Fleets can increase efficiency and reduce cost by using technology to tackle complex routing decisions.
To make pickups and deliveries, trucking companies arrange a set of orders and shipments into routes for a given set of equipment and drivers. Even the smallest fleets might have hundreds – perhaps even thousands – of routing options. With so many choices, mistakes are bound to happen.
Errors in planning routes impact customer service, operating costs, driver satisfaction and many other crucial measures. Because of the time and complexity involved in routing, many fleets use advanced technology to automate and optimize the process.
But route optimization systems aren’t always complete solutions. As with any automated system, important “human elements” can get lost in the solution. A computer might not know how long a customer takes to unload around lunchtime, or that a 70-mile route is 15 minutes faster than a 65-mile route, says Jeff Schuler of Estenson Logistics, a Tempe, Ariz.-based dedicated carriage provider operating 400 trucks.
“We’ve never been able to effectively use [route optimization software] in our day-to-day operations,” says Schuler, the company’s director of information technology. “But we’re still trying.”
Technology vendors, however, say state-of-the-art routing systems have the flexibility to adapt to any fleet’s particular set of business rules. In addition, these systems’ underlying digital maps and mileage databases have become extremely accurate. As a result, routing solutions are highly reliable when the rubber meets the road. Perhaps, but another potential obstacle remains: Convincing fleet planners to accept and act on the recommended solutions, some of which may require a leap of faith.
Myriad of solutions
Route optimization software is designed for two broad markets. One set of solutions assists fleets with multiple stops per vehicle each day, such as less-than-truckload carriers and metro delivery fleets. Another helps truckload carriers that route vehicles within a large geographical freight network.
Even within the individual segments, solutions vary according to the nature of the operation. Systems that optimize routes on a local scale determine the shortest or least-cost route for a combination of ZIP codes or street addresses on a truck-by-truck basis. For local optimization, most carriers use commercial mapping, mileage and routing systems from vendors such as ALK, Rand McNally, ProMiles, Prophesy, Maptuit and Tele Atlas. These systems, when calculating the shortest or least-cost routes, factor in constraints such as tolls and roads with truck and/or weight restrictions. They also integrate with fleet management systems to provide fleet managers with important decision support tools. Instantly, a dispatcher can see the revenue per mile, deadhead miles and other metrics for each possible tractor/load combination.
By contrast, global optimization solutions can plan least-cost routes for an entire fleet or subset of a fleet. Vendors such as Appian Logistics, UPS RoadNet, InterGis and Descartes offer commercial solutions for LTL carriers and operations with multi-stop pickups and deliveries. For example, an LTL carrier determines routes given a set of orders or shipments – each with unique locations, sizes (weight and volume), product types and time windows for pickup and delivery. The global routing systems consider these and many other types of constraints such as equipment capacity and driver availability to determine least-cost solutions.
Global route optimization systems in the truckload market automate the task of planning continuous moves based on a network profitability calculation known as yield. These systems consider the unique business rules and constraints of truckload operations such as hours-of-service and driver home time requirements. The dominant software vendors in this segment are Manhattan Associates and Integrated Decision Support Corp. (IDSC).
One of the struggles trucking operations have had with earlier generations of optimization systems was that managers and dispatchers frequently had to override the recommendations due to errors or limitations in the data those systems relied on.
For example, Penske Logistics in the late 1980s developed a proprietary system called Visual Route Assist. Over the years, Penske has enhanced the route planning capabilities to the point where it’s completely automated, says Tom McKenna, senior vice president of engineering and technology. In years past, experienced dispatchers often had to adjust routes due to flaws in data fueling the optimization system. For example, a dispatcher might see that a certain route travels through a construction zone or sends a driver into rush-hour traffic. Today, Visual Route Assist factors in road and traffic conditions from the start by relying on a digital map, mileage and traffic database from Tele Atlas that is more robust and accurate than any database the system has used in the past, McKenna says.
Tele Atlas draws from more than 50,000 data sources worldwide, such as various state departments of transportation, zoning boards, tax maps, satellite and aerial imagery and the U.S. Postal Service, says Dana Fenner, Tele Atlas’ director of fleet telematics. Tele Atlas supplements and verifies this information with field data collection using a fleet of local data collectors in cars and mobile mapping vans, Fenner says. It even uses feeds from GPS-enabled trucks to target its data collection efforts on geographies of interest to commercial fleets.
