OR/MS Today - June 2003|
On The Right Track
Canadian Pacific Railway earns Edelman Award for innovative operating system that saves millions of dollars while improving customer service
By Peter Horner
In the mid-1990s, Canadian Pacific Railway (CPR) was struggling if not to survive, then certainly to ensure its long-term financial health. Faced with high operating costs, low productivity and rising service requirements, CPR needed, well, CPR.
"We were not at all sure that our traditional operating strategies were up to the challenge of dealing with these issues," CPR President and CEO Robert Ritchie told a capacity audience at the INFORMS Practice Conference in Phoenix in May. "It was apparent to me that we needed a new game plan."
Up until that point, CPR, like all major North American freight railroads, relied heavily on a tonnage-based operating system in which trains run only when enough traffic is accumulated to justify the expense. Although the tonnage-based approach attempts to minimize the total number of trains needed by maximizing their size, in practice it disrupts the efficient utilization of crews, locomotives and equipment. More problematic for CPR, the tonnage-based system yields inconsistent transit times, making delivery service less reliable in an era when customers are demanding and getting better service ... or taking their business elsewhere.
Ritchie's new game plan called for wholesale changes in the way CPR operated its 14,000-mile network that stretched from Vancouver to Montreal, and reached down into major cities in the midsection of the United States. Ritchie, promoted to president and CEO in 1995, envisioned a schedule-based system in which customer needs drove operations a radical concept for freight rail.
"I'm sure you can appreciate that shifting to a schedule-base model from a tonnage-based model was a huge challenge for our organization which for the past 125 years had run on the old model," Ritchie added.
The fixed-schedule approach comes with its own set of baggage, which helps explain why it had not been used in the freight sector. For example, it requires railways to: 1) operate trains with low tonnage when customer demand drops below expectations; 2) systematically forecast traffic levels by day of week to quickly adjust the plan; and 3) employ sophisticated software for the comprehensive and timely analysis of different alternatives.
Neal Foot, CPR's senior vice president of operations, recalls the mid-1990s as a critical juncture in the history of the railway. Not only was the CEO contemplating "a massive paradigm shift" in the way the company did business, but CPR was in the midst of a major restructuring that consolidated all of its senior executives at its Calgary, Canada, headquarters.
"The times were changing," Foot said in a separate interview. "We began spending a lot of effort focused on the customer and the concept of customer needs driving the business rather than us trying to tell our customers what to do."
Was CPR in a fight for survival?
"Not in the sense that we were going to go bankrupt," Foot said. "It was more in the sense that we had to take the company to a different level. We had to be more profitable than we were in the past.
"In today's business climate," Foot continued, "you're either responding to your customers' needs or you're not going to be in the marketplace for very long. A railroad can no longer say to its customers that our train leaves at 5 o'clock and if you have business there, we'll take it. A customer might need service at 3 o'clock in the afternoon and then again later that night. That was what Rob [Ritchie] was talking about, and it drove us to say, 'you know, he's right.' If we're going to be around, if we're going to grow the business, and particularly if we're going to move from a bulk system to a carload system, every carload that we move for a customer becomes important to us. We were going to grow the business carload by carload."
The objectives of Ritchie's new game plan were clear: improve efficiency and utilization of equipment, crews and track, cut costs and most important of all, improve customer service by "perfecting the scheduled railroad." The path to achieve those objectives was less clear, After all, how do you change all of that in an industry not exactly known for rapid turnarounds?
Ritchie admits he didn't have all the answers, but something in his background told him that operations research and management science held the key. In 1986, while running the Intermodal Freight Systems unit at CPR, Ritchie faced a similar high-cost, low-productivity problem. In searching for solutions, Ritchie came across the work of Carl Van Dyke, who had enjoyed success developing sophisticated, OR-based software solutions for the transportation industry, particularly rail. Ritchie didn't get the chance to work with Van Dyke then, but he remembered the work. A decade later, now CEO of CPR and determined to restructure the railway around a schedule-based system, Ritchie brought Van Dyke on board to help turn his vision into reality.
