Sei sulla pagina 1di 30

Journal Pre-proof

A simulation-based optimization approach to design optimal layouts for


block stacking warehouses

Shahab Derhami, Jeffrey S. Smith, Kevin R. Gue

PII: S0925-5273(19)30352-4
DOI: https://doi.org/10.1016/j.ijpe.2019.107525
Reference: PROECO 107525

To appear in: International Journal of Production Economics

Received date : 5 June 2019


Revised date : 13 September 2019
Accepted date : 15 October 2019

Please cite this article as: S. Derhami, J.S. Smith and K.R. Gue, A simulation-based optimization
approach to design optimal layouts for block stacking warehouses. International Journal of
Production Economics (2019), doi: https://doi.org/10.1016/j.ijpe.2019.107525.

This is a PDF file of an article that has undergone enhancements after acceptance, such as the
addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive
version of record. This version will undergo additional copyediting, typesetting and review before it
is published in its final form, but we are providing this version to give early visibility of the article.
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.

© 2019 Published by Elsevier B.V.


Journal Pre-proof
*Title page including author details
Click here to download Title page including author details: Title page.pdf

A simulation-based optimization approach to design optimal

of
layouts for block stacking warehouses

Shahab Derhami1 , Jeffrey S. Smith2 , and Kevin R. Gue3

pro
1
School of Industrial and Systems Engineering, Georgia Institute of Technology, Atlanta,
GA 30332, USA
2
Department of Industrial and Systems Engineering, Auburn University, Auburn, AL
re-
36849, USA
3
Department of Industrial Engineering, University of Louisville, Louisville, KY 40292, USA

September 13, 2019


P

Abstract
al

Storing pallets of products on top of one another on the floor of a warehouse is called block stacking.

The arrangement of lanes, aisles, and cross-aisles in this storage system affects both utilization of the

storage space and material handling costs; however, the existing literature focuses exclusively on lane
urn

depths and their impact on space utilization. This paper fills this gap and studies the optimal layout

design for block stacking, which includes determining the numbers of aisles and cross-aisles, bay depths,

and cross-aisle types. We show that lane depths affect material handling cost in addition to space

utilization and develop a simulation-based optimization algorithm to find optimal layouts with respect

to both of these objectives. We also propose a closed-form solution to the optimal number of aisles in
Jo

a layout. The results of a case study in the beverage industry show that the resulting layout can save

up to ten percent of the operational costs of a warehouse. We present the computational efficiency of

the algorithm and some insights into the problem through an exhaustive experimental analysis based on

various test problems that cover small- to industrial-sized warehouses.

Keyword : block stacking, warehouse design, lane depth, facility layout, space utilization, cross-aisle

1
Journal Pre-proof
*Manuscript
Click here to view linked References

1
2
3
4
5
6
7
8
9
10
A simulation-based optimization approach to design optimal
11

of
12
13
layouts for block stacking warehouses
14
15
Authors’ names blinded for peer review

pro
16
17
18
19 September 13, 2019
20
21
22
23 Abstract
24
25
re-
Storing pallets of products on top of one another on the floor of a warehouse is called block stacking.
26
27 The arrangement of lanes, aisles, and cross-aisles in this storage system affects both utilization of the
28
storage space and material handling costs; however, the existing literature focuses exclusively on lane
29
30 depths and their impact on space utilization. This paper fills this gap and studies the optimal layout
P
31
32 design for block stacking, which includes determining the numbers of aisles and cross-aisles, bay depths,
33 and cross-aisle types. We show that lane depths affect material handling cost in addition to space
34
35 utilization and develop a simulation-based optimization algorithm to find optimal layouts with respect
36
al

to both of these objectives. We also propose a closed-form solution to the optimal number of aisles in
37
38 a layout. The results of a case study in the beverage industry show that the resulting layout can save
39
40 up to ten percent of the operational costs of a warehouse. We present the computational efficiency of
urn

41 the algorithm and some insights into the problem through an exhaustive experimental analysis based on
42
43 various test problems that cover small- to industrial-sized warehouses.
44
Keyword : block stacking, warehouse design, lane depth, facility layout, space utilization, cross-aisle
45
46
47
48
49
1 Introduction
Jo

50
51 Block stacking refers to stacking pallets of stock keeping units (SKUs) on top of one another in lanes on
52
53 a warehouse floor. This storage system is usually operated without using any storage racks, which makes
54
55 it inexpensive to implement but challenging in terms of space planning because lower pallets cannot be
56
reached until the ones stacked on top and in front of them are retrieved. Block stacking is widely used for
57
58 warehousing and cross-docking, especially when pallets are heavy and large. Examples of such environments
59
60 are bottled beverage and home appliance companies.
61
62
63 1
64
65
Journal Pre-proof

1
2
3
4 Block stacking is mainly operated under the shared (also known as random) storage policy, in which,
5
6 unlike the dedicated storage policy, lanes are not dedicated to SKUs and an empty lane is available to all
7
8 SKUs. However, to avoid blockage and relocation of pallets, a lane is temporarily dedicated to the SKU
9
10 that occupies its first pallet position until it is fully depleted. This restricts utilization of the storage space
11 because some unoccupied pallet positions in a partially occupied lane will not be available to other SKUs.

of
12
13 This effect is called honeycombing, and the waste associated with it accumulates in a lane until it becomes
14
15 entirely occupied or emptied [3]. Aisles also contribute to the amount of wasted space because they are used

pro
16
17 for accessibility rather than storage. There is a trade-off between honeycombing and accessibility waste with
18 respect to space utilization: shallow lanes generate less honeycombing waste but impose more aisles on the
19
20 layout, while the opposite is true for deep lanes [3].
21
22 The other influential factor that must be considered when designing a warehouse layout is transportation
23
24 cost. The number of cross-aisles and their configurations (locations, directions, etc.) affect the material han-
25
re-
dling cost. This effect is escalated in warehouses operated under continuous multi-command pick-up/drop-off
26
27 operations, where vehicles (pickers) perform replenishment or retrieval operations continuously in a working
28
29 shift before going to their home/parking—a situation that frequently occurs in block stacking warehouses.
30
In such warehouses, a replenishment or a retrieval operation starts by moving a vehicle from its last drop-off
P
31
32
location to the pick location. Hence, vehicles travel between any points of the warehouse rather than only
33
34 traveling from/to well-defined P/D points and storage locations. In such environments, it is very difficult
35
36 to develop analytical models to formulate and assess the total transportation cost because the distance that
al

37
38 vehicles travel depends on the sequence of the retrieval and replenishment operations and their assignment
39
to the vehicles. Simulation is an appropriate tool to accurately estimate traveled distance for such cases
40
urn

41 while considering the stochastic nature of the problem.


42
43 Material handling is usually performed in block stacking warehouses by unit-load forklifts that perform
44
45 pallet retrieval and replenishment operations continuously based on a load assignment policy. Cross-aisles
46
are used to facilitate travel. However, like aisles, they are not used directly for pallet storage and are
47
48 subsequently considered a waste of storage space.
49
Jo

50 We define bays in a layout as a set of adjoining lanes of the same depth separated by aisles (see Figure
51
52 1). To better utilize the storage space, we assume that bays are back to back and that every pair of bays
53
54 shares an aisle. Bay depths also affect the travel distances. Consider the two layouts presented in Figure 1.
55 One layout has two bays, each being seven pallet positions deep, and the other has the same dimensions but
56
57 consists of six bays, each two pallet positions deep. The total area in both layouts is the same. Assume that
58
59 aisles and cross-aisles are one pallet position wide and that pallets are stacked up to one pallet. The total
60
61 distance a forklift has to travel from the P/D point to replenish a pallet located in the very first position
62
63 2
64
65
Journal Pre-proof

1
2
3
4
5
6
7
8
9
10
11

of
12
13
14
15

pro
16
17
18
19
20
21
22
23 Figure 1: Travel distance vs. space utilization with respect to the number of aisles.
24
25
re-
26 of a lane in the top bay of the left layout (as highlighted in Figure 1) is 44 pallet positions, which includes
27
28 traveling 32 pallet positions in the cross-aisle and 12 pallet positions in the lane (round trip). This distance
29 for a pallet located in the same position at the bottom bay is 14 pallet positions, which includes traveling
30
P
31 12 pallet positions inside the lane and 2 pallet dispositions to cross the aisle. Similarly, the total vertical
32
33 distance to replenish all pallet positions in both layouts is
34
35
36 Layout with deep bays: 9((14 + 12+, . . . , +2) + (44 + 42+, . . . , +32)) = 2898 pallet positions.
al

37
38 Layout with shallow bays: 9((4 + 2) + 2(14 + 12) + 2(24 + 22) + (34 + 32)) = 1944 pallet positions.
39
40
urn

41 A longer distance must be traveled to replenish the layout with deeper bays. However, this layout has
42
43 more storage positions (i.e., better space utilization). The total horizontal travel distance is the same for
44
45 both cases and was disregarded in the comparison. Derhami et al. [11] used simulation to show there is a
46
47 trade-off between space utilization and transportation costs with respect to bay depths. Moreover, studying
48
cross-aisles is meaningless without considering the material handling cost. Hence, both of these objectives
49
Jo

50 must be considered when designing a layout.


