How did a small, owner-managed company from a remote part of Bavaria achieve absolute global dominance in the domain of automation solutions for grocery mixed cased palletizing?
In the 80ies and 90ies Witron was known has a very competent sub-supplier for controls for well established warehouse and distribution solution providers like for example Dematic. Through this deep involvement with the logic of material handling automation Witron over the years developed its own solution know-how and became a competitor to its former customers.
In the early years of this century Witron introduced what shall become the benchmark for grocery mixed cased palletizing. Many very big competitors tried to institute a viable alternative to the Witron solution. And it is fair to say, that overall no competitor was able to establish a successful alternative solution on the market. Although some achieved a remarkable quantity of systems sold, none of the competitors for mixed cased palletizing secured a sustainably profitable business in this domain. The various solution alternatives and especially the many different design iterations within the concepts offered by each Witron competitor speak volumes.
Before doing a deep-dive into the Witron solution and understanding its logic, let’s first take a look at the specific challenges of automated mixed case palletizing.
The challenges
The first challenge is the high throughput of grocery in combination with a large product range. ( Just compare the cash register activities of your typical grocery store with any other shopping place, be it fashion, toys or electronics etc. ) This translates into a distribution warehouse having to handle a 5 to 6 digits large SKU range in very high volumes of typically 1000 to 5000 cases per hour. Whereas a case can be anything from a small box of spices over a carton of tomato soup cans to a tray of yogurt cups.
The second challenge is determined by the combination of SKU characteristics and the store operations where the mixed case pallets are delivered to. First one has to ensure that the eggs are on top and the washing powder is at the bottom of the pallet and not the other way round. Second, the sequence when retrieving the cases for the store replenishment shall match as best as possible the layout and hence SKU locations within the store. So from top to the bottom of the pallet the retrieval of the cases shall reflect a sequence which corresponds to a path through the store without jumps or returning to previous locations. Or in short: the sequence of the cases on the pallet shall allow for an efficient replenishment path inside the shop.
The third challenge is also specific to the grocery industry: margins per SKU and hence the allowable logistics operations costs per case are low. Unlike for example the fashion industry where the margins per item are high, the profitability of grocery relies on high inventory turnover with efficient, low cost operations.
The ‘simple’ task
Hence the ‘simple’ task: a high volume warehouse moving several tons of product per hour while building a perfect sequence for each outbound pallet and at the same time minimizing the operations costs.
After we clarified the challenges to be solved, now let’s have a look at the specific design approaches by Witron.
Inbound of a grocery distribution warehouse is represented mainly by pallets with cases of only one or very few different SKUs. In order to translate this into building outbound pallets with a mix of different SKUs with in average less than 2-3 cases per SKU the respective selection and sequencing of the cases requires the cases to be accessed individually.
The first solution design decision is therefore the choice between a big case buffer store allowing for the individual case access with few inbound / overstock pallets operations or a small buffer store for individual cases with frequent inbound / overstock pallets accesses. Or in other words, you either invest in a big storage and retrieval sub-system for individual cases or you invest in a frequent warehouse internal replenishment process from the inbound / overstock pallets to the buffer storage for individual cases.
Obviously the respective costs for the two alternatives, the costs of the storage and retrieval sub-system for individual cases and the costs of the internal replenishment process from the inbound / overstock pallets, are key factors for this solution design decision.
The Witron solution has by far the lowest costs per storage location of the storage and retrieval sub-system for individual cases. Whereas practically all competitors rely on some kind of mini-load shuttle sub-system, Witron opted for a conventional stacker crane based AS/RS sub-system with four cases deep storage.
Conventional low-tech is king
This technology choice translates into costs per storage location which are at least 30-40% lower than competitor AS/RS sub-system solutions for the individual cases. These lower costs allow for a larger buffer and hence reduced inbound / replenishment pallet traffic. And as every warehouse automation experts knows: less frequent processes not only reduce operation costs but also reduce the error rate.
Lower costs and better process quality
The latter, the process quality, was also a driver for a second very important solution decision by Witron. All cases, regardless of their nature, are always placed on trays individually once they are transferred from the inbound / replenishment pallet to the buffer store. Many competitors tried “tray-less” designs with cases stored and transported directly, without a tray. In light of the high volume operations this seems to be a good idea. It would allow to save the costs for thousands and thousands of trays and also for the loading and un-loading processes to get the cases on and off the trays. To make it short: all these attempts failed. The broad range of SKUs and hence SKU characteristics like shape, material stability (or lack of), center of gravity etc. made a tray-less operation error prone and hence unsustainable in light of the high volume operations.
