12 minute read
Next generation of maintenance
BY MARIO CYWINSKI
Greek philosopher Heraclitus is credited with saying “change is the only constant in life.” In his time, the issues we are facing today in maintenance were not around. However, all this time later, the saying still holds true, and in turn, those companies who do not adapt to ongoing change, will be left behind. Many of the technologies that we are seeing in the world today, were only seen in science fiction movies years ago.
Like the rest of the world, maintenance is constantly changing and evolving. Obviously, the core values remain, but how things are done, continues to change. We are seeing advancements in artificial intelligence (AI), use of data and diagnostics, moves to outsource maintenance, more focus on predictive maintenance, and focus on many other new technolog ies.
MRO reached out to experts come from across the maintenance and asset management sphere, to get a complete look at what the future holds for the world of maintenance. They include Hugh Alley, president, First Line Training Inc.; Richard Kunst, president and CEO, Kunst Solutions; Susan Lubell, principal consultant, Steppe Consulting Inc.; James Reyes-Picknell president and principal consultant, Conscious Asset; and Cliff Williams, author, and asset management exper t.
MRO: What do you see as the biggest advancement in maintenance in the next five years? Ten years?
Hugh Alley: Five years, there will be more retrofittable sensors that will provide diagnostic data about equipment condition that will connect wirelessly. Some equipment manufacturers will start taking data security seriously, and more maintenance operations will be looking for secure connectivity; just being connected will no longer be enough.
In 10 years - equipment manufacturers and maintenance service organizations will increasingly provide uptime as a ser vice, as they leverage the data they gather.
Richard Kunst: Sensors are getting cheaper, faster and have broader capability. Add to that how the bandwidth and speed of data highways has become, and we will see a migration from preventive maintenance to predictive maintenance through the early detection of subliminal abnor malities for quicker, cheaper, and easier repair s. It is now very cost effective to build several layers of redundancy within an automated system, repairs can be made without incurring any actual processing downtime. The coolest aspect is that we will be aware of a potential problem before we even physically realize it.
Susan Lubell: In the next five years, we will continue to see step changes in the use of predictive maintenance for asset health monitoring. As sensors and condition monitoring becomes cheaper to install and can be placed in more hazardous environments, we will be gathering more asset health data on a real time 24/7/365 basis. This analysis will inform our decisions on run, repair, replace for physical assets, providing the information to decide when to maintain physical assets and what the scope of the repair will be.
With the advancements seen in the last few years, 10 years out is too far to predict what new data and information the future technologies will enable.
James Reyes-Picknell: In the short term, competitive pressures are driving a push towards productivity and in all but new plant and equipment, they are struggling with reliability issues.This is causing an uptick in focus on reliability currently.Within the next five years, we will see a reset, going back to basics in both reliability (maintenance tactics) and in work management practices. Processes and systems are in place but are often used poorly and to little effect. The workforce has lost a lot of experience with the retirement of large numbers of people during and shortly after COVID-19. With relatively few baby boomers left, and in most cases, no attempts to capture their knowledge, companies are realizing just how much they depended on them. They are repeating old mistakes and realizing that they are not as systematic in their methods as they may have thought.
In the longer term (10 years) we will see a shift towards outsourcing of maintenance in small, medium and some larger industrial operations. Attempts to fix the problems above will not be entirely successful due to an unwillingness to invest enough in maintenance, and the continued drain of experience. Realizing that they are not as good at maintenance as they want to be, and that they are seemingly incapable of managing changes, they will look to outside expertise - not in consulting, but to delegate the whole function. This will open the door to a big shift, particularly in North America, where outsourcing hasn’t been popular, from in-house management of maintenance and reliability to reliance on external specialists who treat it as a business, and not just a troublesome cost centre.
Cliff Williams: Obviously AI and machine learning will be the advancements – how effective they will be in mainstream maintenance and reliability is the question. Many lack the basics of good maintenance and reliability, don’t appreciate, or use the data they have –or should have, so there is skepticism around how they will take advantage of the advancements. At the Society for Maintenance & Reliability Professionals (SMRP) Conference, there were more AI technology vendors than any others – but what they were selling was the technology and not necessarily the solutions. These will depend on the knowledge and understanding of what drives results in the maintenance and reliability organizations. It will also involve a lot of change in approaches, roles, responsibilities, and we’ve proven time and again to be lacking in change management.
MRO: How do you see AI affecting maintenance operations?
