InformationWeek

Recently, many companies have charged their application development teams with refining existing business processes through the use of specialized optimization tools. These tools are based on sophisticated software optimization algorithms, which traditionally required the skills and expertise of highly trained mathematicians and operations research professionals.

Enterprise resource planning and supply-chain management vendors have begun including these algorithms in their products, adapting them to their customers' industries.

In the process, software optimization has become accessible to mere mortals: developers who may or may not have stellar programming or mathematics skills, but who have keen business acumen and a desire to improve the company's bottom line.

There's no doubt that optimization can be complicated stuff, used to answer complex questions. Optimization is a discipline that uses mathematics and computer science techniques to solve complex business problems whose results depend on hundreds, thousands, or even millions of interconnected variables.

For example, in a geographically dispersed manufacturing concern, researchers might use optimization to determine how many plants are needed, where they should be, and what each of those plants should manufacture.

Or, to take a transportation example, one might use optimization to create the best possible crew schedule, taking into consideration routes that need to be manned.

Milorad Sucur, a consultant for Sabre Inc., a software vendor specializing in the travel and transportation industries, says it has sometimes taken him several months to develop a single optimization model. Holding a Ph.D. in operations research, Sucur is fluent in C/C++, as well as a number of higher-level optimization-specific modeling languages. He works on Unix workstations, where the models are run once they are written.

Despite the highly skilled labor software optimization has demanded, it has always wielded an impressive return on investment. One example comes from United Parcel Service of America Inc. Several years ago, UPS began using the CPlex optimization engine from Ilog Inc., a leading French supplier of optimization engines. CPlex was used to assist in tasks that previously had been performed manually, such as package routing and aircraft scheduling.

Before putting CPlex on the job, the Atlanta shipping company was planning to buy 30 jets to handle a predicted increase in package volume. After solving its scheduling problem with CPlex, UPS determined it could get by with just 18 to 26 new jets.

At $60 million a jet, UPS saved between $240 million and $720 million, estimates Keith Ware, operations research manager at UPS Airways.

That sort of cost savings is not out of the ordinary, analysts say. According to research conducted last year by John Bermudez, group VP of enterprise application strategies at AMR Research, ROI figures for optimization investments applied toward planning and scheduling sometimes run as high as 300%.

In the past couple of years, news of successes such as these has trickled down to companies that aren't necessarily blessed with highly specialized developers. Yet they, too, have managed to capitalize on software optimization, presenting their companies with impressive cost savings and respectable ROI figures.

Take General Mills Inc., the Minneapolis cereal manufacturer, which runs an optimization package called Supply Chain Designer from Caps Logistics Inc., a Baan company. The company uses Supply Chain Designer to make long-term decisions about how to place its manufacturing and distribution sites, according to Jeff Siewert, a General Mills supply-chain analyst.

Currently operating out of eight manufacturing facilities, General Mills used Supply Chain Designer to determine that it could operate much more efficiently by paring down to four manufacturing facilities. This will happen over the course of the next 10 to 20 years, says Siewert, but "we've already saved millions of dollars."

continued...page 2, 3