The effect of worker learning on manual order picking processes

Int. J. Production Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Contents lists available at ScienceDirect

Int. J. Production Economics journal homepage: www.elsevier.com/locate/ijpe

The effect of worker learning on manual order picking processes Eric H. Grosse n, Christoph H. Glock Carlo and Karin Giersch Endowed Chair “Business Management: Industrial Management” Technische Universität Darmstadt, Hochschulstr. 1, 64289 Darmstadt, Germany

art ic l e i nf o

a b s t r a c t

Article history: Received 30 January 2014 Accepted 11 December 2014

Order picking is a time-intensive and costly logistics process as it involves a high amount of manual human work. Since order picking operations are repetitive by nature, it can be observed that human workers gain familiarity with the job over time, which implies that learning takes place. Even though learning may be an important source of efficiency improvements in companies, it has largely been neglected in planning order picking operations. Mathematical planning models of order picking that have been published earlier thus provide an incomplete picture of real-world order picking, which affects the quality of the planning outcome. To contribute to closing this research gap, this paper presents an approach to model worker learning in order picking. First, the results of a case study are presented that emphasize the importance of learning in manual order picking. Subsequently, an analytical model is developed to describe learning in order picking, which is then evaluated with the help of numerical examples. The results show that learning impacts order picking efficiency. In particular, the results imply that worker learning should be considered when planning order picking operations as it leads to a better predictability of order throughput times. In addition, the effects of learning are relevant for the allocation of available resources, such as the allocation of workers to different zones of the warehouse. The results of the numerical analysis indicate that it is beneficial to assign workers with the lowest learning rate in the workforce to the fastest moving zone to gain experience. & 2014 Elsevier B.V. All rights reserved.

Keywords: Order picking Warehousing Learning Human factors Picker-to-parts systems

1. Introduction Global competition and high cost pressure force companies to utilize every option for improving the efficiency of their operations. An area that still contains significant potentials for improvements in many companies is internal logistics and warehousing. Processes in this area have, as a consequence, come more and more under detailed scrutiny in recent years. Order picking, which is the process of retrieving items from their storage locations to fulfill customer orders, is one of the most time- and laborintensive operations in warehousing. Therefore, identifying and realizing potentials for efficiency improvements in order picking is an important lever for increasing the efficiency of operations. Studies indicate that order picking is responsible for more than 50% of the operating costs of a warehouse (Tompkins et al., 2010), which is mainly due to the high amount of manual human work that is involved in this process step. Although more and more

n

Corresponding author. Tel.: þ 49 6151 165281. E-mail addresses: [email protected] (E.H. Grosse), [email protected] (C.H. Glock).

order picking systems are partially automatized today, manual order picking systems are still dominant in practice. Researchers have estimated that up to 80% of all order picking warehouses are operated manually (de Koster et al., 2007; Baker and Perotti, 2008; Napolitano, 2012). The reason why many companies rely on manual order picking is that the cognitive and motor skills of human order pickers cannot be imitated economically by automatic order picking systems (Roodbergen and Vis, 2009). Cognitive and motor skills, however, are human characteristics that may have a high impact on the performance of manual order picking processes, and therefore they should not be neglected when planning order picking operations. To predict and monitor the performance of individuals, various types of learning curves have been developed in the past (e.g., Jaber and Glock, 2013). Learning curves model the performance improvement of an individual or a group performing a task over time as a result of accumulated experience. They facilitate better performance predictability, for example in inventory models that consider learning in the production rate or in setups (Jaber et al., 2009). Learning in order picking has attracted less attention in the past. A recent study of Grosse et al. (2015) revealed that worker characteristics—such as learning—have regularly been neglected, and that performance

http://dx.doi.org/10.1016/j.ijpe.2014.12.018 0925-5273/& 2014 Elsevier B.V. All rights reserved.

Please cite this article as: Grosse, E.H., Glock, C.H., The effect of worker learning on manual order picking processes. International Journal of Production Economics (2015), http://dx.doi.org/10.1016/j.ijpe.2014.12.018i

E.H. Grosse, C.H. Glock / Int. J. Production Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

2

indicators of order pickers (such as the time needed to fulfill particular order picking tasks) have therefore been (wrongly) assumed to be constant. This indicates that there is an opportunity to improve existing order picking models by considering learning in the planning of order picking processes and in studying which aspects facilitate learning. This may contribute to a better predictability of order picker performance and may help to manage the system in such a way that learning is maximized and order picking time and costs are reduced (compare for production e.g., Anzanello and Fogliatto, 2011; Jaber and Bonney, 1999, 2011). For example, a dedicated storage assignment strategy may promote learning as workers get familiar with the location of stored products over time. This may lead to improved picker performance which, in turn, may reduce order lead time. If, in turn, a high number of contract or temporary workers are employed in order picking, only little learning may occur at the order pickers due to the constant labor turnover that takes place. Thus, each time a new worker enters the warehouse, he or she has to become familiar with the operations on site, which may reduce the performance of order picking systems. To study the effects of human learning on order picking efficiency, and to contribute to closing the research gap identified above, this paper develops an analytical model that helps to predict worker performance in order picking systems. The remainder of this paper is structured as follows: The next section reviews the related literature. Section 3 summarizes empirical evidence for learning in order picking and the impact of learning on order picking efficiency, and Section 4 formulates the assumptions made in developing the proposed model. Section 5 develops a mathematical model that considers learning in order picking, and Section 6 presents the results of a comprehensive numerical experiment. The paper concludes in Section 7.

2. Literature review 2.1. Order picking literature The manual picker-to-part process can be described as follows (de Koster et al., 2007): The order picker receives the order usually on a pick list, i.e. a paper-based list that specifies the items to be picked with respect to item identifications, item numbers and item locations. The picker walks (typically using a trolley or cart as a tool) to the shelve locations, picks the required items and returns to the depot, i.e. the place where order picking tours start and end and where collected items are dropped off, packed and shipped. After an order has been finished, the order is usually subjected to a quality control step. Many researchers assumed that order pickers spend most of their time on traveling in the warehouse to fulfill customer orders (e.g., Tompkins et al., 2010). As a result, planning models on the design and control of manual order picking systems have a major focus on the reduction of travel time or, equivalently, travel distance. An exhaustive review of this popular research stream is not within the scope of this paper, and the reader is referred to the reviews of Gu et al. (2007) and de Koster et al. (2007). In the following, we give a short summary of the four most important planning problems that have to be addressed in the design and control of picker-to-part order picking systems, namely layout design, routing, order batching, and storage assignment. These planning problems are usually addressed with the help of analytic models; simulation models, however, have been used more and more frequently in recent years (e.g., Basile et al., 2012; Chackelson et al., 2013). The first planning problem, layout design, is concerned with determining the configuration of the warehouse, which includes defining the number of aisles and cross aisles as well as their dimensions (Petersen, 2002; Roodbergen and Vis, 2006; Roodbergen et al., 2008). In many cases, warehouses have a rectangular shape. In recent studies, however, authors have analyzed alternative layouts for

