Simulation based Performance Analysis of an End-of-Aisle

Automated Storage and Retrieval System

Behnam Bahrami, El-Houssaine Aghezzaf and Veronique Limère

Department of Industrial Management, Ghent University, Technologiepark 903, Gent-Zwijnaarde, Belgium

Keywords: Automated Storage and Retrieval Systems, End-of-Aisle, Simulation.

Abstract: This paper presents and discusses simulation of an End-of-Aisle automated storage and retrieval system,

using FLEXSIM 6. The objective of the simulation model is to analyze and compare results of different

control policies and physical designs. The performance measures considered for the evaluation of each

control policy and layout combination are the total travel time of the crane and the number of storage and

retrieval operations performed. The experiments set up and the corresponding results are discussed.

1 INTRODUCTION

Automated storage and retrieval systems (AS/RS)

are widely used in manufacturing facilities,

distribution centers and warehouses. An AS/RS

system is defined as a storage system that uses

fixed- path storage and retrieval machines (cranes),

running on one or more rails between fixed arrays

of storage racks.(Automated Storage Retrieval

Systems Production Section of the Material

Handling Industry of America, 2005). In such

systems, stock keeping units (SKUs) arrive at an

Input station from which they are moved by a

crane to their storage location. In case of retrieval,

the crane transports the SKUs from their storage

location to an Output station. AS/RS have both

some advantages and disadvantages. The effective

use of an AS/RS may lead to substantial savings for

a company by reducing direct and indirect labor,

energy, maintenance, and building costs (Sarker

et.al., 1995). Apparent disadvantages are high

investments costs (approximately $634,000 for a

single aisle AS/RS), less flexibility and higher

investments in control systems (about $103,000),

(Zollinger, 1999).

In designing an AS/RS, both physical design

and control policies must be carefully considered to

fully take advantages of its positive impact. On

the one hand, the physical design consists of two

aspects. First, the AS/RS type has to be chosen;

this is what is called the system choice. Secondly,

the chosen system must be configured, for example,

by deciding on the number of aisles and rack

dimensions; this is called the system configuration.

On the other hand, the control policies are methods

which determine the actions performed by the

AS/RS (e.g., storage assignment, dwell point

policy), (Roodbergen, 2009). An AS/RS is usually

hired for transporting unit loads (e.g., fully loaded

pallets) within the system; but, in many cases, only

part of the unit-load may be needed to fulfill a

customer’s order. A common option to resolve this

situation is when the AS/RS drops off the retrieved

unit loads at a workstation at the end of the aisle. An

operator at this workstation takes the required

amount of products from the unit- load, and the

AS/RS moves the remainder of the load back into

the storage rack (Figure 1). This system is often

referred to as an End-of-Aisle (EOA) system. If the

unit-loads are bins, then the system is generally

called a miniload AS/RS (Roodbergen, 2009). In this

study various design scenarios for End-of-Aisle

automated storage/retrieval systems are evaluated

via simulation.

LITERATURE

REVIEW

Hausman et al., (1976) initiated evaluation of

performance in AS/RS systems using analytical and

empirical methods. Specifically, they addressed the

problem of optimal storage assignment with

considering two main policies: randomized storage

and class-based storage.

334

Bahrami B., Aghezzaf E. and Limère V..

Simulation based Performance Analysis of an End-of-Aisle Automated Storage and Retrieval System.

DOI: 10.5220/0004927603340341

In Proceedings of the 3rd International Conference on Operations Research and Enterprise Systems (ICORES-2014), pages 334-341

ISBN: 978-989-758-017-8

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

Figure 1: End-of-Aisle AS/RS order picking system

(Hwang et al., 2002).

They showed that significant reductions in S/R

machine travel time are achieved from turnover-

based storage rules such as a class-based storage

policy. Accordingly, Bozer and White (1984)

developed a continues rack configuration to

evaluate single and dual travel times for rectangular

racks and variety of I/O configurations.

