Order picking performance improvements in a manual order picking process

Abstract

The rapidly growing market for online grocery shopping requires improvement of operations and remaining ahead of the competition. The online supermarket is a relatively new market and has other challenges than regular warehouses due to order sizes and the products it handles; both fresh and preserved, but also fragile products. This influences several processes in the warehouse, of which order picking is the most expensive part of the operation. Therefore, ways to improve the order picking performance have been examined, resulting in the following research question:
How can the overall manual picking performance, from a human perspective, in a warehouse be measured and improved?
There are six decision factors for order picking in a warehouse that influence the performance: layout design, storage assignment, zoning, batching, routing and technical equipment selection. Next, the following three types of human factors play a large role in the order picking performance: physical, cognitive and organizational factors. All these factors have been researched individually, but seldom in combination with each other. Therefore, the focus of this research is on the influence of technical equipment on the physical and cognitive factors.
In the warehouse of Picnic, the picking process is a manual procedure, where order pickers walk around with (mainly) pick carts with totes on it. They receive order lines on the scanner which is attached to their wrist. There are significant differences in picking speed between different order pickers. The picking speed depends on experience, but also on the ease of performing a certain action like scanning a product or grabbing a product. The biggest challenges for Picnic are the reduction of the most time-consuming components of the pick process. The duration of each of these components can be measured with a time and motion study.
A Predetermine Time and Motion Study is used to carry out the time and motion study. Order pickers walked their order pick rounds with an action camera attached to their heads to make recordings of the pick process. This footage is analysed by labelling the actions an order picker performs when picking an order line: looking at the scanner, grabbing the product, scanning the product, putting the product in the tote, scanning the tote, walking with the pick cart, walking without the pick cart, walking with the product, inputting the quantity into the scanner, searching for the product, making a mistake, talking to other pickers. The analysed footage is used to determine the times spent on each of the actions. After determination of the times spent per action, they were compared to the Methods Time Measurement (MTM-1) required times to determine the points of improvement.
The relationship between the human factors and the actions of an order pick round have been determined. The order pickers can be divided into three groups; slow, medium and fast. The slow pickers take more time per action, they take more time walking, reading their scanner and picking the product. Fast pickers tend to combine more actions and therewith reduce time. The study also showed that there is more variance within the groups of the slow and medium pickers. The faster pickers are more experienced pickers.
The most time-consuming actions when picking an order pick line are looking at the scanner (±15%) and walking (±25%). Looking at the scanner takes slow pickers almost twice as much time as medium or fast pickers. The slow pickers also spend more of their time walking with their pick cart than the medium and fast pickers. Next, the slow pickers walk slower than faster pickers. This results in more walking time for the slow pickers.
Five alternatives to decrease the time spent on looking at the scanner and walking have been determined:
1. Change the scanner interface, in the current interface the letters are too small and there is too much information displayed
2. Optical head-mounted display, the removal of the action to lift and tilt the arm to be able to read the scanner will decrease the time to read the scanner
3. More totes per pick cart, an increase in pick cart size results in shorter distances between consecutive picks and therewith reducing the total time spent on walking
4. Automated pick carts, when pickers do not have to pull the pick carts anymore, they can walk faster
5. Batching, batching products to totes and totes to pick carts based on location in the pick circuit, makes it possible to make short-cuts and thus reducing the time spent on walking
The interest of both pickers and Picnic are taken into account to determine the best alternative. The human factors of the pickers, like physical inconvenience and learning curve, should be as low as possible. These are determined by comparing the alternatives to the current situation. The costs of implementing the alternative should also be as low as possible. The order lines per hour should increase as much as possible; the number of order lines per hour is determined by using a discrete event simulation model.
The interest of the picker and of Picnic is determined for each of the alternatives. Trade-offs have been made between the human factors, costs and order lines per hour, resulting in quick-wins and long-term alternatives. The quick-wins are the improvement of the scanner interface and the increase of the number of totes per pick cart. The long-term alternatives are the optical head-mounted display and the use of automated pick carts.
The two quick-wins have been implemented in the warehouse and the first results are promising, the increase of the number of totes per pick cart size seems to lead to a decrease in time per order line of 18%. However, this is based on a single pick round so no conclusions can be drawn yet. Also, the implementation of the improved scanner interface seems to work out well, the first reactions of the order pickers are that they can more easily read the scanners. However, also no conclusions can be drawn here since this has not been researched extensively.
The picking performance is measured by determining the time spent on each element of an order line for an order picker. The most promising alternatives improve the picking performance in a warehouse where the operation is not completely automated. The future solutions are the use of automated pick carts and the use of optical head-mounted displays.