order picking systems without conventional parallel pick aisles, such as fishbone and flying-V layouts (Pohl et al., 2011; Çelik and Süral, 2014) or U-shaped layouts (Glock and Grosse, 2012). Routing policies define the sequence in which the order picker retrieves required items. In a rectangular warehouse, routing order pickers is a special case of the Traveling Salesman Problem, which is why an optimal solution to the problem may be found with solution procedures that have been developed in this stream of research (Rattliff and Rosenthal, 1983; Roodbergen and de Koster, 2001). The majority of research has concentrated on developing heuristics for the routing of order pickers, as heuristics often lead to tours that are more convenient for the order picker (Hwang et al., 2004; Petersen and Aase, 2004; Theys et al., 2010). More recently, some authors have considered picker blocking and congestion in routing order pickers through the warehouse (Pan and Wu, 2012; Hong et al., 2012; Chen et al., 2013). A problem that is closely related to the routing problem is commonly referred to as order batching. Order batching helps to better utilize the carrying capacity of the order picker by consolidating or splitting individual orders, which can reduce travel time in many cases (see, for a review, Henn et al., 2012). Some authors developed algorithms to solve moderate-sized batching problems exactly (Gademann et al., 2001). As the order batching problem is NP-hard, most authors have concentrated on developing heuristic procedures for assigning items to batches (Hsieh and Huang, 2011; Grosse et al., 2014). The last optimization problem discussed here, storage assignment, determines how products should be assigned to storage locations. If a random storage assignment is used, items arriving at the warehouse are randomly assigned to an open location in the warehouse. Random storage is most appropriate under dynamic conditions, i.e. in situations where reliable data on item demand frequencies is not available (Tompkins et al., 2010), or in situations where a large product portfolio has to be stored and where warehouse managers desire a high utilization rate for the warehouse. A dedicated storage assignment strategy, in contrast, assigns products to fixed shelf locations based on certain item features, such as demand frequency, part number sequence or demand correlations (Frazelle, 2002; Glock and Grosse, 2012). A popular decision criterion in practice is that items with high demand and turnover frequencies should be located close to the depot (Gagliardi et al., 2008). When the number of items to be stored is high, a class-based storage method could be used. In this case, products are divided into classes and each class is assigned to a dedicated area of the warehouse. Storage within each class is random (Chan and Chan, 2011; Bottani et al., 2012). In a recent study, Grosse et al. (2013) studied the effect of storage reassignment decisions when learning and forgetting occur at the level of the order pickers. The authors investigated how changes in an existing storage assignment impact human learning and picker performance, and evaluated in which cases an existing storage assignment should be changed or not. This paper concentrated only on storage assignment and expressed learning in an aggregated form. More specifically, learning was modeled as a decreasing order picking time that occurs as the number of orders increases. To the best of the authors' knowledge, this paper is the only one that considered learning in an order picking planning model so far. The focus of the literature was instead on learning in industrial and logistics processes, which will be surveyed briefly in the next section. 2.2. Learning literature Learning effects have frequently been the subject of research in recent years. One of the first works in this area is the one of Wright (1936), who observed that unit production costs in airplane assembly reduce as the number of units produced increases, which he attributed to learning. Performance improvement that results from human learning can be modeled with the help of so-

Please cite this article as: Grosse, E.H., Glock, C.H., The effect of worker learning on manual order picking processes. International Journal of Production Economics (2015), http://dx.doi.org/10.1016/j.ijpe.2014.12.018i

E.H. Grosse, C.H. Glock / Int. J. Production Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

called learning curves, which may model learning at the individual, group or organizational level (Jaber, 2013; Glock and Jaber, 2014). Learning that results from repetitive manual work in industrial settings has been studied in empirical works, see for example Ittner et al. (2001), Macher and Mowery (2003) and Nembhard (2000), who studied manual assembly production processes. Further empirical evidence for learning in production can be found in Hinze and Olbina (2009) and Thomas et al. (1986), who studied construction processes, and in Kim and Seo (2009), who analyzed shipbuilding processes. Works that investigated learning in laboratory settings are the ones of Kvålseth (1978), Bailey (1989) and Shtub et al. (1993), for example. Due to the large number of analytical papers on learning curve models in production, the reader is referred to the reviews of Jaber (2013) and Anzanello and Fogliatto (2011). 2.3. Synthesis of both research streams The literature review illustrated that order picking and human learning have both been analyzed frequently in the past. It is surprising, however, that both research streams have been treated widely independently. Given the fact that order picking involves a high amount of repetitive manual work, the specific characteristics of human workers should be considered in the planning of operations processes (Grosse et al., 2015). Some authors already pointed out that human characteristics and behaviors, such as learning or forgetting, can be included in descriptive, simulation, and optimization models used for analyzing and improving operating systems (e.g., Boudreau et al., 2003; Neumann and Village, 2012). This has largely been overlooked by the order picking literature, although there is empirical evidence of learning in order picking, which is addressed in the following section.

3. Learning in order picking: empirical observations 3.1. Experimental investigations Although learning has not been studied in detail in the order picking literature, researchers and managers are well aware that order pickers may become familiar with the item assignment over time when a dedicated storage assignment is in use. Thus, worker familiarity with item locations has frequently been cited as a major advantage of dedicated storage assignments (Jane and Laih, 2005; de Koster et al., 2007; Bindi et al., 2009; van Zelst et al., 2009; Chuang et al., 2012), where worker familiarity implies human learning. Empirical evidence for the effect of learning on the performance of an order picker can be found in the study of Grosse and Glock (2013). The authors collected data in an order picking warehouse and fitted different learning curves to this data. They observed that order picker performance improved as the number of processed orders increased, which reduced the time that was required for identifying/searching and picking items. In the study of Grosse and Glock (2013), learning rates of order pickers varied between 84.25% and 97.67%, which means that the performance of an average order picker doubled after approximately 732 processed orders with an average learning rate of 93.09%. A similar result was obtained for the pick failure/error rate, which was found to decrease in the number of completed orders. In earlier studies, Bishu and Chen (1989) and Bishu et al. (1991; 1992) had investigated how cognitive ergonomic factors of items to be picked, such as color, position, coding and address information, influence human recognition and perception time. In an experimental study, the authors observed that learning occurs in the search for items. In a case study, Chakravorty (2009) observed learning effects after implementing a new material handling system. The author noted that workers gained experience over

