Reality trumps perfection

Just 40 miles from Barcelona is automotive manufacturer Trety, a maker of textile-based vehicle interior components such as headrests and seat fabric for some of Europe's largest automobile manufacturers—including PSA Peugeot Citroën, Renault, and Volkswagen.

Scattered across Spain, Portugal, and the North African countries of Tunisia and Morocco, Trety's nine manufacturing plants form an integrated supply chain. Capital-intensive operations—such as applying foam backing to fabrics, or cutting fabric into "kits" of parts—are centralized, while other plants exploit the lower wage rates prevailing in their local markets to focus on labor-intensive operations such as sewing.

But it's an integrated supply chain with two critical imbalances. First, Trety's customers give just three or four days' notice of their order requirements. Including transportation, Trety's production time is longer than this, leading to inventory holdings of partly finished production within the chain of plants. Worse, the textile manufacturers from which Trety sources fabrics require lead times of weeks or even months—leading to substantial raw material inventory holdings as a buffer between the two planning horizons.

In multiplant supply chain planning scenarios, such imbalances aren't unusual. But as manufacturers strive to meet short lead times with ever-slimmer inventories, supply chain visibility is moving center stage as a way of dealing with it. Advance warning of potential problems can make the difference between success and failure—provided, of course, that manufacturers have the tools to react to those warnings.

For Trety, the challenge lay in using supply chain visibility to balance demand, capacity, work-in-progress, and raw material inventory holdings. When orders arrived from vehicle manufacturers—or their Tier 1 automotive suppliers—often the manufacturing plants that appeared most appropriate from a capacity perspective actually lacked adequate stock of the fabric or partly finished products required to complete the order. This necessitated costly shipments of fabric between plants, or diverted inbound shipments of raw materials.

Mix and volume changes added another unwelcome layer of complexity. A forecast demand for 1,000 headrests evenly distributed across five colors, for example, could easily turn out to be a single requirement for 500 headrests in just one color.

The solution eventually adopted by Trety to combat these challenges involved enhanced supply chain visibility coupled to—and integrated with—a multiplant planning solution from Preactor International. Twice a day, explains Alberto Cárceles, Trety's director of organization and systems, detailed information on material holdings, work-in-progress, machine capacity, and order status from each plant is transmitted to a homegrown ERP system developed by French parent company Trèves Group. From here it is passed to the server-based Preactor application, which updates the schedules for each plant, allocating new incoming orders as appropriate—based not just on capacity, but on fabric holdings and fabric location.

The impact has been significant, explains Cárceles. For a start, there's been a huge reduction in the movement of fabric between plants. Trety also cites better scheduling that allows clearly identifying the impact of unforeseen demand variations from the original customer forecast, and prompting at times extra payments from customers in recognition of higher transport costs and overtime and additional shifts.

Prior to implementing Preactor, says Cárceles, high levels of work-in-progress between different processes and plants buffered against the inefficiencies and slow reaction times that might impede the company's ability to react to sudden changes in demand. Working in a more integrated manner, it has been possible to reduce inventory holdings of partly finished production by 12 percent. What's more, the ability to generate a single production plan for the entire multiplant supply chain has improved customer service levels by 17 percent. Overall, notes Cárceles, the move to Preactor paid for itself in just a few short weeks.

Not surprisingly, explains Mike Novels, managing director at Preactor, other manufacturers want to emulate Trety's results.

"As a 'rules-based' scheduling tool, Preactor allows creating scheduling rules to take account of whatever objectives you want to meet," says Novels. "Plant-specific factors such as capacity or tooling constraints are traditionally used when scheduling plants, but they're not the only game in town. Supply chain and inventory constraints also can be very relevant."

Nor is Preactor itself the only game in town. Milpitas, Calif.-based Solectron Corp., for example, makes extensive use of a supply chain visibility capability from Kinaxis to better serve the needs of customers such as North Reading, Mass.-based Teradyne, the world's largest designer and manufacturer of automatic test equipment for the semiconductor, electronics, and automotive industries, and for which Solectron supplies contract electronics manufacturing services following Teradyne's decision in 2000 to outsource manufacturing.

