A crystal ball for maintenance

Even the best vehicle preventive maintenance programs can go only so far because they are based on historical data, drawn from events that have already occurred. But dial in input from dynamic data generated by events as they happen right onboard vehicles and you have a maintenance system with an uncanny eye on the future. That's the value case being made for the Predictive Monitoring solution launched

Even the best vehicle preventive maintenance programs can go only so far because they are based on historical data, drawn from events that have already occurred.

But dial in input from dynamic data generated by events as they happen right onboard vehicles and you have a maintenance system with an uncanny eye on the future.

That's the value case being made for the “Predictive Monitoring” solution launched earlier this year for the bus fleet of the Metro St. Louis transit system by Chicago-based Accenture Technology Labs.

“Predictive Monitoring presents a significant leap forward in the field of business intelligence,” says Bob Suh, Accenture's chief technology strategist. “Today, most businesses are focused on analyzing data to better understand the past. Predictive Monitoring shifts the orientation by capturing data in real time and using the information to predict business events early enough to take action.”

According to Accenture staff consultant Jim Richmond, the Predictive Monitoring package for fleets is an “innovative, cost-cutting solution that can determine equipment failures before they happen.” He says the system collects historical and real-time data from electronic sensors onboard the vehicle and then uses sophisticated analytic models to predict the future.

“Most equipment maintenance is done reactively, after a breakdown occurs, or routinely on a schedule, whether or not it's necessary,” says Richmond. “By analyzing real-time sensor and enterprise data, the Accenture solution can pre-empt equipment failures before they occur. This reduces equipment maintenance costs and improves maintenance-related business processes.”

He says Predictive Monitoring builds an “information model” for each individual vehicle that, based on mathematical modeling, can predict when components will break down.

According to Richmond, Accenture developed the Predictive Monitoring solution specifically for Metro St. Louis. Accenture served as the system integrator and implementer of the project, partnering with several high-tech firms.

Supporting Metro St. Louis and Accenture on the project are technology partners SmartSignal Corp., for data analytics, Orbcomm, for satellite service, and Quake Global, for sensor translation.

SmartSignal is a Lisle, IL-based firm whose Equipment Performance Improvement (EPI) Center software provides extensive, real-time data analysis.

According to SmartSignal, its EPI software can deliver an “early warning “of equipment problems. “Recent technology advancement enables real-time equipment monitoring and early warning of performance degradation and equipment failure,” says the firm. “The EPI Center analyzes performance, monitors equipment condition, and detects problems such as impending catastrophic failure or degraded operation.”

Developing Metro's Predictive Monitoring solution began with inputting “healthy” operating data to give “a picture of how the equipment behaves when it's humming along,” reports Richmond. This data represented the various modes and functions of the buses' standard operating cycle.

“Once this model is built, we capture real-time data on how the individual vehicle runs and compare it to where the healthy model says it should be,” he continues. “If there's deviation from the healthy model, the system can then do some prognostics and make a prediction as to the type and immediacy of a failure.”


Richmond says Predictive Monitoring models effectively predict failures because equipment is examined closely under different conditions. Using sensor data presents an individualized picture of each vehicle as it's running in actual service — not as it might in a test situation.

“The engine oil temperature for a properly working Cummins ISV engine, for example, is between 140 and 200 degrees, yet that doesn't take into account the differences during warm-up, idle, and full throttle, or whether it's running in a very warm place or a very cold one,” Richmond remarks.

No warning light will come on if that engine stays within the variation for that range. But he explains that the Predictive Monitoring system would be able to detect something is not right for that engine in its particular operating scenario and issue a timely warning, well before a failure could occur.

“Although we can predict with a great deal of certainty when a bearing is going to overheat or a compressor blade will crack, it doesn't eliminate the need for standard maintenance,” Richmond adds.

“We're a progressive transit organization committed to enhancing passenger convenience,” says Dianne Williams, director of communications for Metro St. Louis. She notes that MetroBuses carried over 100,000 daily passengers and traveled 16.7-million revenue miles in fiscal-year 2004.

“While our maintenance facilities are recognized to be among the best in the country,” Williams adds, “there's always room for improvement. We are enthusiastic about the project's early results.”

Indeed, according to Tom Dutton, Metro's director of IT operations systems, the transit agency saw Predictive Monitoring as an avenue to improve system-wide performance by reducing vehicle failures and enhancing maintenance schedules.

“Accenture, which worked on our ERP system, approached us about adopting Predictive Monitoring,” says Dutton. “We're putting in computer-aided dispatch that includes GPS. That gave us the hardware infrastructure enabling us to draw right from the data bus onboard the vehicle.”


Launched step by step with a pilot project that began in January of this year and will conclude this month, 20 of some 600 buses in the fleet have been equipped with sensors that capture operating data from engines and transmissions. The information is stored electronically in data collection units — “black boxes” — installed on the buses.

The 20 Metro buses in the pilot project are all Cummins-powered units. Eighteen are in regular city service and two run express over an Interstate route. “They are all typical of the overall fleet,” Dutton notes.

“The data are sent electronically to computers at Accenture Technology Labs for analysis,” says Richmond. There the information is evaluated by an “analytic engine,” SmartSignal's software. These electronic snapshots of the buses' operational data are compared with the data model created earlier that reflects “normal” operating behavior.

Richmond notes that data can be sent off the vehicle wirelessly, as Metro chose to do, or it can be downloaded manually if a fleet prefers.

When the software identifies potential issues, it automatically notifies Metro St. Louis maintenance managers via e-mails, pages or website alerts. Fleet management then determines the response — deciding when is the most optimal time to schedule maintenance given the type of problem uncovered and customer-service issues.

According to Richmond and Dutton, the system is highly accurate, able to identify anomalies in the operations of engine and transmissions before costly problems develop.

Richmond says that along with reducing breakdowns and maintenance expenses, Predictive Monitoring can help extend vehicle life by customizing — right down to individual vehicles — intervals between scheduled maintenance and major overhauls. “By performing maintenance on each bus when needed, the life of the bus and parts can be extended, which decreases the overall cost of ownership,” Richmond notes.


“A key goal of the project is to determine an optimal time to bring each bus in for service,” says Dutton. “That is, not too soon so that we save maintenance expenses and not too late so that we incur failures or lose fuel economy due to increased emissions.”

As for the day-to-day functioning of the system, Metro has it set up so that fleet managers can access an Accenture web site at any time for real-time review. In addition, alerts are sent via e-mail or as text messages to cellphones to inform Metro when an equipment issue pops up.

“Any equipment issues that are revealed go on a watch list,” Dutton explains. “We have it set so that if an issue remains on that list for 10 days, we receive an alert by e-mail or phone.”

Results to date have been impressive. For example, Dutton reports that the system gave advanced warning when a hydraulic retarder on one transmission began overheating. “While that problem was not severe or dangerous,” he notes, “the early detection prevented a minor problem from growing into a costly repair.

“We've also already eliminated one scheduled radiator change-out, saving $6,000 per unit, by being able to document extending radiator life by 50,000 miles,” says Dutton. “Now we're overhauling radiators once and transmissions twice instead of three times. And those savings add up. “

“We're excited about the long-term benefits we could realize with Predictive Monitoring,” he says. “In fact, we extended the original pilot project schedule on half the buses because those units were giving us data feedback that could help ‘build’ the system further.”

Dutton says Metro is so satisfied with the pilot project's results that it's planning to adopt the Accenture system across the fleet. “We'll roll out Predictive Monitoring in phases — 25 or 30 buses at a time — based on our funding model.”

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.