3

time, which led to shorter lead times, reduced inventory and fewer mistakes in processing an order. As the theoretical knowledge on the effects of learning on order picking is still limited, we conducted a qualitative study in which order pickers and warehouse managers were interviewed on their perception and experience of order picking tasks. The results of the interviews are presented in the next section. 3.2. Expert interviews To gain insight into how workers perceive and warehouse managers observe learning in order picking, we conducted a set of semi-structured expert interviews. Semi-structured interviews allow interviewees to describe and explain their work in their own words. Such interviews rely on an interview guide, but questions may be adjusted based on the responses of the interviewees. Semistructured interviews were used to gather information on how workers perceive order picking tasks, to study which factors facilitate or hinder learning, and to assess the impact of worker learning in order picking (Grosse et al., in Press). Interviewees were selected from different companies that employ manual picker-to-part order picking systems and a dedicated storage assignment. Both order pickers performing the job as well as warehouse managers planning and supervising it were interviewed, which allowed us to gain insights into both perspectives. The expert interviews were not meant to be representative of some population; instead, due to the lack of empirical studies in this area, we aimed on contributing to earlier experimental investigations that observed learning in order picking to support the development of analytical models. Data was collected in 5 (2 order pickers and 3 warehouse managers) semi-structured, 45- to 60 min interviews, accompanied by field observations. The interview guide consisted of a set of questions which were left open-ended to avoid preconception and to create a dialog that supported spontaneous and long answers (Roulston, 2010). Through the chosen wording, the context and the open-ended nature of the questions, we attempted to reduce interviewee and response bias. Additional supporting questions, such as follow-ups and probes, were used to elicit more information on the topic (Kvale, 1996). After the interviews had been completed, audio recordings were transcribed and the transcripts were carefully read and analyzed by the research team. Subsequently, concepts were identified to draw relationships among the interview answers and factors that influence worker learning in order picking (Thomas, 2006). The authors compared, discussed and consolidated their codes, and no significant differences were identified (Miles and Huberman, 1994). From the interviews, the following propositions were derived. Proposition 1. Items with high demand frequency are subject to fast learning The interviewees revealed that order pickers perceive that their familiarity with the storage assignment and the item numbers improves over time. Especially items with high demand frequencies are subject to learning, whereas less familiarity was perceived for items with low demand frequencies. Proposition 2. A picker who is not familiar with item locations needs considerably more time for searching for items than an experienced picker. The interviewees stated that order pickers spend a lot of time searching for items when they are new on the job and estimated that they need about three times longer to fulfill an order than an experienced workmate. The interviewees indicated that working in an order picking warehouse consists mainly of searching items

Please cite this article as: Grosse, E.H., Glock, C.H., The effect of worker learning on manual order picking processes. International Journal of Production Economics (2015), http://dx.doi.org/10.1016/j.ijpe.2014.12.018i

E.H. Grosse, C.H. Glock / Int. J. Production Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

4

in the first few weeks on the job. Searching, thereby, refers both to orientation in the warehouse, such as finding aisles and shelve numbers, as well as to finding items within an aisle or on a shelve. The interviewees mentioned various factors that complicate searching for items, in particular (I) the high number of items stored in warehouses, which is difficult to memorize, (II) the high quantity of item numbers and their structure, which are confusing especially if items that are stored close to each other are labeled with similar item numbers, and (III) deviations from the dedicated storage assignment, i. e. items are not stored in their dedicated locations, which may become necessary if shortages in warehouse space occur.

wy

l

h

wy

Proposition 3. Pick errors decrease due to learning Interviewees pointed out that the pick error rate is higher when the order picker is new on the job and mentioned that learning reduces pick errors. Some companies tried to reduce pick errors by requiring the order pickers to check the orders before they were packed and shipped, which led to a substantial increase in pick time. The field observations that were conducted further indicated that worker experience mainly influences search time. The results of our empirical study indicate that learning is of high relevance in order picking, and that it should therefore be considered in planning order picking operations. In light of these findings, the next section describes the assumptions made in developing a mathematical planning model that considers learning in search time.

4. Problem description In the literature, the time that is required to complete an order in manual order picking is usually split up into four components, which are setup (administrative tasks at the beginning of a pick tour), travel (walking between storage locations), pick (extracting items) and search (identifying shelves and items) time (Tompkins et al., 2010). Travel time is often considered to be the dominant part. In light of the empirical observations presented in Section 3, however, it is reasonable to assume that search time is considerably higher for workers that are new on the job, and that search time reduces as more and more orders have been completed by the order picker. Thus, we model learning in search time with the help of a learning curve, which can express the improvement in productivity due to repetition of similar activities over time (Argote et al., 1990). This paper studies a conventional rectangular picker-to-part warehouse which has frequently been analyzed before (e.g., Petersen and Schmenner, 1999; Hwang et al., 2004) and which is often used in practice (see Fig. 1). Rectangular warehouses consist of several parallel aisles, front and back access aisles, and a depot in the front aisle. Each pick tour starts and ends at the depot. Cross aisles exist only at the respective ends of the parallel aisles, and there are no cross aisles inbetween. As is shown in Fig. 1, l is the length of the parallel aisles, h is the number of storage levels in each aisle, wa is the breadth of an aisle, wy is the breadth of the front and back aisle, and ws the breadth of each storage rack. The characteristic steps of the order picking process have already been described in Section 2.1. In developing the proposed model, the following assumptions are made: – In each storage location, one type of item is stored, and the assignment is dedicated. Replenishment is neglected. – The warehouse is a narrow-aisle warehouse, i.e. the gap that separates two adjacent aisles can be neglected in calculating the travel distance (Caron et al., 1998). – The horizontal travel distance within aisles is negligible, i.e. the pick time does not depend on the height of the shelves or the position of the items on the shelves (Hwang et al., 2004).

back aisle

wa

ws

front aisle

depot Fig. 1. Rectangular warehouse layout.

– Items are assigned to storage locations based on the demand frequencies of the items. This strategy is widely used in practice and has been shown to lead to good results (e.g., Le-Duc and de Koster, 2005, Glock and Grosse, 2012). Items with high demand are thus stored in locations near the depot. The warehouse is divided into zones (A, B and C) according to a within-aisle-storage strategy, i.e. the items with the highest demand frequencies are stored in the aisles closest to the depot (A), followed by aisles with a medium distance from the depot (B). Items with low demand frequencies are stored in the aisles farthest from the depot (C) (Petersen, 1999; Petersen and Aase, 2004). – Two zoning scenarios are studied, (a) several order pickers work simultaneously in the warehouse in all zones (picker blocking is neglected), and (b) one order picker is responsible for one zone. Workforce characteristics (initial search times and learning rates) are varied in the numerical study to evaluate the effect of workforce heterogeneity on average search time. – For routing order pickers through the warehouse, the S-shape routing policy is used, i.e. the order picker traverses each aisle that contains at least one pick completely and returns to the depot after all items assigned to this tour have been picked (Petersen, 1997). This routing policy is widely applied in practice (de Koster and van der Poort, 1998). A constant travel speed of the order picker is assumed (Petersen and Aase, 2004). – The order batching policy in effect follows the principle “firstcome-first-served” (de Koster et al., 1999), and the capacity of the pick device is sufficient to ensure that all items on a pick list can be picked in a single tour. In the next section, a mathematical model is developed to consider learning effects in order picking processes. 5. Model development 5.1. Definitions The following terminology will be used throughout the paper.

