New preventive maintenance model can save companies millions


Philips’ Interventional XR scanner on which researcher Collin Drent examined his sensible upkeep mannequin. Credit score: Philips

72,000 euros. That is what it prices when a chip-making machine breaks down for an hour on account of a malfunction. It is logical that firms wish to stop these sorts of breakdowns in any respect prices. Their answer is preventive upkeep: detecting potential defects in time, in order that they’ll intervene earlier than it’s too late. Predicting the optimum time for upkeep (not too late, however not too early both) nonetheless is just not straightforward, notably as each piece of kit is totally different. Ph.D. researcher Collin Drent discovered the reply to this problem in sensible mathematical fashions and information—plenty of information. He lately obtained his Ph.D. with honors from the division of Arithmetic and Pc Science.

Chip machines are usually not the one high-tech gadgets whose failure can prices handfuls of cash (and typically even lives). Consider trains or planes, CT scanners in hospitals or wind generators at sea. It’s estimated that unplanned machine failures value companies worldwide round 50 billion euros yearly. About half of this is because of breakdowns. A substantial quantity, in different phrases.
“It’s due to this fact essential for corporations to detect doable failures of important elements in time, in order that they’ll restore or change them earlier than it’s too late,” says Collin Drent, a researcher within the Stochastic Operations Analysis group. “However after all, firms additionally do not wish to intervene too early: components are costly, and also you’d favor to make use of them for so long as doable.”
An costly scanner
To find out the most effective time to intervene, Drent researched Philips’ so-called IXR scanner. These costly CT scanners enable docs to carry out image-guided surgical procedure that’s minimally invasive to the affected person.
Drent rapidly discovered throughout his analysis that the usual fashions for preventive upkeep don’t work effectively on this case. “They assume that gadgets are all the identical, and that you could due to this fact predict with nice certainty when sure components should be changed. However that is usually not true. Each CT scanner is totally different, and so is the way in which and place the place it’s used. Take into consideration the temperature or humidity.”
Thankfully, these fashionable gadgets, due to the various sensors they’re geared up with, produce an terrible lot of information. “We are able to use this information to make our fashions smarter and smarter. On this means we’re in a position to make particular predictions for every system and part, regardless that we do not know upfront precisely which elements will affect the ageing course of.”

New preventive maintenance model can save companies millions

This diagram reveals the ageing course of in time of an X-ray tube, a important part of Philips’ IXR system. Drent’s mixed Bayes and Markov strategies predict from sensor information the optimum second (t = 31) for substitute of the tube (the purpose the place the dotted line of the substitute threshold crosses the degradation line). The second is clearly beneath the failure threshold. Credit score: Eindhoven College of Expertise

Drent used two totally different strategies for his evaluation: Bayesian Studying and the Markov choice . “This has two benefits: by combining the training functionality of Bayesian Studying with the Markov choice mannequin, I used to be in a position to make my predictions much more correct. Furthermore, these algorithms are very clear. So we all know precisely what is going on and for what purpose. In that respect, these algorithms complement conventional AI strategies resembling , the place the precise workings stay hidden in a black field.”

In the long run, the researcher managed to scale back the upkeep prices of the IXR gadgets by about 10 to twenty % on common, in comparison with . “And that is actually lots, for those who have a look at how a lot cash goes into sustaining this sort of high-tech gear: the upkeep prices of such machines are usually a minimum of as excessive as the acquisition prices.”
Twins
Drent (28) was instructed through the promotion ceremony that he had been awarded his Ph.D. cum laude. “That got here as a complete shock to me, however my supervisors later instructed me that it was primarily on account of my contribution to the sector and the broad applicability of my technique.”
Along with CT scanners, he checked out 5 different situations. This confirmed that his fashions additionally work in precept for and chip machines, for instance. Drent due to this fact calls his mannequin a “unified framework.”
Though a lot of his analysis is mathematical, the researcher is initially an industrial engineer. And that’s the area the place he’ll return after his doctorate. On the finish of August Drent will begin working as an assistant professor on the Operations Planning Accounting & Management analysis group within the division of Industrial Engineering & Innovation Sciences.
There he’ll be a part of his (similar) twin brother Melvin, who has additionally lately began working as an assistant professor within the OPAC group. Is {that a} coincidence? “Not fairly. My brother and I’ve at all times had the identical pursuits. And we are able to additionally work very effectively collectively. So I am actually trying ahead to my new job.”

Virtual mission control cuts costs at wind farms

Extra info:
Structured Studying and Determination Making for Upkeep. pure.tue.nl/ws/portalfiles/por … 0220701_Drent_hf.pdf

Supplied by
Eindhoven University of Technology

Quotation:
New preventive upkeep mannequin can save firms tens of millions (2022, July 15)
retrieved 16 July 2022
from https://techxplore.com/information/2022-07-maintenance-companies-millions.html

This doc is topic to copyright. Other than any honest dealing for the aim of personal research or analysis, no
half could also be reproduced with out the written permission. The content material is offered for info functions solely.



Source link