The Metropolitan Atlanta Rapid Transit Authority (MARTA) wanted to improve the efficiency, safety and reliability of its tunnel ventilation system assets. It sought to upgrade its asset management system from a manual to a predictive model.
The authority chose to do a pilot to upgrade its tunnel ventilation system’s asset maintenance processes using the Maximo APM – Predictive Maintenance Insights solution deployed on IBM Cloud.
Improves asset condition assessmentby using machine learning to analyze multiple datasets
Extends the lifespan of assetsby more accurately assessing asset maintenance needs
Provides a centralized view of asset performancewith dashboards that consolidate information
Business challenge story
Venturing into predictive maintenance
As one of the 10 largest public transit agencies in the US, MARTA is crucial to the economic vitality of the fast-growing Atlanta metropolitan area. The authority serves 1.7 million residents, provides transit to more than 500,000 passengers a year and employs 5,000 people.
The infrastructure for the authority is equally impressive: over 338 rail cars provide service via 38 stations, and 550 buses carry passengers along 101 routes. The authority oversees six million square feet of facilities-related assets, such as administrative buildings, railway stations, rail yards and supporting equipment.
Managing these assets is a major undertaking — and a high priority for the organization. Not only are safety and regulatory factors tied to asset maintenance, but customer satisfaction and economic considerations exist, as well: a one-hour service outage can cost the authority hundreds of thousands of dollars, not to mention a potential loss in future ridership.
In fact, MARTA is a leader in transit asset management. Recently, it became the first North American public transit agency to receive the ISO 55000 certification, an international standard that demonstrates the authority’s assets are maintained and its systems are safe.
The certification is also is a key consideration for federal funding. Every year, federal representatives audit the authority to make sure assets are maintained up to standards. “It shows that we’re being good stewards with taxpayers’ money,” says Remy Saintil, Director of Facilities Maintenance. “We want to make sure all of our assets are working to their fullest capabilities and our systems are safe.”
The past few years have seen a move within the industry from a schedule-based to a predictive maintenance model. Fueled by a strong interest in using innovative technology to drive growth, efficiencies and cost savings, the MARTA facilities team wanted to test the waters to see if predictive maintenance would improve the reliability of its life-safety systems.
“ I don't know of any other transit agency that is doing the type of innovative work — moving into the realm of AI — we're doing with this project. ”
— Remy Saintil, Director of Facilities Maintenance, Metropolitan Atlanta Rapid Transit Authority (MARTA)
Evolution of a pilot
IBM’s involvement in the project started with a cold call. An IBM representative reached out to Saintil, sparking a series of conversations. “He introduced me to what IBM was doing with predictive maintenance and AI,” says Saintil. “From there, I wanted to see what cost savings and efficiencies it could deliver for the authority.”
MARTA launched a POC using the IBM Maximo APM – Predictive Maintenance Insights solution deployed on IBM Cloud to improve the asset reliability of its tunnel ventilation system. The solution uses machine learning to look for patterns in asset data, usage and the environment, and correlates them with known issues to help predict failures.
MARTA chose to limit the focus of the pilot to its tunnel ventilation system for several reasons. The system provides smoke ventilation using large fans and blowers to move air through the authority’s train tunnels and is critical from both safety and regulatory perspectives. It’s a complex system, with more than 500 components and multiple stakeholders, which makes it ideal for testing and potentially scaling to other systems.
The team inputted historical, preventive-maintenance-related information from MARTA’s enterprise asset management system — such as work orders, service requests, repairs, data alerts and weather data — and started using this information to generate models. For example, if MARTA was able to extend maintenance by a month or two, the technicians could use that time to perform other work.
“It showed the survival curve of certain assets and opened up a lot of doors to things we were missing,” says Saintil. “We were able to get predictions, and with dashboards, were able to see all data in one place. We didn’t have to go searching.”
From there, the project evolved. As MARTA got closer to achieving ISO 55000 certification, the team was able to apply information from the pilot to help develop a transit asset management plan, a federally mandated requirement for transit organizations.
Saintil and his team extended the pilot to include some of the information needed for the plan — such as asset inventory, condition assessments, performance measures and decision support tools.
Opening doors to expansion
After completion of the pilot project, MARTA has moved into full production with the predictive maintenance solution deployed in its tunnel ventilation system. One of the benefits the team anticipates from the implementation is the ability to address asset maintenance issues when needed, rather than relying on the manufacturer recommendations, which can result in over-maintenance.
It also anticipates more accurate condition ratings. In the past, these have been largely subjective, based on visual inspections of a piece of equipment. With the predictive maintenance pilot, MARTA was able to use machine learning to generate more objective condition ratings — which are required and audited annually by federal agencies — on an as-needed basis. “It's almost like it's alive, and it's daily,” says Saintil, “so I can have a living, day-to-day condition rating.”
MARTA can also use the IBM Maximo APM – Predictive Maintenance Insights on IBM Cloud solution to adjust preventive maintenance schedules, which in itself is a cost saving measure. Technicians don't have to do as many preventive maintenance checks and can spend more time tending to other issues and improvements.
Beyond the current project, MARTA is looking to address performance management of the health of the full lifecycle of an asset: monitoring reporting processes that enable predictions for future performance and risk, examining historical evidence, analyzing performance and root cause analysis. With this in place, if an asset breaks down, MARTA can more easily do a root cause analysis and go back and look at the data within the system. It can help MARTA better plan for capital budgets, as well.
Eventually, Saintil would like to expand the program to include not just life-safety systems like the tunnel ventilation system, but also operation-critical and operational support assets. “We're on the leading edge in technology when it comes to public transportation asset management. When we talk with people from other transit agencies, it's like, ‘Wow, we've never heard of anybody doing this.’ Not only that, but the reputation of IBM and Watson and what we’re doing together in this area backs it up. It’s something that's really innovative.”
Metropolitan Atlanta Rapid Transit Authority
Serving 1.7 million citizens in Atlanta, Georgia, and surrounding areas, MARTA is one of the 10 largest transit agencies in the US. It provides rail service with over 338 rail cars serving 38 stations, and bus service via more than 550 buses that cover 1,439 miles of road across 101 routes. MARTA is committed to the safety and satisfaction of its passengers, ongoing improvements to its services and the economic growth of the area it serves. The authority has more than 5,000 employees.
Take the next step
To learn more about the IBM solutions featured in this story, please contact your IBM representative or IBM Business Partner.