You are on page 1of 31

Accepted Manuscript

Human energy expenditure in order picking storage assignment: A bi-objective
method

Daria Battini, Christoph H. Glock, Eric H. Grosse, Alessandro Persona, Fabio
Sgarbossa

PII: S0360-8352(16)30013-4
DOI: http://dx.doi.org/10.1016/j.cie.2016.01.020
Reference: CAIE 4243

To appear in: Computers & Industrial Engineering

Received Date: 12 June 2015
Revised Date: 26 January 2016
Accepted Date: 27 January 2016

Please cite this article as: Battini, D., Glock, C.H., Grosse, E.H., Persona, A., Sgarbossa, F., Human energy
expenditure in order picking storage assignment: A bi-objective method, Computers & Industrial Engineering
(2016), doi: http://dx.doi.org/10.1016/j.cie.2016.01.020

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers
we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting proof before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Human energy expenditure in order picking storage assignment:

a bi-objective method

Daria Battini1, Christoph H. Glock2, Eric H. Grosse2, Alessandro Persona1, Fabio Sgarbossa1*

1 Department of Management and Engineering, University of Padova, Stradella San Nicola, 3 36100 Vicenza

– Italy
2 Institute of Production and Supply Chain Management, Technische Universität Darmstadt, Hochschulstr.

1, 64289 Darmstadt – Germany

Abstract

Order picking is the most time-consuming and labor-intensive activity in warehousing. Due to the need to

frequently handle items, order picking requires high human energy expenditure and poses a risk

environment for workers to develop musculoskeletal disorders. The storage assignment policy in use has a

significant impact on human energy expenditure and fatigue during the picking process, but this impact is

usually not considered in (management-oriented) decision support models for storage assignment.

This paper models and analyzes the integration of human energy expenditure as one dimension of
ergonomics into the storage assignment problem using a bi-objective approach that considers both total order

picking time and human energy expenditure. Time and energy expenditure depend on the main features of

the order picking system, such as item characteristics, item popularity, order profiles, and physical

dimensions of the shelf and locations. Pareto frontiers are constructed to understand the impact of the

storage assignment policy on the objective functions. Subsequently, a quantitative approach is developed to

integrate the energy expenditure rate into the time estimation for a general order picking system based on the

introduction of rest allowance. Finally, the results of the model are analyzed and suggestions for the practical

application of the model are presented.

Keywords: Order picking, storage assignment, ergonomics, human factors, energy expenditure, order

picking time

1

1. Introduction

Nowadays, the optimization of warehouse processes becomes more and more important for

several reasons. First, warehouses are seen as cost centers that tie up large sums of capital.

Secondly, warehouse activities require a high amount of human work both in terms of time and

physical effort. Modern global supply chains have increased the complexity of warehouse

processes in recent years, requiring a great variety of products in small quantities, high service

levels and short delivery times. This poses a challenge to logistics managers, who have directed
their attention to improving the efficiency of feeding systems, the design of supermarkets, or

logistics processes that take place in warehouses (Battini et al., 2009). In light of these

developments, order picking, i.e. the process of retrieving items from the warehouse to complete

customer and production orders, continues to be seen as one of the main drivers of warehouse

flexibility and efficiency.

Despite the various advantages warehouse automation offers, order picking is still characterized

by a high share of manual human work in many companies. Picker-to-part systems, where the

operators walk or drive along the aisles of the warehouse to pick items from storage locations, are
still dominant in industry (de Koster et al., 2007; Napolitano, 2012). Typical manual tasks in picker-

to-part systems involve walking, grasping, lifting, lowering, sorting, pushing, and pulling.

Excessive and repetitive manual materials handling in warehouses, however, renders the order

picking process a risk environment for workers to develop musculoskeletal disorders (MSDs),

among them back injuries, sprains, strains, and/or tears (Punnett and Wegman, 2004). MSDs are

the most common work-related health problem, with 25% of European workers complaining about

backache and 23% about muscular pains, for example (Schneider and Irastorza, 2010). In the US,

MSDs accounted for 34% of all injury and illness cases in 2012, where stock and material movers

had the highest number of MSD cases (BLS, 2013). The economic burden of MSDs, as measured by

compensation costs, lost wages, and lost productivity, are estimated to range between $45 and $54

billion annually in the US (NRC, 2001), and to account for up to 2% of the Gross National Product

in the EU, with €23.9 billion solely in Germany (Schneider and Irastorza, 2010). To reduce the risk

of MSDs, various ergonomic methods and tools have been developed in the past to support
engineers and managers in evaluating ergonomic issues in an industrial context.

To support the design and management of picker-to-part systems, various mathematical decision

support models have been developed over the last decades (see, for a review, de Koster et al.,

2007). The most common objective of these decision support models is to minimize the travel
2

2013). Due to the impact of ergonomic aspects on the performance of order picking. this paper develops an integrated storage assignment method that considers both economic aspects (in terms of order picking time) and ergonomic aspects (in terms of human energy expenditure). which is an important planning problem in order picking. 2010. Thus. 2015. and the assignment of products to storage spaces. Boysen et al. 2013). a promising research opportunity is the integration of ergonomic aspects into decision support models for storage assignment. including ergonomics evaluations in the human operations analysis is a win-win approach due to the interaction between productivity. To guarantee a high level of productivity and efficiency and to make sure that decision support models reflect reality as good as possible. some authors noted that considering ergonomic aspects is not only relevant for reducing health risks. the internal configuration of shelves and aisles. (2011). but neglected the influence of the order picking process on the human operator. (2015).distance or the picking time of an order by determining the facility layout.. Finnsgård et al.. as well as a variable search time and a pick error rate depending on human learning (Grosse and Glock. The main objectives of this research and the methods used to achieve them can be summarized as follows: 1) Investigate how the storage assignment policy impacts total order picking time and energy expenditure in order to estimate the trade-off between time and energy expenditure for different storage assignment solutions. showed that the large majority of these decision support models solely concentrated on the (short-term) economic impact of order picking. To contribute to closing the research gap highlighted above. the routing of order pickers. 2011. To carry out this analysis.. 3 . In addition.. A recent study of Grosse et al. Finnsgård and Wänström. As was demonstrated in Battini et al. motion efficiency and operational safety. 2015). several authors recently called for the integration of ergonomic aspects into decision support models for order picking (Grosse et al. it is necessary to consider human factors in addition to economic aspects in designing labor-intensive manual order picking systems. however. but that it may also improve the performance and quality of order picking significantly (Grosse et al. Existing works in this area include a variable pick and materials handling time depending on materials exposure and its position on the shelf (Neumann and Medbo. a bi-objective model is developed. 2015).