τi Δτ tSep tT i tSio tP i i, j I Jp an

total order picking time for item i (s) average total order picking time per item and day (s) setup time for pick list p (s) travel time to reach the storage position of item i (s) search time for item i for order picker o (s) pick time for item i (to extract items from shelves) (s) item indices with i, j¼1, …, I number of items stored in the warehouse number of items on pick list p aisle with n ¼1, …, N

Please cite this article as: Grosse, E.H., Glock, C.H., The effect of worker learning on manual order picking processes. International Journal of Production Economics (2015), http://dx.doi.org/10.1016/j.ijpe.2014.12.018i

E.H. Grosse, C.H. Glock / Int. J. Production Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

m meter N number of aisles in the warehouse dði; jÞ distance between the storage locations of item i and item j [meter], with i a j travel distance to fulfill all items on pick list p (m) dp p index of pick list p with p ¼1, …, P P total number of pick lists δio pick count of item i for order picker o. δio A ℕ\f0g individual learning exponent for order picker o bo LRo learning rate for order picker o o order picker index with o ¼ 1; …; O O number of order pickers employed in the warehouse z zone index with z ¼ 1; …; Z Z number of zones in the warehouse φt average search time per item and day t (s) βoz binary variable that expresses if worker o is assigned in zone z s seconds t day t with t¼1, …, T T number of days in the planning horizon wy breadth of the front and back aisles (m) wa breadth of the parallel aisles (m) ws breadth of each storage racks (m) h number of storage levels in each aisle l length of the parallel aisles (m) P(A) Probability that an item assigned to zone A is included in pick list p P(B) Probability that an item assigned to zone B is included in pick list p P(C) Probability that an item assigned to zone C is included in pick list p σ Possible combinations for assigning order pickers to zones v Worker travel speed (m/s)

5.2. Order picking time The time τi an order picker needs to retrieve one unit of item i, with i¼ 1,…, I, from its storage location can be expressed as the sum of setup time tSep (time to set up pick list p), travel time tT i (time needed to walk to the storage position of item i), as well as search tSi and pick time tP i per item i.

τi ¼ tSep þ tT i þ tSi þ tP i ðs=itemÞ

ð1Þ

Setup time is assumed to be the same for all pick lists: tSep ¼ tSe 8 p ¼ 1; …; P

ðs=pick listÞ

ð2Þ

Travel time is calculated according to the routing policy in use. The S-shape routing policy, which is used in this paper, can be described as follows:

5

end (Petersen and Schmenner, 1999). If dð0; iÞ is the rectilinear distance between the depot and the storage location of item i, dði; jÞ the rectilinear distance between the storage locations of item i and item j, and dðj; 0Þ the rectilinear distance between the last pick location on pick list p, then the travel distance to complete a pick list p, dp , is set as the sum of rectilinear cross- and within-aisle travel. Assuming a constant travel speed v of the order picker, travel time per pick list p can be derived as tT p ¼ dp v (s/pick list), and the travel time per item i as tT i ¼ tT p =J p

ðs=itemÞ

ð3Þ

Search time is assumed to follow a learning curve, and it is further assumed that learning occurs at the individual item level. Thus, the more frequently an item is picked, the better the order picker remembers the position of this item, and the shorter is the search time. To model learning, we use the learning curve of Wright (1936), which has frequently been used before as it expresses empirically observed learning rates well (e.g. Nembhard and Osothsilp, 2001, Grosse and Glock, 2013). Search time can consequently be calculated as follows:  bo

tSio ¼ tSδio ¼ tS1io δio

ðs=itemÞ

ð4Þ

Here, δio represents the number of times item i has been picked by order picker o. tSδio is the search time per item i for order picker o after δio picks, tS1io is the initial search time for item i when the order picker o is unfamiliar with the storage location (for example because he/she is new on the job), and bo is the learning exponent for order picker o, with 0 obo o 1. The time that is needed to physically extract item i is assumed constant irrespective of the location of the item on the shelf. It is calculated as follows: tP i ¼ tP; 8 i; i ¼ 1; …; I:

ðs=itemÞ

ð5Þ

5.3. Zoning If more than one order picker is employed for completing customer orders, then the warehouse may be divided into zones with one (or multiple) order pickers assigned to each zone. As some authors noted that one of the advantages of zoning is that order pickers get familiar with the designated zone, which implies learning, we model two possible zoning scenarios (de Koster et al., 2012). In the first scenario, several order pickers work simultaneously in the warehouse (in all zones). In the second scenario, order pickers are assigned to zones with one order picker being responsible for one zone (cf. Kováks, 2011). Zones (A, B and C) are determined as defined in Section 4. As order pickers handle more orders for individual items within their zones compared to the case of no zoning, the objective is to find a worker-zone assignment that leads to the lowest average search time. ( O Z 1; if worker o is assigned in zone z ∑ ∑ φoz βoz ; with βoz ¼ 0; otherwise o¼1z ¼1 ð6Þ

0. Initialization: Number all aisles N and begin with n ¼1. 1. Check if aisle an contains at least one item i on pick list p. (a) If yes, go to step 2. (b) If no and n oN, set n ¼n þ1 and repeat step 1. 2. Travers the aisle completely and go to all items in aisle an that are contained on pick list p. 3. Check if pick list p contains more items. (a) If yes, set n ¼n þ1 and go to step 1. (b) If no, go back to the depot.

and φoz expressing daily average search time per item of order picker o in zone z, subject to Σ Zz ¼ 1 βoz ¼ 1; o ¼ 1; …; O

According to this routing policy, the order picker enters only those aisles that contain at least one position on the pick list. The picker enters the aisle on one end and exits the aisle on the other

In this paper, we assume that different order pickers have different learning abilities, which leads to different values for bo and tS1io . It is clear that also in order picking, some workers learn faster than others

Σ Oo ¼ 1 βoz ¼ 1; z ¼ 1; …; Z Both zoning scenarios will be investigated separately in the numerical experiments in Section 6. 5.4. Order picker characteristics

Please cite this article as: Grosse, E.H., Glock, C.H., The effect of worker learning on manual order picking processes. International Journal of Production Economics (2015), http://dx.doi.org/10.1016/j.ijpe.2014.12.018i

E.H. Grosse, C.H. Glock / Int. J. Production Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

6

or some workers are subject to labor turnover and therefore acquire less knowledge than their fellow workers (Stratman et al., 2004; Min, 2007). Individual learning rates per order picker (LRo) can be calculated as LRo ¼ 2  bo (Jaber, 2013). The higher LRo, the slower the individual learns. The reader may refer to Grosse and Glock (2013) for a practical approach for measuring learning rates. To gain insight into the behavior of the developed model, an extensive numerical experiment is conducted in the following.

