July 19, 2017

Sorting It Out : The Case for Count-Back - Three Step Approach to Getting it Right the First Time

Potential Solutions for Under & Over Picking Problems

One the most common forms of distribution center automation involves a sortation system downstream from full case pick areas. Generally, much of the ROI for these systems comes from leveraging the sorter to enable batch picking of cartons during some period of time, such as a pick wave, versus traditional discrete order picking, where each order is picked individually. In batch picking, all the orders for a time period/wave are consolidated so that a picker only needs to stop once at a location for each wave picking all the cases needed for that wave at one time.

 This provides significant efficiency through the reduction of travel time versus discrete order picking, in which numerous pickers all travel to the same SKUs/locations for different orders. The batch picking process, however, can result in significant cases quantities per wave, especially (though not always) for fast moving SKUs. Dozens or even a hundred or more cases might be required in a single pick. Keeping track of the picking count can be a problem especially when there is a large quantity involved. 

In manual systems, case picking errors lead to costs, either from shipping extra cases to customers (rarely reported by them), short shipping (which is almost always caught and leads to deductions and charge-backs), or having the error caught in a DC audit process before the order is shipped, which is better than having a shipping error but must be dealt with manually.

Holste Says... Case pick accuracy would seem highly important in such mechanized/automated system. Unfortunately, in trying to ensure picking accuracy there are usually trade-offs between accuracy goals and productivity reality. What do you say? Click here to send us your comments

In automated sortation systems, errors of under- or over-picking are almost always identified by the system itself, but the costs related to each error remain high. If cases are under-picked, that wave or those pallets cannot be "closed" awaiting the missing carton(s). This can lead to system inefficiency and poor sorter utilization, and a costly process of working through the WMS to have the missing cartons picked and brought to pallet build areas.

Over-picks are recognized by the Warehouse Control System (WCS) and sent to the "reject line," where they are usually mixed up with cartons for which the bar code couldn't be read on the sortation system. Re-processing the cartons on the reject line is time consuming and thus costly. And as may be clear, cartons thought to be missing could in fact be in the reject lane, compounding the time and cost of resolving the errors.

So, case pick accuracy would seem highly important in such mechanized/automated system. Unfortunately, in trying to ensure picking accuracy there are usually trade-offs between accuracy goals and productivity reality.

Traditional Case Picking Options in Pick Modules

Traditionally, distribution center managers could select from one of three choices:

1. Scan each case bar code to verify the quantity being picked. While this provides high levels of accuracy, the obvious downside is a severe hit to productivity due to the effort to pick up and put down the scanner to grab each carton, and time it takes to scan. Wearable scanners are available and can improve operator performance.

2. Scan the location at the beginning to confirm the picker is in the right spot then rely on the operator to correctly count the right number of cartons in his or her head, confirming via key entry at the end that the right quantity was picked. This obviously can enable rapid picking because no other scanning is required, but the downside of course is that the picker is quite likely to make a quantity mistake, especially as the number of cartons required for a given pick in the batch increases.

3. Another approach (which has been around since the beginning of the batch picking concept) is to have preprinted carton labels for each set of picks. The picker puts on the labels and places each carton on the belt, moving on to the next location when the labels for a given SKU/pick are exhausted. The downside is that this approach basically defeats the concept of using wireless technology to begin with, and also is a hit to productivity, being roughly as fast (possibly a little faster) than scanning each carton. We will note though that if the cases are not already labeled (highly unlikely), pick by label may be necessary to identify the cartons on the sorter.

More recently, other options have emerged such as voice technology. The voice technology option can deliver high levels of accuracy with minimal impact to picking efficiency. In this scenario, pickers would count into their headsets as they pick the cartons. One of several methods could be used, such as counting as each carton is picked (1, 2, 3, etc.), or saying "next" or some similar phrase at each pick, having the system keep track of the count. At any time, the picker could ask for where they stood in the count ("You have picked 24 of 33 cases.").