Make no mistake, the operating problem was as huge as the paradigm shift Ritchie envisioned. If your perception of the railroad industry conjures of visions of Casey Jones madly shoveling coal into a boiler as a steam engine chugs up a mountainside, think again. True, rail is a relatively ancient technology, but today's locomotives boast pulling power unimaginable only a few years ago. In addition, rail has evolved into an industry that is extremely dependent on sophisticated computer systems and cutting-edge information technology to tackle a series of complex operational problems that only an operations researcher can truly love.
Consider, for example, the scope of the CPR's planning and scheduling problem. The CPR network includes14,000 miles of track and 250 rail yards. The company employs 15,000 people (including 5,000 crew members), and has a fleet of16,000 locomotives and 65,000 railcars. Every day, CPR's base of 6,000 customers provides about 7,000 new shipments heading to 10,000 distinct pairs of origins and destinations.
And that's just the tip of the problem. In planning its operations, CPR has to account for both the expected (track maintenance, seasonality of freight, emerging new markets, connectivity with other railroads and intermodal transportation providers, etc.) as well as the unexpected (severe weather, derailments, mechanical breakdowns, etc.), any and all of which can throw a monkey wrench into the system.
As Ritchie told the INFORMS audience, "Not only do we need to manage and integrate a complex set of issues and assets efficiently, we have to do it 24/7."
"We needed to develop an integrated operating plan based on a scheduled railway philosophy," Foot adds. "Given the size and complexity of CPR's operation, we needed extensive use of management science and operations research."
Enter Van Dyke and his company, MultiModal Applied Systems. The MultiModal team started by decomposing the mammoth operating plan problem into a series of sub-problems that they then attacked sequentially using MultiRail. MultiRail is MultiModal's unique software solution packed with OR-based tools and techniques including heuristic algorithms, optimization, network flow models and simulation.
Built Out of 'Blocks'
The operating plan was literally built on a foundation of "blocks," and it was the blocking concept that made the schedule-based railway viable. In this context, a block is a group of railcars that move together for some portion of their journey. A block that goes from Point A to Point C can also carry traffic to Point B. For further delivery to Points E or F, a railcar will be routed on multiple blocks over the network. A block sequence is the path from origin to destination.
"It's important to understand that our customers want us to transport carloads, but we need to move entire train loads," Foot says. "Our core challenge is to efficiently aggregate these lines of traffic in our operating plan."
As outlined by Van Dyke, the process of solving the operating plan problem involves several intermediate steps. First, the team develops a traffic forecast. The traffic forecast produces requirements for the design of the blocking plan. The blocking plan is used as the basis for the train-design process. Simulation then analyzes the yard and train workloads by day of week and time of day, and suggests adjustments to improve the overall capacity and utilization of yards and trains. The process is repeated until capacity targets are met. Compliance for service standards is also verified during the simulation process, and changes are again made to the plan until the service standards are achieved. Finally, the resulting train schedules are passed through additional planning tools that develop crew and locomotive cycle plans.
"No single algorithm or model can capture the full complexity of the rail operating plan design problem," Van Dyke says. "Our solution works by tracking the entire problem with the use of many separate algorithms in a holistic framework."
So what are the characteristics of a good train plan?
According to Van Dyke, a good train plan must provide frequent service to meet customer requirements, but it also needs to use as few trains as possible in order to reduce costs. A train should be fast to maximize track capacity and improve service, yet slow to save fuel. A good train plan does not overburden yards by sending too many trains through them at one time, yet bunching trains at yards may reduce connection times.
"Train planners must resolve these somewhat contradictory design criteria," Van Dyke says. "MultiRail provides rapid, interactive feedback on all of these criteria, allowing the planner to focus on perfecting the plan."