The remainder of this paper is organized as follows: the next section reviews the related literature. where the second alternative is often referred to as a dedicated storage assignment. In general. As the focus of this paper is on the storage assignment problem. de Koster et al. items arriving at the warehouse are assigned randomly to open locations in the warehouse (Petersen. the literature on order picking had a major focus on the development of decision support models. for example). The construction of Pareto frontiers supports the analysis of the outcomes. Section 4 models and analyzes the bi-objective optimization problem. Section 3 illustrates the problem context and the main assumptions. 2007).. and thus order picking time. where products are assigned to the next available storage position from the depot (de Koster et al. This approach allows the calculation of an additional rest time that is required to maintain a low level of fatigue. and the assignment of products to storage positions. we give a short overview of the literature on this problem category in the following. products can be assigned to storage locations either randomly or based on certain criteria (such as the demand frequency of an item or the distance from the depot to the storage locations. Literature review Over the last decades. The paper concludes in Section 5. When a random storage assignment is in use. for reviews. Gu et al. and then integrates the time and energy concepts by introducing rest allowance. This is based on the introduction of the rest allowance formulation (Price. A special case of random storage assignment is closest open location assignment. The major advantage of random 4 . the pooling and batching of orders. which can be used to plan the picking process in practice (see. A parametric analysis of its behavior permits the analysis of general results. 2. the determination of the warehouse layout. 2007). The focus of decision support models for order picking is usually on one or more of the following planning problems: the routing of order pickers through the warehouse. 2) Analyze the trade-off between order picking time and energy expenditure by varying typical warehouse parameters. which converts the energy expenditure rate into a relative time value. 3) Integrate energy expenditure into the function of total order picking time. The main objective of decision support models for these problems is to reduce the average distance the order picker needs to travel. 4) Optimize the storage assignment problem based on the new integrated order fulfillment time function. 1997). 1990). such as the shelf-shape or the COI curve.. 2007..

2008). Several works describe the manual order picking process as a risk environment for workers to develop MSDs (e. Battini et al. where products that are often ordered together are stored next to each other (Chuang et al.. In addition. and each class is then assigned to a certain zone of the warehouse. This method has the advantage that operators get familiar with the item assignment over time. The literature on storage assignment discussed above widely focused on developing methods that reduce travel distance and order picking time. 2007. 2013). 2013). 2015a). 2012. which usually reduces the time they require for searching for items in the warehouse (Grosse and Glock. 2013. Chackelson et al. In the case of class-based storage. storage is random.. (2015). fast moving items (so-called A-items) are usually assigned to the zone that is nearest to the depot.. which combines random and dedicated storage. for example according to demand characteristics. 2000. This concept considers that less time is required for retrieving items from easily accessible positions than from positions that are difficult to access.. Within each class area. It has also been demonstrated that long-term muscular pain. as was discussed in the recent review of Grosse et al.. 2012). all items are assigned to fixed locations in the warehouse. In this case. Grosse and Glock.. the well-being of warehouse operators has been studied in the ergonomics literature.. Glock and Grosse. Lavender et al. 2005). Ene and Öztürk. 2012). 2015). If demand rates change over time.storage assignment is high space utilization. In contrast to the management-oriented literature. where frequently-ordered products are assigned to storage locations near the depot (Rao and Adil. Another storage assignment method that aims at reducing travel distance is family storage. Kadefors and Forsman. A popular criterion for dedicated storage assignment is the turnover of products. all products are assigned to a number of different classes. then a turnover-based storage assignment would require a change in the product assignment according to the new demand characteristics.g. Another option is to implement a class-based storage assignment. where high-demand items are stored in the area between a picker’s waist and shoulders (Petersen et al. a storage assignment method that helps to reduce total fulfillment time is the concept of golden zone storage.. depends on the discomfort felt by warehouse operators (Hamber-van Reenen et al. such as the minimization of worker discomfort or the prevention of MSDs. Other objectives. have largely been overlooked in managerial decision support models for the storage assignment problem. In the case of a dedicated storage assignment. 2012. 2013. which helps to reduce travel distance (Manzini et al. such as MSDs. The literature discusses many methods and tools that support engineers in ergonomics 5 . which might increase operation costs and lower performance as operators have to relearn the new assignment (Grosse et al..

g. The formulation introduced by Garg et al. especially those that occur during order picking. Konz. such as predetermined time motion systems (PTMS) (Zandin. Sharp and Rosenberger. cranking. and they are typically applied to study macro-activities without analyzing each elementary activity that composes the examined task in detail. 1982. Warehousing activities. e. cycling. where the total time to perform a task is the sum of a standard predetermined time and the function of the singular movements which compose the operation in question. called Predetermined Motion Energy System (PMES). 2009.. which can be divided into self- report. Rohmert (1973). 1994. Andreoni et al. and the length of the required rest period varies with workload and energy expenditure.. 2002). Based on the energy expenditure approach. the results of these methods are semi-quantitative indices. appropriate rest periods are recommended to avoid fatigue. Its simplicity has led to a wide use in many areas. OWAS. 1978. 1994). 1999. 1990. where each method has its specific limitations.evaluations of manual tasks in industrial contexts. This method is based on the energy expenditure assessment of standard operations execution as a function of oxygen consumption.. Another method for ergonomics evaluations of manual tasks was introduced by Garg et al. and only by using more than one ergonomics evaluation method. Louhevaara et al. the NIOSH lifting equation. advantages and applications (Li and Buckle. 1992. Waters et al. it is possible to have a globally accurate assessment. Rohmert’s method is more appropriate for heavy dynamic muscular works (e. virtual simulations. These methods can be applied using different tools. and Price (1990). and arm movements) and it converts the energy expenditure in rest time with an exponential function of energy and time. while other approaches use a linear 6 .. and the Borg-Scale (Occhipinti. some formulations have been developed to estimate the relative rest allowance.. Waters et al. 1998b). for an easy estimation of the energy expenditure of typical manual tasks. or standing in a fixed position. In general. It is similar to traditional time estimation methods. discomfort. Konz. among others. Battini et al. David. 2005). 1990. 2009). and MSDs (Price. involve all body parts. 1998.g. Generally. 1998a. Borg. (1978) permits the estimation of the energy expenditure of each single task execution. High workloads lead to a significant consumption of metabolic cost. walking. Müller (1953). such as in the medical. Popular ergonomics evaluation methods are OCRA. Konz 1998b. operator observation. such as lowering or lifting an item. causing health risks for the workers (Price. grasping an item from a certain distance. by Spitzer (1951). ISO 8996). military and industrial sectors (Garg et al. (2015b) recently introduced a new technique. (1978). and direct measurement methods.

