Networking Opportunity: The low cost/best service challenge

May 1, 2006
Is it time to redesign your distribution network

The low cost/best service challenge

IT'S not a question of "If," but a question of "When?" Rising fuel and transportation costs will have an impact on the structure of your distribution network.

"A year ago, when [costs] started to climb, people were looking at it as a blip. Now there's a mindset that the increases in transportation and fuel costs are here to stay," says John Sidell, a principal with ESYNC, a supply-chain consulting firm based in Toledo, Ohio.

Other factors having an impact on distribution network design are globalization and import growth, which is pulling some inbound and even outbound facilities nearer to the nation's ports. Companies continue to build greater flexibility into their supply and distribution networks by outsourcing more activities to logistics service providers. More organizations are adding flow-through or cross-docking operations to speed material flow, and they are also working to more effectively combine their facility location and inventory allocation strategies.

Still, where a company locates its distribution centers has always been about transportation. Today's cost increases only add fuel to the fire, forcing supply-chain planners to revisit the calculations that balance transportation costs against facility operating costs and inventory carrying costs.

"It's not one size fits all. There are certain industries that use rail or other inter-modal methods that reduces some of the fuel charges," says Sidell. "But certainly there are some companies that will revaluate [their network] and find that a satellite facility in a more densely populated area will save them dramatically on transportation while reducing overall costs."

The underlying point is that any design has a limited lifespan. As assumptions, market factors and customer expectations change, a distribution network has to change. Managers at Gillette (Boston) refer to this continual process of optimization as "a state of constructive dissatisfaction," reports Maggie Walenty, an inventory modeling analyst for the company, which was acquired last year by Proctor & Gamble (Cincinnati).

Following this philosophy, starting in 1999 Gillette began a series of initiatives to reconfigure its distribution network. For its blades and razors business, the company has relatively few, highly automated, high-volume manufacturing facilities scattered around the world. As part of a broad postponement strategy, it established flexible pack centers in different regions to customize product packaging for local markets. To optimize transportation costs, it also began to co-locate products from multiple divisions in the same DCs, which includes everything from deodorant (Right Guard and Soft & Dri) and toothbrushes (Oral-B) to batteries (Dura-cell) and electric appliances (Braun).

"This distribution strategy provided pretty significant cost savings for us. Full truckload utilization, which would be an indicator of distribution costs, increased substantially, without compromising responsiveness," says Walenty. Her team's responsibilities include determining how a distribution network for a particular product line should be setup, as well as the required safety stock to achieve the desired customer service levels. To help with such tasks they use a combination of distribution network design software from Insight (Manassas, Va.) and inventory optimization software from Optiant (Burlington, Mass.).

"Based on a few years of working with the tools, within the Gillette business we came to the conclusion that iterating between the two really gave you the best answer," says Walenty. The modeling process itself can get complicated but it follows a basic pattern.

"You start out with the business question that you're going to answer. The next thing you do is go out and get the data that's going to support that particular analysis. Then you construct your models and make sure that those models represent the business environment that you're trying to analyze," Walenty explains. Once the models are constructed the analysis begins to determine how changes in inputs—a 20% increase in transportation, for exam-ple—changes the recommendations.

"A good network design project will highlight breakpoints," she adds. "Points where a particular network design is no longer optimal from a cost viewpoint." These breakpoints will determine how often managers have to revisit their network design.

"Every sensitivity analysis is going to reveal the boundaries within which your recommendation is a good recommendation. If those boundaries are large boundaries, then that drives you in one direction. If those boundaries are very tight, you might be back in to review that model a lot more frequently," says Walenty.

A poor model, Walenty explains, will not factor in total landed costs—which includes labor, facilities, taxes, insurance, inventory and transportation— and could lead to bad decisions. A good network optimization model will help managers understand how different factors impact the ideal network, balancing the need to minimize transportation costs against supply-chain reliability requirements. A skilled analyst understands what variables are more significant than others from a strategic standpoint. He or she will know how to translate different cost units—a 40-foot ocean container vs. a per pound truck rate, for example—and specify the necessary level of detail for the software to spit out useable answers.

The algorithms used by the optimization tools tend to provide extreme solutions, cautions Dr. Jeffrey Camm from the University of Cincinnati's Quantitative Analysis and Operations Management Department. Supply-chain planners have to be aware of the risks associated with solutions that offer the absolute best costs and service levels.

"You may have this globally optimal solution that is extremely risky if you look at perturbations in demand," says Camm, which is why he runs a lot of scenarios when he's doing this type of work. "If you back off to a solution that is within 5% of [optimal], that is very robust with regard to perturbations in demand, to my way of thinking you should go with the 5% solution."

In the recent past the challenge wasn't so much the capability of the optimization tools, but the availability of good data, recalls Camm. Now everyone has a lot of data, but it still takes some effort and time to normalize and cleanse it. One hole that often remains, says Camm, is the availability of data on locations where a company does not currently have a facility.

"Some companies seem to be much better at getting data for locations where they currently don't exist than others," says Camm. Logistics partners and suppliers can offer some help. "I think it's a matter of effort. Some companies are willing to spend a lot of effort going to a new location, digging down and taking the time to see what real costs would be there. Some just make a phone call to the Chamber of Commerce," he adds.

