Suddenly spin control
Mastering demand management in the supply chain can require sophisticated forecasting, inventory tools
By Malcolm Wheatley, senior contributing editor -- Manufacturing Business Technology, 6/1/2006 6:00:00 AM
In the spring, a young man's fancy turns to—well, fixing his pickup truck. And as the nation's driveways and sidewalks get warmer, do-it-yourselfers of all kinds decide it's time to tackle those little maintenance chores they put off all winter.
For Elgin, Ill.-based SKF Vehicle Service Market (SKF VSM), the aftermarket parts division of Swedish multinational SKF, all this activity means a sudden uptick in demand for the bearings, seals, U-joints, and other automotive products it sells.
The challenge, explains Matthew Schiele, vehicle service market supply chain manager at SKF VSM, lies in meeting demand both cost-effectively and reliably—and in the face of difficulties imposed by upstream and downstream supply chains. Approximately 70 percent of the 60,000 unique parts SKF VSM sells to distributors and retailers move downstream through its six North American distribution centers (DC), which exhibit either intermittent or 'slow-moving' demand—or both. Worse, not only do stock-outs at centers frequently lead to lost sales, but certain DCs experience what Schiele describes as "will-call" demand: If the product's in stock, a customer will call to collect it.
Life is complex upstream, too. SKF VSM's central operations facility in Hebron, Ky., receives bulk dispatches of products from manufacturing facilities around the world. Each shipment must be broken down, repackaged, and placed in inventory awaiting dispatch to a DC. Supplier lead times, though, can be as long as 26 weeks for some product lines—leaving the business critically dependent upon accurate forecasts to meet demand.
Accurate forecasts were exactly what were missing, claims Schiele, when he took over the aftermarket supply chain role in 2003. A homegrown forecasting system struggled to accurately reflect seasonal demand variations.
"The reaction time was very poor," Schiele recalls. "With an upturn or downturn, it took the system some time to recognize that demand had changed." What's more, he adds, the exponential smoothing algorithms on which the system was based were inappropriate for items with intermittent demand.
Recognizing these limitations, manual workarounds by Schiele's team had become the order of the day, with demand planners spending a lot of time adjusting the forecasts that drove target inventory levels. Yet it was impossible to do this for all 60,000 SKUs every month, leaving the stocking decisions on huge numbers of products based on incorrect forecasts. Packaging operations that were idle during the slow winter months had to scramble through the summer to meet demand and fill the back orders.
Smart tooling
The solution lay in the SmartForecasts package from Smart Software. In addition to improved techniques for dealing with products with regular demand, the system has patented algorithms for forecasting intermittent demand.
After talking to a Smart Software customer using SmartForecasts to tackle similar problems, Schiele arranged an online demonstration using some of SKF VSM's actual data.
"I had my managers sitting in, and we were very impressed," he recalls. Yet the discussions brought to light the fact that SKF VSM was already a Smart Software customer. It had licensed—but ceased using—an early version of the software more than a decade prior.
Once implemented, SmartForecasts proved straightforward to integrate, says Schiele, with the programming work to connect the software to SKF VSM's distributed requirements planning (DRP) application taking just two weeks. Raw data from which to calculate forecasts in batch mode is downloaded, with the resulting forecasts uploaded back to the DRP system to initiate order replenishment and inventory management.
The impact of the new software was immediate. To maintain a target service level of 95 percent, inventory levels had been rising in lockstep with sales, explains Schiele. For example, during the year-to-date prior to implementing the system, inventories had risen 6 percent. SmartForecasts effectively broke the linkage: during 2004, inventory levels fell 3 percent.
"During 2004, we were using the Smart system to 'run lean,' reordering less, and running down inventory levels," says Schiele.
The full benefit was seen in 2005. With service levels still being maintained at a consistent 95 percent, SKF VSM reduced inventory holdings by 16 percent. What's more, use of the Smart Software-generated forecasts improved communications with SKF VSM's major suppliers.
"We're able to give them a better picture of our requirements, and they are able to plan their production accordingly," reports Schiele. "As a result, we're seeing improved delivery performance."
Internal efficiencies have improved, too. "There's been a big change in warehouse operations," he adds. "In the past, we were never ready for the spring rush. Now, instead of being 200,000 units behind sales in our packaging operations, we're 200,000 units ahead."
Acquired challenges
Pittsburgh-based food giant HJ Heinz is another business successfully deploying a demand-management solution to wrest improvements from its supply chain. Its European operations have been built up through an ongoing process of acquisition. As a result, the company operates with disparate ERP systems.
Its manufacturing strategy, notes European demand-planning manager Mike Bonnici, calls for Heinz to concentrate product manufacture at specific sites—with ketchup for the whole of Europe, say, being produced at a single location. The challenge, he explains, is to feed that production site with timely demand forecasts from each country.
One immediate difficulty is that the sales histories on which those forecasts will be based reside on different ERP systems, while approaches to forecasting are equally varied. "The Germans work in tonnes, the Italians bottles, while the U.K. works in packs of a dozen. What the factory needs is a single unit of measure," explains Bonnici.
A collaborative forecasting tool—Mercia Linked Enterprise from Mercia Software, now part of Infor—offers the solution. Information is abstracted from national systems, forecasts derived, and the output fed into production-planning systems to drive manufacturing as well as financial systems.
The result has been a transformation of the forecasting process, says Bonnici.
"The quality of the game has been raised," he says. "Not only have we been able to statistically eliminate forecast bias within national markets—in other words, automatically correct a consistent tendency to forecast high, or forecast low—but the attitude toward forecasting has changed. Instead of people regarding forecasts as a capacity reservation, it's become a genuine estimate of what they think they are going to sell."
Heinz Procurement Director Rob Hemsley concurs. "There's no doubt we're seeing smoother and more predictable purchasing patterns," he notes.
ERP's value-add
"Demand management is evolving," notes Simon Bragg, research director at Dedham, Mass.-based ARC Advisory Group. "What was once essentially forecasting is now tinged with aspects of collaboration, alerts, and real-time order capture."
In particular, adds Bragg, businesses are recognizing the role demand management can play in multi-echelon inventory management—especially in supply chains dogged by the 'bullwhip effect' and other essentially self-induced distortions.
Pittsburgh-based health-care giant Bayer Corp., for example, uses SmartOps' Multistage Inventory Management Planning and Optimization application for just such a purpose.
"Without a common demand signal across every tier of the supply chain, each tier must reforecast demand based on orders from its immediate downstream counterparts," says Arthur Brohinsky, senior VP of strategic business development at SmartOps. "It's very easy for this to trigger a bullwhip effect."
According to Robert Rudski, president of Pittsburgh-based Greybeard Advisors, "SmartOps is a clear example of a technology adding value to an ERP backbone." Rudski, as Bayer's chief procurement officer, also was closely involved in the evaluation process that led to SmartOps. "It's a solution that has its roots in advanced academic research, turned into software," he says.
Once, of course, much the same thing could have been said of the exponential smoothing algorithms that were junked by SKF. But then, that's progress for you.
Plan for demand
02/01/2003Solutions Spotlight
01/31/2004High-tech Darwinism
11/01/2003