Furthermore, Han et al., (1987) illustrated that by

sequencing the retrieval orders, the throughput

performance of the AS/RS can be improved. This

can thus reduce the interleaving travel time between

storage and retrieval locations in a dual command

cycle.

For simulation models in AS/RS, most of the

researchers only evaluated some of the physical

design aspects in combination with a limited

number of control policies. Meller and

Mungwattana (2005) applied simulation to evaluate

the benefits of different dwell point policies. The

results indicate that the position of dwell point

has an negligible effect on system respond time

when the AS/RS has high utilization. Randhawa and

Shroff (1995) examined the effect of different

sequencing rules on six layout configurations (with

varying I/O-point, item distribution over racks, rack

configuration and rack dimensions). Based on a

limited number of experiments they mentioned in

their study that the position of I/O point at the

middle of the rack results in a higher throughput.

Vanderberg et al., (2000) have developed a

simulation study and examined various aspects of

AS/RS control policies: storage location assignment

policies, request selection rules, open location

selection rules and urgency rules. Considering

randomized storage and class-based storage, they

concluded that using a FCFS sequence for the

retrievals by implementing urgency rules result in

better expected time. Randhawa et al., (1991) used

simulation to evaluate single and dual I/O point

configurations in a unit load AS/RS and their

model demonstrated that the dual dock layout

maximizes the throughput.

In an End-of-Aisle AS/RS, current retrieval

operation become next storage operation, since

loads are returned into the system after items have

been picked. A mathematical model to investigate

the performance of an EOA order picking systems

have been proposed by Bozer and White (1996).

They provided some approximate expressions for

the expected travel time for systems performing

under peak demand, and presented a design

algorithm in order to calculate the minimum

number of miniload aisles needed to meet a given

throughput requirement. Foley and Frazelle (1991)

have developed an exact solution for the S/R

machine travel time under the FCFS retrieval

policy in order to evaluate the throughput for the

miniload AS/RS. Hwang et al., (2002) proposed the

design of miniload AS/RSs in combination with

Automated Guided Vehicles. To recognize the

optimal number of loads transferred by each AGV

to machines, a non-linear model and heuristics have

been developed in their research.

To the best of our knowledge, there is no

simulation model for the End-of-Aisle system,

therefore in this study a multitude of physical design

aspects and control policies are simultaneously

included for the End-of-Aisle system.

SIMULATION MODELING

THE SYSTEM

3.1 Assumptions

In the system analyzed, a crane serves a single aisle

with storage racks placed on one side of the aisle.

All storage locations are identical in size and each

location can hold one unit load. The total capacity

of the system is 100 unit loads. Each unit load

(e.g., pallet) contains a number of boxes of one

item type. Although the pallet sizes are constant,

the size of the boxes on the pallets are different for

different item types. The crane’s pick-up and

deposit time is calculated according to the size and

number of boxes inside the pallet; for pallets with a

higher number of boxes and larger boxes, the time

for pick- up and deposit is higher. Moreover, the

times for manual loading and unloading boxes from

pallets are calculated according to the boxes’

dimension.

The turnover of each itemtype is known in

advance and is the same for all scenarios tested.

The crane scheduling rule in this study is FCFS

(first come, first serve) for both storage and

retrieval operations. The capacity of the crane is

SimulationbasedPerformanceAnalysisofanEnd-of-AisleAutomatedStorageandRetrievalSystem

335

one pallet and the crane can move horizontally and

vertically simultaneously. Crane acceleration and

deceleration are assumed instantaneous. The actual

travel time equals the maximum of the horizontal

and vertical travel time (Chebyshev distance metric).

