ED. NOTE: This story was originally written and published on July 22, 2003
Back when we first reported on Houston Distributing Company's leap of faith into the complex world of automated picking for pre-sold orders, the jury was still out on whether beer distributors could achieve reasonable returns on such an expensive investment, or even if it would work at all.
Indeed, there were times when Houston Distributing president & ceo Bo Huggins was having second thoughts. "It was an extremely complex set of variables we were dealing with in the early days," Bo told us a few months ago. "I'm not going to tell you it was easy going. We were faced with basically running manual and automated picking simultaneously until we could get the kinks out of the automated system. We were overcoming a very high learning curve."
Chesbay Distributing president Chuck Ricci, the second beer distributor to install an automated picking system, agrees. "We started just after Houston," he said. "We were basically the guinea pigs. We learned things from Houston and they learned things from us. The first six or seven months were challenging."
But both distributors were willing to take that leap of faith because they knew that warehouse and, more significantly, delivery labor expenses were the largest cost centers in their companies. They would rather get those major costs under control now while margins and volume were still holding up than down the road where both are uncertain. "We always look ten and twenty years down the road," Bo said. "We would rather get the efficiencies in place now."
Though both Miller/Coors distributors, Chesbay and Houston operate in very different markets and each required a different sort of system. Now the Reyes family is beginning an installation at Chicago Beverage Systems.
I visited HDC last month and the system was humming full bore, building and picking orders in the early afternoon for staging. To say that the system is a marvel to behold is an understatement. Conveyors speed cases of beer from the Vertique storage towers to a centralized area, where either layer picking clamps or giant robotic arms gently drop the cases on pallets according to each presold order.
Fully Automated System
HDC exec. v.p. and general manager Perry Parsons took me on a tour of the picking operation. He explained that the beer is only touched by a warehouseman to unload the long-hauls from the brewery, to replenish the tower filling system, and to take the finished palletized order to the route truck. Warehousemen don't ever see a load sheet, they never hand pick product, they don't stretch wrap it, they don't label it, and they don't have to think about how to build the mixed case pallet to get the most cases in a bay. Vertique's picking and loading management software, VIPER, takes care of these details automatically.
Let's follow a case of beer from order to delivery to give you a sense of how this works. So as to not show favoritism, I'll choose a SKU which is certain to offend nobody, my father's and Manny Sanchez' old import, Mamba 10/22oz bottles. Mamba is the sort of SKU which is certain to cost distributors more money to handle than they make because it was one of those low volume, high touch SKUs. It's the type of SKU that Vertique might make profitable for you to handle again.
08:04 A.M. - WAREHOUSE. A warehouseman unloades a trailer from the brewer, placing it in the warehouse. He enters the bin location, SKU, amount, and code date into the inventory system. He notices that it is a new SKU, so he measures a case with a tape measure and enters the product information into the VIPER system so it will know how to handle it later.
08:26 AM - LUCKY PENNY SUPERMARKET. Across town, presaleman Tony Stackem is taking an order - excuse me, I mean making a sale in a grocery store. He notices that there is only one dusty bottle of Mamba left on the shelf. He thinks about replacing the SKU with a higher volume craft brew, but remembers there is a trip to Africa incentive for repeat Mamba orders. He enters a case of Mamba into his handheld. He completes the order and transmits it via cellular connection to the company's route accounting system.
11:53 A.M. - COMPUTER ROOM. The distributorship's RAS system processes the order, calculates pricing and discounts, etc. At this point several things can happen depending on how the distributor handles orders. Let's say that this is on a static bulk grocery route and the distributorship stages those early in the day.
So the RAS dumps the order into Vertique's VIPER software, which then uses rules that the distributorship sets up about weight limits, time stops, etc. and virtually builds the bays. The order that the Mamba is on is a 70 case order on one pallet headed to the Lucky Penny.
02:40 P.M. - WAREHOUSE. A warehouseman checks the electronic marquee and notices that Mamba 22oz bottles are getting low in their Vertique storage tower. He has a vaguely confused expression because he doesn't recall ever seeing this product before because it is so rarely ordered. Luckily, the marquee also tells him which bin it's in the warehouse. He takes a tow motor to a dusty far corner reserved for unwanted neons and neglected SKUs and picks up a pallet and drives it over and places it on one of the de-palletizers, which strips the pallet layer by layer and places the cases on a conveyor which whisks the cases up to the Mamba Vertique tower, filling it to the brim. There the cases wait until they are called upon by an order. Think of the Vertique towers as giant vending machines which drop cases on an as-needed basis. They have a series of levers which drop cases level-to-level as a case is released at the bottom. The energy they use comes from this source called gravity, which to-date has never failed.
(Note: In reality, a low volume SKU like Mamba probably wouldn't warrant it's own Vertique tower, but would just be racked and placed on orders manually. But this is for illustration).
