Keeping the Promise

When a customer places an order, a manufacturer checks inventory to see what’s available and calculates when to promise to deliver the goods.

In the semiconductor industry, though, it’s common to promise fulfillment dates based on product that’s in planning or production as well as on what’s in stock, says Gerry Dundon, vice president of planning and logistics at Analog Devices.

That’s because customer lead times can vary greatly. One customer may need a shipment in one or two weeks, while another asks for delivery three months off.

“We calculate ATP (available to promise) over time, not only against inventory,” Dundon says.

Based in Norwood, Mass., Analog Devices makes integrated circuits used for signal processing in entertainment systems, medical devices, cell phones, digital cameras, and other products. It operates wafer fabrication plants in the United States and Ireland, a wholly-owned assembly and testing facility in the Philippines, and relationships with numerous contract manufacturers.

Along with the challenge of calculating ATP over time, Analog faces other issues that stem from variable supply and volatile demand.

Sometimes, having predicted that 85 to 90 percent of the units in a product lot will pass final testing, the company finds that only 70 percent of the product is usable. On the demand side, there are periods when many customers place orders, then cancel or postpone them, making that product available for someone else.

Intelligent Scheduling

“In an ATP environment you need to be able to intelligently schedule,” Dundon says. “Equally, you need to be able to intelligently reschedule order backlogs as certain factors change the supply or demand side of the equation,” Dundon says.

Another challenge arises when the economy grows stronger and production can barely keep up with demand. In those times, Analog wants to set business rules for pre-allocating product among customers. Rather than fill orders first-come, first-served—leaving some customers empty-handed—the goal is to satisfy all customers at least in part and fill out the rest of their orders over time.

Analog is an SAP shop. Before 2004, it used the order scheduling and rescheduling capabilities within SAP’s R3 enterprise resource planning (ERP) system to calculate ATP. The software functioned well, but wasn’t flexible enough to deal with varying supply and demand or to pre-allocate inventory, Dundon says.

In 2003, Analog was contacted by Adexa, a Los Angeles-based software provider whose Enterprise Global Planning System (eGPS) includes 13 supply chain management modules for manufacturers.

“I was impressed right off the bat with Adexa’s knowledge of how the semiconductor industry works in general. They definitely understood the types of challenges we were trying to overcome,” says Brad Martin, Analog’s planning systems development manager.

Building an ATP Solution

Once they heard about Analog’s ATP challenges, developers at Adexa offered to use portions of the eGPS suite to build an ATP solution. Closely integrated with SAP’s ERP system, the new solution would provide the functionality and flexibility the manufacturer was looking for.

Analog and Adexa started working together in late 2003; by early 2004 Analog was using a prototype of the system to develop an allocation plan. Other adjustments followed, but “the system has been fairly stable since late 2004,” Martin says.

Adexa recently launched a version of Analog’s solution to the general market, calling it Precision ATP.

A major goal of Precision ATP is to automatically calculate when a manufacturer can promise a product to a customer, based on all product in its manufacturing pipeline and inventories.

“Manufacturers want to do this calculation very quickly, preferably in seconds,” says Kameron Hadavi, vice president of strategic marketing and alliances at Adexa. “If a manufacturer tells a customer it will take three days to respond, the customer will go to somebody else who can fulfill the order. And if the manufacturer plucks a fulfillment date out of the air, then fails to meet it, the customer will not come back.”

Many semiconductor manufacturers schedule fulfillment dates manually. “They’ll go to a spreadsheet that is the output from their planning system, and also where their inventory is. They’ll make their best guess and say, ‘there’s the scheduled date,'” Hadavi says.

When several customer service reps work from separate printouts, conflicts crop up easily. “You never know if something you’re ordering now has been promised by somebody else right next door,” says Hadavi.

At Analog, about half the orders come in via electronic data interchange, about five percent via the web, and the rest through phone, fax, or e-mail to a customer service rep. No matter how they come in, orders are entered or transmitted into R3.

The SAP system holds data on the company’s inventory, and receives data on incoming supply from Analog’s manufacturing system. All that information is available to Adexa, which runs on a separate server.

“When an order comes in, a live call is made from SAP to Adexa,” explains Elizabeth Derwin, demand logistics manager at Analog. “Then, a schedule comes back, and the order is saved.”

At any given time, Analog is working with approximately 50,000 customer requests for specific products on specific dates. “More than 95 percent of those requests will be scheduled automatically by the system, without anybody looking at or touching the order,” Dundon says.

For the remaining five percent, Analog doesn’t have enough product in the pipeline to fill the order. Rather than tell R3 to promise what is currently available and follow up with more pieces over the next few weeks, Adexa puts the request on hold for 24 hours. During that time, R3 reports the surge in demand to Analog’s planning system to determine if the company can speed production and fill the entire order sooner.

If so, “Adexa automatically schedules from that,” Dundon says.

As supply and demand fluctuate, Adexa reschedules orders according to a consistent set of business rules. It does this in a more sophisticated way than R3, according to Dundon. Adexa uses two target delivery dates: the customer request date (CRD) and, if Analog can’t promise to meet that request, the realistic commitment date.

If supply increases, Adexa performs “a kind of two-path rescheduling, where it tries to match all the commitments made to customers and looks for surplus supply that can be pulled into certain lines to match the CRD,” Dundon explains.

If supply worsens, the system pushes back delivery dates according to established rules, including different priorities assigned to different customers.

Smart Allocation

In times of tight supply, Adexa uses allocation rules to decide how to divide products among different customer categories.

“For example, a particular product might have a rule that no more than 50 percent of the supply can, at any point in time, be ordered by the distribution channel (distributors, rather than end-users),” Dundon says.

Though SAP’s R3 is essential for managing many important functions throughout Analog’s operations, for the specific task of calculating ATP, a specialized application running off to the side has proven more effective, Dundon says.

Precision ATP is part of a new strategy at Adexa to tailor solutions for specific needs, rather than simply market the full supply chain planning suite to every customer. Over the past two years, customers have developed a sharper focus on the problems they face in their businesses, Hadavi explains.

“Customers say, ‘This is my pain; tell me how to relieve it.’ They don’t want to hear that if they purchase a whole suite, the solution not only takes care of their specific pain, it will take care of everything else as well,” he says.

Along with Precision ATP, Adexa recently introduced another focused solution called Sales Inventory and Planning. This product integrates forecasting, demand planning, inventory optimization, material and capacity planning, and supply planning.

One of the most important benefits Analog has gained from Adexa’s ATP solution is a greater ability to meet commitment dates.

“We now meet commitment dates more than 95 percent of the time, and we’re definitely seeing improvements as a result of that,” Dundon says. “We’re also seeing far less volatility in projected delivery dates on orders.”

In addition, “our scheduling and rescheduling rules are consistent with the current rule set that we’re applying in the planning domain,” he says. “We were never able to synchronize that in the past.”