Penske Logistics also employs a proprietary solution called CellComm to increase the accuracy of its routes. CellComm is a software program that works on GPS-enabled cell phones to track and update routes in real time. The GPS data provides a baseline to compare actual routes versus planned routes, such as determining if a route took longer to run than planned. In the next route planning session, fleet managers can adjust the routes based on actual experience.
Even with all the advancements in route planning with better digital map data and GPS positioning, the tool works best in the hands of an experienced person, McKenna says. “We tell our dispatchers, ‘you’ve got to make sure the route is right,’ ” he says.
Today’s optimization systems are so robust that a fleet could continually change routing to maximize efficiency, but frequent changes to routes may result in dissatisfied drivers and customers.
True Serve Co. operates a 327-truck fleet that distributes retail goods to True Value hardware stores nationwide. Once a year, the Harvard, Ill.-based company uses a route optimization system from Appian Logistics to redesign delivery routes to its co-op membership of independent retailers, says Greg Swanson, transportation manager of operations.
The company looks for cost savings by consolidating routes as new members are added during the year or as members drop out. After Appian Logistics generates a solution, Swanson works with managers at the company’s 12 distribution centers to come up with transportation solutions that are acceptable to members.
“Our membership is resistant to change, so we limited it to once every year,” Swanson says. “We don’t do anything unless salespeople and members agree to it.”
Often, the software puts together routes that do not register in peoples’ minds until they actually accomplish it, Swanson says. Recently, the software found a way to eliminate a route in North and South Dakota, with no drop in the volume of shipments in that area.
“The transportation manager in that area said, ‘I think we can make that work,’ ” Swanson says. As a result, the company is saving 1,200 miles per week in that area. “That translates into big dollars.”
Whether businesses use route optimization daily or periodically, the return on investment from using these tools should be clear-cut. Reducing mileage or eliminating a route and a vehicle are obvious, hard cost savings. Yet in many instances, the routes these systems produce may not seem intuitive to fleet planners and dispatchers, who may feel inclined to outsmart the system.
Old Dominion Freight Line (ODFL), a less-than-truckload carrier based in Thomasville, N.C., has area supervisors that continuously follow up with freight planners at each terminal to make sure they are using routes calculated by the company’s Descartes Routing & Scheduling system.
“When we talk to planners in the terminal, we tell them to not ‘over-think’ it,” says Dennis Phelps, director of P&D operations. For example, planners may think they can create a better route than the computer solution by moving a few shipments to a different route. “They may be correct,” Phelps says. “They could make it a better solution between two routes, but not between 40. You have to look at it globally.”
Prior to using Descartes, ODFL used an in-house solution that created routes by placing pickups and deliveries into ZIP code “buckets.” But this practice sometimes meant freight destined for the same ZIP code was loaded into trailers in the wrong sequence for delivery. As a result, a driver would have to return to the terminal to back-strip some of the freight and resequence it.
By contrast, the solution from Descartes routes by street addresses, eliminating the “rework” of freight, reducing miles between stops, cutting fuel costs and improving productivity – all while accomplishing the same amount of work with fewer drivers, Phelps says.
Any routing software requires manual file maintenance and upkeep to run at optimum performance, Phelps says. But most supervisors now see the benefit in disciplining themselves to use the software – and to focus more time on executing the plan and managing exceptions.
“What used to take three hours a night, now takes 45 minutes to an hour,” Phelps says.
Truckload carriers that use route optimization software often face similar challenges in executing the software’s “optimal” recommendations.
Truckload carrier J.B. Hunt has used optimization software since 1992. The company recently decided to upgrade to the latest release of Manhattan Associates’ Carrier Management Solution. As part of the upgrade, management saw an opportunity to revisit its business rules and clean up many of its systems and processes, says Mark Drewry, senior engineering consultant.
The upgrade also provided an opportunity to increase the number of users that were trained on the solution and ensure that team members knew how to best leverage the tools. “The world is too complex for people to do it in their head,” Drewry says.
During training, Drewry spent a lot of time explaining the system’s recommendations to users. He also spent time listening to planners’ feedback on the recommendations, and tweaking the system so that everybody believed their input would affect the outcome.