Historically some of the network design tools have not factored in enough information about the transition and clean-up costs associated with closing facilities, says Camm. What companies have to do for displaced workers can vary dramatically from country to country. Such details need to be included in any modeling project, as does information about where certain fixed assets are located, and which leases may or may not be up for renewal.

From Proposal to Reality
After the data has been crunched and the analysis has been done, if the ideal distribution network differs radically from a company's current network, managers need to establish a transition plan. The models have to be rerun with portions of the existing system "locked in" to identify a manageable series of changes that will eventually lead to the ideal network after several years.

"Nobody just shuts down everything and starts over, or closes a significant number of facilities all at once," says Camm. "You need some sort of roadmap in going from where you are to where you want to be."

When it comes to making the actual transition, companies need to allocate adequate resources, even if they are reducing the overall number of facilities in their distribution network.

"You have to build the infrastructure inside of the company to support that change. For a while you need more operators, you need more IT staff," says Patrick Sedlak, a principal with Sedlak (Highland Hills, Ohio). People need to be pulled out of their current jobs to focus on the development of new facilities.

"It's challenging to run your current network and develop a new one at the same time," he explains. In addition to dedicating enough resources, Sedlak advises clients to give themselves enough time to evaluate all of the options for both the network modeling process and the implementation phase. Rapidly growing companies often have the biggest challenge because any changes to their distribution network need to support their growth plans, however indeterminate those plans may be.

"It seems elementary, but it's hard for firms to predict where they're going to grow next and be able to sustain the growth, or commit to a growth plan," says Sedlak. "Companies tend to determine the best location for their DCs after they see how growth has changed their business."

Looking Beyond Optimal Locations
Once the facility location decisions have been made, the analytical work isn't done. Following the lead of the inventory management team at Gillette, supply-chain planners need to narrow their focus from aggregate levels and product families and work on figuring out how much of each SKU should be in which locations to avoid shortages and lost sales.

"People don't take that next step after they figure out how big their building should be and where they should be located," says ESYNC's Sidell. "Understanding what SKU mix you have, what the demand is for those SKUs by region, and then carrying the appropriate amount of stock reach for each of those items is a real important follow-on strategy to doing a network optimization."

In addition to maximizing inventory turns, such efforts will minimize transportation spend on internal transfers. After all, the whole point of optimizing a distribution network is to maximize location benefits, and not pay to move product back and forth between facilities on opposite sides of the country.

Network Optimization Methodology

1. Project Foundation.

  • Determine the project scope, modeling strategies and data sources.

2. Data Collection

  • Network—definition of plants, DC's, customers, product lines.
  • Facilities—Manufacturing and DC costs, capacities, bill of materials, run rates.
  • Transportation—Costs and customer service limits
  • Demand.

3. Baseline Validation and Model Development

  • Validate current costs (base-line) for savings comparison during the optimization analysis phase.
  • Import database and build tables using network modeling tools.

4. Scenario Development and Optimization Analysis

  • Based on data collected, analyze several scenario models to determine optimization points (for example, adding or closing facilities, varying service policies, planning for growth)
  • Conduct "What If?" analyses.

5. Recommendations

  • Present recommendations that optimize inventory levels, transportation costs and customer service.
  • Develop tactical "project roadmap" for deployment of recommendations.

Source: ESYNC (Toledo, Ohio)

Distribution Network Redesign: Is It Time?

Here are five major reasons why companies embark on a distribution network redesign.

1. Mergers and acquisitions. When asked what triggers a network redesign, the immediate response of Dr. Jeffrey Camm from the University of Cincinnati's Quantitative Analysis and Operations Management Department, is that "it's almost always an acquisition. One company buys another, and you have two supply chains." In fact, the potential "synergies" from combined operations are a big factor supporting the rationale behind many acquisitions.

2. A matter of policy. "We have companies today that model every change to their supply chain in our tool before they execute on it," says Eric Nilsson, senior director of solutions management for SSA Global (Chicago). SSA's customers, who are primarily 3PL suppliers and retailers, use its supply-chain design tools (formerly known as CAPS) to model all lane and supplier changes.

"What they're able to do is determine their costs very accurately of those changes before they ever make them, even if they think they know what it is," says Nilsson. Other SSA customers use SSA's design tools to review their network every quarter or every six months.

3. A new product introduction. When a company launches a whole new product line, with a whole new group of customers that need to be served, managers will need to setup a distribution network that best serves the product line and the new customers.

4. Because it's time. "If things don't change, if a company grows 5% over the next three years, makes no additional acquisitions and they don't change their product line, chances are the network they put in three years ago is still pretty effective and efficient for them," says John Sidell, a principal with ESYNC, a supply-chain consulting firm based in Toledo, Ohio. But how many companies are that stable these days?

Sidell tells the story of a recent customer that hadn't taken a close look at its distribution network for eight or nine years. Following the appropriate analysis, they determined that customer requirements had changed significantly, as had their domestic product sources. By closing two out of six facilities and adding a cross-dock operation, they saved $1.5 million per year, with no change in service levels.

5. Sourcing changes. A company just starting to source product from Asia might not have needed a West Coast facility in the past because it didn't have customers in the region. Because of the product flow out of the ports, managers may find that it makes economic sense to put a facility out-side of Los Angeles, or in Reno or Las Vegas, where the inventory taxes are less onerous.

Latest from Technology & Automation