and then extend the model to the case with multiple picks per order (Section 4. The storage locations have fixed and equal dimensions. 2015. we deduce that. each item is stocked in a unique location. 2015). and especially in storage assignment models. 7 . but restrict our analysis to a single aisle consisting of a single (possibly long) shelf at a time.formulation. Petersen. Even though the return routing policy often leads to long travel distances. Mohapatra et al. 1997). This paper uses the formula introduced by Price (1990) and adapts the equation developed by Müller (1953). to our best knowledge. In this paper. and the operators can pick items from each of them. with the depot on the left side and example location numbers. the rest allowance formulation of Price (1990) is used to develop an integrated approach that considers both order picking time and energy consumed during the picking activities. A scenario as the one described here frequently occurs when each order picker is assigned to one specific aisle. From the literature review presented above. 2008). As widely used in industrial problems where more than one objective has to be optimized (Abedi et al. 3. Pan and Shih. the application of a bi-objective model permits to compare and analyze the storage assignment solutions based on two functions: the first one considers the total order picking time and the second one estimates the energy expenditure. Figure 1 illustrates the two- dimensional structure of the shelf. we consider a warehouse with multiple parallel aisles. no prior work included the rest allowance concept in warehouse design and management.3).. Then. where the latter permits us to consider ergonomic aspects in the model.4). There are different vertical and horizontal storage levels (see Section 4. We first model the case where an order consists of a single pick (Section 4. it is suitable for routing order pickers in our case given the location of the depot and the fact that only a single aisle without any cross-aisles is considered. where order pickers leave the aisles on the side where they entered it (cf. or when each order is divided into sub-orders that are fulfilled per aisle to avoid picker blocking (cf. The operators pick orders using a return routing policy. The depot (I/O point) is located at one end of the shelf. In this aisle.3).. Problem description This paper studies the storage assignment problem for a picker-to-part order picking system and directs special attention to the energy expenditure of the order picker during the picking process.

but completely different in terms of energy expenditure. products can be assigned to storage locations according to two different storage assignment methods. which is expressed as the sum of travel and order picking time. 8 . many solution methodologies have been developed. namely time-based and energy-based storage assignment. In the case of a time-based storage assignment. which represents the set of optimal solutions. the objective is to minimize the total time required to fulfil an order. For bi-objective optimization problems as the one discussed here. which forces the decision maker to find a trade-off between both objectives. the objective is to minimize the total energy the operator needs to spend to fulfil an order. 0601 location jk … 0201 0202 0203 … … jk 0101 0102 0103 0104 … 0132 Figure 1. out of which most are based on the study of Pareto frontiers. In the scenario investigated in this paper. storage assignment models are developed in the following that minimize two objective functions: a) total order picking time and b) total energy expenditure for order fulfillment. (1978). Generic shelf with storage locations and depot (I/O point). In fact. such as the traditional direct measurement. The time to complete an order is estimated using well-known time measurement techniques. Based on the assumptions stated above. The time that is needed to retrieve an item from its storage location and the energy that needs to be spent for performing the pick may differ from location to location. In the case of an energy-based storage assignment. The result is a bi-objective optimization problem. picking from two different levels may be equal in terms of time. work sampling and PMTS. where two different objective functions have different individually optimal solutions. while energy expenditure for the completion of an order is assessed using the formulations introduced by Garg et al. and for different popularity functions in terms of order lines pick frequency.

1. calculated as 9 .….…. with .…. Definitions The following notations are used throughout the paper: Indices = total number of stocked items = total number of horizontal locations = total number of storage levels = total number of lines per order = item index with i = 1. which is the number of lines per order (in case of multiple picks per order) Product demand parameters = popularity of item i.J k = storage level index with k = 1. each line is dedicated to one item i stocked in location jk based on . expressed as pick frequency = cumulative popularity function Storage assignment variable = binary variable.4. Model development 4.K storage assignment policy index with t for the time-based storage assignment policy and e for the energy-based one order line index.n = horizontal location index with j = 1. with = weight used in the weighted sum method in bi-objective optimization Time notations = walking speed [m/s] = horizontal distance [m] between the depot and location jk = horizontal (walking) time [s] required to reach location jk from the depot.

= shelf-shape parameter = rest allowance [%] for a specific storage assignment policy Energy notations (for the energy expenditure formulations. considering multiple picks per order. calculated as: = horizontal (walking) time [s] required to reach the location where the item assigned to line is stocked (in case of multiple picks per order) = time [s] required for the vertical movement along the shelf to reach and physically extract one or more items from the location where the item assigned to line is stocked (in case of multiple picks per order) = total order picking time [s]. estimated as: . = time [s] needed for the vertical movement along the shelf to reach and physically extract one or more items in location jk. calculated as = energy expenditure [kcal/task] for horizontal movements to reach the location where the item assigned to line is stocked (in case of multiple picks per order) 10 . estimated as: = total energy expenditure [kcal/task] to pick an item stocked in location j. (1978)) = energy expenditure [kcal/task] for walking from the depot to location jk = energy expenditure [kcal/task] for standing posture during = energy expenditure [kcal/task] for physically retrieving items from storage level k = energy expenditure [kcal/task] for standing posture during = energy expenditure [kcal/task] for horizontal movements to reach location jk. The measurement value of time depends on the number of items to pick = time [s] required to complete the overall picking process (for walking and performing the pick) for an item stocked in location jk. for a specific storage assignment policy = average total order picking time values resulting from the Monte Carlo simulation. please see Garg et al. estimated using time measurement techniques. = energy expenditure [kcal/task] for vertical movements to pick an item stocked in location jk.

