How Predictive Maintenance Is Reducing Risk in Multifamily Operations
Early detection is helping operators spot problems before they escalate.
At The Bainbridge Cos., predictive maintenance has altered the way the firm takes care of the more than 44,000 apartments across its portfolio. Rather than waiting for equipment to fail, the company leverages data such as run time, vibration, temperature and pressure to spot trouble early on. “That means fewer emergency calls, longer equipment life and better overall experience for residents,” noted the firm’s vice president of facilities and capital projects, Kevin Kochersperger.
READ ALSO: The New Rules of Amenity Maintenance: 3 Shifts Apartment Operators Need to Make
Throughout the multifamily industry, companies are optimizing IoT sensors, artificial intelligence and machine learning to help head off the possibility of water leaks, HVAC breakdowns, equipment malfunctions and more before they occur, boosting resident peace of mind.
Staying ahead
Predictive maintenance differs from reactive maintenance, in which operators respond to problems after they occur, and preventive maintenance, in which they routinely check for problems that might occur. Predictive maintenance enables multifamily owners and operators to be proactive. They no longer need to repair or replace equipment when it breaks, or make regular checks that may or may not turn up a problem, said Albert Lord, founder & CEO of multifamily investor-manager Lexerd Capital Management. It helps owners and operators stay ahead of failures, leveraging real-time operating signals from equipment monitors and analytics to flag risk, allowing teams to address any maintenance issue before it snowballs into a much larger woe.
Among the actual instances of predictive maintenance at The Bainbridge Cos.:
- An unusual vibration trend revealed a bad bearing before it caused compressor failure, avoiding costly repairs and operational downtime.
- Pressure irregularities indicated that a water pump was in decline. Repairing it avoided a complete shutdown and negative resident impact.
- A roof exhaust fan’s overheating was spotted when it pulled higher amps than normal. Swapping out the motor prevented the fan from becoming a fire hazard.
- When vibration data ascended for two weeks, it signaled a failing pool pump. Its bearings were replaced before the units seized and led to a pool closure.
- Afternoon AC unit energy spikes signaled a failing capacitor. The resulting fix thwarted a full system failure amid a heat wave, Kochersperger said.
Key data
One of the places predictive maintenance can make the most difference is in spotting water leaks and getting them stopped before they do great damage, Lord said. In the U.S., non-weather-related water damage is associated with $13 billion to $16 billion in yearly insured losses. Leaks often remain hidden until they grow large enough to cause a costly event.
Point-of-leak systems can flag presence of water, often with data like “Leak under washer in 4J,” tipping building staff to respond in seconds. “That faster detection and clearer location data can reduce size of a loss, which is why brokers increasingly use these programs as a risk-mitigation story with insurance underwriters,” Lord said.
A 26-story Manhattan, N.Y., apartment tower reduced its insurance premium by $300,000 after the broker persuasively argued leak detection materially slashed costs, he added.
Jeff Klotz, CEO of The Klotz Group of Cos., also endorses water-leak predictive maintenance. Installing sensor systems linked to centralized alerts slashed water damage claims and emergency mitigation costs. “The lesson is simple: predictive maintenance works best when it protects high-impact risk categories first,” he said.
To reap the maximum benefits from predictive maintenance, operators must key in on specific types of data, said Colleen Needham, assistant vice president & regional property manager for Draper and Kramer.
They include repeat maintenance on the same units or systems, utility consumption anomalies, seasonal performance patterns, run-time hours for HVAC and lighting systems, elevator callback frequency and equipment age versus repairs.
“If we’re not tracking repeat repairs and the costs for that asset either monthly, quarterly or annually, we’re missing the insights a predictive system needs to determine how best to optimize maintenance spending,” she said.
Honghao Deng, Butlr Technologies CEO, points to another type of useful data: Number of doors swings into a common area public bathroom, which otherwise might be tended to reactively or through routine cleaning. Responding based on people entering can help ensure maintenance is truly cost effective rather than based on simple scheduling.
Predictive maintenance is based on AI, which will likely be able to provide additional predictive insights in the future, Needham noted. Leveraging AI will be invaluable in generating capital replacement schedules, automating work order triage, creating energy optimization and reducing insurance risk based on maintenance data trends.
Operators will also be able to use predictive maintenance to improve budgeting and inventory accuracy, said Angie Atkins, senior vice president of community management with Thompson Thrift.
In addition, the technologies could be relied upon “to assist in centralization efforts by scheduling and allocating work to team members best skilled for the task, creating a quicker resolve and response,” she said.
Surmounting hurdles
Operators face a number of challenges in ensuring predictive maintenance achieves real-world success. “Many properties have incomplete work order notes, tools that do not talk to each other and alerts that go off too often,” Lord said. Teams that get results often do simple rollouts letting them select a single pain point, establish clear objectives and leverage initial successes that earn buy-in before they expand, he added.
Comprehensive training that is not only thorough but evolves over time will also be crucial to success, Atkins said. “Understanding what data is required and how it is collected can help inform operators and other potential stakeholders,” she added.
The greatest obstacle isn’t hardware, Klotz said. It’s the habitual tendency of teams to simply react. “Predictive maintenance requires discipline, dashboards and accountability,” he reported. “Discipline and accountability are the hardest to implement in a team that’s been reactionary for so long . . . It’s less about installing sensors, and more about changing workflows, mindsets and behaviors.”
Whether predictive maintenance merits its cost isn’t always evident on the profit-and-loss statement in Month 1. “The payoff shows up in fewer emergencies, staff overtime hours, repeat tickets and resident disruptions,” Lord said. “Over time, that also protects turns, preserves occupancy and keeps teams from burning out.”
Operators who take predictive maintenance the furthest will view it as a process, not a gadget. They will begin with a single problem area, link alerts to a real-world work order workflow and adjust the rules so the team trusts what it sees. Said Lord: “When that loop is in place, predictive maintenance stops being a tech project and becomes a simpler operating model: Fewer surprises, better planning and more consistent service for residents.”
