Is Maintenance efficiency and effectiveness still a challenge for most industries? It seems so. McKinsey tracked maintenance performance across the most technologically advanced companies like offshore platforms, navy and military aircraft maintenance over a ten year period (2012 to 2021).
What did they find?
They found that the trend of average scores on maintenance effectiveness achieved by these companies who have adopted relatively advanced maintenance practices’ remained simply flat. No improvement was found in reducing maintenance costs, increasing uptime, reducing failures and reducing unplanned outages.
Across all industries this lack of progress is evident and that is a matter of real concern, especially when there is an intense pressure on industries to improve their Return on Assets and Return on Capital Employed, specially after being affected by COVID pandemic, which unfortunately cannot be done without improving maintenance engineering to eliminate or reduce failures.
The pressure is so high that US Navy is willing to pay more to manufacturers of Light Amphibious Warships who guarantee reduction of failures and unplanned outages of their warships.
So, do industries need to try out something new?
Good practices that drive maintenance performance are well known though not practiced rigorously. Some of the good practices are as follows:
CBM: Condition Based Maintenance (CBM) was developed in the 1970s and had kept evolving by introducing new techniques of prediction. However, I feel that the prognosis part of CBM is still not practical or well developed. Whatever it is, I consider this as a fundamental pillar of Maintenance Engineering. However, Predictive Maintenance is not to be equated to CBM though it plays a vital role in CBM.
RCM: Reliability Centred Maintenance (RCM) was developed from 1980s to counter the risks of failures by taking care of its consequences. It has found its way into industries, where CBM forms the keystone of RCM practice. One of the important contribution of RCM is that it delineates failure patterns into six distinct patterns, where the sixth pattern (infant mortality) is the most prevalent across all industries. That is 68% to 90% of all failures falls in this pattern. The concern is that there are no objective and well developed methods to address this common and dominating failure pattern other than employing predictive maintenance. Though predictive maintenance, when applied correctly detects failures in time, it fails to bring down the failure rate to an acceptable level.
RBI: Based on RCM, Risk Based Inspection approaches matured during the 1990s and are in use in petrochemical industries. It is used for assessing the condition of static equipment.
MP: Though Maintenance Planning and scheduling have been around for decades and is widely used through the application of Computerised Maintenance Management system we are still not sure how to optimise the planning system and use it to achieve the three general objectives of Maintenance Engineering — a) reduce cost b) reduce failures c) eliminate unplanned stoppages.
- IOT and AI: Presently, this technological strategy is attracting a lot of interest. This is basically an enhancement of Predictive Maintenance with a liberal dose of Artificial Intelligence that aims to predict faults quickly, accurately and well in time and eliminate the human element in maintenance engineering. Advanced analytics techniques aims to forecast failures using information sources and data that were not previously accessible, or even available or unsafe to collect with hand held data collectors. Therefore, it is now possible, for example, to combine information in shift handover reports, production schedules, and even changes in the weather to predict when equipment failures are likely to occur. Industry 4.0 or Maintenance 4.0 is a catch-all term for a big range of technologies and approaches, including the networked sensors and devices that make up the Internet of Things, big data and advanced analytics approaches, and new digitally-enabled manufacturing techniques. Many of those things have potential applications in maintenance. That could have a huge impact. We know from experience that implementing a predictive maintenance system can reduce production losses by more than 20 percent while also cutting maintenance costs by over 10 percent. So by using this advanced version of Predictive Maintenance can we hope to better existing baselines to justify return on investments?
With the implementation of new technology, the revised realistic and achievable goals can be— a) 50% reduction of maintenance costs from their present level b) 80% reduction of failures c) sustained uptime of 98% or above and d) 100% elimination of unplanned outages. Though early, such achievement through the application of new technology of AI based IOT, in its present form seems unlikely.
The important insight from the above discussion is — the fundamental conundrum in the maintenance engineering still remains unaddressed that is — what do you do about the dominating Type 6 failure pattern (infant mortality) which have three fundamental characteristics — a) randomness b) infant mortality and c) non-linearity? Preventive maintenance/IOT/AI cannot address all the three characteristics. Though CBM and Predictive Maintenance do address the first characteristic that is – randomness it does not address the other two characteristics. The only strategy that would address this nagging puzzle of maintenance engineering is a combination of precise application of CBM with a reasonably accurate Prognosis (overcoming the challenge of the third characteristic -‘non-linearity’) and DOM (Design Out Maintenance). Though CBM is very well developed, Prognosis and DOM are still in their infancy. That is unfortunate.
However, there are other constraints that need to be addressed before implementing the new technology of IOT, such as:
1. Data: Accuracy of equipment data, equipment history data, detailed data on historical downtime, trend of asset performance, trend of condition of the assets, process data, and linking of different types of data for effective prognosis — without which deep analysis and strategising isn’t possible. Unfortunately, accuracy availability of such data along with manpower resource to critically examine data are lacking in most industries.
