Cloud Platform Predicts Elevator Failures With 90% Accuracy

ThyssenKrupp a leader in the urban mobility has developed, MAX a real-time cloud-based predictive maintenance solution built on the Microsoft Azure Machine Learning platform to solve problems of elevator downtime by proactively identifying issues before they occur and increasing elevator availability.

If you are a tenant of a commercial office building or been a guest or resident in a hotel or condo multi-unit condo building that frequently experiences elevator outages and overcrowding due to the lack properly working elevators, then you clearly understand the significance of reducing elevator downtime. Elevators break down on minimum average of at least four times a year, and with more than 12 million lifts in operation globally, that results in nearly 190 million hours of downtime each year.

Improving elevator downtime using predictive maintenance

Predictive maintenance, machine learning, and data analytics have become common buzzwords in virtually every industry but often is hard to understand their applications in real-world settings. In the case of elevators, the process involves monitoring every movement inside the elevator through the deployment of sensors and monitoring devices and then transmitting this data to a cloud which deploys machine learning to predict failures and outages before they occur.

“We are monitoring every movement of the elevator. We keep track of kilometers traveled, the number of times doors open and close, alerts from control systems, error codes, and many other factors including things such as the number of deliveries to predict when and what kind of maintenance is required to prevent failures,” said Dr. Rory Smith, director of Strategic Development, ThyssenKrupp Elevator Americas in an interview with In-Building Tech.

ThyssenKrupp says that MAX can predict with 90% accuracy that an elevator is going to fail in 5 days and is refining its algorithm to increase the time ten days. ThyssenKrupp’s, elevators are used worldwide in more than 100,000 buildings and facilities. The company maintains more than 337 elevators at Microsoft’s 8 million square foot corporate headquarters and has deployed MAX in 69 high use elevators.

The company has spent two years developing its algorithm by collecting data from nearly 80,000 elevators and has invested more than four years into the project. The entire project entailed developing capabilities to communicate in the cloud and deploying modems and processors for gathering data inside of elevators which are notorious for being the worst locations to carry signals needed to transmit data to the cloud.

Using predicting maintenance to uncover elevator outages before they occur is essential to both clients commercial real estate owners because significant disruptions result in unhappy tenants and increased the cost to vendors who are often in blanket maintenance contracts.

“While there is no way of eliminating maintenance. Predictive maintenance better solves outage problems and prevent wasted resources and time,” Smith said.

Before machine learning, the industry generally relied on a checklist of items and hoped that one of them would solve the outage problem. However, with predictive maintenance, the diagnosis can be far more precise and enable technicians to eliminate services that are not effective which leads to significant savings in time, and labor costs said, Smith.

“We now know that maintenance wise the industry has been doing things that haven’t been effective. As we mature with this, we will have a list of things to do from machine learning. We will know whether there is a need to go today instead of next week and can inform out technicians exactly what to do while they are out there to prevent future outages,” Smith added.

If you are building owner, you might be wondering how does the company transmit data to the cloud since signal strength inside of elevators is perhaps the weakest.

The company also uses a dual antenna system which gives 2-3 decibel gains; one antenna is used to send and receive while the other only receives which doubles signal strength. While ThyssenKrupp’s modem was using 3G, the company has upgraded to a 4G LTE connectivity modem because its lower frequencies make it easier to penetrate buildings and its larger bandwidth enables the communication channel to transmit data other than just related elevator movements.

The additional capacity to send more than just elevator data in real time such as video transmission will serve as means driving innovation in other industries such hospital services, emergency response and public safety, and hotel guest management said, Smith.

The company is researching how to improve cellular connectivity inside of elevators, he added. – Inbuildingtech