6. Numerical experimentation 6.1. Determination of parameters The parameter values used in the numerical experiment are based on empirical findings and results taken from the literature. The layout of the warehouse consists of 10 picking aisles (2 in zone A and 4 in zone B and C, respectively) with 1500 pick locations in total. The dimensions of the warehouse are measured as wy ¼3 m, wa ¼2 m, ws ¼ 1 m, h¼3 and l¼ 25 m as in Fig. 1. In addition, it is assumed that order pickers use a trolley with capacity of K¼50 items for transporting products. The maximum number of items per order was set to 25, such that the size of the pick list can vary as a result of the order batching strategy. Orders were generated randomly with different demand probabilities per item class. Item demand characteristics were set as follows: P(A)¼0.52, P(B)¼0.36, P(C)¼0.12, with P(A), for example, being the probability that an item assigned to zone A is demanded in an order. We simulated order picking operations for a timeframe of T¼300 days with 150 to 200 customer orders per day. We studied three different types of order pickers with LR1 ¼ 0:95, LR2 ¼ 0:90 and LR3 ¼ 0:85 (Grosse and Glock, 2013). Further, the following parameter values were assumed: v ¼ 1 m/s, tP¼10 s, tSe¼ 60 s tS1io ¼120 s for o¼ 1, 2, 3 and i¼1,…, I. Workforce characteristics (initial search time and learning rate) were varied in the numerical experiment to evaluate the effects of workforce heterogeneity on average search time. The numerical experiment was implemented in Java and run on a laptop with core i5 processor and 4 GB RAM. 6.2. Results To facilitate the discussion of our results, this section is divided into three subsections that discuss the effect of learning on order picking efficiency in two scenarios (see Section 4), i.e. (1) several order pickers work simultaneously in the warehouse (in all zones), and (2) one order picker is responsible for one zone.

In this line of thought, comparing the results of our model to the case where learning is neglected (which is the case in most available order picking models), it can be seen that considering learning would lead to a better model predictability and a better allocation of available resources. For example, comparing the difference in daily average search time per item for order picker 3 between day 1 and day 300 (see Table 1), the assumption of a constant value for search time per item (here 120 s.) compared to the case that worker learning is considered (here 30.38 s. after 300 days) leads to a significant overestimation of the time needed to fulfill pick orders. Regarding the average total order picking time per item and day, Δτ, our numerical experiment obtained a value of Δτ ¼55.14 for the case with worker learning and a value of Δτ ¼149.50 for the case without learning. In a next step, we varied workforce characteristics (initial search time and learning rate) to study the effects of workforce heterogeneity

120

o=1

o=2

o=3

110 100 90 80 70 60 50 40 30 20

Fig. 2. Average search time per item and day for three different types of order pickers.

Table 1 Numerical results for three different types of order pickers. o

φ1

bo

φ10

φ50

φ100

φ150

φ200

φ300

1 2 3

120 120 120

0.95 0.90 0.85

104.16 90.52 78.33

90.82 67.35 50.26

85.19 60.16 41.40

82.30 54.93 36.68

80.35 52.57 33.68

77.76 48.62 30.38

, = 2

120

6.2.1. No worker zone assignment in use As illustrated in Fig. 2, the daily average search time per item φd , which is calculated as the sum of search time for all items demanded per day, divided by the number of items demanded per day, decreased over time for each type of order picker. Numerical results of the experiment are presented in Table 1, which summarizes the daily average search time per item for the days 1, 10, 50, 100, 150, 200 and 300 for the three different order pickers. The slight variations in daily average search time per item that were observed can be explained by the modeled learning effect on the level of each individual item, as the number of items per order can vary and the learning effect per item decreases with an increasing number of retrievals due to the chosen Wright learning curve. The effect of ABC zoning on average search time is illustrated in Fig. 3, which shows the daily average search time per item for each item group (A, B and C) exemplarily for order picker 2. As can be seen, learning has the highest impact on items in zone A as those items are retrieved most frequently.

110 100 90 80 70 60 50 40 30 20

Fig. 3. Average search time per item and day for different warehouse zones for order picker 2.

Please cite this article as: Grosse, E.H., Glock, C.H., The effect of worker learning on manual order picking processes. International Journal of Production Economics (2015), http://dx.doi.org/10.1016/j.ijpe.2014.12.018i

E.H. Grosse, C.H. Glock / Int. J. Production Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

on average search time. tS1io was set to 60 s. for o¼1 to model an experienced worker, while the other parameters were kept as defined in Section 6.1. As can be seen in Fig. 4, order picker 3 reaches a lower daily average search time per item than order picker 1 after 74 days, whereas order picker 2 does not reach the performance of order picker 1 within the planning horizon of 300 days. In the following, we study the case where order picker 1 is an experienced (full time) worker, and the position of order picker 2 is staffed with varying temporary personnel that is only employed unregularly, for example for one month every three months. The parameters for this scenario are: LR1 ¼ 0:95, LR2 ¼ 0:90, tS1io ¼60 s. for o¼ 1, and tS1io ¼ 120 s. for o¼2. Fig. 5 illustrates the daily average search time per item and day for both types of order pickers. As can be seen, temporary workers do not have the chance to gain the experience of full time workers due to their limited tenure. Warehouse managers have to consider the portion of temporary workers in the workforce. Temporary workers can be assigned more flexibly and have considerably lower wage rates; however, their average order picking time may be significantly higher as compared to permanent workers due to lower learning opportunities. In this case, if learning is important for improving the performance of the system, it may make sense to invest in a stable workforce and to avoid high labor turnover. If temporary workers need to be employed, companies should verify whether worker training could be of help in improving learning. In addition, enabling order pickers to gain

120 110 100 90 80

o=1

o=2

o=3

70 60 50 40 30

7

experience may not only improve performance, but also improve quality in terms of reduced pick errors (Grosse and Glock, 2013). These results have several implications for warehouse managers. Firstly, although learning may be neglected in order picking planning models, workers learn in practice (where learning characteristics depend on the working conditions). Neglecting this effect may lead to wrong decisions in the planning of the order picking process, such as an unfavorable worker assignment, and warehouse managers may predict an incorrect operating time for a given set of orders. If learning is considered in planning order picking operations, in turn, then the accuracy of the plan increases, which may also have a positive effect on customer satisfaction. Secondly, the results indicate that learning in order picking should be promoted actively, for example by maintaining a dedicated storage assignment, by avoiding (unnecessary) worker turnover or by investing in worker training. Such managerial actions may help to utilize learning as good as possible.