2. Planning:The insights from the analysis and strategising are to be backed up by equally robust processes of CBM, Planning and Scheduling, Spare Planning and Quality Workmanship. For most industries gaps exist in these processes, which are difficult to fix quickly.
3. RCM: Without proper development of CBM, Prognosis, DOM through RCM process, massive technological intervention of IOT is surely headed for failure. The fact is CBM, Prognosis, DOM and RCM can’t be outsourced like many other organisational processes since these processes form the core of modern maintenance engineering practice.
4. Management: Very often management is a great hurry to accomplish the desired goals. It is indeed tempting to try and introduce new analytics and IT tools in a “big bang’ way like an ERP system implementation. Unfortunately, ‘big bang’ approach does not work in Maintenance Engineering. We can’t apply technology in a vacuum and expect great results.
Why is that?
This is because Maintenance Engineering would always involve significant amount of human element and application of human consciousness for it to be successful. I don’t see how the human element can be eliminated by IT tools and techniques, which Maintenance 4.0 envisages.
Hence, a sure and safer way to introduce advanced maintenance technology is the time honoured “pilot-and-rollout approach”. This is because it takes time to develop and hone new skills and new ways of measuring performance and new ways of seeing and working. Usually the ‘pilot-and-rollout approach’ is followed by large-scale maintenance and reliability transformation during the “roll out” stage, which is generally fast paced, only if the “roll out” is supported by appropriate stream of value added knowledge and fine tuned analytical process.
In my experience, ‘pilot and roll out approach” is the only sustainable way to drive organisation change through Maintenance Engineering 4.0 to bring about significant changes to transform business results on an on-going basis with minimal effort.
1. What Industry 4.0 can do for Maintenance https://www.mckinsey.com/business-functions/operations/our-insights/ask-an-expert-what-industry-40-can-do-for-maintenance?cid=soc-web
2. Navy willing to pay more for maintainable ships : https://www.defensenews.com/digital-show-dailies/navy-league/2021/08/04/navy-willing-to-pay-more-for-more-maintainable-ships/#.YQuF1xDGCqs.twitter
3. Thermodynamic Degradation Science by Alec Feinberg, 2nd Edition, 2019, DfRSoft/Wiley
4. Complex System Maintenance Handbook Editors- Kobbacy & Murthy, Springer, 2010
5. Design in Nature, Adrian Bejan & J. Peder Zane, Anchor Books, 2013.
6. Irreversible and Reversible Degradation httpsrmcplrapid.com/2021/08/03/irreversible-reversible-degradation/
7. Impact of R&M Improvement through DOM httpsrmcplrapid.com/2021/07/15/impact-of-rm-improvement-through-dom/
By Dibyendu De, Director, RMCPL, 6th August 2021
While analysing an equipment for degradation it becomes necessary to understand as to whether an observed degradation of an equipment’s condition is irreversible or reversible. This post deals with this issue and is illustrated by a case.
An irreversible degradation process would call for either a complete overhaul of a machine or modular replacement of a sub-assembly or replacement of a spare.
On the contrary, a reversible degradation would need minor adjustment of machine or process parameters to correct the situation like greasing, topping up of oil, bolt tightening or minor cleaning, change of fluid level etc.
So the first decision to make is whether degradation of a system is reversible or irreversible so that appropriate follow up decisions on the right maintenance actions may be taken.
The easiest way to do that is by observing the graphical trend of a parameter that monitors the on-going condition of an asset, for example, vibration readings in velocity (in rms or peak).
So, when a graphical trend of a selected parameter is monotonically rising (moving in one direction without fluctuating) it indicates an irreversible degradation. Similarly, when a graphical trend of a selected parameter is not monotonically rising (that means when it fluctuates or changes direction) it indicates that the degradation is reversible. This to my mind is the easiest way to understand the nature of degradation to determine the right maintenance actions. And this insight is based on thermodynamic principle (1st and 2nd Law of Thermodynamics).
Case to illustrate the insight:
I happened to visit a power plant in northern India to examine a vertical ACW pump. It was vibrating at 54 mm/sec, which was abnormally high. When I looked at the trend I saw the unmistakeable monotonic rise of the vibration from a level of 3 mm/sec to its present value in six months time. So I reasoned that the white metal bearing was slowly wearing out through rubbing and as the clearance increased so did the vibration. Clearly, the degradation was irreversible which called for a replacement of the spare, i.e. the bearing.
As soon as the bearing was changed and the clearance was set through “run in”, vibration came down dramatically from 54 mm/sec to 1.5 mm/sec. It was nothing short of a magic to the engineers and managers of the plant.
But all that was really done was to model the degradation by invoking fundamental physical laws or principles.