1 0 0 0. 20-60 means that 20% of the items are responsible for 60% of all picks.7 G 20-60 0. where.4 0.6 0.4 G 20-90 0.9 0. 20-70. Popularity functions expressing item demand characteristics. for a specific storage assignment policy. 20-90 (see Figure 2). namely 20-60.6 G 20-70 0.3 0. we define the two main objectives of our model as minimizing total order picking time.8 1 % of items Figure 2. for example. In order to carry out the bi-objective analysis.8 0. 4. E: (1) (2) where: (3) 11 .2. 1 % of cumulative orderliness picking frequency G 0.5 G 20-80 0. . and total order picking energy expenditure. = average total energy expenditure values resulting from the Monte Carlo simulation. = energy expenditure [kcal/task] required for the vertical movement along the shelf to reach and physically extract one or more items in location where the item assigned to line is stocked (in case of multiple picks per order) = total energy expenditure [kcal/task] considering multiple picks per order.2 0. Basic model: the case of a single pick per order To define item demand characteristics. 20-80. let us assume several cumulative popularity functions as in Bender (1981).2 0.

(1) and (2). in the sense that no solution in the set is dominated by any other feasible solution for time and energy. such as a minimum value of energy expenditure. we adapt Eqs. and we apply the so-called weighted sum method to Eq. Let us define the utopia points and as the solutions of the single objective functions modelled by Eqs. (1) and (2) to provide two non- dimensional objective functions and . (1) and (2) refer to the case where an order consists of a single item. Other constraints could be added to the model in order to consider certain upper limits. which is one of the most common approaches in multi-objective optimization: (6) (7) (8) 12 . the optimal storage assignment is found by minimizing the total order picking time. We use a non-dominated optimization approach to identify a set of feasible solutions that are equal-rank optimal. In this study. the most frequently demanded items should be assigned to locations with low values of required energy . 2004). Traditionally. or other lower limits. we did not include such constraints to develop a general model that permits analyzing all possible solutions and deriving general results for the problem. (4) Note that Eqs. Using the utopia points. (5). The set of optimal solutions is estimated using the weighted sum method based on the following function transformation (Marler and Arora. such as a maximum total order picking time for each operator. which would make it necessary to consider the influence of the energy expenditure level on order picking time. In turn. Pareto frontiers are analyzed. with the consequence that the most frequently demanded items are located near the depot (see Section 2). if the objective is to minimize the energy expenditure of the overall picking process. The bi-objective function we intend to minimize is thus formulated as: (5) To investigate the storage assignment problem with a bi-objective approach.

where only energy expenditure is considered in the optimization. we compare the objective values with respect to the minimum value that results from optimizing only the respective (single) objective with different priorities. as illustrated by curves EF1 and EF2 in Figure 3.where is the weight defined by the decision maker. If a Pareto frontier has a horizontal or vertical shape. each solution of the Pareto frontier consists of two values: one related to time and one related to energy. based on his/her own preference. Pareto frontiers with a curve shape as in EF3 in Figure 3 represent the case where there is a strong interdependency between the objectives. Constructing the Pareto frontiers permits us to understand the trade-off between the optimal solutions for the time and the energy objective. where only time minimization is considered as an objective. and energy optimization for energy (2)). the shape of Pareto frontiers supports understanding the impact of each approach on the adopted solution. Examples of Pareto frontiers for bi-objective (F1 and F2) optimization. Two special cases are . or the influence is a minor one. we vary the value of the weight from 0 to 1. one objective either does not influence the other at all. and . In contrast. EF2 F2 EF1 EF3 F1 Figure 3. 13 . and they are normalized to the relative optimal value obtained using the single objective (time optimization for time (1). To build the Pareto frontier. Typically. In the following. In other words.

96 meters (m).63. which is motivated by a scenario observed in practice. 1.2 kg per picked item in this example and base our calculations on average values for the assumed parameter values. which is defined as the ratio of twice the length of the entire aisle from one end to the other end to the average vertical movement time to reach a location: (9) 4.25. Other parameters useful for the bi-objective analysis are related to the energy expenditure formulations. bending or stretching (Petersen et al. 2011. Numerical examples and Pareto frontiers analysis The proposed model is applied to a numerical example. Order picking time and energy expenditure could be correlated in many ways. and it is set depending on the height of the different levels to 6. require less time for being retrieved as compared to items that require kneeling. 0. is normalized to one picked item.To generalize the results of our model. This is due to the fact that the location of items on the shelves affects pick time and energy expenditure. 2. 1.5. Prior research has 14 . such as those located between the operator’s waist and shoulder (golden zone). 5. This can also be assumed for the energy expenditure required to pick an item. Finnsgård and Wänström. the shelf-shape parameter S is 4. As can be seen. 1. where each location is 0. but it is interesting to observe that in case items are picked from different storage levels.3. 2013).5. 2005. 1978). 0. .7. The observed vertical movement pick time per level.. we assume a male operator weighting 75 kg and a load of 0.30.. 1. The warehouse studied in this example consists of a single shelf with six different stocking levels. Consequently. 4.. Figure 4 illustrates the different values of time and energy expenditure required to pick an item from each storage location. and 6 seconds. The shelf dimension is measured with the help of the so- called shelf-shape parameter S. Finnsgård et al. as more energy is required to retrieve items from lower locations due to the lowering of the entire body with squat or stoop movements (Garg et al. and 32 locations for each level. the energy expenditure is completely different also if the execution time of the pick activity is the same.305 m wide.58. the darker cells have higher values of time and energy expenditure. Easily accessible items. namely 0. in this situation. The walking speed is set to 1 m/s.94. at different heights. we define the Pareto frontiers by varying the popularity function and the shelf dimensions. Keeping in mind that the energy expenditure is highly individual.

