Why RF Picking isn't Always Better

ABSTRACT

If you spent close to $85,000 or more to outfit a dozen pickers with RF gear wouldn't you expect to get something out of it? The truth is, not everyone does…

The success of Radio Frequency terminals, barcode labels and what these technologies have done together to benefit the warehousing industry, has been widely published. While there are a number of companies experiencing dramatic improvements in picking productivity and picking accuracy as a result of these tools, there are a significant number of organizations that fail to see much benefit from this technology at all. Some even fall back to the old ways of dealing with paper picklists after seeing productivity drop after installing RF.

Why do some companies get a return out of this investment and others get disillusioned? The answer lies in how a firm views the technology. The successful ones look at RF as a means to an end. The less successful seem to feel that RF alone will inherently improve their operations. Stated simply -you get out of RF what you put into it. For some, no matter how much you put into it, it still is not the answer.

The Progress Group encourages its clients to consider RF as a tool to improve productivity. We believe you should review five issues before choosing to use RF picking in your facility:

• What are the short cuts used by your order pickers today to improve their picking productivity that they might not have available to them with a new RF system?
• How out of control are your operations?
• How can you use reengineer how you do picking around RF's capabilities?
• Does picking with RF buy you something somewhere else in the facility?
• How significant is the RF/WMS system response time relative to the duration of an entire pick tour

Man versus Machine: Who's really smarter?

Contrary to popular belief, there are a lot of really smart pickers out there. Also contrary to popular belief, there are a number of really dumb RF implementations out there. Before you choose any RF-based picking system you need to inventory how smart your pickers are, and determine what the impact will be if you let an RF system determine each picker's pick tour. If short cuts are somewhat important you need to look for a more sophisticated system. If they are really important you may want to reconsider RF altogether.

On a paper picklist you can see all of your picks before you begin a tour

Let's talk more about "short cuts"? Because paper picking provides a picker visibility of the entire pick list before he or she starts a pick tour, order pickers enjoy tremendous flexibility as far as sequencing their pick tour. The better pickers (especially case and piece pickers) have learned multiple ways to take advantage of this flexibility to keep their pick tours as short as possible.

With most RF systems, pickers only have visibility of one pick at a time. Thus they have to rely on the intelligence behind the RF program to make up for this limitation. Unfortunately, many picking shortcuts are not supported by any system and only a few are supported by the more sophisticated systems. This can, in certain instances, translate into lower picking productivity when RF is installed.

On an RF handheld pickers only have visibility of their next pick·

TPG has compiled from its experience in warehouse design and WMS implementation a list of paper picking "best practices" that many RF systems have trouble emulating:

• Choosing a "base pick" with which to begin a pick tour to avoid individual case or piece picking
• Choosing substitute picks to avoid "digging" for items
• Skipping certain aisles to avoid picks in congested aisles/locations
• Skipping picks to avoid traveling to excessively large picks that can't be easily carried
• Reverse picking
• Spreading pickers across zones or within a zone to avoid congestion and/or excessive incomplete orders

What is Reverse Picking?

Experience has shown us that the most important of the shortcuts seems to be reverse picking. To understand what we mean by reverse picking, it is useful to understand how most RF/warehouse management systems assemble and communicate a pick list to a picker, and this requires an example.

While this issue relates equally well to case picking situations, consider an area within a warehouse that is used for storing and picking small, piece pick items into shipping cartons. There are 300 bin shelving locations in this zone, organized into six aisles of 50 locations each. To minimize travel distances pickers should follow a serpentine pick path in that zone, traveling up one aisle and down the next as shown in Figure 1.0.

To accomplish this in most WMS or RF-based systems, each location will need a Location Sequence Number. This sequence is used by the RF-based system to determine which pick on a picklist to serve an operator next. Thus, to create the serpentine path, the locations would be numbered in increasing sequence in the first aisle and then in reverse sequence down the second aisle and so on as Figure 1.0 shows. The software directs operators to perform those picks having the lowest Location Sequence Number first.

