InformationWeek

Supply Chain Designer presents users with Windows-based wizards that walk you through the basics of creating a model. For especially thorny problems, General Mills turns to a Caps Logistics consultant with whom the team has worked for a number of years. Working with the consultant, Siewert says he has "picked up the lingo" well enough to be able to model all but very complex problems on his own.

Similarly, Miller Brewing Co. in Milwaukee has had a great deal of success applying optimization algorithms to its capacity-planning projections without the luxury of an operations research staff on hand.

Using a supply-chain optimizer called AspenMIMI from Aspen Technology Inc., Miller applies optimization technology to producing an optimal annual production schedule. This schedule answers such questions as which breweries should make which brands of beer, and which distributor should ship them.

Given the breadth of the company's brewing enterprise, this is no mean feat: Miller Brewing consists of eight breweries, each of which houses 12 or 13 packaging lines. The company produces 80 brands of beer and outputs more than 2,200 brand and packaging combinations, which collectively ship to more than 1,800 distributor locations.

"The time it takes us to produce our annual production plan has gone down from more than a month to less than a week," says Art Hansen, Miller's capacity-planning manager.

The company's most recent success using AspenMIMI's optimization capabilities came last summer, when Miller entered into an extensive production and distribution contract with Pabst Brewing Co. "We spent a lot of time working with MIMI, determining if this was going to be an advantageous deal for us," Hansen says. "Without it, we wouldn't have been able to analyze the situation with as high a level of accuracy as we did."

One way to think of optimization is that it "replaces what was previously someone's gut feeling," says Bill Scull, VP of marketing at Ilog. And in today's highly competitive marketplace, there's no lack of gut feelings to replace.

"There's no question that optimization has only just scratched the surface," says Larry Lapide, VP and service director for supply-chain strategies at AMR Research, a market-analysis firm that focuses on E-business infrastructure. Already, Lapide says he's begun to see many examples of optimization applications that move beyond traditional planning and scheduling uses.

Examples of optimization applications that he's spotted recently include yield management (do I ac-cept this order now, or wait?), product life-cycle management (how do I phase products in and out most efficiently?), and warehouse management (what is the best route to take through the warehouse on a given day?).

If optimization is being applied to such a wide variety of problems, it's largely because the high-level optimization vendors that actually write the engines and develop the modeling languages have been hard at work creating products that are easier for independent software vendors to integrate into their vertical ERP or E-business solutions. Larger ERP vendors such as J.D. Edwards, Baan, Oracle, PeopleSoft, and SAP don't usually write their own optimization engines, but license them.

At its lowest level, "optimization actually is one of those things that you need a rocket scientist to perform," says David Dobrin, an analyst with Benchmarking Partners, an analyst firm that focuses on the business-to-business marketplace.

Many E-business vendors feel the same way: Models are built on top of a generic optimization engine for the specific needs of the customer or market to which the company caters. For example, Aspen Technology, i2 Technologies, and Manugistics Group deliver optimization modules to the so-called "process industries." Siebel Systems Inc., a leader in customer-relationship management, recently announced a partnership with Ilog that adds optimization capabilities to its configuration products. E-commerce enabler Trilogy Software Inc., meanwhile, keeps its framework open so it can work with a variety of third-party optimization libraries and routines.

If putting optimization in the hands of nonspecialists has had some recent success, it's due in large part to the efforts by optimization vendors to hide the complexities inherent in the discipline. "Making optimization more accessible is what all optimization vendors are trying to do," says Robert Daniels, director of software optimization vendor Dash Associates.

For vendors, this means insulating the users from as much nitty-gritty C/C++ code as possible. Instead, optimization vendors develop high-level modeling languages where the developer describes the problem in simple English-like terms.