6.2.2. Worker zone assignment in use This section studies the case where order pickers are assigned to zones, where one order picker is responsible for a single zone (compare Section 5.3). The question that arises here is which order picker should be assigned to which zone if order pickers have different learning parameters. Daily average search time for assigning different types of order pickers to different zones of the warehouse are summarized in Table 2. In our example, order pickers may be assigned to zones in six different ways (σ), see Table 3. As can be seen, the worker assignment combination σ3 leads to the lowest average daily search time in all zones for the planning horizon of T¼300. Combination σ5, for example, would lead to an increase in average daily search time of about 10%. This implies that order pickers with low learning rate (corresponds to fast learning) should be assigned to zone A, which are responsible for the highest share of warehouse turnover. In a next step, we studied the influence of workforce characteristics on the assignments of workers to zones by assuming that one order picker is an experienced one, with a lower initial search time but a higher learning rate, and that two order pickers are new on the job. The following parameters are assumed in addition to those introduced in Section 6.1: o ¼ 1; 2; 3 with

20

Fig. 4. Development of search time for order pickers with different experience levels.

120

o=1

Table 2 Numerical results for assigning each order picker to a specific zone. o

φ1

LRo

Zone

φ10

φ50

φ100

φ150

φ200

φ300

∅φ

1

120

0.95

A B C

94.53 103.45 109.12

80.79 89.41 95.38

75.76 83.97 89.60

73.00 80.99 86.56

71.15 78.97 84.45

68.58 76.14 81.47

75.41 83.52 89.09

2

120

0.90

A B C

72.78 88.95 97.03

53.22 65.61 74.64

46.66 57.79 66.12

43.23 53.57 61.34

41.01 50.89 58.24

38.05 47.20 54.28

46.70 57.53 65.58

3

120

0.85

A B C

55.57 74.25 88.77

34.25 47.28 58.39

27.94 38.78 47.89

24.83 34.57 42.76

22.89 31.95 39.46

20.38 28.45 35.24

28.51 39.22 47.94

o=2

110 100 90 80 70 60 50

Table 3 Average search time for combinations assigning order pickers to zones.

40 30 20

d Fig. 5. Average search time per item and day for a temporary and a full time worker.

Combination/zone

σ1

σ2

σ3

σ4

σ5

σ6

A B C Σ∅φ

1 2 3 180.88

2 1 3 178.17

3 2 1 175.13

1 3 2 180.21

2 3 1 193.89

3 1 2 177.60

Please cite this article as: Grosse, E.H., Glock, C.H., The effect of worker learning on manual order picking processes. International Journal of Production Economics (2015), http://dx.doi.org/10.1016/j.ijpe.2014.12.018i

E.H. Grosse, C.H. Glock / Int. J. Production Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

8

Table 4 Effects of workforce experience on worker zone assignment decisions. Combination/∅φ

σ1

σ2

σ3

σ4

σ5

σ6

tS1i (o ¼1) ¼ 120 tS1i (o ¼1) ¼ 60 tS1i (o ¼1) ¼ 30

180.88 150.20 129.09

178.17 143.30 120.45

175.13 137.05 113.06

180.21 149.41 128.11

193.89 156.05 131.35

177.60 142.33 119.34

research should seize this line of thought and integrate more human factors and ergonomic issues in analytic models of order picking to gain further insights into how human characteristics impact order picking processes (and vice versa).

References LR1 ¼ 0:95, LR2 ¼ 0:90 and LR3 ¼ 0:85, tS1io ¼ 60 for o¼ 1, and tS1io ¼120 for o ¼2 and 3. For the new scenario, the results of our numerical experiment show that worker assignment combination σ 3 still leads to the lowest average search time, even if tS1io is reduced to 30 for o ¼1, see Table 4. This implies that workers with the lowest learning rate (i.e. highest learning ability) in the workforce (with the assumed workforce parameters) should be assigned to the fast moving zone to gain experience, even though it would be more intuitive to assign the experienced worker(s) to zone A. This illustrates that if learning is not considered in planning order picking operations, workers may be assigned to incorrect zones, which may reduce the efficiency of the warehouse.

7. Conclusion This paper modeled worker learning in order picking processes. First, the relevance of worker learning in manual order picking was evaluated with the help of a literature study and complemented by expert interviews. In the literature, the impact of human learning on order picking efficiency has mostly been neglected in planning order picking operations. The results of the qualitative study showed that worker learning in order picking is highly relevant in practice, and that it is most often observed in the time that is needed for searching for items. Further, it is dominant for fast moving items. Based on the insights gained in the literature review and the qualitative study, an analytical model was developed to study the impact of worker learning on order picking efficiency. The learning effect was modeled on the level of the individual item, which seems to be an appropriate method to consider learning in order picking planning models. The results of a numerical study showed that worker learning has a significant impact on order picking efficiency. Thus, it should be considered in managing order picking warehouses. The advantages of order picking models that consider learning are a better predictability of lead times (e.g., for customer or production orders), a better allocation of available resources (e.g., workforce), and an explicit promotion of learning by identifying design factors that facilitate learning (e.g., dedicated storage assignment, investments in training or full time workforce). With the help of planning models as the one developed in this paper, workers can be employed in such a way that their individual characteristics are utilized best. This paper has limitations. Due to the lack of extensive empirical evidence, more studies are needed to ensure a proper estimation of model parameters, especially those related to learning. The results of this paper depend on the assumptions that were made in formulating the model. This paper showed, however, that learning should be considered when planning and optimizing picker-to-part operations in warehouses as models become more realistic. There are various opportunities to extend the model developed in this paper. Future research could, for example, study which factors are crucial to develop a logical storage assignment to promote worker learning. In addition, it may be worth evaluating the coherence between the costs of worker training measures and an improved learning rate in an investment model. Finally researchers could integrate the effects of boredom and forgetting in order picking planning models. Apart from this, future

