Many predictive maintenance providers focus only on the tip of the iceberg, leaving hidden risks and unrealized value.
Predictive maintenance has become a buzzword for good reasons. Detecting and diagnosing failures early using wireless vibration or other machine health sensors is helping manufacturers avoid tens of thousands of hours of downtime each year. However, an uncomfortable truth about the field of predictive maintenance is that finding machine failures is just the tip of the iceberg; what lies beneath the surface is a much bigger problem that must be addressed to eliminate the icebergs that affect your plant’s bottom line. When choosing a predictive maintenance partner, it’s important that they understand and can uncover these underlying issues to realize the full value of your investment.
However, many predictive maintenance companies overlook the underlying issues because these factors are difficult to detect, and in some circumstances, addressing them conflicts with their business models. What lies below the surface is machines being damaged by a wide range of factors like hard starts & stops, resonance, cavitation, over greasing, excessive temperatures due to debris build up, moisture ingress, deadheading, aeration, grade changes, or running at critical speeds, pumps in parallel that are fighting each other, periods of high viscosity when cold, and excessive preventive maintenance. Predicting failures before they occur helps plan maintenance and avoid unplanned downtime, but the equipment will only function for a fraction of its rated life unless the underlying source of the failure is understood and addressed. Studies have shown that a typical pump in a Petrochemical plant lasts anywhere from 20% to 50% of its rated life.
Most predictive maintenance companies share a laundry list of fault modes that they cover and suggest that they do root cause analysis. But most don’t measure machines frequently enough to evaluate anything beyond the failure event (the tip of the iceberg). And further, their analyst teams are often spread so thin that outside of a pilot where they are convincing a customer to buy their product, they can’t do anything more than fault identification and simple diagnosis. They are not doing the detailed work required to understand what led to failure, what events shorten the machine life, and what to do about them.
Many predictive maintenance companies have little incentive to address underlying issues because uncovering quick wins from failure prevention is easy for customers to appreciate. This focus on visible problems allows the solution to appear effective, but it falls short of delivering true value. But not all predictive maintenance companies are short-sighted. One way to distinguish between the companies just focused on the tip of the iceberg, and those that are addressing the whole problem is to learn how they measure machines. This will tell you if they can see the whole iceberg or just the tip.
Measuring every few hours or once a day, often referred to as standard definition, is frequent enough to catch failures and do basic diagnostics. Measuring full-spectrum data sets on a time scale of tens of minutes, known as high-definition monitoring, exposes the damaging events that lead to failure. This is because most of the machine behaviors that cause damage occur over short periods of time and are complex. And this is what uncovers the full iceberg lurking beneath the surface.
Effective monitoring requires consistent, minute-by-minute, full-spectrum data acquisition, commonly referred to as high-definition monitoring.
Some predictive maintenance companies are beginning to recognize the importance of high-definition monitoring; however, their products are not designed to support it. Consequently, they rely on frequent measurements of limited parameters, such as RMS, peak values, and temperature, a long with basic health assessments. This approach creates ambiguity, as these summary metrics lack sufficient diagnostic detail, leaving maintenance teams unsure of the underlying causes of damage.
So, how do you know if a provider offers standard or high-definition monitoring?
We’ll be diving into this in the 2nd blog of our “Rethinking Predictive Maintenance: Beyond the Myths” blog series.