Repeated History
Coding Horror had a story on LHR’s baggage system. I remember reading an analysis of the DEN baggage system in Dr. Dobbs’ way back in the mid ’90s. In spite of all the issues with implementation, I recall that the biggest conclusion of the report was that it wasn’t built with the airlines’ actual operational behavior in mind. Airlines that use a hub-and-spoke system schedule in “complexes” – lots of flights arriving and departing around the same time. The reason for this is so that passengers can make connections – you tend to have a lot of planes on the ground at the same time. One of the biggest flaws at DIA was that they simply couldn’t put enough baggage carts in the system to handle the peaks effectively – the system was engineered to handle average loads but not peak loads. The system would simply get backed up every time a wave of flights would hit the ground.
My first job out of high school involved a similar situation: I was working in a brand-new warehouse that included a state of the art warehousing system. This was essentially a cart on a 2-dimensional rail system – the cart would ride on rails and could go up and down; on each side of the cart were an enormous number of shelves; probably 4 stories tall and 100 yards long. An operator (the “picker”) would navigate the cart around to fill orders; the order form would be printed with locations for each item, so filling an order was simply a matter of navigating around to each location. There were 4 rails, so that meant 4 pickers and 4 banks of shelves. It was the most amazing thing I’d ever seen. I worked in receiving and was incredibly jealous of the chosen 4 who got to run these things.
Unfortunately, this system never actually worked. Orders would commonly have items on multiple aisles, so instead of a traditional warehouse, where one picker would carry the order through the aisles to fill it, a single order would be handled by up to 4 people before being pushed down to shipping. In addition, you needed at least one expediter; someone who could usher orders between aisles and ultimately to the shipping department when the orders were finally filled, so orders were touched by a minimum of 2 people, and possibly more than 5 (there might be more than one expediter working on busy days) before getting to shipping.
Stocking was a problem; certain items were hugely popular, so the one solution was to stock those items on multiple aisles, but then you introduced a scheduling problem – you wouldn’t really improve efficiency unless you could get orders filled by getting handled by the fewest number of pickers.
The worst process was capacity. The business was very seasonal, with a huge crunch starting in November and tapering off a couple weeks before Christmas. But since you had only 4 pickers, that area became a huge bottleneck. In addition, the pickers had to fill their stock from the incoming shipments on the receiving docks, so they had to stop filling orders for a time and restock their shelves. The system couldn’t be parallelized at all.
In the end, management simply moved the most popular items onto the warehouse floor in a traditional warehouse setting, and have workers fill those orders in the normal way. The multimillion dollar warehousing system was being used only for the less popular, less profitable items.
Automation is the key to a more efficient business, but sometimes the solutions we devise are a net drag on efficiency.
— Gordon Weakliem
Comment
Commenting is closed for this article.