All of business is a trade-off of one kind or another. For example, carrying enough inventory to always be able to quickly respond to unexpected demand surges brings with it an unnecessary—and in today’s economy, unacceptable—cost. The trick, therefore, is to find the optimal balance between multiple objectives, which requires identifying trade-offs. According to news from IBM, understanding trade-offs is about to get much easier.
An article that ran in Supply Chain Digest reports that IBM recently announced what it calls a substantial breakthrough in supply chain optimization technology: so-called “multi-objective” optimization capabilities that will enable companies to better understand trade-offs between different objectives, such as cost and service. According to Dr. Michael Watson of IBM, the multi-objective capability enables users to optimize across two or more objectives at the same time.
Executives know that when they make decisions in the supply chain, the issue isn’t cost, service, or capital investment as isolated elements because it’s a combination of those factors that drives their decisions, Watson says. What’s now possible using new mathematical optimization techniques, is the ability to analyze these different objectives all at the same time.
These capabilities were announced as part of a number of new features in the latest release of IBM’s LogicNet Plus XE, a network optimization tool. While it’s first being rolled out in IBM’s network optimization tool, the Supply Chain Digest article reports that the multi-objective capability is expected to also appear in IBM’s inventory optimization and factory scheduling software, as well as its optimization tool kit, CPLEX, which is embedded in supply chain planning software solutions offered by other suppliers as well as used by individual companies to solve specialized supply chain optimization problems.
The problem in the past, Watson says, was that companies had to run a single objective optimization, and then manually run scenarios with a few data points on different service levels. Besides the manual nature of the process, which takes time, the result is really a very incomplete curve--and risks the result not showing important step changes or other insights between the few data points selected, Watson says.
However, using the new technology, optimization software can now build a trade-off curve that shows how different objectives play against each other, Watson says. For example, these trade-off curves can allow companies to see areas where there might be a big step jump between one part of the curve and an adjacent part, or easily identify where there might be an exponential type rise, such as in an objective like cost as it approaches a 100 percent service level, he said.
While there are many potential uses, one example is to compare the total cost of a supply chain network versus the capital investment required to get to that point, Watson says. A slightly less than optimal network in terms of operating costs may--in some cases--require substantially less capital investment, which would make it the best total decision. Or, Watson said, a user could look at what the optimal network might be given different levels of capital investment ($10 million versus $20 million, versus $30 million, and so on).
In the end, I’m intrigued by the possibilities as well as the potential impact for users. And while improved optimization engine capabilities are one thing, adding the use of cloud computing to the equation could really make things interesting. I’m looking forward to hearing more on the story as the technology develops. Are you?