Research in motion

Some fleets analyze operational data to solve comples business problems.

High costs, competitive rates, changing freight demand and schedules mean that even small mistakes in operations can result in big losses. In the past, fleet owners solved such problems through a combination of experience, intuition and luck. But increasingly, fleets are turning away from the source of many mistakes – conventional thinking – and incorporating operations research to solve complex problems.

Operations research basically means using the information you can gather on your operation to improve efficiency and effectiveness. It has existed at some level in trucking since trucks hit the road. The goal of operations research is to find not just a good solution – but the “optimal” solution.

What has made operations research more powerful is the rise of information technology. Virtually everything your fleet does leaves a “trail” of data. Today, savvy fleets use this data to determine how changes in variables could achieve a principal objective – maximizing utilization or revenue, for example. Operations research achieves through relatively quick mathematical computations what fleet owners once only could accomplish – if they were lucky – through disruptive and often expensive changes in equipment or operations, followed by months of tracking results. Today, operations research often can allow fleets to make adjustments in time to change the outcome – something fleets never could have done just a few years ago.

While operations research may sound complicated, many software developers incorporate advanced operations research within easy-to-use tools that fleets can deploy at various levels within their organizations. When combined with teamwork and a company culture of continuous improvement, operations research can become a powerful tool.

The following case studies are examples of how fleets are using operations research to solve complex challenges, helping them to become more efficient and profitable.

Southeastern Freight Lines
Using conventional accounting, executives at a less-than-truckload carrier easily could determine the best lifecycle strategy for tractors used in their local pickup-and-delivery operations. If they determined the lifecycle to be 10 years and 400,000 miles, their objective would be to average 40,000 miles per year for every tractor. This concept seems simple on paper, but making it a reality is a major challenge.

In October 2006, Southeastern Freight Lines organized a new process improvement team. One of the team’s first projects was to “even out” the miles on its P&D power units, says David Foster, vice president of maintenance for the LTL carrier. The Columbia, S.C.-based company operates 2,200 tractors of various makes, models, ages and mileages distributed among 70 terminals.

“From an optimization standpoint, we would prefer all units meet our mileage criteria for their age.” says Lem Sample, manager of equipment utilization. “In a perfect world, all units would mile out and age out at the same time.”

Sample assigns tractors to terminals after evaluating different metrics that include their age and mileage, the utilization rate (average miles/day) for tractors at each terminal, and the demand for power units at each terminal. Trucks in Miami, for example, average 69 miles per city trip, whereas in Amarillo, Texas, the average is 243 miles per trip. Demand for power units varies from one terminal to the next as freight volumes change and drivers come and go.

The constraints in the utilization problem include vehicle maintenance and the minimum number of trucks each terminal requires to meet its customer service standards. The fleet must maximize vehicle reliability while shuffling vehicles around to different terminals to balance utilization. A tractor might not be able to be immediately assigned to a different location because it is scheduled for preventive maintenance work, Foster says.

“If we had a low-mile truck that was due a major service, can we pull it and get it to a shop facility before we reassign it?” Sample says. “With 1,200 P&D trucks, that process gets very convoluted. If I’ve got an overpowered service center, I look at units that are due service. I may take a unit with the most recent service, or take a unit that needs service, and reassign that unit. Or, based on information from our maintenance field engineers, I might reassign a problematic unit to a shop location to improve reliability at the non-shop as well as reduce outside vendor expenses.”

This complicated problem needs to be solved daily, Sample says. Southeastern Freight Lines has assembled “reams” of data in spreadsheets and added mainframe computer support to automatically track the metrics involved in its decision making.

“We don’t have a nice statistical model, but the principles of operations research are inherent in the decision making,” Sample says. For example, the company uses a measurement system to identify tractors that are over or under mileage according to age (in months). A tractor that is 12 months old, for example, should have at least 40,000 miles. This metric – along with “remaining time-to-service” and the current demand for tractors at each terminal – forms the basis for moving equipment, Sample says.

Sample works closely with Foster and other members of the process improvement team to exchange the low-mileage units with high-mileage units across all facilities. As a result, the company is able to meet its goals for customer service and equipment reliability, utilization and resale value.

Pitt Ohio Express
In 2005, Pitt Ohio Express took an unconventional step for an LTL carrier. The company began to incorporate Sprinter vans into its P&D fleet of straight trucks and tractor-trailers. Two of the objectives for buying Sprinter vans were to increase efficiency for the delivery of smaller-size LTL shipments and to offer a career path for non-CDL drivers, says Jim Fields, chief operating officer of the Pittsburgh-based company.