51
52 In this paper, we develop a simulation-based optimization algorithm to find the optimal number of aisles,
53
54 cross-aisles, and bay depths in the layout of a block stacking warehouse. Our algorithm optimizes two
55
objectives: (1) maximizing space utilization; and (2) minimizing material handling costs. We develop a
56
57 closed-form solution model to determine the optimal number of aisles in the layout and use it along with an
58
59 experimental analysis to restrict the search space for the algorithm. Implementing the algorithm for a case
60
61
62
63 3
64
65
Journal Pre-proof

1
2
3
4 in the bottled beverage industry shows that significant improvement is possible.
5
6 The remainder of this paper is organized as follows. In section 2, we review the related research and
7
8 depict the gap in the literature along with the contributions of this paper. Then, we present the proposed
9
10 algorithm in section 3. We describe the three main components of the algorithm, develop a model to find
11 the optimal number of aisles in the layout, and show how the model is used to tighten the search space for

of
12
13 our algorithm. Finally, we present a case study and extensive numerical analysis in section 4 and evaluate
14
15 the results.

pro
16
17
18
19 2 Related research
20
21
Most of the papers studying warehouse layout focus on conventional warehouses with storage racks [37, 17, 2,
22
23 18]. They mostly aim to design a layout to minimize the transportation cost (or distance) for order picking
24
25
re-
[19, 20, 30, 38, 7, 4, 14]. Interested readers are referred to [9] for further details on this research. The
26
27 other objectives considered in warehouse layout design problems are operational costs [39, 28, 43], product
28
allocation [27, 32, 25], storage space utilization[10, 6], warehouse throughput [29, 31, 23], and operating
29
30 policies [8, 41, 36, 21, 1].
P
31
32 There is a stream of research focused on the effect of cross-aisles on transportation costs. Roodbergen and
33
34 de Koster [33] developed multiple heuristics to find the shortest path for order picking when multiple cross-
35
aisles exist. Vaughan and Petersen [40] proposed a heuristic to find the shortest path for order picking and
36
al

37 studied the effect of adding cross-aisles on transportation costs. They showed that the number of cross-aisles
38
39 needed to maximum transportation efficiency depends primarily on the length of the storage aisles relative
40
urn

41 to the length of the cross-aisles. Roodbergen and Vis [34] proposed an analytical model to approximate the
42
43 average length of an order picking route for two routing policies in a layout with one block (two cross-aisles).
44 Their approximation can be used as an objective function in a nonlinear model to obtain the optimal number
45
46 of aisles. Roodbergen et al. [35] developed an analogous approximation for a layout with multiple blocks.
47
48 Only a few papers have studied the design of block stacking layouts. Kind [22] considered the trade-off
49
Jo

50 between honeycombing and accessibility waste to find the optimal lane depth. He proposed a model to
51 estimate the lane depth that optimizes this trade-off. However, he did not provide any insights into how
52
53 to derive the model. Matson [26] developed another model to approximate the optimal lane depth under
54
55 instantaneous replenishment (i.e., infinite storage rates). Her model was appropriate for warehouses receiving
56
57 products from suppliers.
58 Goetschalckx and Ratliff [15] showed that if multiple lane depths are allowed, the set of optimal lane
59
60 depths follows a continuous triangular pattern. They developed a dynamic programming algorithm to select
61
62
63 4
64
65
Journal Pre-proof

1
2
3
4
5
6
7
8
9
10
11

of
12
13
14
15

pro
16
17
18
19
20
21 Figure 2: Space utilization vs. travel distance [11].
22
23
24 the set of optimal lane depths from a set of finite allowable lane depths to minimize the occupied floor
25
re-
26 space. Larson et al. [24] proposed a heuristic to design a class-based layout. Their three-phase algorithm
27
28 characterizes the aisle directions and storage zone dimensions, determines the storage types, and assigns the
29 required storage space to each storage zone.
30
P
31 Derhami et al. [12] proposed a closed-form solution to find the lane depth that maximizes volume utiliza-
32
33 tion under finite storage rate constraints. Their model can be used for warehouses located in a manufacturing
34
35 facility. They showed that using the traditional lane depth model with a finite replenishment rate produces
36 lane depths up to twice as deep as they should be, but the resulting loss of storage space might not be
al

37
38 significant. This is because the space utilization curve, as a function of lane depth, is relatively flat as the
39
40 lane depth increases.
urn

41
42 Derhami et al. [13] showed that the traditional lane depth model [12, 26, 22] underestimates accessibility
43
waste when used to design a layout because it computes accessibility waste only for the time that a lane is
44
45 occupied. They developed a new waste function to compute the total waste of storage volume as a function
46
47 of bay depth and used it in a mixed integer program to find the optimal bay depths. Their model allows
48
49 multiple bay depths in a layout and minimizes the waste of space by assigning SKUs to their preferred bays.
Jo

50
Derhami et al. [11] used simulation to study the trade-off between space utilization and transportation
51
52 costs with respect to bay depths. Using a common bay depth assumption, they simulated multiple layouts
53
54 with various numbers of aisles (i.e., different bay depths). Figure 2 shows the simulation results for a
55
56 warehouse with 50 SKUs. As the graph shows, utilization of the storage volume increases as the number
57
of aisles increases from two to four (i.e., bay depth decreases), where it reaches its peak. It then starts
58
59 declining as the bay depth becomes shallower because accessibility waste increases. On the other hand, the
60
61
62
63 5
64
65
Journal Pre-proof

1
2
3
4 total travel distance improves as the bay depth decreases. The improvement becomes modest after a certain
5
6 point, and then the travel distance remains steady. Hence, transportation costs improve at the cost of lower
7
8 space utilization.
9
10 Most of the research on the design of block stacking systems focuses exclusively on the utilization of the
11 storage space [26, 12, 13] rather than transportation costs or both. As discussed, bay depths and cross-

of
12
13 aisles affect both material handling costs and space utilization. Designing a block stacking layout includes
14
15 determining bay depths and the number of aisles and cross-aisles; therefore, both space utilization and

pro
16
17 transportation costs must be considered. Considering a conventional block stacking layout similar to the one
18 presented in Figure 3, this research answers the following questions:
19
20
21 • How many cross-aisles should a layout have?
22
23 • How many aisles should a layout have?
24
25
re-
26 • How deep should the bays be?
27
28 We develop a simulation-based optimization algorithm that finds optimal layouts with respect to space
29
30 utilization and material handling costs. This paper contributes to the current warehouse design literature
P
31
32 in three primary ways:
33
34 • We find the optimal layouts with respect to two main objectives of the layout design problem (space
35
36 utilization and material handling costs). This allows the model to determine the optimal number of
al

37
38 aisles and cross-aisles while accounting for the effects these variables have on one another;
39
40 • We develop a closed-form solution for the optimal number of aisles in a layout. In contrast with
urn

41
42 the traditional lane depth model, which optimizes the trade-off between block width and depth, it
43
44 minimizes the total wasted storage space in the entire layout rather than a block; and
45
46 • The simulation-based modeling allows the algorithm to capture the stochastic variations that exist in a
47
48 real world situation. The stochastic variables considered in the simulation model are production rates,
49
Jo

50 demand rates, production batch quantities, and forklift travel times. Moreover, the simulation model
51
is an appropriate tool for accurately assessing the transportation cost in continuous multi-command
52
53 pick-up/drop-off operations; analytical models are unable to accurately estimate the total distance
54
55 traveled by vehicles in this context because it depends on the sequence of retrieval and replenishment
56
57 operations, which is stochastic, and their assignments to the vehicles.
58
59
60
61
62
63 6
64
65
Journal Pre-proof