Figure 1.0 - Example Piece Picking Zone with a Serpentine Pick Path

This approach works very well and generates an efficient pick path, so long as there are items on a picklist in every aisle. The problem arises when there are not picks in every aisle, and operators are forced to skip aisles. To see why, consider the situation in Figure 2.0.

In this scenario a picklist containing seven line items (i.e. SKU's A,B,C,D,E, F and G) needs to be picked. Based on how the items have been slotted, one pick is in the first aisle; two are in the second; two are in the fourth, and two are in the sixth.

Most paper based order picking systems will print the line items on a pick list in Location Sequence which is different from Location Sequence Number sequence. This is typically a worse sequence to follow than the one suggested by most RF systems. Fortunately, experienced pickers ignore the system generated sequence and create their own. In this example, an efficient picker would disregard the printed sequence and follow the pick path shown in Figure 2.0, picking the line items according to the following sequence: A, B, C, E, D, F and G.

Figure 2.0 - With a Paper Picklist an Operator Can Sequence a Very Efficient Pick Path

With most RF-based systems, however, the operator will be directed along a path like the one shown in Figure 3.0. Notice how this path is almost 50% longer! The reason is that the system has no understanding that if a picklist skips an aisle the system should reverse the pick sequence and serve picks in the next aisle going from the largest to the smallest Location Sequence Numbers.

This pick path is almost 50% longer than the one paper pickers might follow·
Figure 3.0 - RF can Sometimes Result in a Much Less Efficient Pick Path

Another point is worth noting here. Sometimes WMS and RF vendors will advertise that they use X,Y, and Z coordinates to assign operators their next task. This technically true statement is often times misinterpreted by clients to mean that situations like the one described above will not occur. This interpretation, unfortunately, is often not true.
Many WMS systems use X,Y, and Z coordinates along with other techniques to assign an operator his or her next closest task, but a picklist is considered one task made up of many sub-tasks. Usually, systems use the first pick on the picklist to represent the location of the picklist "task" relative to other tasks such as replenishments, cycle counts, loading and such. These systems do not use this logic once the operator has begun the picklist. Instead, they use the Location Sequence Number to determine the operator's next stop.

What this means is that, when searching for a new WMS or RF system, you should walk through an example like the one above to get a clear definition for what the software will and will not do.

Reverse Picking is just one example of a much larger set of short cuts that your operators may be using. You really need to inventory these techniques along with the skill and performance levels of your people before you decide to move to RF. You may find that your people are already delivering excellent productivity and accuracy. If that is true, RF might not offer much improvement.

RF is Great When Things Are Out of Control

This brings up the second issue to evaluate prior to installing RF. If a facility is not blessed with lots of knowledgeable and conscientious pickers, RF seems to really pay off. There are a couple of reasons for this. First, while there are companies that paper pick and have inventory and picking accuracy of 98% or above, there are many places that do not(i.e. inventory and/or picking accuracy can be below 85%). Bad inventory accuracy and mispicks can lead to a huge amount of rework and wasted time that is often not tracked until someone's Income Statement begins to suffer.

While not all paper picking operations have bad inventory accuracy; many, many RF operations have good inventory accuracy. If your operation is suffering heavily from this problem, RF can pay dividends many times over.

Other reasons warehouses tend to get out of control include excessive turnover, a reliance on seasonal/temporary labor, or having many inexperienced pickers. If this is the case, RF can help in these situations as well. How this happens is best explained using the example outlined earlier.