Initially, the load planners did not know how to incorporate Sprinter vans into the daily schedules and routes, Fields says. So to maximize the value of the new vans, the company started an operations research project to determine which shipments to allocate to vans and how many to use for each terminal, says Steve Milcoff, the company’s director of account analysis.

The project began by acquiring shipment-level data for each terminal through Microsoft Access from various databases. Pitt Ohio’s operations research team also looked at the stop and drive times for each type of freight from the data that was available through the fleet’s onboard computers from PeopleNet.

“This was done for a sample of data, and then the results were analyzed using statistical software to identify the underlying trends,” Milcoff says. The trends in freight and route characteristics became the basis for the mathematical algorithms developed for the Excel-based planning tools that were distributed to Pitt Ohio terminals. Load planners at terminals use the tools to determine how many of each type of vehicle to use each day.

To develop the tools, the operations research team worked with the company’s data experts and operations personnel to account for the constraints and nuances not fully captured in hard data, Milcoff says. Examples include the time windows of delivery for each type of freight and special conditions where Sprinter vans could not be used, such as delivering freight on a pallet and backing up to a dock.

To use the Excel-based planning tools, load planners input the inbound bill count and tonnage information for the day. Planners confirm all the underlying assumptions and then initiate a nonlinear program solver to calculate the optimal equipment usage for the next-day outbound service, Milcoff says.

The following day, planners review a report to compare the actual equipment used with what the program determined to be optimal, and look for ways to improve the solution. Human intervention is needed to make adjustments as the day progresses, Fields says.

“(The planning tool) gives us guidelines as to how to start the day off,” Fields says. “As the day moves forward, things change.”

Since 2005, Pitt Ohio has added 47 Sprinters across its network of 22 terminals. The company also is in the process of using operations research to improve efficiency of its complex overnight linehaul system, and to optimize its pricing and profitability.

“A large part of each project is teaching the organization about operations research concepts and gaining buy-in with the methods,” Milcoff says. “The processes and subsequent continuous improvement efforts are really starting to take hold. As they become part of the daily operation of the terminals, operations research will have a strong foundation for identifying potential improvements and, most importantly, suggesting implementation strategies for turning theory into practical reality.”

DistTech
In the first quarter of 2006, DistTech – a dedicated contract carrier of bulk liquid commodities – launched a new offering to lower delivery costs for shippers. The offering, called LTL Bulk, was designed to incorporate numerous shippers into shared vessel space on the company’s multicompartment tank trailers.

DistTech’s trailers have segregated delivery systems to prevent products from being mixed while loading and unloading, says John Rakoczy, chief operating officer of the Newbury, Ohio-based fleet. DistTech therefore can deliver liquids in a dedicated “less-than-tankload” environment. Shippers that participate are charged by vessel space instead of the full cost of a dedicated delivery.

The objective for LTL Bulk is to minimize the delivery cost-per-gallon for a set of shipments by combining routes. To solve this problem, DistTech formed a marketing team consisting of one member from IT, two region operations managers (depending upon geography), one logistician and one safety professional. This team mines the current DistTech business base to penetrate new and existing shipper business units for opportunities.

The concept of LTL Bulk first must be sold to prospective customers at a high level within the organization, with a focus on value and capacity, Rakoczy says. After the sales process is in motion, the data behind the shipment information is gathered from shippers’ logistics groups. DistTech planners choose initial products and destinations based upon customer delivery date requirements. This information is obtained through an interface with customers’ enterprise resource planning (ERP) systems.

The data for the initial products and destinations then are stored in a proprietary DistTech application that finds the optimal miles-per-gallon of delivery for loads with compatible products, Rakoczy says. This application interfaces with the company’s ERP system, TMW Suite, to query for tractor/trailer unit information and positions, which allows for the right equipment to be selected for a given load.

Constraints for LTL Bulk include the lack of flexibility of shippers’ existing business rules to change P&D schedules to accommodate the service, Rakoczy says.

“Understandably, rules were developed to accommodate historic transportation methods,” he says. “The LTL Bulk concept encourages and perhaps mandates that a shipper review current practices for productivity gains.”