1
2
3
4
5
6
7
8
9
10
11

of
12
13
14
15

pro
16
17
18
19
20 Figure 3: Components of a conventional block stacking layout.
21
22
23 3 Designing optimal layouts
24
25
re-
The major factors in designing a block stacking layout are the numbers of aisles and cross-aisles, bay depths,
26
27 and cross-aisle types (unidirectional vs. bidirectional). Our algorithm takes all these design factors as
28
29 decision variables and attempts to find the set of layouts optimizing space utilization and transportation
30
costs. It simulates different potential layouts by varying values of the above variables and produces a Pareto
P
31
32 frontier with respect to two objectives: maximizing space utilization and minimizing transportation costs.
33
34 It has three steps: layout scenario generation, pre-simulation assessment, and simulation.
35
36 The algorithm first generates different layout scenarios for simulation. The layouts are different from
al

37
38 one another in terms of the numbers of aisles and cross-aisles and the cross-aisle types. We use analytical
39 and empirical approaches to bound the number of aisles and cross-aisles in the layout scenario generation
40
urn

41 and to efficiently limit the number of layouts being evaluated by the simulation without disregarding any
42
43 optimal solutions. In the next step, the layouts are assessed with respect to the total travel distance, and key
44
45 measurements, for example, the travel distances between all potential origins and destinations, are computed
46
and stored to prevent having to re-calculating them when executing any event in the simulation. Finally,
47
48 the generated layouts are evaluated by a steady state simulation and compared in terms of space utilization
49
Jo

50 and material handling costs. Figure 4 demonstrates the flowchart of the proposed algorithm, and Table 1
51
52 presents the definition of parameters and variables used in the algorithm. The components of the algorithm
53
are described in detail in the following sections.
54
55
56
57
58
59
60
61
62
63 7
64
65
Journal Pre-proof

1
2
3
4
5
6
7
8
9
10
11

of
12
13 Table 1: Table of notation
14 amin minimum number of aisles allowed in layout generation
15 amax maximum number of aisles allowed in layout generation

pro
16 cmin minimum number of cross-aisles allowed in layout generation
17 cmax maximum number of cross-aisles allowed in layout generation
18 Qi production (storage) batch quantity of SKU i
19 Pi production (storage) rate of SKU i
20 λi demand (retrieval) rate of SKU i
21 Hi height of a pallet of SKU i, in units of distance (e.g., meters. feet, etc.)
22
Zi stackable height of SKU i, in units of pallets
23
Sw warehouse width, in units of pallets
24
25 Sh
re-
warehouse height, in units of distance (e.g., meters. feet, etc.)
26 na number of aisles
27 A aisle width, in units of pallets
28 xi assigned lane depth to SKU i, in units of pallets
29 I set of all SKUs
30 x̄ common bay depth, in units of pallets
P
31 Ns number of SKUs stored in the warehouse
32 Sl warehouse length, in units of pallets
33 n∗a optimal number of aisles
34 x̄∗ optimal common bay depth, in units of pallets
35 α parameter to set amin
36 β parameter to set amax
al

37 (xo , y o ) coordinate of the origin for a shipment


38 (xd , y d ) coordinate of the destination for a shipment
39 Nv total number of vehicles required
40
Du total distance loaded forklifts traveled in the simulation
urn

41
Dl total distance unloaded forklifts traveled in the simulation
42
43 Ts simulation period
44 Tw simulation warm-up period
45 V average speed of a forklift
46 nc number of cross-aisles
47 C cross-aisle width, in units of pallets
48 Nb number of bays in the simulated layout
49 Nl number of lanes in a bay
Jo

50 WijH honeycombing waste generated at the jth lane of bay i in simulation


51 Lmin minimum number of lanes between two subsequent cross-aisles
52 Lmax maximum number of lanes between two subsequent cross-aisles
53
54
55
56
57
58
59
60
61
62
63 8
64
65
Journal Pre-proof

1
2
3
4
5
6
7
8
9
10
11

of
12
13
14
15

pro
16
17
18
19
20
21
22
23
24 Figure 4: Flowchart of the proposed model.
25
re-
26 3.1 Layout scenario generation
27
28
29 The optimization problem at hand is a multi-objective optimization problem with two objectives: maximizing
30
space utilization and minimizing transportation costs. The first step of our algorithm is generating layout
P
31
32 scenarios for simulation. The number of aisles and cross-aisles and the cross-aisle types are decision variables
33
34 and change over different layout scenarios. To perform a complete search of the solution space and obtain a
35
36 comprehensive Pareto frontier, all potential layouts must be simulated. If the lower and upper bounds on the
al

37
numbers of aisles and cross-aisles are denoted by amin , amax , cmin , and cmax , respectively, the total number
38
39 of layouts generated for simulation by assuming a common bay depth is (amax − amin + 1)(cmax − cmin + 1).
40
urn

41 Hence, the ranges of [amin , amax ] and [cmin , cmax ] determine the number of layouts that must be evaluated
42
43 by the simulation and affect the computational difficulty and solution quality of the algorithm. These ranges
44
45 are usually too wide; this imposes too many potential layouts for simulation. The number of potential
46 layouts grows exponentially as the size of the warehouse increases because more aisles and cross-aisles are
47
48 required. However, the simulation process is computationally intensive, and evaluating all possible scenarios
49
Jo

50 is not computationally practical.


51
52 We narrow the search space by bounding the number of aisles and cross-aisles allowed for generating
53 layout scenarios. The bounds must be tight enough to prevent an extensive computational burden from
54
55 simulating too many scenarios but large enough to ensure that no potentially optimal solution is omitted.
56
57 Once the bounds are set, the layout scenarios are generated using all possible combinations of the allowed
58
59 values for the number of aisles and cross-aisles. In the next sections, we describe an analytical approach to
60 bound the number of aisles in the layout generation and describe how we do the same for the cross-aisles.
61
62
63 9
64
65
Journal Pre-proof

1
2
3
4
5
6
7
8
9
10
11

of
12
13
14
15

pro
16
17
18 Figure 5: Changes in the inventory of SKU i over its cycle time, Pi > λi .
19
20
21 3.1.1 Bounding the number of aisles
22
23
To avoid simulating too many layout scenarios with only marginal differences, we use a common bay depth
24
25
re-
policy. That is, all bays in the layout should have the same depth.This setup is widely used in practice
26
27 because it discards the need for lane depth assignment to SKUs. This policy significantly shrinks the search
28
29 space by disregarding many layouts that have the same number of aisles but slightly different bay depth
30
profiles. For example, if bay depths are allowed to be between 5 to 25 pallets, then for every single layout of
P
31
32 (amax − amin + 1)(cmax − cmin + 1) layouts, all variations of bay depths between 5 to 25 pallets for which
33
34 the sum of bay depths and dedicated space to the aisles equals the warehouse length must be considered for
35
36 simulation. This results in an enormous number of layouts differ only slightly from one another.
al

37
38 Thus, setting the number of aisles in each scenario gives the bay depths (note that the dimensions of the
39 layouts are kept fixed in all scenarios). We develop a closed-form solution to obtain the optimal number of
40
urn

41 aisles to maximize space utilization under the common bay depth constraint and use it to determine amin
42
43 and amax . This analytical model assumes deterministic production and demand rates, but this assumption
44
45 is relaxed later in the simulation model.
46 Consider SKU i produced in batches of Qi pallets and stored at a rate of Pi pallets per unit of time.
47
48 Assume it is retrieved at a rate of λi pallets per unit of time, where Pi > λi and the replenishment starts
49
Jo

50 when the inventory reaches zero. Pallets of this SKU are Hi feet high and can be stacked up to Zi pallets.
51
52 The changes in the inventory of this SKU are shown in Figure 5. Assume set I that consists of Ns SKUs is
53 stored in a warehouse whose length and width are S l and S w , respectively, and its aisles are A pallets wide.
54
55 The following lemma determines the optimal number of aisles and lane depth for the warehouse.
56
57
58
59
60
61
62
63 10
64
65
Journal Pre-proof

1
2
3
4 Lemma 1. The optimal number of aisles in the warehouse is obtained by
5
6 r
7 S l Ns
n∗a = . (1)
8 4S w A
9
10
11 Proof. As shown in [13], the average wasted storage volume generated in the layout is given by

of
12
13 Sh X X 1  
14 W̄ = AS h S w na + xi + (Qi (S h − Zi Hi ) − Zi Hi )(Pi − λi ) − λi (2S h − Zi Hi ) , (2)
15 2 2Pi Zi
i∈I i∈I

pro
16
17
where S h is the warehouse clear height (in units of distance, i.e., inches, feet, etc.), na is the number of
18
19 aisles, and xi is the assigned lane depth to SKU i. Variables xi and na are the only decision variables in (2);
20
21 therefore, the optimal bay depths are obtained by optimizing (2) with respect to xi and na . Note that these
22
23 two variables are dependent; that is, the sum of aisle widths and bay depths equals the warehouse length.
24
25
re-
The constant part of (2) is trivial in optimization and can be removed. Assuming a common bay depth,
26 denoted by x̄, we have
27 X
28 xi = Ns x̄, (3)
29 i∈I
30
P
31 where Ns is the number of SKUs. Replacing (3) in (2), the optimal common bay depth is obtained by solving
32
33 the following constrained optimization problem:
34
35  
1
36 Minimize AS h S w na + S h Ns x̄, (4)
2
al

