The Wall Street Journal reported recently on UPS’s project Orion, a ten year, $300 million technology build intended to optimize delivery routes for the tens of thousands of trucks the parcel giant has on roads across the globe daily. According to WSJ, each UPS driver makes an average of 120 stops per day representing a staggering 6.70190 possible alternate routes. UPS expects to save $300 million to $400 million annually with this technology. Your transportation plan may not approach a fraction of the complexity facing UPS but nevertheless, the variables at work in your network are probably still beyond the computing power of even the sharpest human minds. The good news is, you don’t have to spend $300 million to harness similar heuristic algorithmic technology to optimize your shipments.
We call Orion an “optimizer”, in the sense that it’s supposed to provide the “optimal” answer, and find the best path between millions of pickup and delivery locations that change daily. In reality, their algorithm uses what we call “heuristics” to calculate a very good answer, which can be found in a reasonable time with current computing power (because truly evaluating all possibilities to find the optimal answer would take too long to be practical).
There are some powerful, cloud-based transportation optimizers available to companies outside the Fortune 50, and for a very modest price. (See for example, LoadFusion Optimizer, the 2013 Transportation Product of the Year.) There’s really no excuse for declining to deploy such a solution. Even with a relatively small private or dedicated fleet, a transportation organization will leave money on the table if it opts to leave freight and route optimization to manual processes and the (limited) computational power of a human brain. Without algorithmic assistance, the resulting routes will undoubtedly be more miles than they should be and/or the shipper will need more trucks than would be ideal. Moreover, it’s almost impossible to incorporate all ancillary constraints such as customer pickup/delivery windows and driver hours of service, into a manually created route plan. If you primarily use common carriers, you can still save money using strategies such as aggregating LTL shipments into larger LTL shipments, creating truckload shipments with stop-offs, or using pool distribution strategies.
Like any organization working towards successfully automating supply chain optimization, UPS is encountering some issues as they roll out their ambitious solution. For example the Orion algorithm often produces routes that drivers know are unrealistic -but that’s a normal occurrence when newly embarking on an transportation optimization initiative. UPS will need to continue to refine it, blending human common-sense with the brute computing force of the algorithm, building in more real world constraints and ensuring the tool provides useful, practical results. The task is immense but UPS will almost surely get it done.
What we’ve seen at many of our LoadFusion implementations is that clients may have rolled out some optimization solution several years ago, and it was OK for a moment in time, but it’s grown moldy and doesn’t work well anymore since the business rules haven’t been adjusted in many years. Doing optimization with a SaaS provider represents a better option for shippers of all sizes because it engages the provider over the long haul, and a SaaS provider can work with the shipper to continue to refine the algorithm, keeping it up to date with new business rules as the environment changes.
In sum, it’s safe to assume our readers can’t afford a $300M solution like UPS. However LoadFusion Optimizer (and others) already offer the market a similar approach—leveraging computer power, and heuristic algorithms to find cost savings, but with reasonable, practical solutions.