“It’s more of an art than a science,” Drewry says, adding that the real success came by convincing fleet planners that the solution does all the “no-brainer” work.
“On any given day, a certain percentage of dispatches occur that make perfect sense considering cost, utilization and service,” Drewry says. “We decided to let the solution assign those trucks and loads, and allow the planners to focus their attention and efforts on exceptions that needed manual review or intervention.”
The fleet went from managers using 15 percent of the recommendations to 69 percent, and some use the recommendations constantly. Today, nearly 8 percent of J.B. Hunt’s over-the-road units are auto-preplanned, and since the upgrade, the company has seen a 3 percent improvement in on-time service and a 4 percent decrease in rework.
Too small to help?
For truckload carriers, technology vendors say that optimization solutions generally do not make financial sense for fleets with less than 250 trucks, or those that average less than 300 miles per shipment. By contrast, because of the complexity of planning routes for fleets with multiple deliveries per vehicle each day, some fleets with as few as 10 vehicles can benefit from route optimization technology.
In July 2004, Berger Brothers – a manufacturer of commercial and residential drainage products – began a project to reduce its shipping and warehousing costs. The Feasterville, Pa.-based company uses a 12-truck private fleet for local deliveries (250 miles roundtrip) and LTL carriers for longer-distance shipments, which range between five and 15 trucks per day. One of management’s first steps was implementing Appian Logistics’ solution to build better schedules and routes for its shipments. The company also uses Appian to build loading and delivery sequences for the LTL shipments it outsources.
To date, routing savings at Berger Brothers have been substantial, says Richard Falconio, the company’s director of technology.
“(Appian) allowed us to build better trucks that were complete and full, and give us more of a direct route, rather than a manually organized route,” Falconio says. “It also reduces the time that our shipping people would spend routing trucks. They can now route a truck in two minutes.”
Routing is the most critical operational decision. Considering the cost of man-made errors, technology that automates at least some planning decisions can be a worthwhile investment.
Routing solutions weigh complexities that humans can’t
Each route optimization system has unique features, but they all use similar algorithms and procedures to produce an “optimal” solution. Steve Brown, president of InterGis, explains the key elements as they relate to his company’s core routing product, Visual Control Room. The system is ideal for fleets with vehicles that make more than five stops per day, Brown says.
To solve the complex routing problems of such fleets, the tool uses two components: a core optimizer and a business rule optimizer. The core optimizer solves a minimization problem. Typically, the variable you are trying to minimize is total cost, Brown says. The program adds the cost variables of each possible combination of vehicles and orders, including pickups and deliveries. Cost variables can be calculated in a variety of ways, such as on a per-mile basis for each vehicle. The system then converges on the set of order and vehicle combinations that has the lowest total cost.
Simultaneously, the second component – the business rule optimizer – is running while the core optimizer searches for the least-cost solution. This component contains the set of all constraints and business rules, such as equipment capacity and time windows for pickups and deliveries. Users can turn these rules on or off, emphasize or de-emphasize, or define their own business rules. The two components work together to converge on the least-cost solution that is feasible.
Using these programming techniques, route optimization software can solve extremely complex problems almost instantaneously. James Stevenson, executive vice president of Appian Logistics, says the company’s routing solutions can, in two minutes, solve a routing problem with 200 stops, including pickups or deliveries.
“That is extremely fast from a computation standpoint,” Stevenson says. Speed in route planning is especially crucial for businesses that offer next-day delivery, such as a building materials supplier. If the business sets an order cutoff at 5 p.m., it could have all orders arranged into routes before its evening work crew arrives at 6 p.m. to load trucks, he says.
Another complex problem the software can solve quickly is managing the imbalance between a fleet’s outbound and inbound shipments. For example, LTL carriers deliver in the morning and pick up in the afternoon. Originally, there may be more scheduled deliveries than scheduled pickups, which might arrive throughout the workday. Fleet managers have to determine how to route their deliveries so trucks can end up at a strategic location to pick up new orders.
Routing systems can plan delivery routes automatically based on historic pickup volumes, Stevenson says. Without this level of planning, a dispatcher may only send out 30 trucks to make deliveries without considering that he needs 37 trucks out in the field for pickups. Without the software, fleets have no ability to determine the cost of such actions, compared to what their cost savings could be by routing differently, he says.