A unit-load AS/RS can perform in two ways,

namely in a single command cycle or in a dual

command cycle. In a single command cycle the

crane operates either a single storage or a single

retrieval operation. The storage cycle time then is

equal to the sum of the time to pick-up a load at the

input station, the time to travel to the storage

location, the time to place the load in the rack

and the time to return to the input station. The

retrieval cycle time can be defined similarly. A dual

command cycle can be defined as performing both

a storage and a retrieval request simultaneously in a

single cycle. In this case, the cycle time is defined

as the sum of the time to pick up the load, the

time to travel to the storage location and store the

load, the empty travel time (interleaving time) from

the storage location to the retrieval location and the

time to pick the unit-load and transport it to the

output station. In this study the crane can perform

both single and dual command. This means when

both storage and retrieval orders are in the queue,

the crane performs in dual command but when only

a storage order or a retrieval order is available, the

crane moves in single command.

If the pallets’ arrival rate is very high, the

system will be completely operating on dual

command, while by reducing the arrival rate,

single- and dual- command cycles will be combined

in the system. To provide a better basis for

analyzing the system, in this study five scenarios

for arrival of pallets to the system are assumed.

Arrival rates are 25, 40, 60, 80 and 100 per hour

respectively.

To study the End-of-Aisle system described, we

simulated storage and retrieval for a number of

different scenarios. In the following sections, we

give an overview of the factors included.

3.2 Configurations of Rack

Since cranes can move vertically and horizontally

simultaneously, a good balance between rack height

and length can help to reduce travel times. In this

study three configurations are assumed. The first

configuration consists of 10 bays, and 10 levels; the

second has 5 bays and 20 levels; and the third one

has 20 bays and 5 levels (Figures 2, 3, 4).

Figure 2: Rack with 10 bays and 10 levels.

Figure 3: Rack with 5 bays and 20 levels.

Figure 4: Rack with 20 bays and 5 levels.

3.3 Input/Output Station

An input station is the location where incoming

products are received from other parts of the

warehouse by for example, automated guided

vehicles, conveyors, or forklift trucks, before they

are stored. An output station is where retrieved

products are unloaded before being moved to

shipping. In this End-of-Aisle model, it is

assumed the output station is at the end of the aisle.

For the input station however, two different

positions are assumed, one in the first bay of the

rack and a second one in the middle of the rack

(Figure 5).

ICORES2014-InternationalConferenceonOperationsResearchandEnterpriseSystems

336

Figure 5: Location of input station: in the first bay (a), in

the middle of the rack (b).

3.4 Storage Assignment Policies

In this model two storage policies are

considered, i.e. random storage and class-based

storage. For random storage, every incoming pallet

is assigned to a location in the racks that is selected

randomly from all eligible empty locations with

equal probability (Figure 6). For class-based storage

the racks and pallets are divided into K classes

based on their turnover frequency. Pallets with

higher demand frequency are assigned to class I

while pallets with smaller demand frequency are

assigned to class II and so on. The position of

class I is the best location close to the Output

station. For class II the position is the second best

location near the Output station (Figure 7).

Figure 6:

Random-based

storage.

Figure 7: Class-based storage.

3.5 Dwell Point Policy

When the crane is idle, its location is called as the

dwell point. Choosing a proper dwell point can

minimize the expected travel time for the next

operation. In this model three dwell points are

considered: at the Input station, in the middle of

the aisle and at the Output station respectively

(Figure 8).

Figure 8: Position of crane: in the Input station (a), in the

middle of the rack (b), in the Output station (c).

SIMULATION

RESULTS

The simulation results are given in the following

tables. For each storage policy – i.e. random

storage and class-based storage different physical

designs are analyzed for five arrival rates of pallets.

Specifically, for each case the performance of the

system is evaluated according to dwell point, input

station and rack configuration. Each simulation is

run for 3600 seconds. The measures of performance

are: total travel time of the crane and total number

of storages and retrievals performed by the system.

To evaluate the performance of the model, the

simulation was initiated with storage rack utilization

at zero (rack is completely free). Considering the

nature of random storage, the simulation results are

based on 40 replications for each scenario, and the

average results are reported. Results of the

experiments can be read from Table 1 and 2.

In the following subsections we will discuss the

results.

4.1 Total Travel Time of the Crane

The following plot indicates that the total travel

time increases significantly when the number of

SimulationbasedPerformanceAnalysisofanEnd-of-AisleAutomatedStorageandRetrievalSystem

337

Table 1: Total crane’s travel and total number of storage and retrievals for Random Based Storage.

