As customer expectations rise and multiply, order picking flexibility becomes more important. By using human labor in picking processes businesses can fulfill individual requests (such as customer-specific packaging) while maintaining delivery capacity. That’s why labor still plays an important role in order picking in most distribution systems.
Nonetheless, transparency in picking systems remains poor as performance measurement is typically limited to the system-level. Key performance indicators such as picks per employee per hour are typically derived from mean values rather than individual employee figures, resulting in a lack of visibility to individual picking performance. Furthermore, the allocation of picking orders and their respective demands on employees is at present problematic due to a lack of technical capabilities.
Finding Star Players
An employee’s individual strengths and weaknesses are often not taken into account in deciding who fulfills which order. This is typically determined on a first-come, first-served basis. Take for example the case of an employee picking high-value, temperature-sensitive medication. He received this assignment arbitrarily when he pulled the order from the top of the stack. Would the distribution system have been better served had this employee “traded” orders with a colleague who sometimes needs help retrieving heavy, full boxes contained in some orders, but whose orders virtually never contain errors?
Within the scope of our research project, “EfPick” (Assessment and improvement of efficiency in manual picking processes), researchers at the University of Stuttgart are developing a method by which the performance of individual pickers can be objectively quantified. Additional project aims include improved ergonomics and establishing a methodological basis for the strategic assignment of picking orders to specific employees. Factors include individual order properties and unique performance profiles defined for each picker. Six companies are participating in the project and providing support either in the form of picking data or software development expertise.
Characterize Orders
An objective picking performance assessment requires consideration of items contained in the orders, including SKU characteristics such as weight, volume and storage location. These are derived from the corresponding data retrieved within the scope of the data collection. Performance is assessed at the most basic observable unit contained within the picking data of a given company for the highest possible accuracy in data analysis.
For example, analyses at the SKU-level are preferable to those at the order line-level, as performance-affecting variables measured at the former’s level correlate more closely to picking time. This is evidenced by the former’s higher correlation coefficient value (Figure 1) and is logical, as an order line may contain one or more SKUs.
Because SKU-specific attributes can influence the time required to pick (compare picking a laptop to picking a 45-lb-barbell), the impacts of such characteristics must be reflected in performance assessments as well as in plan times. The effects of such variables – in this case weight – on picking performance can be analyzed on an individual basis for each employee using regression analysis.
Each point along the curves shown in Figure 2 represents an expected picking time for an order given a known order weight.
Additional performance-affecting variables such as total volume and total required walking distance and their respective effects on picking time can also be calculated on an individual basis without excluding stochastic variables. The different curves observed for each picker form the foundation for individual performance profiles, which will at a later stage in the project be integrated into order assignment optimization.
Based on these profiles, it becomes apparent which employees work most accurately and which are the fastest. Using such information, it would be possible to steer an order containing fragile items to a picker with an exceptionally low damaged item rate. Urgent orders could be assigned to the fastest pickers and these individual profiles could be used in determining which supplemental training might be most useful for individual employees. Additionally, much of the guesswork could be removed from employee scheduling because the plan times are tailor-made for the workers and take into account picking order-specific factors, e.g. total required walking distance, into consideration for increased precision.
Performance Timing
Aside from data analysis, the definition of standard process times for individual picking orders represents an essential element of both performance assessment and personnel planning in picking systems. Within the scope of the same project, the University of Stuttgart is currently developing process blocks based on MTM / UAS (Methods-Time Measurement / Universal Analysis System). These blocks are later to be combined according to picking order variables in order to return independent standard times required for the fulfillment of individual picking orders.
The picking process analyses necessary for the development of such process time blocks also identify waste such as unnecessary movement and poor ergonomics resulting from poor workspace design. In addressing such inefficiencies, the value-added portion of picking time is expected to increase while unnecessary strain placed on employees diminishes. Such optimizations can take very simple forms, such as placing disposal bins for empty boxes closer to packing stations in order to reduce walking distances.
The detailed collection of variables affecting picking performance and quality, as well as the calculation of key figures on the level of the individual employee, represents a contribution to logistics control.
From the standpoint of the companies involved, this research project offers an opportunity to improve efficiency. The more transparency afforded, the greater the improvement. By matching picking order characteristics to people, performance and efficiency can improve—thus enhancing business profitability.
André Siepenkort (left) and Matthew Stinson (center) are researchers and doctoral students at the Institute of Mechanical Handling and Logistics at the University of Stuttgart in Germany. Dr. Stefan Gerlach conducts research at the Institute of Work Science and Technology Management at the University of Stuttgart.