Note the route that an inexperienced picker would take to pick Order #123 if they adhered exactly to the sequence suggested by the paper picklist. Again, most paper picking and order management systems print out picklists in location sequence. If you assign locations from front to back, an inexperienced operator's pick tour would look like the one in Figure 4.0 below:

An inexperienced picker who follows the pick list literally ends up with a pick tour almost 20% longer than the RF picker·
Figure 4.0 - If you follow the pick list- literally- the result can be an inefficient tour

This pick path is almost 20% longer than the RF suggested pick path. Some companies try to get around this by trying by labeling the locations in one aisle in ascending order and the locations in the next aisle in reverse sequence, repeating this pattern to force the picks to come out on the picklist in a serpentine sequence, but we find that this confuses inexperienced pickers even more and ends up generating roughly the same pick path.

This example is but one of several ways RF creates and enforces a standard level of operating efficiency among pickers. Pickers, whether they are inexperienced or experienced, pick at about the same accuracy and at about the same rate(+/- 20%) in RF-based environments. In a paper based world you can see huge differences in productivity amongst pickers. At one client, the best picker more than doubled the productivity of her coworkers in some zones because she could take 3 or 4 orders and pick them all at once. She could keep in mind what picks were where and avoid a lot of wasted time spent paper shuffling to figure out what her next pick should be. Her pick tours were significantly shorter than her colleagues because for every 3 or 4 orders she only had to pass through the zone once. Her colleagues would make one tour for each order.

This again emphasizes the point that if your operation is full of experienced pickers that take advantage of short cuts such as multi-order picking, you might see little improvement or actually be penalized by RF. If, on the other hand, your facility is constantly plagued by inexperienced people who come and go, RF could be a boon. Operations that are out of control really seem to benefit from this technology.

Do Not Pave Cow Paths: Reengineer Your Picking Operation First

The number one reason why RF projects are not successful is that companies fail to rethink how to pick and ship orders in the new RF environment. Most organizations more or less pave the cow path of the past, electing to continue the precedent of standard order picking, while disregarding the greatest advantage of the better RF systems -their potential to support new methods for picking orders.

Another client implemented RF-based order picking and saw picking productivity fall almost 50%. They were ready to throw it out, until they were persuaded that this technology could enable them to pick in entirely new ways. For example, they could multi-order pick; they could also batch pick and sort the picks into orders at the end of the pick tour, and either of these two methods appeared to be more efficient than order picking. These alternatives were also superior to order picking using paper pick lists.

Very few like to hear this, but you have to spend some time up front analyzing the value of these different pick methods before you implement the technology. RF is like anything else: you get out of it what you put into it. Determining the feasibility and value of batch picking or multi-order picking requires order and cube information, and some simulation. This is not easy, but it should be done. If after this analysis, RF still seems to offer little benefit, you probably can rule it out. This of course assumes that your operations are not out of control.

Does Picking Via RF Buy You Something Somewhere Else?

Another factor that needs to be assessed when considering RF picking is that while RF picking may cost productivity in picking, it may buy additional productivity somewhere else. Two particular scenarios where this is true come to mind:

• Warehouses with lots of workers who move around a lot such as forklift operators
• Warehouses with forward pick areas that are replenished from bulk

So far we have discussed scenarios in which the picker is generally located in one area, and all that person does is pick. However, in almost every warehouse there isa subset of workers that are sent all over the facility to perform a variety of different tasks. Usually these employees operate forklifts picking product in one area and delivering it to another. In the case of a public warehouse this could be a "swing" team that moves to wherever the work is.

While RF Picking may lengthen a subset of picking tasks in these warehouses, it may increase the overall productivity of workers by ensuring that they are fed with a continual stream of work. No longer will forklift drivers be found sitting around waiting to be told what to do next. No longer will they be driving around looking for work or looking for a location to putaway something. Nor will they work on tasks that aren't as important as some other task. The system will send them to wherever the most important work is. Supervisors can also have the system create work for them such as cycle count tasks during "valleys" in the work flow.

Today the better RF and warehouse management systems support task management rules which let supervisors describe priorities for this mobile work force. The system farms out the work according to the rules established by the supervisors. These rules usually take the form of:

• Work of This Type (e.g Picking, Cycle Count, Replenishment, Putaway, etc)
• In This Zone of the Warehouse
• Is This Priority?