Dwell

Point

Input

station

B*L

Arrival rate (pallets/h)

25 40 60 80 100

TT S

R TT S

Input f

20*5 1758 35 3017 57 4326.4 77 4571.5 78 4603.9 82

Input m

20*5 1155 36 2013 58 3043.2 84 3563.8 103 3733.5 110

put f

20*5 1798 35 3075 57 4366.2 77 4619.2 78 4679.6 82

put m

20*5 1154 36 2025 58 3042.3 84 3554.7 103 3732.6 110

20*5 1778 35 3037 57 4346.2 77 4591 78 4627.8 82

20*5 1136 36 1994 58 3023.9 84 3534.3 103 3710.3 110

Input f

10*10 866 36 1496 58 2248.5 84 2963.4 108 3245.7 122

Input m

10*10 526 36 970 59 1380.7 83 1779.5 109 2049.3 131

put f

10*10 885 36 1516 58 2148.3 84 2822.6 108 3265.9 122

put m

10*10 525 36 969 59 1379.6 83 1778.3 109 2047.5 131

10*10 875 36 1506 58 2258.2 84 2973.3 108 3255.9 122

10*10 517 36 961 59 1371.4 83 1770.1 109 2037.3 131

Input f

5*20 415 36 735 60 1094.6 83 1440.5 109 1685.6 131

Input m

5*20 223 36 441 60 583.96 84 739.89 109 863.22 132

put f

5*20 425 36 745 60 1104.4 83 1450.3 109 1696.4 131

put m

5*20 222 36 420 60 579.46 84 738.34 109 861.62 132

5*20 420 36 740 60 1099.5 83 1445.4 109 1689.8 131

5*20 219 36 417 60 579.46 84 735.39 109 858.58 132

Table 2: Total cranes travel time and total number of storages and retrievals for Class-Based storage.

Dwell

Point

Input

station

B*L

Arrival rate (pallets/h)

25 40 60 80 100

TT S

R TT S

Inpu

20*5 1568 35 2583 56 4044.3 81 4561.7 100 4888.1 118

Input m

20*5 1091 36 1923 58 2830.4 83 3259.6 104 3608.5 124

put f

20*5 1564 35 2622 56 4084 81 4586.4 100 4947.9 118

put m

20*5 1087 36 1906 58 2820.3 83 3287.1 104 3604.6 124

20*5 1587 35 2570 56 4064 81 4553.6 100 4907.9 118

20*5 1078 36 1903 58 2810.4 83 3270.8 104 3585.5 124

Input f

10*10 758 36 1311 59 1992.8 84 2616.2 107 2944.2 128

Input m

10*10 574 36 991 59 1402 84 1834.5 109 2083.7 130

put f

10*10 735 36 1331 59 2012.5 84 2636 107 2968.3 128

put m

10*10 570 36 987 59 1398.3 84 1830.3 109 2079.6 130

10*10 757 36 1252 59 2002.5 84 2626 107 2958.1 128

10*10 564 36 981 59 1392.5 84 1824.4 109 2073.1 130

Input f

5*20 351 36 623 59 943.71 84 1348.8 109 1509.1 132

Input m

5*20 255 36 442 60 589.27 84 756.28 109 849.68 133

put f

5*20 328 36 633 59 953.48 84 1358.6 109 1518.8 132

put m

5*20 251 36 438 60 585.22 84 759.94 109 845.64 133

5*20 356 36 628 59 948.50 84 1353.6 109 1513.9 132

5*20 250 36 437 60 584.16 84 758.88 109 844.58 133

ICORES2014-InternationalConferenceonOperationsResearchandEnterpriseSystems

338

arriving pallets to the system change from 25 to 60,

but for higher rates of incoming pallets from 60 to

100, the total crane’s travel time rises slowly and

become more stable.

Figure 9: Total crane’s travel times for different arrival

rates.