Alley: AI is about multi-dimensional pattern recognition. It demands very large datasets, which are only going to be available to organizations managing large quantities of equipment. This will create challenges for small organizations, which won’t have the analytical capacity, or the quantity of data needed. This will open opportunities for specialist maintenance services to aggregate data, through the better understanding provided by the data, they will be able to offer superior uptime to smaller operations. It will also open opportunities for manufacturers to provide monitoring services, and they will aggregate the performance data from the equipment they build, allowing them to tune where they make improvements and to provide better guidance about maintenance needs. The manufacturers will therefore be able to develop ongoing revenue streams from sharing the results of a customer’s data with them.
Kunst: First, AI is going to be a big benefit to user manuals to insure people properly start and operate machines. In many cases we fail to optimize the capability of our equipment because we get used to using only cer tain functions. Second, AI will be critical to early detect and resolve potential issues. With AI we can be constantly running diagnostics in the background, while the machine is operating, stealthily detecting, and seeking abnormalities and quietly switching to a redundant process while alerting the organization what and where to perform maintenance.
Lubell: The vast quantities of data gathered are staggering. We will need AI techniques to build analytical models and frameworks to harness and analyze the data, turn it into information, and allow us to make business decisions. The maintenance operations space is no different. The increased collection of asset health monitoring data needs to be analyzed to be useful for business decisions and this will require AI.
Reyes-Picknell: As companies shift more towards installed sensors and non-human data gathering the quality of data will improve. This will enable more reliable performance of predictive/prognostic tools that are presently hampered by poor quality maintenance data, that often lacks the granularity needed to be used for reliability analysis. Deterministic algorithms that are in use will become more valuable as the quality of data feeding them improves.
With more sensors and data, the door will open to data scientists who can apply more general analytic tools to finding relationships, correlations, and helping to identify where we may find causation of those correlations. We will still need to intervene in making those latter determinations as the analytics don’t “know” what’s going on aside from the numerics. This will all enable shifting towards greater reliance on condition monitoring and may reveal the correlation of excessive inspection work and overhauls to lower reliability to those who remain skeptical today. Reliability engineers will be increasingly needed, and in turn, they will need enhanced analytical/data science backgrounds.
MRO: Do you see any maintenance activities that will be eliminated in the future?
Alley: Routine monitoring activities will become less necessary as it becomes cheaper and easier to connect sensors wirelessly. Time-based preventative maintenance will be replaced by condition-based predictive maintenance.
Kunst: If we gauge our maintenance efforts in the form of frequency and severity, I envision that severity will significantly decline due to better early detection of potential problems, before they can create a catastrophic consequence such as a machine breakdown. However, it would be reasonable to expect more minor maintenance requests being performed.
The more data you feed AI the more effective your AI will assist you. Like CNC machining, where you can enter “tool wear parameter” so the machine will tell you when to replace a specific tool. We need to take a fresh look and develop parameters that we do not actively monitor, temperatures, amp draw, and vibrations, we have the detection technology, but have not created upper and lower control limits.
Lubell: With the increased focus on predictive maintenance and asset health monitoring on a real time continuous basis, maintenance activities will shift to more analytics and interpretation of the data and information. The fundamental understanding and performance of maintenance activities won’t go away; however, the emphasis and time spent will shift to more analytic interpretation and deciding how to respond followed by more targeted repair strategies.
Reyes-Picknell: Already we see less inclination to perform repairs at the component and assembly level in favour of swapping out parts and assemblies. Some assemblies still get repaired, but less by in-house shops. That trend will continue with two possible outcomes. For those who invest in the ability and knowledge needed to do effective troubleshooting, repair times are likely to shrink and that could potentially contribute to a drop in workload overall.
For those who do not invest in those abilities and knowledge, it will lead to higher usage of parts and assemblies, as they are swapped out, not knowing if they need to be.
We already see that in the realm of electronics, where circuit cards and modules are swapped out to hasten repairs, without knowing if they are flawed. That will expand into the mechanical world, because of the loss of exper ience and expertise that is well in progress today. Sadly, what will be disappearing in many cases, is that ability to troubleshoot effectively.
Williams: No, I hope not, we still see organizations that don’t do what they should be doing, the hope is that we don’t eliminate any activities, but progress to using the right ones for the specific situation.
MRO: What can facilities do to make themselves more technology-ready in the short term? Long term?
Alley: It’s all about developing your people. Maintenance staff will always need the wrenching skills, but they will need to know how to tap into the insights from large quantities of data. Those skills are unlikely to be available “on the street.” Hiring the skills, you need was more possible 30 years ago, when controls for most machinery and equipment were mechanical. If you knew how hydraulics and mechanical systems worked, you could fix just about everything. In the coming years that won’t work; every company’s specific needs will be different because their specific mix of equipment and products will be unique. Consequently, companies will need to be able to learn these skills as they apply to each of their specific machines. They’ll also need to be able to retain and teach what they’ve learned as staff change over time.