37
38 s.t.
39
40 2na x̄ + na A = S l , (5)
urn

41
42 x̄, na ∈ Z+ . (6)
43
44
45
Constraint (5) guarantees that the sum of bay depths and aisle widths is equal to the warehouse length.
46
47 Solving (5) for x̄ and substituting x̄ into the objective function (4) produces an unconstrained optimization
48
49 model whose objective function is
Jo

50
51  
h w 1
52 Minimize AS S na + S h (S l − na A). (7)
53 4na
54
55
Differentiating (7) with respect to na , setting the results equal to zero, and solving for na gives n∗a . Expression
56
57 (7) is continuously differentiable, has only one extreme point, and its second derivative with respect to na is
58
59 non-negative. Hence, it is a unimodal function and n∗a is its global optimum.
60
61
62
63 11
64
65
Journal Pre-proof

1
2
3
4 Variable n∗a must be an integer. To have an integer value for n∗a , we evaluate the two nearest integers
5
6 smaller and greater than n∗a in (7) and select the integer that produces a smaller waste of space. Once n∗a is
7
8 determined, the optimal common bay depth is obtained by
9
10
S l − n∗a A
11 x̄∗ = . (8)

of
12 2n∗a
13
14 When x̄∗ is not integer, arranging the layout into exactly 2n∗a equally deep bays is not possible. In this case,
15
we create 2n∗a bays with the depth of bx̄∗ c pallets and split the remaining S l − n∗a (2bx̄∗ c + A) pallet positions

pro
16
17
18 equally among all bays.
19
From an operational standpoint, deep lanes limit storage space availability, as it takes longer to fully
20
21 deplete or replenish a deep lane. Storage space availability becomes more important when the SKU variety
22
23 is high. In such a situation, it is important to have sufficient storage space available (empty) at any time to
24
25
re-
provide storage lanes for incoming SKUs and allow flexibility in production planning without any concerns
26
27 about storage space availability. The following proposition discusses another property of n∗a for the layout
28 design problem.
29
30
Lemma 2. n∗a maximizes storage space availability.
P
31
32
33 Proof. Proof is provided in Appendix A.
34
35
36 Using n∗a , we set amin = αn∗a and amax = βn∗a for layout scenario generation. Parameters α and β are
al

37
set through a numerical experiment presented in section 4.1.
38
39
40
3.1.2 Bounding the number of cross-aisles
urn

41
42
43 For each layout scenario, two alternatives are considered: unidirectional and bidirectional cross-aisles. Bidi-
44
45 rectional cross-aisles are twice as wide as the unidirectional cross-aisles. Hence, in the layouts with unidirec-
46 tional cross-aisles, we put twice as many cross-aisles as the respective layouts with bidirectional cross-aisles.
47
48 This allows a fair comparison between the two alternatives in terms of space utilization. For example, a
49
Jo

50 layout with three bidirectional cross-aisles is compared to the layouts with six unidirectional cross-aisles.
51
52 Cross-aisles are evenly spaced from each other, and the distance between any two subsequent cross-aisles is
53 the same. Unidirectional cross-aisles are added to the layout in pairs with opposite directions.
54
55 Cross-aisles facilitate material handling in a warehouse. Developing a closed-form model to estimate the
56
57 transportation cost (or total travel distance) as a function of the number of cross-aisles (similar to the method
58
59 used for the number of aisles) is not possible because the shipment operations are performed continuously,
60 and the travel distance for a shipment depends on the sequence of operations assigned to a vehicle (i.e., the
61
62
63 12
64
65
Journal Pre-proof

1
2
3
4
5
6
7
8
9
10
11

of
12
13
14
15

pro
16
17
18
19
20
21
22
N
23
W E
24
25
re- S

26 Figure 6: Components of the layout and their relative configurations.


27
28
29 last drop-off location of a vehicle must be known). For this reason, we use an experimental approach with
30
P
31 simulation in section 4.1 to analyze the total travel distance with respect to the number of cross-aisles and
32
33 determine cmin and cmax .
34
35
36 3.2 Pre-simulation assessment
al

37
38 We compute the rectilinear shortest distances between all potential origin and destination pairs in the layout
39
40 and store them for simulation use. This is to avoid recalculating travel distances every time an event is
urn

41
42 initiated. For each pair of layout scenarios that only differ in cross-aisle type (i.e., the same numbers of aisles
43
and cross-aisles but one with unidirectional and the other with bidirectional cross-aisles), we calculate the
44
45 total travel distance between all potential pick-up and drop-off locations and consider only the layout with
46
47 the smaller total travel distance for simulation. The total space dedicated to the cross-aisles is the same in
48
49 both alternatives; therefore, the layout with a smaller total travel distance is the non-dominated solution.
Jo

50
Figure 6 presents the relative locations of the main components of the layout. The locations of production
51
52 lines, vehicle parking, and outbound docks, as well as the number of P/D points, are given and can differ
53
54 from those presented in Figure 6. Although the main purpose of this model is not to optimize the location
55
56 of these components, one can analyze their arrangements by simulating different scenarios while keeping the
57
58 remaining design factors fixed. The following distances are required in the simulation:
59
60 • Distances between all storage lanes.
61
62
63 13
64
65
Journal Pre-proof

1
2
3
4 • Distances from storage lanes to the production line and vice versa.
5
6
7 • Distances from storage lanes to outbound docks and vice versa.
8
9 • Distances from storage lanes to the vehicle parking and vice versa.
10
11

of
12 • Distances from outbound docks to the production line.
13
14 • Distances from outbound docks to the vehicle parking.
15

pro
16
17 • Distances from the vehicle parking to the production line.
18
19 A rectilinear distance between two locations is obtained by adding the distances traveled along the x-axis
20
21 and y-axis. Considering both unidirectional and bidirectional cross-aisles adds an extra level of complexity
22
23 to the calculations because the directions of the move and the cross-aisle must match. As Figure 6 illustrates,
24
25
re-
we use a Cartesian coordinate system and assume that the origin is located at the southwest corner of the
26 layout. Instead of designing a web of P/D locations to find the shortest path, we take advantage of the
27
28 special shape of the layout to simplify calculations.
29
30 The shortest path between two points is given by the path that travels through one cross-aisle and at
P
31
32 most two aisles (the aisles that the origin and destination points are located in). This is highlighted in Figure
33 6 by two paths connecting an example shipment between two P/D locations in two different bays. All other
34
35 alternatives that produce the same travel distance are not evaluated for the sake of computational efficiency.
36
al

37 The shortest path for each move is a direct path that passes either through the closest cross-aisle located
38
39 north of the origin and allows moving toward the destination or a similar cross-aisle located south of the
40 origin (see Figure 6). Hence, the distances of the origin and destination points to the eligible cross-aisles (the
urn

41
42 closest cross-aisles located north and south of the origin that allow moving toward the destination) must be
43
44 taken into account for each move.
45
46 Consider a west-to-east move in Figure 6. The travel distance along the shortest path through the
47 cross-aisle located north of the origin is
48
49 
Jo

50 

|y c − y o | + y c − y d + xd − xo if y c ∈
/ ∅,
51 N
DW →E = (9)
52 

53 ∞ otherwise,
54
55
where y c is the y-coordinate of the closest cross-aisle located north of the origin heading east, xo and
56
57 y o are the coordinates of the origin, and xd and y d are the coordinates of the destination. If an eligible
58
59 cross-aisle does not exist, the distance is set to infinity for this path.
60
61
62
63 14
64
65
Journal Pre-proof

1
2
3
4 S
Similarly, the travel distance along the shortest path from the southern cross-aisle, DW →E , is obtained
5
6 by replacing the coordinates of the respective cross-aisle in (9). The shortest path is then obtained by
7
8
9 N S
DW →E = min{DW →E , DW →E }. (10)
10
11

of
12
Note that DE→W is not necessarily equal to DW →E because unidirectional cross-aisles may impose longer
13
14 travel distance along the y-axis. Expressions (9) and (10) are valid in the layout in Figure 6 for all distances
15

pro
16 between the storage lanes. They are also valid for distances between the storage lanes and pick-up points,
17
18 outbound docks, and vehicle parking if they are located on the short sides of the warehouse. Similarly, they
19
are also valid for the distances between any pair of outbound docks, pick-up points, and vehicle parking if
20
21 they are located on the short sides of the layout. If either of these components are on the long sides of the
22
23 warehouse, the shortest distance between the storage lane and that component is simply obtained by
24
25
re-