we define the colored spectrum for the energy expenditure values in Figure 4b.399 kcal/task.demonstrated (Garg et al. and we use the objective weighting method to define the set of non-dominated solutions composing the Pareto frontier (Marler and Arora. their values are more correlated for the case of a high value of the shelf-shape parameter (i. in particular. at 0. 1978) that the energy expenditure required for retrieving items from lower positions is higher than for retrieving items from higher locations. as was described earlier. is about 0. at 1. we normalize the function values with respect to the value of each objective function resulting from optimizing the single-objective formulations (1) and (2). To facilitate reading the colored spectrum. In our case. we classify all locations into five groups.. Colored spectrum of time and energy expenditure needed to retrieve items from specific locations.263 kcal/task. Figure 4a shows the different colored zones based on the ratio between the time required to pick an item from each location and the maximum time necessary to pick the item from the farthest location. Based on the same assumption. we calculate each feasible solution for the bi-objective function (5).96 meters. Since the total order picking time and the total energy expenditure to pick an item from a location consider both the horizontal (walking along the shelf) and vertical movements (grasping items).. a longer shelf). the energy expenditure required for picking an item from the lower position. a) Colored pick time spectrum T 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 06 ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### LEGEND 05 ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### from 0 to 20% of maximum value 04 ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### from 20 to 40% of maximum value 03 ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### from 40 to 60% of maximum value 02 ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### from 60 to 80% of maximum value 01 ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### from 80 to 100% of maximum value b) Colored energy expenditure spectrum E 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 06 ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### LEGEND 05 ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### from 0 to 20% of maximum value 04 ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### from 20 to 40% of maximum value 03 ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### from 40 to 60% of maximum value 02 ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### from 60 to 80% of maximum value 01 ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### ### from 80 to 100% of maximum value Figure 4. 15 . The time for performing the pick is equal to 6 seconds in both cases. 2004). is about 0.e. while picking an item from the higher position. As explained before.25 meters above the floor. Subsequently.

In this case. respectively. Regarding the optimal solution for the time-based and energy-based storage assignment models formulated in Eqs. (1) and (2). and 90% of all orders. it can be seen that the maximum difference is about 2% in time and 2. the industrial manager can remove all solutions from the Pareto frontier that lead to excessive energy expenditure. Cumulative popularity function G Optimal energy-based solution Optimal time-based solution Figure 5. Few changes result for the 20-90 16 . The results are shown in Figure 6. there are no significant differences between the popularity functions.5% higher than for the solution that results for the energy- based assignment. the Pareto frontiers are very similar both with regard to shape and normalized optimal values. if a constraint on the maximum value of energy expenditure is necessary.5% in energy. 20-70. We can affirm the same coherence for the energy-based assignment: Energy expenditure within the time-based storage assignment is 2. 70%. also in this case. the order fulfillment time in employing the energy-based storage assignment is 2% higher than the total time that results if the time-based assignment is used. In fact. The solution to be implemented would then be selected from the remaining ones. the Pareto frontiers are analyzed for varying values of the shelf-shape parameter ( ). Pareto frontier for the numerical example. 80%. 20-60. In general. we note that there are no significant differences between the Pareto frontiers defined by different popularity functions. 20-80 and 20-90). In a next step. It can be seen that. One aspect that has to be kept in mind when applying this analysis to an industrial context is that it may be necessary to consider a maximum energy expenditure to limit accumulated fatigue and thus injury risk for the order picker. where 20% of the items are responsible for 60%. In fact.Figure 5 illustrates the Pareto frontier for alternative values of the cumulative popularity function (namely.

This means that the optimal solutions defined by Eqs. the differences between the single- objective optimal solutions are greater. the order picking time and the energy expenditure for performing the pick are stronger correlated.. This means that for shelves with short aisles or with a high vertical movement time. longer aisles). where the energy expenditure resulting from travelling along the aisles is more significant than the energy expenditure required for performing the actual pick.popularity curves. no differences in the curves of the Pareto frontiers were found for varying values of . For higher values of (i. up to a maximum value of about 6% in time and about 7% in the energy objective. In particular. we conclude that shelves with a low shape parameter lead to higher differences between the single-objective optimal solutions. This means that moving from one solution to another in the middle region of the Pareto frontier involves a significant reduction of one objective in favor of the other.e. which indicates that the different solutions are more strongly influenced by the weights of the single-objective functions. where the Pareto frontiers are more curved. In general. the Pareto frontier is very limited and close to the origin of the graph (see the lower part of Figure 6). As a result. 17 . As to the impact of the shelf-shape parameter . (1) and (2) for the time and the energy objectives are similar. the optimal solution obtained with the time- based approach leads to 6% higher energy consumption than the solution that results from optimizing the energy-based model.

Shelf-shape S = 1 Shelf-shape S = 2 Shelf-shape S = 4 Shelf-shape S = 8 Cumulative popularity function G Figure 6. Pareto frontiers for varying values of the shelf-shape parameter . 18 .

After picking the last item E. the risk of injuries may be reduced in this way.The results highlight the importance of considering the impact of the storage assignment policy on other objectives than time. E in sequence while travelling along the aisle. As can be seen. an energy-based assignment may increase order picking time. The next section extends the proposed model by studying multiple picks per order. 1992). . a time-based storage assignment may lead to an efficient order picking processes in terms of time. but reduces oxygen consumption. .and energy-based objectives in storage assignment models is necessary to implement efficient and sustainable order picking processes. but may increase accumulated fatigue due to higher energy expenditure. and that the rest allowances of the operators need to be considered to make sure that an appropriate work-life- balance is maintained in practice. In addition. Ayoub. 4. On the other hand. 0601 A C location jk … 0201 0202 0203 … … jk 0101 B 0102 0103 0104 … D E 0132 Figure 7. Completing an order with multiple picks in an aisle requires that the operator walks to the location farthest from the depot. which leads to higher order picking cost. where the operator starts from the I/O point and picks the items A. D.4 Model extension: the case of multiple picks per order This section extends the model developed above to account for multiple picks per order. the operator goes back to the depot (I/O). too (cf. for a generic order with multiple picks with lines. Consequently. Scheme of the order picking system with multiple picks per order. and that he/she then returns to the depot carrying the picked items (or pushing a trolley). The results suggest that integrating time. fatigue and necessary rest times of the operators. B. and energy expenditure. as there is evidence that exceeding acceptable energy expenditure and fatigue limits may be linked to a person’s injury potential. formulations are adjusted as follows: (10) 19 . Figure 7 illustrates a scheme for the case of multiple picks per order. C. the total order picking time.