The voice advantage is that it is "hands free," meaning high levels of productivity could be achieved because the count verification is coming through the voice process while the cartons are being placed on to the belt.

Another potential approach that a few companies have used would be to put a fixed scanner(s) along or at the end of the pick belt in each module. This is becoming more feasible because the cost of such scanners has dropped dramatically in recent years.

In this set-up, the warehouse control system (WCS) or possible the WMS itself would receive carton count information as the cases on the pick belt passed by the scanner. If it saw more than expected, one set of actions would take place, perhaps to the level of stopping the pick belt - though some in the material handling industry say that action would cause too many overall system efficiency issues.

If too few cases of a given SKU were seen, there would have to be some communication method to alert pickers or supervisors that such a shortage had occurred. That could be a visual display in the pick module itself, an alert to a supervisor, or some other method.

But, given the complexity of this workflow dynamic, and the level of integration that is required between the WMS and WCS, in practice means this is an option very few companies would embrace.

The Case for Count-Back

Developed in the 1990s by food companies using 3PLs to run their distribution centers as a means to reducing inventory discrepancies that were linked to errors in full case picking, count-back involves requiring the picker to perform a cycle count at the conclusion of each full case pick. That count involves simply entering the number of layers and loose cartons left on the pallet into a wireless terminal, enabling the WMS to understand how many cartons are there based on maintaining a pallet profile for each SKU. If the count matches what the WMS expects, the picker moves on. If it doesn't match, the picker needs to resolve the discrepancy, or possibly a supervisor needs to go and check the previous picker's pallet.

While some may see count-back as an unproductive waste of time, in reality the activity takes just a few seconds, and can be done in (3) easy steps:

1. Picker selects cases per RF pick instructions and places on pallet or pick belt
   
   
2. Picker is prompted to enter number of layers and loose cases that are left on storage pallet
   
   
3. WMS compares pallet quantity to what is expected, releases next pick if OK, requires action if count is off

So for companies that are invested in RF and, for whatever reason, are unlikely to add voice technology in the near term, count-back could offer some improvement in case picking accuracy by quickly identify when cartons have been under-picked, in time to easily add the missing cartons back on to the pick belt. Under-picks in the end are far more costly and problematic than over-picks, and these could be almost eliminated using count-back - at least from pallet flow rack pick locations (more on that below).

Count-back would identify that over-picks have occurred, but by that time it is largely too late to correct the error - the cartons are already on the belt and moving towards the sorter. Still, the cycle count would identify the error and re-correct the current case count in the location, needed among other reasons so that the next count-back process is accurate. In addition, the conveyor system should be able to identify and divert extra cartons for put back into the pick locations, and/or generate an automatic WMS task for doing so, better automating the process than most have in place today.

So, it seems like count-back could be a good thing to solve this pick-to-belt accuracy issue, with a couple of caveats:

1. The WMS system is likely to require modifications to enable count-back. 

2. The process works well for full pallets, typically positioned in pallet flow rack lanes in a pick module. It could also work for full pallets stored in reserve areas that are picked via order picker trucks and placed onto a conveyor system. It could be made to work in "half pallet" storage locations by maintaining a profile of half pallets as well. But, it wouldn't likely work very well in case flow pick lanes, where it may be difficult to see and/or count the number of cases remaining in the location.

However, picking errors are much lower when picking just a case or two for slower movers such as those that may be carried in the case flow racking. In other words, even if case flow rack locations could not be accommodated, count-back might still reduce picking errors dramatically in pick-to-belt applications with downstream sortation. Given the 20/80 rule, an 80% solution may be much better than no solution.

Final Thoughts

Count-back remains a little understood approach that exists largely in the food manufacturing sector. While it may not be a panacea for the issues of pick errors in batch pick and sort systems, it just might be a piece of the puzzle for some companies to consider.

Any reaction to this Expert Insight column? Send below.

Your Comments/Feedback