26 DS→N = DN →S = y o − y d + xo − xd . (11)
27
28
29
This is because aisles are bidirectional (to provide enough space to the forklifts to easily maneuver when
30
P
31 replenishing/retrieving lanes), and the cross-aisles located in front of those locations are bidirectional for the
32
33 ease of transportation. Expression 11 is also used to calculate distances between pairs of outbound docks,
34
35 pick-up points, and the vehicle parking if either the origin or the destination is located on the long sides of
36
al

the layout.
37
38
39
40 3.3 Simulation model
urn

41
42 The core of the model is an event-oriented simulation of pallet storage and retrieval operations along with
43
44 material handling for a given layout. The output reports performance metrics pertinent to space utiliza-
45
46 tion and transportation costs. The model has four sources of variation: production rates, demand rates,
47
production batch quantities, and vehicle transportation times. The company that we studied for our case
48
49 study uses automated guided vehicles (AGV) for material handling. Accordingly, the simulation model is
Jo

50
51 built assuming operations are handled by AGVs. If vehicles are operated by humans, human factors such as
52
53 learning affect the performance of the system because order picking is a repetitive and cognitive task that
54
involves human learning [5, 16, 42]. Incorporating the human learning factors in simulation is an extension
55
56 to our model that provides more accurate prediction of the performance of the system for such cases.
57
58 The simulation model consists of nine procedures: three events to simulate a replenishment operation,
59
60 three events for a retrieval operation, two events for a vehicle release, and a warm-up event. In a replenish-
61
62
63 15
64
65
Journal Pre-proof

1
2
3
4 ment operation, a vehicle is sent to pick up a produced (or inbound arrival) pallet from the production line
5
6 (or inbound dock) and deliver it to a storage lane. In a retrieval operation, a vehicle is sent to pick up a
7
8 pallet from a storage lane and deliver it to an outbound dock. The simulation events are as follows:
9
10 • Production pick-up: The closest available vehicle to the pick-up point (production line or inbound
11

of
12 dock) is dispatched to pick up a waiting pallet, and a “lane drop-off” event is scheduled taking into
13
14 account the travel distance. The storage lane is determined for the pallet.
15

pro
16 • Lane drop-off: The dispatched vehicle picks up the pallet from the production line or inbound dock
17
18 and starts traveling to the assigned storage lane. A “replenishment” event is scheduled considering the
19
20 travel distance.
21
22 • Replenishment: The pallet is stored in the target lane, and a “release vehicle” event is scheduled at
23
24 the simulation time plus epsilon time unit.
25
re-
26
• Outbound pick-up: The closest vehicle to the pick-up lane is dispatched to pick up the requested SKU
27
28 from its assigned lane. A “retrieval” event is scheduled taking the travel distance into consideration.
29
30 An outbound dock is assigned for drop-off.
P
31
32
• Retrieval: The dispatched vehicle picks up the requested pallet from the lane and starts traveling to
33
34 the assigned outbound dock. A “truck drop-off” event is scheduled considering the travel time.
35
36
al

37 • Truck drop-off: The requested SKU is delivered to the assigned outbound dock and a “release vehicle”
38 event is scheduled at the simulation time plus epsilon time unit if there is no waiting pick-up request.
39
40
urn

41 • Release vehicle: The empty vehicle is sent to parking. A “Park vehicle” event is scheduled considering
42
43 the travel distance.
44
45 • Park vehicle: The released vehicle is parked and becomes available.
46
47
48 • Warm-up: This event is executed once at a given time and resets all variables used for performance
49
Jo

evaluation (not the control variables) to their initial values.


50
51
52 Figure 7 shows the simulation flowchart and steps. The simulation model uses an event list to queue and
53
54 execute all scheduled events based on their execution times. We assume the production and outbound plans
55 are known at the beginning of each working shift (i.e., which SKUs and how many are produced/shipped),
56
57 though the exact times are stochastic. They are added to the simulation event list at the beginning of
58
59 each working shift. Outbound events are sequenced based on the truck schedules, truck capacity, and SKU
60
61
62
63 16
64
65
Journal Pre-proof

1
2
3
4
5
6
7
8
9
10
11

of
12
13
14
15

pro
16
17
18
19
20
21
22
23
24
25
re-
26
27
28
29
30
P
31 Figure 7: Flowchart of the discrete event-based simulation model.
32
33
34 demands. The “Release vehicle” event is scheduled at the simulation time plus epsilon time unit to allow
35
36 awaiting pick-up requests to be processed before an idle vehicle is sent to parking.
al

37
38 Along with many performance metrics pertinent to space utilization and transportation costs, such as
39 total distance traveled by vehicles and vehicle utilization, the simulation model computes two major perfor-
40
urn

41 mance metrics reflecting the objective functions: the required number of vehicles (Nv ) and the percentage of
42
43 wasted space (W ). We purposefully defined these two metrics such that they are sufficiently tangible for the
44
45 decision making and easily convertible to a dollar amount. Multiplying Nv and W by the unit vehicle cost
46 and unit space cost yields the total transportation cost and space cost for the layout. Space cost includes
47
48 direct and indirect space-related costs such as rent, insurance, lights, building amortization and repairs.
49
Jo

50 The material handling cost includes the costs of operating vehicles such as forklift costs and amortization,
51
52 maintenance, and labor.
53 Nv is obtained by summing the distance traveled by all vehicles in the simulation and then dividing the
54
55 result by the total distance that a vehicle can travel. That is,
56
57
58 Du + Dl
Nv = , (12)
59 V ∗ (T s − T w )
60
61
62
63 17
64
65
Journal Pre-proof

1
2
3
4
5 Table 2: Preliminary experimental study to fine tune parameters: parameters used in the algorithm and the
6 minimum and maximum numbers of aisles and cross-aisles observed in the solutions of the Pareto frontiers.
7 Parameter values Pareto frontier solutions
Test problem
8 ∗
na amin amax cmin cmax M in(na ) M ax(na ) M in(nc ) M ax(nc )
9 10 SKUs 2 1 3 2 5 2 3 2 5
10
50 SKUs 3 2 5 2 10 3 5 2 9
11

of
100 SKUs 5 4 8 2 12 5 8 2 5
12
13 150 SKUs 6 4 9 2 15 6 9 2 9
14 200 SKUs 7 5 11 2 17 7 11 2 5
15 300 SKUs 8 6 12 2 20 8 12 2 7

pro
16
17
18 where Du , and Dl are the total loaded and unloaded distances traveled by all vehicles, T s is the simulation
19 time, T w is the warm-up period, and V is the average speed of a vehicle. The percentage of wasted space
20
21 is obtained by summing the total wasted space in lanes and dividing it by the total space-time. The total
22
23 wasted space in the warehouse is the sum of waste from honeycombing and accessibility [10]. The volume
24
25
re-
dedicated to the aisles and cross-aisles comprises the accessibility waste [13] and honeycombing waste is
26
generated in the lanes as they are being replenished and retrieved[3]. Hence the percentage of wasted space
27
28 is obtained by
29 PNb PNl
S h (na AS w + nc CS l )(T s − T w ) + i=1 j=1 WijH
30 W = , (13)
(S w S l S h )(T s − T w )
P
31
32
33 where nc is the number of cross-aisles, C is the cross-aisle width, Nb is the number of bays, Nl is the number
34
of lanes in a bay, and WijH is the honeycombing waste at the jth lane of bay i. All simulation scenarios are
35
36 compared with respect to these two objective functions, and non-dominated solutions produce the Pareto
al

37
38 frontier.
39
40
urn

41
42 4 Numerical experiments
43
44
45 In this section, we perform an experimental analysis to fine tune the parameters of the algorithm. Then, we
46 test the algorithm on a case in the beverage industry. Using data from the case study, we design various test
47
48 problems to analyze the computational efficiency of the model and obtain some insights from the generated
49
Jo

50 Pareto frontiers on small- to industrial-sized test problems.


51
52 The proposed model was coded with Python 2.7 and run on a cluster whose nodes are equipped with ten
53 Intel Xeon processors E5-2660 (2.6GHz) and 128 GB of RAM. We ran the algorithm in parallel on ten cores.
54
55 Following the simulation setups proposed in [11], we set the warm-up period to one month, the start-up
56
57 inventory to zero, and the simulation time to 8 months. To reduce the number of replications, we used
58
59 common random numbers across scenarios.
60
61
62
63 18
64
65
Journal Pre-proof