After a sometimes costly learning-curve practically all competitors now also use some kind of tray based approach. So, is the big and relative inexpensive case storage sub-system the only remaining key differentiation of the Witron approach?
There is more
When we looked at the challenges, we established the demand for a perfect sequencing of the cases for building the mixed outbound pallet. Again the combination of this requirement with the very high volume operations translates into a real challenge. Any cases arriving out of sequence or not arriving at all at the palletizing station translate into a ‘broken’ mixed pallet. Such a pallet could become unstable or can’t be build anywise.
The sequence of cases on the pallet is not only determined by the SKU characteristics (remember, the eggs on top and the washing powder at the bottom) and the shop replenishment / layout but of course also by the requirement for a stable pallet. If a case is at the wrong place or missing entirely, the pallet can become unstable like a Yenga-tower with the wrong piece removed.
As most competitors of Witron rely on a high dynamic miniload-shuttle technology for the case buffer, they also use it for the sorting of the cases to match the building pattern of the outbound pallet. Witron on the other hand uses stacker cranes for the case buffer, allowing only for a rough pre-sort of the required SKUs, if at all. Instead an additional vertical sorting buffer is installed closely to the mixed case (robotic) workstation. Again, at first glance, Witron opted for a less efficient and less elegant solution.
Not elegant, but much better
Typically there is a n:1 or n:m relationship between the exit points – e.g. lifts – of the case buffer and the mixed case palletizing workstations. Firstly, because not all SKUs are in stock in every aisle. And secondly, because the sorting performance per exit point is not synchronized with the performance of the workstation(s). As a logical consequence there can’t be an exclusive 1:1 conveying connection between an exit point and a workstation. Instead several diverting points, merges and sometimes even loops are required for the conveying of the cases from the case buffer to the various palletizing workstations.
Even with a reliability of 99,99% or higher for every component, be it the lift, a diverter, a merge etc., the total reliability of the connection between the case buffer and the workstation(s) quickly deteriorates to somewhere between 99,95 % and 99,9% with just a single digit number of different components building that connection. Reliability means an error free operation, of course without equipment breakdown but also without any wrong orientation of a case, something stuck, a barcode unreadable, a timing error etc. etc.
Its a numbers game
Remember that we established the typical required outbound capacity to be 1000 to 5000 or even more cases per hour? With 5000 cases/hr a reliability of 99,99% will translate into one error every 2 hours. But with a total connection reliability of 99,95% we are talking about an error every 24 minutes and with 99,9% we are already down to every 15 minutes.
Suddenly the combination of a low-tech AS/RS system without meaningful sorting capacity but instead a second buffer with sorting function right next to the mixed case palletizing workstation makes a lot of sense. Witron designed this sorting function to be a buffer with a capacity of a bit more than the cases required for an outbound pallet.
The following graphic illustrates the operation of this end-of-the-line sorting buffer. First there is one sorting buffer per workstation and the connection between this buffer and the workstation is therefore very simple, hence reliable. Second, as the graphic shows, the cases building an outbound pallet are already completely in the buffer before the actual building of the pallet starts. This allows for the mixed case pallet building algorithm to compensate (= adjust the building pattern) for any error which occurred prior to the sorting buffer.
The magic wand
It’s basically as if you had a magic wand that could erase any error, or at least eliminate any impact of an error that occurred before the buffer. When one considers the aforementioned figures of several errors per hour, it is obvious that this additional sorting buffer directly next to each workstation pays off big time.
Operators of large mixed case palletizing warehouses confirmed to me that the Witron solution is second to none for reliable, high-performance operations. Especially for large, complex systems a reliable solution translates also in less risks and smooth ramp-up / system hand-over for the system supplier himself. Now we understand why for Witron these complex systems represent a sustainable profitable business.
One last trick
Which also has to do with the last ‘trick’ Witron uses to ensure the reliability of their mixed case palletizing systems. Actually some customers are not happy about this ‘trick’ because they are given no choice: Witron links the delivery of such a system to the mandatory ordering of Witron’s own service technicians. This ensures from day one that the system is maintained by competent technicians in close coordination with the supplier. And of course, this also translates into a continuous, profitable income for Witron.
In summary, Witron’s dominance in this area is based on:
- large, but cost-effective case buffer
- tray-only solution
- sorting buffer directly next to each workstation
- mandatory Witron technicians on-site
Please allow for these final words: It is now probably obvious that I consider the Witron approach to be the best solution in this area so far. Some might suspect a connection between me and the owner of Witron, because of the same family name. Rest assured, this is pure coincidence. I am neither related to, nor connected in any way to Walter Winkler. I actually never had the opportunity to even meet Mr. Winkler.