A couple of weeks ago, we ran some thoughts about “wave management,” or the process in a distribution center of grouping orders in some fashion for release to the floor.
Again, just to get everyone on the same page, a “wave” is basically an automated grouping of orders by some criteria that is released to the floor for processing as a set. Grouping attributes might be a set of carriers, a group of stores in retail, high priority orders, orders requiring a specific type of value-added services, or even a specific customer or three if they order in large enough quantities. In general, a wave should consist of something like 45 minutes to 2 hours worth of work.
The question was where and when the wave approach made sense – and whether it was time for some better order release intelligence that avoided the problems with waves (mostly around dwell time between waves being completed and started).
We generated a lot of reader feedback, including this comment from Greg Stubbs,
Fulfillment Services Manager for Royal Caribbean/Celebrity, who succinctly stated one aspect of the issue we raised: “These concepts have been wrestled with for decades in manufacturing operations, with the result being what's now often called 'lean manufacturing'. It is sad that there has been so little spill over to supply chain operations, despite the very significant gains that are possible when an organization gets away from batching.”
But there is a need for waves in many situations, right? Or that whatever the downside of waves are, they are more than compensated for by the labor savings from in fact “batching” – being able to go to a single location and pick product for multiple orders. Many DC automation systems are fundamentally built around this concept. In many scenarios, such as a standard carton sortation system, you may need to use waves to enforce some constraint, such as the number of diverts available (48 diverts for this wave of 48 store orders) or pallet positions.
For example, as David Schneider of Pep Boys writes: “In our process we plan the outbound loads first through an outbound routing optimization. Orders go into the TMS and we plan what the outbound routes look like. We take the plan out of that and load it into the WMS, route dispatch time and stop order. The orders that load into the nose of the loads that dispatch first are in the first "wave, then perhaps the nose loads of the next group of trucks along with the tail orders of the first group of loads, and so on.”
But Alan Reigart of St.Onge says that the waves approach often does not lend itself to good order flow and labor balancing: “We have found that most, if not all approaches to wave planning and the associated work release within the scope of most commercially available WMS technology lack the capability to systemically balance work across the various areas of a DC. Simply stated, WMS applications facilitate the creation of the wave plans based on the wisdom of the system configuration, however, the application do not programmatically determine or recommend the most appropriate means to process the demand.“
We’ll publish all these letters and more in full in the next couple of weeks.
Ok, so here’s my take.
- Many companies using “discrete order picking” today could benefit from some batching.
- Too many companies probably use a one-size fits all approach to order release (e.g., only straight order release, only full waves). You should want the flexibility and the tools to release orders the best way based on the day, the group of orders needing processed, etc.
- More companies need to do a more rigorous job of really analyzing how orders will or are flowing to the floor, and what is the most efficient way to release orders or create wave criteria. We look at overall DC/material flow, but rarely at this specific aspect.
- The WMS providers have opportunity to differentiate themselves by more flexible “wave” release, especially for complex DCs, that allow “mini” waves across areas and do a better job of intelligently balancing the flow of work and picks.
Finally, for many DCs, I think we will evolve into a more intelligent order release that is somewhere in-between straight release and waves, with orders flowing to the floor based on needs (e.g., carrier appointment in 30 minutes) in a way that is in many ways consistent with “lean” concepts, but which intelligently takes advantage of opportunities for small batching, task interleaving, etc.
Can it be done? I really think so. We’d really love some more of your thoughts on this topic? |