1
2
3
4 4.1 Tuning the parameters
5
6
7 We conducted a numerical experiment to set amin , amax , cmin , and cmax such that the computational effort
8 is minimized while no potentially optimal solution is removed from the search space. We generated random
9
10 test problems with various sizes (10 to 300 SKUs) and tested the algorithm by setting wide ranges for the
11

of
12 parameters without considering the resulting computational times. The goal was to analyze the solutions in
13
14 the Pareto frontiers and use them to efficiently tighten the range of the input parameters.
15 Define the minimum and maximum numbers of lanes between two subsequent cross-aisles by Lmin and

pro
16
17 Lmax , respectively. Then, cmin = (S w + Lmax )/(Lmax + 2C) and cmax = (S w + Lmin )/(Lmin + 2C), where
18
19 C is the width of a unidirectional cross-aisle in units of pallets. We tested the following setups: α = 0.8,
20
21 β = 1.5, Lmin = 10, and Lmax = 40. Table 2 shows the resulting amin , amax , cmin , and cmax for each test
22
problem and the minimum and maximum numbers of aisles and cross-aisles observed in the solutions of the
23
24
25
Pareto frontier for each test problem.
re-
26 As Table 2 shows, the smallest number of aisles observed in all solutions is n∗a for all test problems,
27
28 although smaller numbers were allowed in the experiment. This is because decreasing the number of aisles
29
beyond n∗a deteriorates both objective functions. Therefore, the efficient value for α will be one for the final
30
P
31 experiments. Increasing the number of aisles reduces space utilization but improves the total travel distance.
32
33 However, as highlighted in [11] and shown in Figure 2, the improvement rate declines as the number of aisles
34
35 grows and becomes negligible once the layout has many aisles. Beyond this point, savings in transportation
36
al

costs do not justify the loss of storage space. Taking this into account, we set β = 1.4.
37
38 The maximum number of cross-aisles observed in the Pareto frontiers is smaller than cmax for almost
39
40 all test problems except the 10 SKU test problem, for which it is equal to cmax . This shows that cmax
urn

41
42 can be reduced without risking the removal of potentially optimal solutions. The largest Lmin that yields a
43
44 cmax equal to or larger than the maximum number of cross-aisles observed in all Pareto frontiers is obtained
45 by setting Lmin = (0.1)S w . This setup limits cmax while preserving all solutions (including the 10 SKUs
46
47 problem, for which this condition yields cmax = 5). We set cmin = 2 because we assumed, for ease of
48
49 transportation, that one cross-aisle exists next to the long side of the warehouse.
Jo

50
51
52 4.2 A case study in the beverage industry
53
54
We used the algorithm to design the warehouse layout for one of the production facilities of a leading supplier
55
56 of bottled beverages in North America. The company produces various types of bottled beverages and runs
57
58 dozens of plants and warehouses across the U.S.A., Canada, and Mexico. The storage system in all of these
59
60 facilities is mainly block stacking. The selected facility produces more than 100 SKUs and stores them in a
61
62
63 19
64
65
Journal Pre-proof

1
2
3
4
5 Table 3: Results of a case study in the beverage industry.
6 Current layout Optimal layout
7
8 Bay depths (pallets) (32 , 72 , 82 , 93 , 101 , 112 , 124 , 133 , 151 ) (222 , 218 )
9 Number of aisles 11 5
10 Number of cross-aisles 4 16
11 Cross-aisle type Unidirectional Unidirectional

of
12 Wasted space (%) 47.69 65.13
13 Required number of vehicles 38.58 34.43
14
15

pro
16
17
18
19
20
21
22
23
24
25
re-
26
27
28
29
30
P
31
32 Figure 8: Pareto frontier of the case study vs. the current layout, solutions format:(na ,nc , cross-aisles type).
33
34
368,000 ft2 storage area. Material handling is performed by unit-load forklifts. The main characteristics of
35
36 the current layout and the results from simulating it are presented in Table 3.
al

37
38 The Pareto frontier generated by the algorithm is presented in Figure 8. Quantifying the objectives with
39
40 their costs produces an optimal layout presented in Table 3. Like most facilities located in suburban areas,
urn

41
the material handling cost in our studied facility is greater than the space cost. As a result, the optimal
42
43 solution is the layout that most reduces the travel distance by using 16 unidirectional cross-aisles in contrast
44
45 with the current layout, which has four cross-aisles of the same type. This solution dedicates four times the
46
47 current amount of space to the cross-aisles and consequently has lower space utilization compared to the
48
current layout, but it improves the total travel distance by approximately 10%. The optimal layout has fewer
49
Jo

50 than half the aisles of the current layout. This shows that the benefit of shallow lanes on transportation
51
52 costs did not justify the cost of lost space utilization. Implementing the proposed layout would save more
53
54 than half a million dollars annually on the total operational costs of the studied facility.
55
56
57
58
59
60
61
62
63 20
64
65
Journal Pre-proof

1
2
3
4
5 Table 4: Computational experiments.
6 Warehouse # of simulated # of layouts Avg. Simulation time Computational
Problem amin amax cmin cmax
size (ft) layouts in Pareto frontier per scenario (sec) time (sec)
7
10 SKUs 200 × 400 2 3 2 8 14 2 16 36
8 50 SKUs 400 × 720 3 5 2 10 27 6 32 123
9 100 SKUs 400 × 920 5 7 2 10 27 10 54 160
10 200 SKUs 760 × 1400 7 10 2 10 36 9 237 977
11 300 SKUs 920 × 1600 8 12 2 11 50 12 337 1746

of
400 SKUs 1080 × 1840 9 13 2 11 50 12 518 2676
12 500 SKUs 1200 × 2080 10 14 2 11 50 8 536 3084
13 600 SKUs 1320 × 2200 11 16 2 11 60 10 544 3421
14 700 SKUs 1440 × 2400 12 17 2 11 60 8 658 4177
15 800 SKUs 1600 × 2600 13 19 2 11 70 8 1034 11520
900 SKUs 1720 × 2880 14 20 2 11 70 12 1057 16080

pro
16
1000 SKUs 1840 × 3000 14 20 2 11 70 14 1159 17155
17
18
19
20
21
22
23
24
25
re-
26
27
28
29
30
P
31
32
33
34 Figure 9: Pareto frontier of the 500 SKUs test problem, solutions format:(na ,nc , cross-aisles type).
35
36
al

37 4.3 Experimental analysis


38
39 We randomly generated 12 test problems using the characteristics of the SKUs in our case study such as
40
urn

41 production rates, demand rates, and production batch quantities. The test problems cover small (10 SKUs)
42
43 to industrial-sized (1000 SKUs) warehouses and are used to evaluate the computational performance of the
44
proposed algorithm and obtain some insights into the Pareto frontier of the problem.
45
46 Table 4 shows the computational times and the number of scenarios simulated for each test problem.
47
48 The simulation times ranged from 16 to 1159 seconds depending on the size of the problem. The total
49
Jo

50 computational time of the algorithm grows from 36 seconds (for the smallest problem) up to 17155 seconds
51
for the largest test problem. The total computational times are smaller than the sum of the simulation times
52
53 for all scenarios because the simulation scenarios were run in parallel. The computational times show that
54
55 although the model simulates multiple scenarios in a steady state, it is capable of finding the Pareto frontier
56
57 in a reasonable time for large-sized test problems.
58
Figures 9 and 10 demonstrate the Pareto frontiers for test problems with 500 and 1000 SKUs, respectively.
59
60 These test problems present extremely large warehouses that store a high number of SKUs and have high
61
62
63 21
64
65
Journal Pre-proof

1
2
3
4
5
6
7
8
9
10
11

of
12
13
14
15

pro
16
17
18
19
20 Figure 10: Pareto frontier of the 1000 SKUs test problem, solutions format:(na ,nc , cross-aisles type).
21
22
23 inventory turnover. For example, forklifts in the 1000 SKUs test problem perform, on average, 3470 picks per
24
25
re-
hour. This explains the relatively large number of vehicles in these test problems. The following observations
26 and insights are inferred:
27
28
29 • Solutions in the Pareto frontiers can be clustered into two groups. The first group covers a few solutions
30
at the top left of the Pareto frontiers that impose a small amount of wasted space at the cost of high
P
31
32 transportation costs. The layouts in this group have few cross-aisles and aisles. The second group
33
34 consists of a larger set of solutions located at the bottom of the Pareto frontiers. These solutions
35
36 mainly incur lower transportation costs and higher wasted space as a result of including more cross-
al

37
38 aisles and aisles in the layout. The decision maker should select an optimal solution from the first
39 group when the unit space cost is considerably greater than the unit transportation cost. Otherwise,
40
urn

41 the optimal solution is among the second group.