which says that each item is stocked in a unique location. (1978). (10). which are functions of the expected energy expenditure. For this purpose. from 1 to 10. characterized by the following time and energy expenditure values: . as the average of the total order picking time values calculated for each scenario using Eq. In addition. .000 orders with different numbers of lines. 20-70. we study the scenario described in Section 3. but also to limit the computation time. vertical movements . we employ a Monte Carlo simulation. we generated a random number using a uniform distribution over [0. we generate 10. 20 . This way. which involved in-depth discussions about model assumptions and results. and we associate it to each line of each order. Again. (11). we use Price’s (1990) formulation that was already introduced above. each line of each order is linked to a single location. To generate data for the simulation. the model was built on findings from the literature (see Section 2) and tested with a comprehensive amount of simulation runs. as the average of the energy expenditure values calculated for each scenario using Eq. and MSDs. Assuming a fixed storage assignment policy (i. We impose this number of scenarios in order to have enough data to perform an accurate analysis. and it corresponds to a particular location. . This value is equal to the value of the particular popularity functions under analysis. horizontal movements . discomfort. In the simulation. .. with 192 locations in total. as explained before. we also consider the appropriate rest periods in order to prevent the effects of fatigue. 20-80 and 20-90. In the following. vertical time . time-based or energy-based). which consists of a shelf composed of 6 levels and 32 locations for each level. 1]. horizontal time . The model was validated using the experience of the research team. This is based on the assumption introduced in Section 3. Energy expenditure results from the execution of the order picking activities and is estimated using the equations introduced by Garg et al.e. . (11) In this situation. we calculate the integrated total order fulfillment time as the sum of two terms: the expected total order picking time . and the rest allowance formulations of Price (1990). based on the popularity functions 20-60. In order to estimate the integrated total order fulfillment time and energy expenditure.

(15).. removing all storage assignment solutions with an excessive energy expenditure value. Konz 1998b. which illustrates the savings calculated with the help of Eq. 1990. the integrated total order fulfillment time. the percentage savings. and varying numbers of lines per order. again for a varying shelf-shape parameter . The average energy expenditure rate . can be derived as follows (Garg et al. Each graph represents a specific value of the shelf-shape parameter and illustrates the savings for varying the cumulative popularity function . and vice versa. which is used for the rest allowance estimation. 1978): (12) Consequently. (15) The results are shown in Figure 8. the rest allowance can be calculated as follows: (13) Eq. a maximum permissible energy expenditure can be considered as a constraint. is calculated.3 . . 1998a. (13) requires that the energy expenditure is expressed in kcal/min. between and of the two storage assignment policies. Also in this step of the study. 21 . can be calculated as: (14) To gain insights into the behavior of the model. ISO 8996).000 data sets studied in the simulation.both the time-based and the energy-based storage assignments are analyzed. Consequently. including also the rest allowance. was assumed to be equal to 4. as in Price (1990). varying popularity functions. . Konz. and as well as are calculated for all 10. then the time- based storage assignment outperforms the energy-based one. as indicated by international standards and previous research (Price. In general. if the savings are positive.

22 . curves using the integrated approach for varying popularity functions. number of lines per order and the shelf-shape parameter. Shelf-shape S = 1 % order lines Shelf-shape S = 2 % order lines Shelf-shape S = 4 % order lines Shelf-shape S = 8 % order lines Cumulative popularity function G Figure 8.

Moreover. the rest period is a function of the energy expenditure rate. In case a 20-90 popularity function is used. and consequently the average walking distance is shorter. where walking typically requires less energy than picking. In the case of picking activities. Conclusion This paper investigated the storage assignment problem in manual picker-to-part order picking systems and considered human energy expenditure in addition to order processing time. First. the time-based policy outperforms the energy-based one. we observe that for a low numbers of lines per order. with no dependence on the number of lines per order. 5. the time-based policy is more convenient as the probability to pick items from the closest location is higher. For medium values of S. in contrast. but more time-consuming. with low impact on energy expenditure. only a few researchers have developed integrated approaches. while walking is usually less energy-expensive. such as picking or bending. if the number of lines exceeds a threshold of about 3. the energy-based storage assignment policy leads to better results than those resulting from the time- based one. an integrated approach that considers the human energy expenditure required for picking an order in 23 . the time-based policy gives lower values of integrated total order fulfillment time. which is typical for warehouses with long shelves or a low number of picked items per line. a recent paper showed that works that integrate ergonomic aspects into managerial decision support models in order picking are rare.. These outcomes are affected by the energy expenditure required to perform elementary manual activities. and called for more research on this topic (Grosse et al. In fact. a bi-objective analysis was carried out. the energy expenditure rate is relevant because metabolic cost is high and pick execution time is quite low. the difference between the two policies is not significant. 1990). and Pareto frontiers were defined for understanding the trade-off between a time-based and an energy-based storage assignment. as indicated by the rest allowance equation (Price. Subsequently. In turn.The results demonstrate that for high values of S. In case of lower values of S. This is the case as the time required to reach the farthest location dominates the total order picking time. 2015). walking and picking involve different levels of physical effort at the operator’s side. such that it depends on the ratio of energy expenditure to execution time. In addition. Although ergonomic issues in manual materials handling have quite often been studied in human factors engineering in the past.

rest allowances and injury risk.g. Lim 2011. it is preferable to adopt an energy-based storage assignment policy. as it influences the distance the order pickers have to carry items or push or pull (a potentially heavy) trolley. on the availability of operators to gain further 24 . Strategies for reducing travel distance. the concept of energy expenditure may be useful.. This made it possible to compare the time-based and energy-based storage assignments and to investigate under which conditions the time-based storage assignment outperforms the energy-based one. Changing the shape of the popularity curve by increasing the percentage of total pick frequency for the first 20% of items. from 20-60 to 20-90. and following the rest allowance.g. mainly due to the high impact of energy expenditure during picking on the total energy consumption. with respect to the number and shape of aisles considered). We generally affirm that for a low number of lines per order. which can also have a major influence on energy expenditure. One example is the routing of order pickers through the warehouse. a parametric analysis was carried out in which we varied the dimensions of the shelf. for example. respectively. 2012). The order profile affects the energy and time assessment of each location and consequently it could lead to different results.a storage assignment model was developed. future research should further focus on the link between energy expenditure limits. In addition. which reduces the total order fulfillment time. Moreover.. could be an appropriate measure to improve the ergonomic performance of the warehouse as well. Bartholdi et al. in particular. especially the number of items to be picked per line of an order. the use of a rest allowance function (Price. Further research could analyze the impact of orders profiles. 2009. In this line of thought. this policy becomes convenient also for larger numbers of lines per order. a part of our future work will be to study the effect of storage assignment methods on the risk of developing MSDs using biomechanical modelling of exposure levels and. 1990) permitted the inclusion of ergonomic aspects in the operations time assessment in a quantitative way. (1978). because it involves lower rest allowances. the order profile and the popularity curve. especially if combined with a low shelf-shape parameter. Future research could also study different warehouse layouts than the one considered here (e. Again. From the energy expenditure concept for performing manual activities introduced by Garg et al. on the relative efficiency of different storage assignment policies. such as bucket brigades (e. it is necessary to investigate how other relevant design aspects in warehousing can impact the performance of human operators and thus the order picking system. which might lead to insights into how a warehouse should be designed to reduce energy expenditure. Finally.