02:45 P.M. All bulk orders are in and have been virtually built. They are approved and ready to go. The Lucky Penny order goes to the picking system.
An empty pallet is pulled along a gantry as conveyors bring the cases ordered by the Lucky Penny. Those cases have been deposited on the conveyors by the appropriate Vertique towers (remember, think vending machine). The computer has built the pallet so that SKUs with identical case dimensions are grouped together in layers (usually 7 or 8 case layers).
This is a revolutionary characteristic and something that a Tygard Claw can't accomplish on its own. Creating mixed SKU layers is one of the reasons Vertique is able to cube out truck bays so nicely and not sacrifice route truck capacity when loading by stop.
The conveyors bring those identical-dimension-but-not-necessarily-same-SKU cases together in the right sequence so that a layer picker scoops them up and deposits the layer on the pallet. It repeats the process, building layer upon layer of our 70 case order to the Lucky Penny.
At some point in building the pallet, it becomes impossible to build identical dimension layers, and then these "onesies" and "twosies" need to be placed on top. Mamba, having a case dimension known only to obscure breweries in far western Africa, is one of these onesies.
At this point, Houston's operation differs from Chesbay's. Houston employs four giant robotic arms, similar in appearance to huge King crab claws, one of which actually picks up the case of Mamba, applying gentle pressure according to the weight and dimensions of the case, from the conveyor and gently places it on the top of the pallet with a lover's touch.
How does it know? The computer tells it how high the pallet should be, so it just knows. This is truly a remarkable spectacle to see. But at about a quarter mil a piece, they aren't for everybody.
Perry told me that before the computer was configured correctly, a robotic arm picked up a case of bottles and crushed it through a full pallet of beer from the top to the wood. Those robotic claws have some strength on them. He said it was like watching fireworks with beer. It was a messy mistake, but almost worth the spectacle.
Chesbay uses a human in lieu of a robotic claw. While plenty strong, this person hasn't crushed a pallet yet. He stands where the robotic arm would be and takes the case from the conveyor and drops it on the top of the pallet. The cases come to him in the correct sequence, so he doesn't even have to think about it.
Now the pallet goes into a stretch wrapper and an automatic labeler places a sticker on the side of the pallet which shows account information, truck/route assignment, bay assignment, any out-of-stocks that were cut from the order, etc.
Now the pallet is picked up by a warehouseman and either staged or placed on the truck.
07:26 A.M., Next Day, LUCKY NICKEL SUPERMARKET. A merchandiser checks in the beer and places the bottles of Mamba in the cold box, making certain to rotate because quality control is so important to the Ivory Coast brewers who craft this malt liquor by hand.
Cutting Warehouse Costs
Obviously, when you load trucks by stop rather than by SKU, it takes more warehouse staff, more time, and more space. Picking by order or stop is so much more labor-intensive and complex that many distributors have tried it and abandoned the idea as a failure. Warehouse night staff who are willing and able to tackle the complex tough work are hard to find, and turnover is high so training becomes a difficult issue.
With an automated or semi-automated system, these points become moot. Both Houston and Chesbay added one technical person to run the VIPER software, but cut their labor costs by 30%-50%. Keep in mind that this is despite the fact that they were loading some routes by SKU before, and now virtually all routes are loaded by stop.
So the warehouse savings are there. But what about maintenance? There are many moving parts in this thing and if any one thing goes wrong, the entire system slows or stops. HDC uses maintenance people from its fleet maintenance department to complete all routine work. Perry said that they haven't had problems now that the system is working, and just like any other automated system in other industries, you just have to keep enough spare parts in inventory so any component can be replaced in a matter of an hour. Contingency conveyor routes and duplication of pallet entry and exit points prevent complete system shutdown. Also, both Chesbay and Houston utilize backup generator power in the event of a black out.
This is perhaps the most surprising benefit of automated warehouse picking systems, though not surprising to long time readers of the DPL.
We've often written of the inherent tradeoff in delivery labor costs and warehouse labor costs when it comes to loading trucks by stop or by SKU..delivery is higher when you load by SKU because the driver is opening two to twenty bays to get the cases he needs for a stop. Warehouse costs are higher, though, because pickers have to painstakingly build each pallet by hand or with a Tygard Claw.
By contrast, loading trucks by SKU allows the warehouseman to load partial or full pallets into the truck, and then throw onesies and twosies in. It's far less labor intensive in the warehouse, but a tremendous strain on the driver.
We all know these two facts:
1. Drivers typically make about twice the hourly rate of warehouse pickers.
2. CDL licensed drivers and competent warehouse pickers are hard to find, and both have high turnover.
The answer: reduce the need for both, with a stress on reducing driver hours, particularly overtime.