Additional constraints include shipper-specific needs such as sales cycle (month-end push and sales demands) and rerouting of units after departing a loading site to cover “emergency run-out” situations. The program DistTech developed is flexible enough to allow manual override to accommodate a specific customer need or a unique business condition, Rakoczy says.

A number of customers are participating in the LTL Bulk program, and Rakoczy expects it to continue to be a successful, unique offering for shippers of bulk chemicals.

Shaw Transport
As the private fleet for large carpet and flooring manufacturer Shaw Industries, Shaw Transport started an operations research project to decrease its net operating cost-per-mile. Asking its corporate board to increase its rates was not an option.

Shaw Transport operates dedicated truckload lanes out of North Georgia to regional distribution centers. Five regional supervisors have the responsibility to match backhauls for drivers to return to North Georgia. When loads are not available, supervisors often have to send drivers back empty to keep their trucks moving.

“That is a very expensive process,” says Randy Black, e-business manager for Shaw Transport. In 2007, Black and other fleet managers began a Six Sigma project to analyze revenue by lane to find opportunities to increase backhaul revenue from third-party shippers. Six Sigma – a data-driven approach to improving business processes – was first championed by Motorola and General Electric.

The metrics Shaw used to analyze its lanes included the percentage of overall revenue by lane, and the percentage of internal versus external revenue. The company looked at these metrics in a control chart over a six- to seven-week basis to analyze trends and changes.

“We are trying to avoid making point-to-point decisions, such as ‘where were we this time last year or this time last month?’ ” Black says. “You can’t see a good picture if you are not looking in the change of process over time.”

The analysis revealed, based on the number of trucks going into an area, where the best fit was for trucks coming back into North Georgia, Black says. Some lanes, for example, showed 60 percent of their revenue was internal revenue. These lanes were being used primarily to haul raw materials inbound to Georgia.

“In some of the lanes, we realized that we can’t do anything about it,” Black says. “(The reporting tools) allowed us to look in little small pieces. You become overwhelmed unless you can dissect and manage. Without these reporting tools, there is no way to judge the health of a lane as a whole.”

By talking to supervisors, Black and other managers also were able to understand what constraints were limiting the growth of backhaul revenue, such as arrival times to certain areas. By improving its route schedules to accommodate more backhaul opportunities, the fleet was able to increase its overall revenue by 25 percent.

“That was a very big win for us internally,” Black says. “Every dollar we generate is a reduction in net operating cost at the end of the month.”


Optimization for all sizes Software developers turning sophisticated tools into bite-size functionality
Traditionally, commercial software applications that incorporate operations research have been tailored to large fleets with significant information technology budgets. Today, these applications continue to become more affordable, with software vendors seeking to expand their products to smaller carriers.

Fleets that transport parcels and less-than-truckload shipments, for example, frequently use route optimization tools to determine pickup-and-delivery schedules and sequences.

Sanimax is a large firm that collects and processes used cooking oil and animal byproducts. The company has a Restaurant Services division that operates a small fleet of 27 vans and trailers that perform a range of P&D services in Ontario, Canada. The fleet averages 25 to 30 stops per day for 7,000 foodservice customers, including Burger King and McDonald’s.

The Sanimax fleet visits the majority of customers on a monthly basis, and its routes and schedules are fairly static. But when the company began to use the Descartes Routing and Planning Solution in September 2007, Stephan Campagna – transportation supervisor for the Restaurant Services division – realized how much he had been missing. In terms of hourly workers, the fleet averaged about 30 days of work a week; since using the tool to create new routes, the fleet cut back to 24 days of work per week.

By contrast, truckload carriers that use so-called optimization software systems must solve a very different set of problems, which include matching tractors and drivers to loads, and purchasing fuel at the lowest cost.

In September 2007, TMW Systems acquired IDSC, a developer of optimization software for truckload carriers. Since the acquisition, TMW Systems plans to accelerate its development of an offering for smaller carriers through an ASP platform, says Ben Murphy, vice president and general manager of optimization solutions.

“We will see scaled-down optimizations for 10- to 15-truck carriers,” Murphy says. Fleet planners will be able to select a number of trucks and loads from their dispatch software, select their goal – such as “minimize empty miles” or “maximize revenue per truck per day” – and then submit the information to an optimization engine through the Internet. The results will come back directly to their screen.

“Our intention is to continue to take these applications and algorithms and migrate toward the ASP platform to address a larger base of smaller carriers,” Murphy says. “People now are more focused on the type of freight they are hauling and how to minimize cost.”