Meaning maintenance leaders will need to become better instructors, and better at the skills of developing scientific thinking. These are skills companies can start developing now, and they will have almost immediate payback. For leaders with these capabilities, the new technologies will present lots of opportunity.
Kunst: Get your user manual into a digital footprint, make sure all corrective actions are logged digitally. Of course, building your data foundation still requires a manual input so conducting your physical daily total productive maintenance (TPM) checks and logging the abnormalities is the fodder for creating a more effective AI.
Lubell: One of the areas that is starting to get more attention is asset and maintenance information management. We need to think ahead to what questions we want to answer in three or five years and structure our data to allow us to do this. Ask yourself, how complete is your asset register? Do you know what assets you own? Do you know the value that that they bring to your organization? Consider cost, risk, and performance in terms of safety, health, environment, economics, regulatory obligations, social and governance. Infor mation is an asset by itself and needs to be managed as an asset, collecting data without a specific purpose quickly becomes overwhelming to manage and often ignored.
Reyes-Picknell: In terms of IT and OT knowledge and skills, most operations have this already. Many have already deployed technologies that are underutilized or not utilized at all. The technologies that are available, have in many cases, exceeded the ability to use them to much effect. In fact, these systems are often generating work just keeping themselves going. The danger we face is not in lack of technical capability, rather it is in jumping to technological solutions before they really know what they are doing to do with them.
I see a bit of “ready, shoot, aim” going on. There is need to do more “aim” and sooner, before deciding on technology and its rollout. It will be necessary to find and/or develop some expertise in production processes, equipment, and industrial systems (not just the IT), working together with maintainers, operators, and reliability engineers to help guide the need for and deployment of technology where it can do some good.
Williams: Plan replacements based on latest technology – get an understanding of what the technology can bring and how it fits in with their organization – identify the opportunities through baseline measures of where they are today and set goals to be achieved through technology. Educate and inform all who will be involved and manage the change.
MRO: Can you speak about the assets and those which need updating the most?
Kunst: As our appetite for data continues to expand and grow, we need to increase our wireless capability within our premises. For many companies it will likely require a review of their electronic policies, as smartphones continue to be an extension of the human capability, preventing them to accompany people into work environments will be limiting the effective use of AI and enhancing employee effectiveness.
Reyes-Picknell: There is a lot of aged infrastructure in our “basic industrial” capability (old plants and equipment). Investment in these more basic industries has largely been directed overseas, allowing our capacity to age and degrade. The impacts of COVID-19 on supply chain were a wake-up call about globalization, too much reliance on just-in-time and excessive lean. The world is far less stable that it was just a few years ago. To sustain our capabilities, so that we are less vulnerable to global calamity and disruptions, we need to sustain that basic industrial capability and some of its capacity.
W illiams: In true asset management style, it would be the systems we use, the processes we employ, the measures we make, and the culture we build – it’s all about people. Updating equipment while employing the flawed practices won’t work, not having the right data or having a way of turning that data into knowledge and eventually into wisdom should be the starting point where we can then, with confidence, understand the asset that needs updating –and not before.
MRO: Anything to add?
Alley: Like many sectors, the world of maintenance is in the middle of losing large numbers of extremely capable technicians as baby boomers retire. A lot of knowledge is walking out the door. For many firms with limited maintenance management systems, this loss can be substantial. I have one client where the entire maintenance staff has turned over in two years and the new supervisor is having to rebuild the entire knowledge base of the facility.
Organizations that can capture that knowledge before it leaves will have a significant leg up on the competition.While new technology can help, the underlying discipline of developing standard work depends on leaders who are committed to the objective of delivering reliable uptime, and who are able to convince business leaders of the necessity of maintaining their assets well.
Lubell: We are seeing huge advances in the use of technologies, sensors, and data analytics to advance predictive maintenance and asset health monitoring. It’s important to make conscious decisions on what and how we will use the data collected, how we will analyze it, and the maintenance decisions that it enables. Otherwise, we will be drowning in data and starving for information, when we want to be in a position of gather ing maintenance and asset data and turning it into information that allows for business decisions.
Reyes-Picknell: History repeats itself because of organizations failing to learn the lessons of the past. Don’t under-estimate the impacts of demographics on your workforce, your in-house expertise, your capabilities, and your capacity. Waiting too long to capture what knowledge remains will prolong the pain of reduced capacity and increase the likelihood of repeated mistakes. The younger workforce is smart, tech-savvy, and eager to learn, but quickly losing the experienced guidance you may still have.
Mario Cywinski is the Editor of MRO, Plant, and Food and Beverage magazines, and a member of the Automobile Journalists Association of Canada. Contact him at mcywinski@mromagazine.com