The Kinaxis supply chain visibility capability in question, called RapidResponse, is dubbed a "glass pipeline" by Kinaxis CTO Dave Haskins. "When a company outsources manufacturing to an electronics manufacturing services company, it loses visibility to some of the supply chain links that it had before—such as purchase commitments that Tier one suppliers are placing on their Tier two suppliers, for example. As a result, the response time to answer queries from customers can deteriorate, because—compared to when manufacturing was in-house—it takes much longer to get answers back from the contract manufacturer. The RapidResponse pipeline restores that visibility," he says.

Essentially an ERP-agnostic series of data feeds from along the supply chain—drawing together information such as inventory levels, work-in-progress, forecasts, outstanding orders, and backlogs—the collaborative relationship that RapidResponse enables between Solectron and its customers is a close one.

According to Beth McKone, Solectron VP of worldwide site program management, "The information flow takes place in a moment. We're using an application to create a virtual enterprise that combines aspects of the customer and ourselves."

Take a scenario in which a supplier has an interruption in production, McKone adds—e.g., a fire or a major machine breakdown. "I can run a simulation, as can Teradyne, allowing us to predict what might happen. Together we can make decisions about what we're going to do about the delays," she says. "With both of us sharing the information about demand and supply, we're seeing a bigger picture—and making better decisions based on that picture."

Previously, asserts Robert Kenney, a supply chain manager for Teradyne, it took up to four days to aggregate incoming supply chain information—sometimes leading to erroneous assumptions.

"Because of that delay, we were making the wrong decisions. We had inventory out of position within the pipeline. We couldn't see it and we couldn't move it, which caused tremendous delays to our customers." Now, just as Solectron can respond more quickly to demand changes from Teradyne, so too can Teradyne respond more quickly to its customers—in just a day, in fact, a reduction from up to 14 days beforehand.

The greater the scope of the solution—in terms of the customer-supplier linkages that Solectron embraces—the greater the resilience to supply chain shocks. To combat unforeseen component shortages, notes Kathleen Ward, Solectron's director of demand management, Solectron can divert supplies of components from one plant—and one customer contract—to another plant and another customer contract if necessary. It's not a frequent occurrence, but one that is occasionally very useful.

"Of course, we do an analysis to determine the impact of shifting those materials from the originally intended customer, and make sure we have conversations with the originally intended customer—as well as the supplier—before anything happens," Ward stresses.

Omaha, Neb.-based ConAgra Foods also makes use of a multiechelon inventory optimization tool from SmartOps to set the policies governing inventory levels in the company's downstream stocking locations—numbering in total more than 80.

The supply chain visibility itself comes from ConAgra's enterprisewide SAP system, which embraces the plants, warehouses, and distribution centers. But running alongside SAP's Advanced Planning and Optimization engine is the specialist SmartOps tool, supplementing SAP's own take on what inventory to hold, and where.

Partly, explains Bob Masching, ConAgra's VP of sales & operations planning, the objective is to reduce the overall amount of inventory in the business. ConAgra Chief Executive Gary Rodkin has made a public commitment to investors to reduce working capital over a three-year horizon, and the investment in the SmartOps tool is seen as underpinning that objective.

But alongside the working capital objective is a determination to create a downstream supply chain that is leaner and more nimble than the one in place within the business today—an emphasis not just on less inventory overall, but on having the right inventory in the right location at the right time.

"With inventory targets that are continually tuned to buffer us against uncertainty—recognizing the multiple lead times and multiple nodes within our physical supply chain—we expect to increase our customer-service levels even as we reduce the working capital throughout the network," says Masching. "The inventory targets derived from SmartOps allow us to better respond to the supply and demand variability that we experience."

Better still, he adds, by identifying the real lead times within the business—and the real levels of variability and uncertainty that are experienced—insight gained from SmartOps can drive improvement plans, allowing ConAgra to eliminate adverse events in its supply chain altogether.

"The challenge of inventory optimization is that you're not setting targets for the perfect world—you're setting targets for the real world," says Martin Barman, a senior VP for SmartOps. "The data you collect from each stage in the supply chain helps you model, understand, and ultimately deal with that real world."