Anzanello, M.J., Fogliatto, F.S., 2011. Learning curve models and applications: literature review and research directions. Int. J. Ind. Ergon. 41 (5), 573–583. Argote, L., Beckman, S.L., Epple, D., 1990. The persistence and transfer of learning in industrial settings. Manag. Sci. 36 (2), 140–154. Bailey, C.D., 1989. Forgetting and the learning curve: a laboratory study. Manag. Sci. 35 (3), 340–352. Baker, P., Perotti, S., 2008. UK Warehouse Benchmarking Report. Cranfield School of Management, Cranfield University, Cranfield, UK. Basile, F., Chiacchio, P., Del Grosso, D., 2012. A control oriented model for manualpick warehouses. Control Eng. Pract. 20 (12), 1426–1437. Bindi, F., Manzini, R., Pareschi, A., Regattieri, A., 2009. Similarity-based storage allocation rules in an order picking system: An application to the food service industry. Int. J. Logist. Res. Appl. 12 (3), 233–247. Bishu, R.R., Chen, Y., 1989. Learning and transfer effects in simulated industrial information processing tasks. Int. J. Ind. Ergon. 4 (3), 237–243. Bishu, R.R., Donohue, B., Murphy, P., 1991. Cognitive ergonomics of a mail order filling company: Part 1—influence of colour, position and highlighting on recognition time. Appl. Ergon. 22 (6), 367–372. Bishu, R.R., Donohue, B., Murphy, P., 1992. Cognitive ergonomics of a mail order filling company: Part 2—influence of shelf coding and address information on acquisition time. Appl. Ergon. 23 (2), 115–120. Bottani, E., Cecconi, M., Vignali, G., Montanari, R., 2012. Optimisation of storage allocation in order picking operations through a genetic algorithm. Int. J. Logist. Res. Appl.: Lead. J. Supply Chain Manag. 15 (2), 127–146. Boudreau, J., Hopp, W., McClain, J.O., Thomas, L.J., 2003. On the interface between operations and human resources management. Manuf. Serv. Oper. Manag. 5 (3), 179–202. Caron, F., Marchet, G., Perego, A., 1998. Routing policies and COI-based storage policies in picker-to-part systems. Int. J. Prod. Res. 36 (3), 713–732. Çelik, M., Süral, H., 2014. Order picking under random and turnover-based storage policies in fishbone aisle warehouses. IIE Trans. 46 (3), 283–300. Chackelson, C., Errasti, A., Ciprés, D., Lahoz, F., 2013. Evaluating order picking performance trade-offs by configuring main operating strategies in a retail distributor: a design of experiments approach. Int. J. Prod. Res. 51 (20), 6097–6109. Chakravorty, S.S., 2009. Improving distribution operations: implementation of material handling systems. Int. J. Prod. Res. 122 (1), 89–106. Chan, F.T.S., Chan, H.K., 2011. Improving the productivity of order picking of a manual-pick and multi-level rack distribution warehouse through the implementation of class-based storage. Expert Syst. Appl. 38 (3), 2686–2700. Chen, F., Wang, H., Qi, C., Xie, Y., 2013. An ant colony optimization routing algorithm for two order pickers with congestion consideration. Comput. Ind. Eng. 66 (1), 77–85. Chuang, Y.-F., Lee, H.-T., Laih, Y.-C., 2012. Item-associated cluster assignment model on storage allocation problems. Comput. Ind. Eng. 63 (4), 1171–1177. de Koster, R., Le-Duc, T., Roodbergen, K.J., 2007. Design and control of warehouse order picking: a literature review. Eur. J. Oper. Res. 182 (2), 481–501. de Koster, R., Le-Duc, T., Zaerpour, N., 2012. Determining the number of zones in a pick-and-sort order picking system. Int. J. Prod. Res. 50 (3), 757–771. de Koster, R., van der Poort, E., 1998. Routing orderpickers in a warehouse: a comparison between optimal and heuristic solutions. IIE Trans. 30 (5), 469–480. de Koster, R., van der Poort, E.S., Wolters, M., 1999. Efficient orderbatching methods in warehouses. Int. J. Prod. Res. 37 (7), 1479–1504. Frazelle, E.A., 2002. World-class Warehousing and Material Handling. McGraw-Hill, New York. Gademann, N., van den Berg, J.P., van der Hoff, H.H., 2001. An order batching algorithm for wave picking in a parallel-aisle warehouse. IIE Trans. 33 (5), 385–398. Gagliardi, J.-P., Ruiz, A., Renaud, J., 2008. Space allocation and stock replenishment synchronization in a distribution center. Int. J. Prod. Econ. 115 (1), 19–27. Glock, C.H., Grosse, E.H., 2012. Storage policies and order picking strategies in Ushaped order-picking systems with a movable base. Int. J. Prod. Res. 50 (16), 4344–4357. Glock, C.H., Jaber, M.Y., 2014. A group learning curve model with and without worker turnover. J. Model. Manag. 9 (2), 179–199. Grosse, E.H., Glock, C.H., 2013. An experimental investigation of learning effects in order picking systems. J. Manuf. Technol. Manag. 24 (6), 850–872. Grosse, E.H., Glock, C.H., Jaber, M.Y., 2013. The effect of worker learning and forgetting on storage reassignment decisions in order picking systems. Comput. Ind. Eng. 66 (4), 653–662. Grosse, E.H., Glock, C.H., Ballester-Ripoll, R., 2014. A simulated annealing approach for the joint order batching and order picker routing problem with weight restrictions. Int. J. Oper. Quant. Manag. 20 (2), 65–83.

Please cite this article as: Grosse, E.H., Glock, C.H., The effect of worker learning on manual order picking processes. International Journal of Production Economics (2015), http://dx.doi.org/10.1016/j.ijpe.2014.12.018i