. (2015a).. M. Eisenstein.... International Journal of Industrial Ergonomics. D. H. 1077-1093. Moreover. 56(4). Order picking system design: the storage assignment and travel distance estimation (SA&TDE) joint method. This will extend the presented research to include. International Journal of Production Research. (2009). further research is needed to improve the energy expenditure formulation.. Springer Berlin Heidelberg. 591-598).. Bi-objective optimisation for scheduling the identical parallel batch-processing machines with arbitrary job sizes. References Abedi. R.. 713-728. International Journal of Production Research. M. (2011).. Mazzola. Bartholdi. M. Persona.. similar formulations may be developed using electromyography. Ciani.. A. O. unequal job release times and capacity limits. Andreoni. E. M.. 121(1). 25 .J. Lim. (2015).D.F. Ergonomics. which enables researchers to collect data on actual muscles' activities and the applied strength during the elementary order picking activities. Method for movement and gesture assessment (MMGA) in ergonomics. Battini. 41(1). Seidgar. Zambetti. 35(7-8). In Digital Human Modeling (pp. (2009).. 53(6). Y. G. M. J. A useful approach may be to consider both the ergonomic evaluation and the economic aspects of the problem in an integrated manner. and Preatoni... Ayoub.. and Sgarbossa. and Sgarbossa. Finally. M. Costa. Deterministic chaos in a model of discrete manufacturing.. Persona. H. M. (1992). Faccio. Romero. the economic cost impact of injuries. 53(4).. (2009). D. & Bijani. A. International Journal of Production Economics. D.. & Sgarbossa. Problems and solutions in manual materials handling: the state of the art. A. F. D. human errors and absenteeism for different exposure levels. Faccio. New methodological framework to improve productivity and ergonomics in assembly system design. 293-299. Naval Research Logistics. 1680-1711. Design of the optimal feeding policy in an assembly system. F. M. F. Persona. F.insights into how the order picking process affects worker well-being. Battini. for example. 233-254. Battini. Fazlollahtabar. 30-42.. Calzavara.

G. Laih. A.gov/iif/oshcdnew. Caron. www. 713-732... 60(5-8). L. D. International Journal of Production Research. Item-associated cluster assignment model on storage allocation problems.. Finnsgård.-F. C. A. C. C. 787-797. USDL-13-2257. 1-22.. (2013). Le-Duc. Wänström.. (2007). International Journal of Production Research.. (2000).. Design and control of warehouse order picking: A literature review. 51(20). Storage location assignment and order picking optimization in the automotive industry. S. M. and Roodbergen. Persona. Finnsgård. European Journal of Operational Research. G. 7253-7274. 1171-1177. Y. 49(24). Evaluating order picking performance trade-offs by configuring main operating strategies in a retail distributor: A Design of Experiments approach. Chuang.-C.-T. (2013). and Öztürk. and Kauderer. literature review and research agenda. Ergonomic methods for assessing exposure to risk factors for work-related musculoskeletal disorders. Ene.. (2012).. 55(3). Borg. G. 63(4). Chackelson. Marchet. Ciprés. Delorme. 1789-1798. Boysen. 481-501. H. 51(6). International Journal of Production Research. C. M. 107-120. T. (1998). Part logistics in the automotive industry: Decision problems. Y. 190-199. R. W. and Lahoz. F. Optimal layout in low-level picker-to-part systems. A. Impact of materials exposure on assembly workstation performance.. N. David. F. C. Medbo. 377-381. 36(3). D..htm Caron. (1982). J. (2011). G. X. A. (2015). Emde.Battini.. Marchet. Occupational Medicine. (2012). Medicine and Science in Sports and Exercise. Bureau of Labor Statistics (BLS) (2013). Routing policies and COI-based storage policies in picker-to-part systems. K.. 101-117. Errasti. International Journal of Production Research. International Journal of Advanced Manufacturing Technology. Nonfatal occupational injuries and illnesses requiring days away from work. Ergonomics in assembly line balancing based on energy expenditure: a multi-objective model.. European Journal of Operational Research.. A.. (2005). S. International Journal of Production Research. (2015b). 182(2). Factors impacting manual picking on assembly lines: An experiment in the automotive industry.. F. Hoeck. 2012. Computers and Industrial Engineering. and Wänström. C. F. and Perego.. 38(1). A. Psychophysical bases of perceived exertion. and Perego.P. De Koster. Lee. N. 26 . International Journal of Production Research. Dolgui. and Neumann.bls. 14(5). 6097-6109.. 242(1). & Sgarbossa..