42
43
• Increasing the number of cross-aisles from two (bidirectional or four unidirectional) to three improves
44
45 transportation costs significantly with only a marginal loss in space utilization. The transportation
46
47 costs continue to decrease but with marginal improvements as the number of cross-aisles increases,
48
49 while the utilization of the storage space deteriorates considerably. Therefore, adding too many cross-
Jo

50
51 aisles not only reduces space utilization but also does not necessarily improve the transportation costs,
52 as pickers must traverse the cross-aisles as well.
53
54
55 • Fixing the number of cross-aisles, transportation costs (or total travel distance) decrease as the number
56
of aisles increases (consequently, bay depths decrease); however, the resulting loss in space utilization
57
58 may not justify the improvement. This is in line with the findings of [11].
59
60
61 • The locations of production lines and outbound docks (P/D points) significantly affect the material
62
63 22
64
65
Journal Pre-proof

1
2
3
4 handling cost, especially when the warehouse length-to-width ratio is large. In such cases, lanes are
5
6 arranged along the long side of the layout as presented in Figure 6. Our experiments show that
7
8 arranging the production lines and outbound docks on the long sides of the warehouse, as presented
9
10 in Figure 6, may result in up to 25% savings in material handling costs.
11

of
12 • The improvement gained in transportation costs by adding a new aisle is not as significant as those
13
14 gained by adding a new cross-aisle.
15

pro
16 • Unidirectional cross-aisles appear more frequently in the Pareto frontiers. However, the effects of the
17
18 cross-aisle type on traffic congestion must also be taken into account.
19
20
• The number of solution points in a Pareto frontier increases as the size of the warehouse grows. This
21
22 is mainly because the range of the parameters increases.
23
24
25
re-
26 5 Conclusions
27
28
29 In this paper, we developed a simulation-based optimization algorithm to simultaneously optimize utilization
30
of the storage space and transportation costs in the layout of a block stacking warehouse. We developed a
P
31
32 closed-form solution to find the optimal number of aisles to maximize utilization of the storage space. Our
33
34 algorithm finds the optimal numbers of aisles and cross-aisles and the cross-aisle type.
35
36 A case study in the beverage industry shows that considering both space utilization and transportation
al

37
38 costs leads to a different layout than when focusing exclusively on only one of these objectives and the
39 obtained layout significantly improves the operational costs. Exhaustive computational experiments show
40
urn

41 that the model finds the Pareto frontier in a reasonable time for large-sized test problems. Although adding
42
43 new cross-aisles improves the total travel distance, the improvement rate decreases as more cross-aisles are
44
45 added, while the reduction in utilization of the storage space continues. Hence, adding new cross-aisles
46 beyond some level does not justify the loss of storage space even if the transportation unit cost is higher
47
48 than the space unit cost.
49
Jo

50 The solutions in the Pareto frontiers can be arranged into two clusters: layouts with high space utilization
51
52 but less efficiency in terms of transportation and those with high transportation efficiency but lower space
53
utilization. This can help the decision makers to identify the optimal layout considering transportation and
54
55 space unit costs. Layouts in the first group contain two to three cross-aisles, whereas the solutions in the
56
57 next group have relatively many cross-aisles (up to ten).
58
59 The number of aisles and consequently bay depths affects the material handling cost in addition to space
60
utilization. We found that increasing the number of aisles in a layout leads to a reduction in travel distance;
61
62
63 23
64
65
Journal Pre-proof

1
2
3
4 however, this reduction becomes less significant as the number of aisles grows. We also found that adding a
5
6 new cross-aisle to the layout results in a greater improvement in material handling cost than adding a new
7
8 aisle.
9
10
11

of
12 References
13
14 [1] E. Ardjmand, H. Shakeri, M. Singh, and O. S. Bajgiran. Minimizing order picking makespan with
15

pro
16 multiple pickers in a wave picking warehouse. International Journal of Production Economics, 206:
17
18 169–183, 2018. doi: 10.1016/j.ijpe.2018.10.001.
19
20 [2] P. Baker and M. Canessa. Warehouse design: A structured approach. European Journal of Operational
21
22 Research, 193(2):425–436, 2009. doi: 10.1016/j.ejor.2007.11.045.
23
24
25
re-
[3] J. J. Bartholdi and S. T. Hackman. Warehouse & Distribution Science: Release 0.98. The Supply Chain
26 and Logistics Institute, Atlanta, GA, 2017.
27
28
29 [4] M. Bortolini, M. Faccio, M. Gamberi, and R. Manzini. Diagonal cross-aisles in unit load warehouses
30
to increase handling performance. International Journal of Production Economics, 170:838–849, 2015.
P
31
32 doi: 10.1016/j.ijpe.2015.07.009.
33
34
35 [5] J. Boudreau, W. Hopp, J. O. McClain, and L. J. Thomas. On the interface between operations and
36
al

37 human resources management. Manufacturing & Service Operations Management, 5(3):179–202, 2003.
38 doi: 10.1287/msom.5.3.179.16032.
39
40
urn

41 [6] L. F. Cardona and K. R. Gue. How to determine slot sizes in a unit-load warehouse. IISE Transactions,
42
43 2018. doi: 10.1080/24725854.2018.1509159.
44
45 [7] L. F. Cardona, D. F. Soto, L. Rivera, and H. J. Martı́nez. Detailed design of fishbone warehouse layouts
46
47 with vertical travel. International Journal of Production Economics, 170, Part C:825–837, 2015. doi:
48
10.1016/j.ijpe.2015.03.006.
49
Jo

50
51 [8] L. Chen, A. Langevin, and D. Riopel. A tabu search algorithm for the relocation problem in a
52
53 warehousing system. International Journal of Production Economics, 129(1):147–156, 2011. doi:
54
10.1016/j.ijpe.2010.09.012.
55
56
57 [9] R. de Koster, T. Le-Duc, and K. J. Roodbergen. Design and control of warehouse order picking: A
58
59 literature review. European Journal of Operational Research, 182(2):481–501, 2007. doi: 10.1016/j.ejor.
60
2006.07.009.
61
62
63 24
64
65
Journal Pre-proof

1
2
3
4 [10] S. Derhami. Designing Optimal Layouts for Block Stacking Warehouses. PhD thesis, Auburn University,
5
6 Auburn, AL, 8 2017.
7
8
[11] S. Derhami, J. S. Smith, and K. R. Gue. A simulation model to evaluate the layout for block stacking
9
10 warehouses. In 14th IMHRC Proceedings: Progress in Material Handling Research: 2016, Karlsruhe,
11

of
12 Germany, 2016.
13
14
[12] S. Derhami, J. S. Smith, and K. R. Gue. Optimising space utilisation in block stacking warehouses.
15

pro
16 International Journal of Production Research, 55(21):6436–6452, 2017. doi: 10.1080/00207543.2016.
17
18 1154216.
19
20
[13] S. Derhami, J. S. Smith, and K. R. Gue. Space-efficient layouts for block stacking warehouses. IISE
21
22 Transactions, 51(9):957–971, 2019. doi: 10.1080/24725854.2018.1539280.
23
24
25
re-
[14] A. H. Gharehgozli, Y. Yu, X. Zhang, and R. de Koster. Polynomial time algorithms to minimize total
26
travel time in a two-depot automated storage/retrieval system. Transportation Science, 51(1):19–33,
27
28 2017. doi: 10.1287/trsc.2014.0562.
29
30
[15] M. Goetschalckx and D. H. Ratliff. Optimal lane depths for single and multiple products in block
P
31
32
stacking storage systems. IIE Transactions, 23(3):245–258, 1991. doi: 10.1080/07408179108963859.
33
34
35 [16] E. H. Grosse and C. H. Glock. The effect of worker learning on manual order picking processes.
36
al

37 International Journal of Production Economics, 170:882–890, 2015. doi: 10.1016/j.ijpe.2014.12.018.


38
39 [17] J. Gu, M. Goetschalckx, and L. F. McGinnis. Research on warehouse operation: A comprehensive
40
urn

41 review. European Journal of Operational Research, 177(1):1–21, 2007. doi: http://dx.doi.org/10.1016/


42
43 j.ejor.2006.02.025.
44
45 [18] J. Gu, M. Goetschalckx, and L. F. McGinnis. Research on warehouse design and performance evaluation:
46
47 A comprehensive review. European Journal of Operational Research, 203(3):539–549, 2010. doi: 10.
48
49 1016/j.ejor.2009.07.031.
Jo