E.H. Grosse, C.H. Glock / Int. J. Production Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎ Grosse, E.H., Glock, C.H., Jaber, M.Y, Neumann, W.P., 2015. Incorporating human factors in order picking planning models: Framework and research opportunities. Int. J. Prod. Res. 53 (3), 695–717. Grosse, E.H., Dixon, S.M., Neumann, W.P., Glock, C.H., Using qualitative interviewing to examine human factors in warehouse order picking: Technical note. Int. J. Logist. Syst. Manage. (in Press). Gu, J., Goetschalckx, M., McGinnis, L.F., 2007. Research on warehouse operation: a comprehensive review. Eur. J. Oper. Res. 177 (1), 1–21. Henn, S., Koch, S., Wäscher, G., 2012. Order batching in order picking warehouses: a survey of solution approaches. In: Manzini, R. (Ed.), Warehousing in the Global Supply Chain. Springer, London. Hinze, J., Olbina, S., 2009. Empirical analysis of the learning curve principle in prestressed concrete piles. J. Construct. Eng. Manag. 135 (5), 425–431. Hong, S., Johnson, A.L., Peters, B.A., 2012. Batch picking in narrow-aisle order picking systems with consideration for picker blocking. Eur. J. Oper. Res. 221 (3), 557–570. Hsieh, L.-F., Huang, Y.C., 2011. New batch construction heuristics to optimise the performance of order picking systems. Int. J. Prod. Econ. 131 (2), 618–630. Hwang, H., Oh, Y.H., Lee, Y.K., 2004. An evaluation of routing policies for orderpicking operations in a low-level picker to part system. Int. J. Prod. Res. 42 (18), 3873–3889. Ittner, C.D., Nagar, V., Rajan, M.V., 2001. An empirical examination of dynamic quality-based learning models. Manag. Sci. 47 (4), 563–578. Jaber, M.Y., 2013. Learning and forgetting models and their applications. In: Badiru, A.B. (Ed.), Handbook of Industrial and Systems Engineering, 2nd ed. CRC Press, Boca Raton, FL. Jaber, M.Y., Bonney, M., 1999. The economic manufacture/order quantity (EMQ/ EOQ) and the learning curve: past, present, and future. Int. J. Prod. Econ. 59 (1), 93–102. Jaber, M.Y., Bonney, M., 2011. The lot sizing problem and the learning curve: a review. In: Jaber, M.Y. (Ed.), Learning Curves—Theory, Models, and Applications. CRC Press, Boca Raton, pp. 265–291. Jaber, M.Y., Glock, C.H., 2013. A learning curve for tasks with cognitive and motor elements. Comput. Ind. Eng. 64 (3), 866–871. Jaber, M.Y., Bonney, M., Moualek, I., 2009. Lot sizing with learning, forgetting and entropy cost. Int. J. Prod. Econ. 118 (1), 19–25. Jane, C.-C., Laih, Y.-W., 2005. A clustering algorithm for item assignment in a synchronized zone order picking system. Eur. J. Oper. Res. 166 (2), 489–496. Kim, I., Seo, H.L., 2009. Depreciation and transfer of knowledge: an empirical exploration of a shipbuilding process. Int. J. Prod. Res. 47 (7), 1857–1876. Kováks, A., 2011. Optimizing the storage assignment in a warehouse served by milkrunlogistics. Int. J. Prod. Econ. 133 (1), 312–318. Kvale, S., 1996. InterViews: an Introduction to Qualitative Research Interviewing. Sage, Thousand Oaks, CA. Kvålseth, T.O., 1978. The effect of task complexity on the human learning function. Int. J. Prod. Res. 16 (5), 427–435. Le-Duc, T., de Koster, R., 2005. Travel distance estimation and storage zone optimization in a 2-block class-based storage strategy warehouse. Int. J. Prod. Res. 43 (17), 3561–3581. Macher, J.T., Mowery, D.C., 2003. “Managing” learning by doing: an empirical study in semiconductor manufacturing. J. Prod. Innov. Manag. 20 (5), 391–410. Miles, M.B., Huberman, A.M., 1994. Qualitative data analysis: an expanded sourcebook. Sage, Thousand Oaks, CA. Min, H., 2007. Examining sources of warehouse employee turnover. Int. J. Phys. Distrib. Logist. Manag. 37 (5), 375–388.

9

Napolitano, M., 2012. 2012 warehouse/DC operations survey: mixed signals. Mod. Mater. Handl. 51 (11), 48–56. Nembhard, D.A., 2000. The effects of task complexity and experience on learning and forgetting: a field study. Hum. Factor.: J. Hum. Factor. Ergon. Soc. 42 (2), 272–286. Nembhard, D.A., Osothsilp, N., 2001. An empirical comparison of forgetting models. IEEE Trans. Eng. Manag. 48 (3), 283–291. Neumann, W.P., Village, J., 2012. Ergonomics action research II: a framework for integrating HF into work system design. Ergonomics 55 (10), 1140–1156. Pan, J.C.-H., Wu, M.-H., 2012. Throughput analysis for order picking system with multiple pickers and aisle congestion considerations. Comput. Oper. Res. 39 (7), 1661–1672. Petersen, C.G., 1997. An evaluation of order picking routeing policies. Int. J. Oper. Prod. Manag. 17 (11), 1098–1111. Petersen, C.G., 1999. The impact of routing and storage policies on warehouse efficiency. Int. J. Oper. Prod. Manag. 10 (10), 1053–1064. Petersen, C.G., 2002. Considerations in order picking zone configuration. Int. J. Oper. Prod. Manag. 22 (7), 793–805. Petersen, C.G., Aase, G., 2004. A comparison of picking, storage, and routing policies in manual order picking. Int. J. Prod. Econ. 92 (1), 11–19. Petersen, C.G., Schmenner, R.W., 1999. An evaluation of routing and volume-based storage policies in an order picking operation. Decis. Sci. 30 (2), 481–501. Pohl, L.M., Meller, R.D., Gue, K.R., 2011. Turnover-based storage in non-traditional unitload warehouse designs. IIE Trans. 43 (10), 703–720. Rattliff, H.D., Rosenthal, A.S., 1983. Orderpicking in a rectangular warehouse: a solvable case of the traveling salesman problem. Oper. Res. 31 (3), 507–521. Roodbergen, K.J., de Koster, R., 2001. Routing methods for warehouses with multiple cross aisles. Int. J. Prod. Res. 29 (9), 1865–1883. Roodbergen, K.J., Sharp, G.P., Vis, I.F.A., 2008. Designing the layout structure of manual order picking areas in warehouses. IIE Trans. 40 (11), 1032–1045. Roodbergen, K.J., Vis, I.F.A., 2006. A model for warehouse layout. IIE Trans. 38 (10), 799–811. Roodbergen, K.J., Vis, I.F.A., 2009. A survey of literature on automated storage and retrieval systems. Eur. J. Oper. Res. 194 (2), 343–362. Roulston, K., 2010. Considering quality in qualitative interviewing. Qual. Res. 10 (2), 199–228. Shtub, A., Levin, N., Globerson, S., 1993. Learning and forgetting industrial skills: an experimental model. Int. J. Hum. Factor Manuf. 3 (3), 293–305. Stratman, J.K., Roth, A.V., Gilland, W.G., 2004. The deployment of temporary production workers in assembly operations: a case study of the hidden costs of learning and forgetting. J. Oper. Manag. 21 (6), 689–707. Theys, C., Braysy, O., Dullaert, W., Raa, B., 2010. Using a TSP heuristic for routing order pickers in warehouses. Eur. J. Oper. Res. 200 (3), 755–763. Thomas, D.R., 2006. A general inductive approach for qualitative data analysis. Am. J. Eval. 27 (2), 237–246. Thomas, H.R., Asce, M., Mathews, C.T., Ward, J.G., 1986. Learning curve models of construction productivity. J. Construct. Eng. Manag. 112 (2), 245–258. Tompkins, J.A., White, Y.A., Bozer, E.H., Tanchoco, J.M.A., 2010. Facilities Planning, 4th ed.. John Wiley and Sons, Hoboken, N.J.. van Zelst, S., van Donselaar, K., van Woensel, T., Broekmeulen, R., Fransoo, J., 2009. Logistics drivers for shelf stacking in grocery retail stores: potential for efficiency improvement. Int. J. Prod. Econ. 121 (2), 620–632. Wright, T.P., 1936. Factors affecting the cost of airplanes. J. Aeronaut. Sci. 3 (4), 122–128.

Please cite this article as: Grosse, E.H., Glock, C.H., The effect of worker learning on manual order picking processes. International Journal of Production Economics (2015), http://dx.doi.org/10.1016/j.ijpe.2014.12.018i