Grosse.. E. 1-21. W. International Journal of Production Research. S. 653-662. Konz. 36(6). H. Ferguson. H. 170 (Part C). (2000). Work/rest: Part II–The scientific basis (knowledge base) for the guide. (2012).. ISO. Kadefors. International journal of industrial ergonomics. ISO 8996. Journal of Occupational and Environmental Hygiene. 22(1). International Journal of Production Economics. Marras. Journal of Manufacturing Technology Management. 27 . Y. Glock. Glock. 661- 674. 53(3). 24(6). Jaber.. 25(4). E. S. Computers & Industrial Engineering. and Forsman. S. (2015). (1978). Y. (2015). 2014. International Journal of Industrial Ergonomics. Glock. International Journal of Production Research.C. E. H. M. Geneva (2014) Jones. Ergonomics — Determination of metabolic heat production. 850-872. E. F. E. C. Lavender. Gu. G. Incorporating human factors in order picking planning models: Framework and research opportunities. L.Garg. B.. Grosse. Glock. S. R. M. P. and Herrin... (2012). C. Work/rest: Part I–Guidelines for the practitioner. (2007). Ergonomic Evaluation of Complex Work: A Participative Approach Employing Video–Computer Interaction. 695-717. E. D. Prediction of metabolic rates for manual materials handling jobs. D. Grosse. 882-890.. Goetschalckx. and Neumann. The American Industrial Hygiene Association Journal. and Yang. R.. (1998a). The effect of worker learning on manual order picking processes. Storage policies and order picking strategies in U-shaped order- picking systems with a movable base. A. (2004). C.. 39(8). M. H. and Battiste. 67-71. (2013). Golden Retrieval.. Splittstoesser. International Journal of Industrial Ergonomics. Grosse. and McGinnis. Konz. J. Research on warehouse operation: A comprehensive review. Jaber. C. H. H. H. 37-41. An experimental investigation of learning effects in order picking systems. The effect of worker learning and forgetting on storage reassignment decisions in order picking systems. European Journal of Operational Research. C. 50(16). 435-445. A. S. A. 9(7). H. 4344-4357. E. W.. Chaffin. 177(1). 66(4). Exemplified in a Study of Order Picking. Industrial Engineering. T. G. 22(1).. H. (2013). Glock. M. 73-99. H. 450-459. (1998b). Developing Physical Exposure-based Back Injury Risk Models Applicable to Manual Handling Jobs in Distribution Centers. Grosse..

P. 2012 warehouse/DC operations survey: Mixed signals. (1992). P. Current techniques for assessing physical exposure to work-related musculoskeletal risks. 1098-1111. Petersen. Shih. 26(6). 41(9). with emphasis on posture-based methods.): Warehousing in the global supply chain. 205-215. S. (1997). International Journal of Industrial Ergonomics. Experimental Physiology. and Heiser. K. E. T. Operations Research. OCRA: a concise index for the assessment of exposure to repetitive movements of the upper limbs. (2004). Mohapatra. National Research Council (NRC) (2001). 48-56. R.. National Academy Press. G. 59(6). 53(6). Washington DC.. Occhipinti. Musculoskeletal disorders in the workplace . Evaluation of the throughput of a multiple-picker order picking system with congestion consideration. 1539-1545. S. Multi-objective process planning and scheduling using controlled elitist non-dominated sorting genetic algorithm. J. V. Napolitano. Y. G. M. The physiological basis of rest pauses in heavy work. P. Neumann. 997-1012. Y. Ergonomics. 25(10). A.-H. S. (2010). A. Modern Materials Handling 51(11). 17(11). 541-548. C. Siu.. (1953). (2008). C. (2015). and Helminen. 1290-1311. International Journal of Operations & Production Management. 40(5). 369-395. 38(4). D. OWAS: a method for the evaluation of postural load during work.. Cellular bucket brigades. Marler. 71-85. Structural and Multidisciplinary Optimization. and Tiwari. (2005). L. Centre for Occupational Safety. Order-picking by cellular bucket brigades: A case study. and Medbo. Survey of multi-objective optimization methods for engineering..Li. Pan. International Journal of Operations & Production Management. Müller.-H. Springer.. An evaluation of order picking routeing policies. London. R.. P.. C. P. narrow bins. M. 28 . Improving order picking performance utilizing slotting and golden zone storage. 674-695. (Ed.F.C. 1712-1735.F. (2011). (2012). G. In: Manzini. Petersen. Nayak. Hinkkanen. 379-389. 55(2). T. W.. International Journal of Production Research. Lim. (2012). R. and Buckle. Institute of Occupational Health. Computers & Industrial Engineering. K. Ergonomic and technical aspects in the redesign of material supply systems: Big boxes vs.. Lim. Suurnäkki. 42(5).low back and upper extremities. Arora. Louhevaara. Ergonomics. Kumar. (1999). pp. (1998). J.. E.

E. New York. S. H. 311-317. (1994). US Department of Health and Human Services. A. Ergonomics. 158-162. J.. A. P.. G. ARMY RESEARCH INST OF ENVIRONMENTAL MEDICINE NATICK MA MILITARY NUTRITION DIV. 2. CRC Press. (1951). Rao. X. H. (2008). National Institute for Occupational Safety and Health. and Adil. 4(3). 637-648. (2013). W. T. Centers for Disease Control and Prevention. Work-related musculoskeletal disorders: the epidemiologic evidence and the debate. A. W. Blatter. Division of Biomedical and Behavioral Science. Waters. 51(5). K. H. Sharp. D. Physiologische Grundlagen für den Erholungszuschlag bei Schwerarbeit. Journal of Electromyography and Kinesiology. L. European Agency for Safety and Health at Work. B. and Irastorza. 21(4). 94-110). and Bongers. M.. Luxembourg. Does musculoskeletal discomfort at work predict future musculoskeletal pain?.. (2002). Calculating relaxation allowances for construction operatives—Part 1: Metabolic cost.. (2006). M. Rohmert. M. USARIEM/MN-MISC-06-31). H. Punnett. 51(16). & Knapik. 14(1). and Wegman. M.facts and figures. OSH in figures: work-related musculoskeletal disorders in the EU . 13-23. and Garg. K. Class-based storage with exact S-shaped traversal routeing in low- level picker-to-part systems.. International Journal of Production Research. Applied Ergonomics. J. R.Price.. Hamberg-van Reenen. Putz-Anderson. Applied ergonomics. V. Zandin. REFA- Nachrichten. (2010). MOST work measurement systems. Common military task: materials handling (No. B. Applications manual for the revised NIOSH lifting equation (pp.. S. (1990). D. Problems of determination of rest allowances part 2: Determining rest allowances in different human tasks. 29 .. van der Grinten. Public Health Service. Rosenberger. Schneider. Spitzer. van Mechelen. 37-42. van der Beek. (2004).. (1973). 4979-4996. M. P.

Highlights:  Considers human energy expenditure in order picking  Develops a bi-objective approach for a class-based storage assignment problem  Estimates rest allowance based on energy consumption  Compares traditional class-based storage and bi-objective storage assignment 30 .