Last month, the Field Service Blog reported on Intel’s plans for the so called “Internet of Things.” (Cisco, GE and a number of other tech companies are pushing similar initiatives with equally impressive names.) The Internet of Things is essentially a plan to put wireless sensors on every electromechanical product. According to Intel, anything that is driven by electronics, and maybe even some that aren’t, will be connected to the internet so that performance and reliability data can be monitored, managed and maintained. Strategies like the Internet of Things have been described as an equipment manufacturer’s paradise where, according to the author (Nick Kolakowski), “Companies could analyze that data for additional insights into how to make those products more efficient.”
Whether or not this paradise turns out to be real is the subject of much debate. On paper it sounds like a great idea but in practice customers are raising some important questions. What’s the problem with the fully-connected strategy? Data, lots and lots of equipment data and no good way to connect all that data to what’s really going wrong out in the field. Whether looking at construction equipment or CT scanners the number of electronic control units (ECU) on modern machines is increasing exponentially. In fact, in just the past three years one diesel engine maker has increased the number of ECUs on their engines from 4 to 40. Looking at the semiconductor industry, it’s not uncommon for the ECUs on those machines to generate an error code every second and there are so many control units that each machine has hundreds of thousands of error codes. Regardless of equipment type, the growing use of ECUs to control performance is creating a big data problem for manufacturers and maintainers alike.
One high tech company recently shared that only 20% of their error codes can be traced back to a known failure mode. Furthermore, comparing combinations of error codes to specific failure modes achieved less than 30% accuracy and that when looking at their entire equipment “fleet,” the top 100 failure modes accounted for only 20% of equipment malfunctions. As a result diagnosing equipment problems now consumes 30-35% of total equipment downtime, a fact not lost on their customers. Even with sophisticated prognostics and health management (PHM) systems and integrated built-in tests (BIT) service technicians and call centers are finding it increasingly difficult to troubleshoot equipment, and nearly impossible to determine which error codes are important and which are not. That means that all this extra performance data isn’t increasing uptime or improving reliability.
Clearly the Internet of Things appears to be a little less perfect than what Intel, Cisco and others want us to believe. However, that doesn’t mean that detailed equipment measurements are a waste of time. Rather the problems described above indicate that better tools are needed to evaluate, troubleshoot and repair complex equipment. Whether it’s the service organization, call center, engineering or logistics new approaches are needed to get the most out of all this new equipment data. For the Internet of Things to be a success, companies and their customers must see tangible benefits in terms of equipment performance and reliability.
Moving forward this blog will continue to look at some of the specific challenges that face equipment manufacturers and owner/operators as they embed more ECUs into their product designs and extract more data from the machines. We hope you’ll check back frequently, ask questions, and participate in the conversation regarding best practices for optimizing equipment performance and reliability.