Walgreens is a national chain of drugstores, selling both pharmaceuticals, personal care products, and sundries. The distribution center in Anderson, SC is a “fourth generation” DC, with lots of automation and excellent engineering.
The DC ships product to about 500 Walgreens stores in the southeastern US. Each store receives one shipment a week.
Many of their most popular SKUs are picked from flow rack (the “AKL” area of the warehouse). Customer orders are assigned to totes (plastic boxes), which circulate on a conveyor that is outside the aisle, but has branches by which a tote can enter any aisle.
A tote will enter an aisle if it requires picks from that aisle. A typical aisle has three different induction points (where a tote can enter): beginning of the aisle, one-third of the way along the aisle, or two-thirds of the way along (approximately). Then the tote will be manually pushed along by an order-picker, filled at the appropriate locations, until pushed downstream to the next order-picker, and finally exiting on the conveyor at the end of the aisle.
Each aisle is managed by pick-to-light, and order-pickers try to move as many totes as possible through the sequence of bays for which they are responsible.
How can order-picking in the flow rack be made more convenient? Currently order-pickers wait until their zone of the aisle fills with totes. Then they push this train of ten or so totes down the aisle. This requires a lot of walking back and forth because of the congested dead-roller conveyor.
Would it be more efficient if they pushed a smaller train? What is the right size? Three totes? Four? Eight? The more totes they move at a time, the more local back-and-forth walking is incurred; but the less often workers have to walk back to the induction point to get more totes.
What is the best pick-to-tote scheme for queue management? What is the productivity gain? Be prepared to make recommendations to the client and to defend them with detailed arguments.
For extra credit:
Here are the slides (111MB) from the class presentation by the client.
H ere is the data (345MB).
The data is copyrighted and is to be used only for this project.