players, outages typically consume between a third and half of the overall maintenance budget and can reduce annual production volume by 5 to 10 percent.
However, the flipside of this carries a huge amount of potential since good management of shutdowns, turnarounds and outages (STO) can lead to cost improvements of up to 30%.
The impact of STO goes beyond just cost and productivity though. For oil and gas companies, the environmental impacts and fines can be drastic for activities like flaring. Layer on top of this is the safety and utilization of the human resources that manage large and complex facilities.
The Silver Bullet?
The shape of this silver bullet might be distinctly binary in nature. It's increasingly likely that the answer to the ongoing problem of reoccurring shutdowns, turnarounds and outages might be hiding in the lines of code that form the existence of artificially intelligent machines that use machine learning and neural networks to analyze masses of big data.
The amount of data produced by assets is overwhelming and difficult to navigate - there is a limit to the amount of data a human brain or even a spreadsheet can handle. Past a certain threshold, more advanced processing and analysis capability is required.
The power of AI lies in its ability to analyze and synthesize massive amounts of data and produce actionable insights and predictions about the future behaviour of an asset or a plant. And when it comes to managing shutdowns - there's no tool more useful than the window into the future that predictive analytics can provide.
Predictive analytics is exactly that - the ability to see into the future of asset performance. This AI enabled control and management means that maintenance - and by extension, shutdowns - can be shifted from reactive and unplanned to proactive and optimized.
So how can predictive analytics lead to reduced shutdowns and increased uptime, in practice? Let us count the ways..
1. AI creates value from data you already own
The beauty of predictive industrial analytics is that is unlocks the latent potential of the data that would otherwise lay dormant. Every asset you own or operate is a treasure trove of data just waiting to be analyzed and optimised.
Given that IDC predicts that the global datasphere will grow from 33 Zettabytes in 2018 to 175 Zettabytes by 2025, it's safe to say that data management is going to be one of the great challenges and opportunities of the information age. And the companies who commit themselves to data mastery will be in a far better position that those who don't.
This is one of the reasons artificial intelligence has such immense potential. Thanks to its ability to compute and synthesize data at volumes previously unimaginable, we're able to unlock insights that would have otherwise been unattainable.
"..it's safe to say that data management is going to be one of the great challenges and opportunities of the information age."
And in the case of shutdowns - those insights can be the difference between a fully functional piece of machinery or a weeklong operations outage.
2. AI can find the connections humans can’t
In most cases, the availability of data isn't the problem - but rather the ability to find utility for it. Most organisations have a wide variety of data points available for analysis. And while human operators can retrieve this data, they can't usually consider more than a small number of independent variables at any one time.
AI gives us a powerful tool to consider huge numbers of variables to find correlations between them and with high degrees of accuracy to predict when equipment failure may occur. By comparing expected operation to actual operation, anomalies detected can provide meaningful, actionable insight, often within minutes.
3. AI takes you from reactive to predictive
After the industrial revolution, the old maintenance paradigm was simple:
wait until something breaks
make sure you have plenty of spares inventory
react only when necessary
Once the third wave of the industrial revolution hit in the 1950s (characterized as the 'digital revolution') we slowly shifted to more proactive models where maintenance was planned or automated to preventatively mitigate breakages and outages from occurring. This approach too, came with its challenges and the efficiency risk of excessive over planning and maintenance.
As the industrial revolutions rolled out in step with the technology available to help drive them, it's no surprise that Industry 4.0 has ushered in a new age of maintenance - this time led by industrial analytics powered by Artificial Intelligence.
Writing a new ending
Thanks to the power of the technology now available, the story of shutdowns can have a new ending. It's becoming more and more apparent that predictive analytics may just be the best way forward for companies transform their asset management.
If the industrial revolutions of the past are anything to go by - there's no doubt of the transformative impact that AI will have on industries across the globe.
Like every bold new technological advancement, implementing AI into a company is not without its challenges.
Wondering what the challenges of implementing Artificial Intelligence really are? Download our whitepaper today: Exploring the Challenges of Implementing AI.
Social MediaShare this article