The Vertique system addresses both, because you can load all of your cases by stop while still reducing warehouse staff. Plus, drivers get in earlier, have less strain, are safer (not moving around the truck with open bays), and are generally happier.
Chet Willey is Distributor Operations Manager for Miller Brewing Company. He has a background in industrial engineering and has worked at Miller for 22 years, so he isn't exactly new to the game. Chet has experience conducting time studies and knows pretty much to the minute how long it takes to deliver a case of beer in certain circumstances.
He toured Houston's operation and even road with drivers. He told DPL, "Every driver I spoke with said they would never go back to the old way [loading by SKU]." Houston was able to take 10 routes off about 50. And their hand route drivers are getting in two hours earlier.
Houston estimates that Vertique saves about 10 to 15 minutes per stop on hand routes (non-bulk), or 3 hours a day per route. With those numbers they can add 2 to 3 stops per day and take drivers off the road.
DPL expressed concern that trucks would lose their load-out capacity when loading by stop, particularly with so many SKUs represented. But all indications point to very efficient cube-outs of bays. Chesbay plans routes so that the driver makes them sequentially, so orders can be stacked in reverse order on top of one another in a bay (or in the case of a back loader, behind one another).
Houston's high number of chain stores with their time stops requires that every stop's order be exposed to the driver. Therefore, Vertique's VIPER software accounts for this and virtually loads the trucks accordingly.
Miller's Chet Willey indicates that a considerable amount of the value derived from this system comes from the VIPER loading software. "I imagine it will give ULM a run for its money," he said. VIPER is like ULM or FleetLoader in that it takes the cases for tomorrow's loads, looks at the fleet of trailers you've got available, looks at the rules you've set up for it (Lucky Penny has to go on Johnny's route because his aunt is the back door lady there, etc.), and then virtually constructs each bay with the driver's ease of access in mind. It can load by stop, by SKU, or both.
Jeff says that they are also introducing the capability to load carts in the same way it loads pallets. Many distributors have experimented with using six wheel Magliner carts with back-loading box trucks. The carts shave delivery time and provide easy access to accounts with small back or front doors, but they are labor-intensive to load. One A-B distributor said his drivers loved the carts but they had to abandon the program because their warehouse couldn't get the trucks loaded on time.
Chet Willey at Miller says that you would typically have to double your warehouse crew to load on carts, so a Vertique cart loading system could add significant savings on both ends.
Return on Investment
Many distributors are concerned about spending so much money on a system like this. What exactly is the pay pack?
It's differs according to company size, number of SKUs, how often each SKU is handled, route design, and market characteristics, among other factors. Says Jeff Stingel, "The system is designed so that you can start with a semi-automated system, which costs less, and then add components later." He said it depends on the comfort level a distributor has with pay back periods. HDC is predicting a 5 year pay back period, but that number keeps coming down as more efficiencies are discovered.
Jeff points out that you can achieve a one year pay back if you utilize lease options or other financial mechanisms. Also, capital expenditures have better tax treatment now than ever.
Is it right for you?
You do indeed need to have some scale to justify an investment of this magnitude. Most people we spoke with indicated that a 3 million case a year distributor could benefit, but not much lower. Also, the more SKUs you have, the more benefit you get in both warehouse and driver efficiencies.
However, I wouldn't rule out A-B distributors. Just because many A-B distributors have fewer than 100 SKUs, they would still benefit by loading by account and considering the carted loads. Several A-B distributors we spoke with have recently initiated loading-by-stop in their warehouses manually. A semi-automated system might help get those trucks loaded faster, particularly during peak summer volume.
We would think that the biggest problem with implementing a project of this scale and complexity is the loss of key executive focus. Installing such a system requires a tremendous amount of management's time and energy, which would be diverted from other endeavors like growing sales. This is not something you install in your warehouse and flip a switch. Things can and will go wrong. It still takes an incredible amount of planning-not just in the warehouse but in routing changes-testing, and a necessary amount of disruption, during all of which you still have to be selling and delivering beer without a hitch.
Additionally, the equipment to run this system needs to be put somewhere, preferably close to where you load your trucks. That takes space, something that is in short supply in the warehouses of many companies. Jeff says that most systems can be placed in a distributorship's current staging/picking area. However, I've seen some of your staging areas. They aren't exactly roomy like a suite at the Savoy.
If you are considering this system and are planning a new warehouse, best to make plans for ample space. Also consider the extra space for spare parts inventory. If you are going to need to add on to your warehouse to implement the system, that will obviously hurt your return on investment and lengthen your payback period.
Having said that, now that Vertique has installed or is in the process of installing systems in a big chain market (Houston), a smaller market (Chesbay), and a large metro market (Chicago), they've now got experience in various types of distributors. It might be time to pull the trigger. -HCS