Each operator or group of operators has a "rules hierarchy" assigned to them consisting of a prioritized set of rules similar to those above that the system uses to decide what an operator will work on next. If there is a large subset of mobile workers and they can be sectioned off from areas and staff who are not using RF you may want the flexible work force to use RF. The average time spent per pick could go up, but the utilization of this group will increase making the investment worthwhile.

Another situation in which implementing RF picking in one area could benefit another is when a warehouse has a separate reserve and forward pick areas. In these situations, the forward pick area might be too dense to permit RF picking, but a high level of picking accuracy is still required. The picking accuracy in a forward pick area is often driven by the accuracy of the putaway and replenishment tasks that go on from the reserve area. Consequently, you might want to implement RF picking and replenishment in the reserve area but not the forward pick area. The benefits are the same but cost much less.

Weighing the Cost of Real-time Accuracy Against the Cost of Productivity

Finally there are many warehouses in which even after the short cuts are assessed, the stability and experience of the workforce are evaluated, the other methods of picking still don't overwhelmingly suggest RF is a valid solution. But there still can be valid, even compelling strategic reasons to consider RF. Such a candidate might be a company whose warehouse is not out of control but isn't achieving the shipping accuracy that it needs to. Other examples can be found in industries such as third-party logistics, wholesalers, health care, or spare parts, where companies want to implement RF to improve customer service by leveraging the real-time inventory and order status information that RF systems provide.

In organizations like this implementing RF is clearly a strategic decision, not a productivity initiative. Like many strategic investments you don't know exactly what the up side will be (you hope that it is good…), but you should evaluate the down side before you make the investment. Someone needs to know at what cost this information or inventory accuracy will come.

To do this, TPG recommends clients look at the current ratio of picking to walking time and then forecast what it will be in the future RF environment. If this ratio changes significantly, another method such as pick-to-light should be considered if the accuracy is really important but the cost of lost productivity is just too great.

To generate this forecast, keep in mind that most RF systems take 1 to 2 seconds to display the location where the pick needs to be made. Two to three more seconds are required to holster and unholster the RF gun between pick locations. At the pick face, 2 to 3 seconds are needed to scan the location, enter the quantity picked if different from the suggested quantity, and press ENTER to confirm the pick. This adds up to an increase of 5 to 8 seconds per pick at the pick face that RF requires over paper order picking.

Waiting on Pick Location: 1 to 2 secs Entering Pick Information: 2 to 3 secs Holstering the RF terminal: 2 to 3 secs
Figure 5.0 - RF Picking takes 5 to 8 seconds more per pick than paper

If you are picking cases onto a pallet and touring through an entire facility before dropping off an order, 5 to 8 seconds per pick might not be a significant amount of time. If, however, the environment requires picking into a shipping carton from dense carton flow rack, this could be a huge productivity hit. By looking at the ratio of picking time to traveling time before and after RF, you can identify instances where choosing to go with RF will be costly.

Is RF the Silver Bullet?

There are a number of misconceptions to be cleared up regarding the use of Radio Frequency in warehouses. First, RF adds a tremendous amount to most warehouse environments, but not all. Second, it is not a silver bullet. It seems to benefit operations tremendously that are "out of control" but at most companies a reasonable amount of analysis needs to be done up front to make the RF decision. Third, RF can be implemented selectively within a warehouse. It is not an all-or-nothing technology. There may be some areas in a warehouse that warrant its use and others that do not. Finally, experience shows that failing to assess the value of this technology prior to implementation invites a significant amount of risk. There are a number of horror stories to complement the wealth of success stories. Do your homework before you buy.

Acknowledgements

Special thanks go to Ozburn-Hessey Logistics and Zethcon Corporation for providing pictures to illustrate this article.