50
51 [19] K. R. Gue and R. D. Meller. Aisle configurations for unit-load warehouses. IIE Transactions, 41(3):
52
53 171–182, 2009. doi: 10.1080/07408170802112726.
54
55 [20] K. R. Gue, G. Ivanovć, and R. D. Meller. A unit-load warehouse with multiple pickup and deposit points
56
57 and non-traditional aisles. Transportation Research Part E: Logistics and Transportation Review, 48(4):
58
59 795–806, 2012. doi: 10.1016/j.tre.2012.01.002.
60
61
62
63 25
64
65
Journal Pre-proof

1
2
3
4 [21] X. Guo, Y. Yu, and R. B. de Koster. Impact of required storage space on storage policy performance
5
6 in a unit-load warehouse. International Journal of Production Research, 54(8):2405–2418, 2016. doi:
7
8 10.1080/00207543.2015.1083624.
9
10 [22] D. Kind. Elements of space utilization. Transportation and Distribution Management, 15:29–34, 1975.
11

of
12
13 [23] T. Lamballais, D. Roy, and M. de Koster. Estimating performance in a robotic mobile fulfillment system.
14 European Journal of Operational Research, 256(3):976–990, 2017. doi: 10.1016/j.ejor.2016.06.063.
15

pro
16
17 [24] T. N. Larson, H. March, and A. Kusiak. A heuristic approach to warehouse layout with class-based
18
storage. IIE Transactions, 29(4):337–348, 1997. doi: 10.1080/07408179708966339.
19
20
21 [25] J. Li, M. Moghaddam, and S. Y. Nof. Dynamic storage assignment with product affinity and ABC
22
23 classification—a case study. The International Journal of Advanced Manufacturing Technology, 84(9):
24
25
re-
2179–2194, 2016. doi: 10.1007/s00170-015-7806-7.
26
27 [26] J. O. Matson. The analysis of selected unit load storage systems. PhD thesis, Georgia Institute of
28
29 Technology, Atlanta, GA, 1982.
30
P
31 [27] M. Moshref-Javadi and M. R. Lehto. Material handling improvement in warehouses by parts clustering.
32
33 International Journal of Production Research, 54(14):4256–4271, 2016. doi: 10.1080/00207543.2016.
34
1140916.
35
36
al

37 [28] C. H. Mowrey and P. J. Parikh. Mixed-width aisle configurations for order picking in distribution
38
39 centers. European Journal of Operational Research, 232(1):87–97, 2014. ISSN 0377-2217. doi: 10.1016/
40
j.ejor.2013.07.002.
urn

41
42
43 [29] I. V. Nieuwenhuyse and R. B. de Koster. Evaluating order throughput time in 2-block warehouses
44
45 with time window batching. International Journal of Production Economics, 121(2):654–664, 2009. doi:
46 10.1016/j.ijpe.2009.01.013.
47
48
49 [30] O. Öztürkoğlu, K. Gue, and R. Meller. A constructive aisle design model for unit-load warehouses with
Jo

50
multiple pickup and deposit points. European Journal of Operational Research, 236(1):382–394, 2014.
51
52 doi: 10.1016/j.ejor.2013.12.023.
53
54
55 [31] J. A. Pazour and R. D. Meller. An analytical model for A-frame system design. IIE Transactions, 43
56
(10):739–752, 2011. doi: 10.1080/0740817X.2010.549099.
57
58
59 [32] F. Ramtin and J. A. Pazour. Product allocation problem for an AS/RS with multiple in-the-aisle pick
60
61 positions. IIE Transactions, 47(12):1379–1396, 2015. doi: 10.1080/0740817X.2015.1027458.
62
63 26
64
65
Journal Pre-proof

1
2
3
4 [33] K. J. Roodbergen and R. de Koster. Routing methods for warehouses with multiple cross aisles. Inter-
5
6 national Journal of Production Research, 39(9):1865–1883, 2001. doi: 10.1080/00207540110028128.
7
8
[34] K. J. Roodbergen and I. F. A. Vis. A model for warehouse layout. IIE Transactions, 38(10):799–811,
9
10 2006. doi: 10.1080/07408170500494566.
11

of
12
13 [35] K. J. Roodbergen, G. P. Sharp, and I. F. Vis. Designing the layout structure of manual order picking
14
areas in warehouses. IIE Transactions, 40(11):1032–1045, 2008. doi: 10.1080/07408170802167639.
15

pro
16
17 [36] K. J. Roodbergen, I. F. Vis, and G. D. T. Jr. Simultaneous determination of warehouse layout and
18
19 control policies. International Journal of Production Research, 53(11):3306–3326, 2015. doi: 10.1080/
20
00207543.2014.978029.
21
22
23 [37] B. Rouwenhorst, B. Reuter, V. Stockrahm, G. van Houtum, R. Mantel, and W. Zijm. Warehouse
24
25
re-
design and control: Framework and literature review. European Journal of Operational Research, 122
26
(3):515–533, 2000. doi: http://dx.doi.org/10.1016/S0377-2217(99)00020-X.
27
28
29 [38] L. M. Thomas and R. D. Meller. Analytical models for warehouse configuration. IIE Transactions, 46
30
(9):928–947, 2014. doi: 10.1080/0740817X.2013.855847.
P
31
32
33 [39] L. M. Thomas and R. D. Meller. Developing design guidelines for a case-picking warehouse. International
34
35 Journal of Production Economics, 170, Part C:741–762, 2015. doi: 10.1016/j.ijpe.2015.02.011.
36
al

37 [40] T. S. Vaughan and C. G. Petersen. The effect of warehouse cross aisles on order picking efficiency.
38
39 International Journal of Production Research, 37(4):881–897, 1999. doi: 10.1080/002075499191580.
40
urn

41
42 [41] N. Zaerpour, R. B. de Koster, and Y. Yu. Storage policies and optimal shape of a storage system.
43 International Journal of Production Research, 51(23-24):6891–6899, 2013. doi: 10.1080/00207543.2013.
44
45 774502.
46
47
48 [42] J. Zhang, F. Liu, J. Tang, and Y. Li. The online integrated order picking and delivery considering
49
Jo

pickers learning effects for an o2o community supermarket. Transportation Research Part E: Logistics
50
51 and Transportation Review, 123:180–199, 2019. doi: 10.1016/j.tre.2019.01.013.
52
53
54 [43] S. Zhou, Y. Y. Gong, and R. de Koster. Designing self-storage warehouses with customer choice.
55 International Journal of Production Research, 54(10):3080–3104, 2016. doi: 10.1080/00207543.2016.
56
57 1158880.
58
59
60
61
62
63 27
64
65
Journal Pre-proof

1
2
3
4 A Proof of lemma 2
5
6
7 Proof. Once a lane is fully occupied, S h xi storage volume becomes unavailable for the period that the lane
8
9 is partially or fully occupied. As described in [13], the total lane-time that SKU i occupies in Ti1 is
10
11  
1

of
12 ((Iimax − 1) + (Iimax − Zi xi − 1) + (Iimax − 2Zi xi − 1) + · · · + (Iimax − Ki Zi xi )) , (A.14)
Pi − λi
13
14
15 where Iimax is the maximum inventory level for SKU i and obtained by

pro
16
17
18 Qi (Pi − λi )
Iimax ≈ , (A.15)
19 Pi
20
21
22 and Ki is the number of required lanes for storage and is given by
23
24
25 Ki ≈
re-Qi (Pi − λi )
. (A.16)
Pi Z i x i
26
27
28 Similarly, the total lane-time that SKU i occupies in Ti2 is
29
30  
1
P
31 (Iimax + (Iimax − Zi xi ) + (Iimax − 2Zi xi ) + · · · + (Iimax − Ki Zi xi )) . (A.17)
32 λi
33
34
Summing (A.14) and (A.17) produces the total lane-time that SKU i occupies in its cycle time. Multiplying
35
36 the result by S h xi gives the total occupied volume-time, and multiplying it by λi /Qi , the cycle time of SKU
al

37
38 i, gives the average unavailable storage space occupied by SKU i
39
40
S h xi S h (Qi (Pi − λi ) − 2λi )
urn

41 Ui = + . (A.18)
42 2 2Pi Zi
43
44
Adding unavailable storage space dedicated to the aisles gives the total unavailable storage space in the
45
46 warehouse
47  X X S h (Qi (Pi − λi ) − 2λi )
h w Sh
48 Ū = AS S na + xi + . (A.19)
49 2 2Pi Zi
Jo

i∈I i∈I
50
51 The variable part of (A.19) is equivalent to the variable part of (2). Therefore, assuming a common bay
52
depth, n∗a minimizes (A.19).
53
54
55
56
57
58
59
60
61
62
63 28
64
65

Potrebbero piacerti anche