Q & A: DAVID HARRIS
Firm Hopes to Find Success in Predicting Failure
What if you could predict one week ahead of time and with 80% accuracy that your assembly equipment will break down or that a component in your infusion pump is going to fail? Zero Maintenance International (Chicago, IL, USA) has developed True Predictive Maintenance (TPM) to enable OEMs to do precisely that. Based on pattern recognition technology, TPM extracts meaningful information from a machine's operational data and applies a family of solutions discretely or in combination to a machine population. Company president David Harris sees great potential for TPM in the medical electronics sector.
Q: Tell us about the genesis of True Predictive Maintenance.
A: I have participated in a lot of product development work, and about eight years ago, I started to involve service personnel in the development process. That was something of a radical notion at the time, and to some extent, it still is. I wanted to get some understanding of the discontinuity of how machines are supposed to act in the field as opposed to how they actually do perform. In this information age, it occurred to me that a gap could now be bridged. Machines are becoming more intelligent and are able to store a history of what they do, and they are able to communicate that information. So about 18 months ago, we started full-scale development of our suite of pattern recognition applications, called True Predictive Maintenance.
Q: There must be more to the concept than applying the technology behind computer spell checkers.
A: True Predictive Maintenance is built on our proprietary skill of knowing how and when to combine various methods drawn from applied artificial intelligence tools, which are known collectively as pattern recognition technology. The raw technologies we employ are well known in the military, robotics, and diagnostics fields, but bringing them together in this application is new. We had to learn what kind of data we wanted to look at, and then we had to generate a way to bring together multiple streams of disparate data into a database suitable for analysis.
Q: How disruptive is the process for your customer?
A: Typically, we sit down with the customer and establish what types of data are available. If the company can put a box of transactional and operational data on the table, that's great. If not, they will need to accumulate two to three months' worth of information. We can then do our data-set evaluation in two to four weeks. Fine-tuning and automating our tools takes four to six weeks.
Q: What are some specific benefits for device OEMs?
A: Medical products in the field benefit by a reduction in downtime and service costs. Servicing MRI equipment may involve swapping out a part costing thousands of dollars and require the use of a couple of highly skilled engineers. If you wait until the part fails, the downtime could result in the loss of substantial revenue. Nor do you want to keep a component costing $10,000 in stock.
Q: How does TPM differ from routine preventive maintenance?
A: Mean-time-between-failure standards are developed in laboratory environments and are generally a decent guide. If you are talking about rotating components, say turbines in the aircraft industry, for example, engineers will tell you that the length of use is a really good guideline. But if you talk to someone in the medical arena about an x-ray tube, you will hear a very different story. If the tube is supposed to have a life cycle of 100,000 uses, the actual field experience may range between 80,000 and 150,000 uses. The mean time between failure is not terribly useful.
Devices today have more intelligence on board: you can dial into them and obtain data. Regulatory bodies will increasingly demand that the data are used to advantage in cases where machines are information capable. In my opinion, once a company is able to apply TPM to improve the performance of its products in the field, to increase their uptime, or to prevent errors that are now predictable, the regulatory bodies will start asking people who are not making use of this technology, why not?
Back to the Mar/Apr table of contents | Back to the EMDM main page
