By: Byrd Analytics, LLC
Elevator companies world wide are exploring what they call elevator analytics, as a means of predicting when elevator components will fail. You may have seen a recent TV commercial from IBM showing how Kone Elevator Company is utilizing IBM’s Watson system to do just that.
The purpose of these systems is to minimize the elevator contractors cost of maintaining your elevator equipment. By reducing the frequency of their technician’s visit to the site, they can reduce manpower, which is a tremendous expense. They will tell you that they are passing the savings on to you, but you will likely not see a reduction in your monthly statement. No, they will tell you that they are not going to increase your monthly invoice by the total amount with your next contract. Just like government rebates on home improvements, the entity that pockets the bulk of these savings is the contractor.
As a long time employee of one of the largest elevator companies in the world, this writer was able to witness, first hand, the attempted implementation of Predictive Maintenance more than 20 years ago. The company made a valiant effort to predict failures through the use of counters installed on the equipment, but to no avail. A truly visionary concept, the system just didn’t work as predicted. Though the procedures and practices remain in place to this day and are utilized by this company, they are no longer used to predict failure. They are however, used to track the maintenance history of the equipment and with this history, can prescribe routine maintenance frequency, thus reducing callbacks, falling short of the ultimate goal of eliminating callbacks..
In this case, the company and it’s customers did in fact experience a net benefit from the program by reducing callbacks, so it was not a total loss. This will not always be the case, depending on what is being offered (sold).
While on the surface, predictive maintenance just makes sense. All manufacturers know the rough life-cycle of a particular component. Whether it be a switch or a PC board, everything has a limited life. When it comes to elevators, the number of cycles or times a component is actuated is what is taken into account for estimating it’s life span. While this theory may work in a closed environment, there is not enough data to reliably predict the life of an elevator component in forever changing real world conditions.
The elevator environment varies widely and by the minute. From office buildings to apartment buildings to hospitals. From factories to cement plants to oil refineries. East coast USA to west coast to mid west. North America vs South America. Europe vs Asia. Cold climate vs hot humid climate. Monday vs Friday, and so on and so on. That said, maybe it will some day possible to tap into all of this data, but by that time, elevators as we know them will be obsolete. There is also the variable that no person or machine can predict, the human interface.
All it takes is for a few people to have some bad days to skew this whole theory of “Predictive Maintenance”. Every person on this planet experiences stresses that will affect how they interact with an elevator. Something as simple as a child’s bad grade or an argument with a spouse. What about a job you just hate, a boss that is always on your case or the road rage on the way to and from work. These are all factors that no computer will ever be able to predict but are at the root of some of the most destructive behaviors an elevator sees on a daily basis.
While the attempt by IBM’s Watson is an admirable one, it will be many years before there is a database or a collection of databases capable of reliably predicting elevator failures before they happen. Even in the world of IoT, how humans interact with elevators will never be duplicated.
Bottom line, when it comes to elevators, there is no replacement for good old fashioned, hands on, routine maintenance to minimize elevator downtime.