As in virtually all OR/MS-type projects, data plays a pivotal role. The specific traffic forecasts the first step in the process and the one that influences all the steps that follow were made possible by combining traffic data from the previous year and the previous month with a high-level revenue forecast produced by CPR's sales and marketing department. The forecasting system gave MultiModal direct access to details regarding CPR's traffic volume, reflecting both marketing projections and seasonality effects.
"The traffic data forms the fundamental foundation for the entire service-design process," Van Dyke says.
Working with MultiModal, the railway developed an operating plan that was tightly matched to traffic patterns, optimizing the routing and classification plan for each railcar movement while determining which trains to run. The team tailored the plan for recurring daily fluctuations in car volumes and developed contingency plans for high- and low-volume days. Shortest path-based algorithms within the MultiRail application identified opportunities to reduce equipment-miles, train-miles and train-hours, thereby reducing operating cost and transit time, and improving service reliability for shippers.
MultiRail generates up to one million trip plans at a time, and produces a viable solution in a matter of minutes. According to Ritchie, the "perfectly" executed game plan also produces some eye-popping numbers for CPR:
Even more impressive, the model-driven operating plan for a scheduled railroad reduced CPR's cost base by 300 million Canadian dollars. A subsequent audit of fuel and crew costs identified another 200 million Canadian dollars in savings attributable to the change in operating practices, pushing the total cost savings to more than half a billion Canadian dollars.
Edelman: An Added Bonus
Now you can add another $10,000 (U.S.) to the pot. That was the first-place prize in the 2003 Franz Edelman Award for Achievement in Operations Research and the Management Sciences, the international competition widely considered the "Super Bowl of OR/MS Practice." Canadian Pacific Railway rode off with the award following an impressive presentation in Phoenix. The award presentation, sponsored by INFORMS and CPMS, the society's practice section, capped a yearlong competition that saw six companies reach the finals.
The members of the prize-winning team, whose presentation was entitled "Perfecting the Scheduled Railroad: Model-Driven Operating Plan Development," included Phil Ireland, Rod Case and John Fallis of Canadian Pacific Railway, and Van Dyke, Jason Kuehn and Marc Meketon of MultiModal Applied Systems, Inc.
"We believe we have the best scheduled railway in the industry," Ritchie said in accepting the award on behalf of CPR. "This award highlights that achievement. The scheduled system brings huge benefits in efficiency and productivity, as well as much better service to our customers.
"Our job, however, is never over. Just yesterday, I learned that an awful lot of opportunities still exist for us in the area of revenue management and contract negotiations. We're continuously looking for new ways to improve, and operations research and management science are going to be crucial to our success in the future."
Asked what he knows now about OR/MS compared to what he knew before embarking on the project that transformed his company, Ritchie couldn't help but chuckle. "I was aware of some of the capabilities of operations research and management science, but I didn't know it would be this successful.
"I think other CEOs are missing the boat if they don't integrate operations research and management science into their operations," he continued. "We find these people bring a different skill set that allows you to look at your problem in a different way. That's particularly true in an industry such as transportation that tends to produce conflicts between optimized operations on one hand and customer service on the other. When you have an over-arching model of your operations that everyone can understand, it allows you to migrate toward a solution that's best for the company and its customers."
When he heard Van Dyke say that no other user of MultiRail had made such extensive and effective use of the system, Ritchie had a quick response. "I can say that without a doubt that the shift to management science and operations research techniques has transformed Canadian Pacific Railway into a more agile, highly cost-effective and profitable company."
Peter Horner is the editor of OR/MS Today. Barry List, director of public relations for INFORMS, contributed to this article.
OR/MS Today copyright © 2003 by the Institute for Operations Research and the Management Sciences. All rights reserved.
Lionheart Publishing, Inc.
506 Roswell Rd., Suite 220, Marietta, GA 30060 USA
Phone: 770-431-0867 | Fax: 770-432-6969
Web Site © Copyright 2003 by Lionheart Publishing, Inc. All rights reserved.