4.2 Random based Storage Vs.

Class-based Storage

Based upon the crane’s travel time and the number

of storages and retrievals carried out by the system,

the analysis demonstrates that

class-

based storage

offers a better performance than random based

storage. The comparison between both types is

given in Figure 11 and Figure 12.

Figure 10: Total travel times for random and class-based.

Figure 11: Number of storages and retrievals for

random and class-based.

4.3 Rack Configuration

The result obtained for different configurations

indicate that with increasing in the number of bays,

the total travel time increases and the total number

of storages and retrievals increase. figures (13, 14).

Figure 12: Total crane’s travel time.

Figure 13: Total number of storages and retrievals.

4.4 Input Station Location

According to the results it can be observed that

the best location for dock, is at the middle of

the rack. The given diagrams illustrate the total

travel distance and total number of retrievals and

storages

Figure 14: Total crane’s travel time.

SimulationbasedPerformanceAnalysisofanEnd-of-AisleAutomatedStorageandRetrievalSystem

339

Figure 15: Total number of storages and retrievals.

4.5 Dwell Point Location

The results achieved by the simulation model,

reveal that the position of dwell point has negligible

impact on both travel time and number of storage

and retrievals. The following figures represent the

comparison for three dwell point locations.

Figure 16: Total crane’s travel time.

Figure 17: Total number of storages and retrievals.

5 CONCLUSIONS

This paper presents a simulation based performance

analysis of an End-of-Aisle automated storage and

retrieval system. Results of different control policies

and physical designs were compared. In particular,

the efficiency of the system for two types of

storage, namely class-based and random based,

with combinations of different dwell points, Input

locations and various rack configurations, were

analyzed. This preliminary analysis shows that

combination of class-based storage with an input

station in the middle of the rack, seems to provide

a superior performance. The analysis with other

strategies is ongoing.

REFERENCES

Automated Storage Retrieval Systems Production Section

of the Material Handling Industry of America (2005).

Sarker, B. R., Sobhan B., 1995. “Travel time models in

automated storage/retrieval systems: A critical

review”. Int. Journal of Production Economics 40173,

184–193.

Bozer, Y. A., and White, J. A., 1984, “Travel time models

for automated storage/retrieval systems”. IIE

Transactions, 16 (4), 329- 338.

Bozer, Y. A., White, J. A., 1996. “A generalized design

and performance analysis model for end-of-aisle

order- picking systems”. IIE Transactions 28 (4), 271–

280.

Foley, R. D., Frazelle, E. H., 1991. “Analytical results for

miniload throughput and the distribution of dual

command travel time”. IIE Transactions 23 (3), 273–

281.

Graves, S. C., Hausman, W. H., Schwarz, L. B., 1977.

“Storage-retrieval interleaving in automatic

warehousing systems”. Management Science 23 (9),

935–945.

Han, M. H., McGinnis, L. F., Shieh, J. S., White, J. A.,

1987. “On sequencing retrievals in an automated

storage/retrieval system”. IIE Transactions 19 (1), 56–

66.

Hausman, W. H., Schwarz, L. B., Graves, S. C., 1976.

“Optimal storage assignment in automatic

warehousing systems”. Management Science 22 (6),

629–638.

Hwang, H., Moon, S., Gen, M., 2002. “An integrated

model for the design of end-of-aisle order picking

system and the determination of unit load sizes of

AGVs”. Computers & Industrial Engineering 42,

249–258.

Kees Jan Roodbergen , Iris F. A. Vis., 2009. “A survey of

literature on automated storage and retrieval systems”.

European Journal of Operational Research 194, 343–

362.

Meller, R. D, Mungwattana. A., 2005. “AS/RS dwell point

strategy selection at high system utilization: A

simulation study to investigate the magnitude of the

benefit”. International Journal of Production

Research 43(24), 5217-5227.

Randhawa, S. U., McDowell, E. D., Wang, T. W., 1991.

“Evaluation of scheduling rules for single- and dual-

dock automated storage/retrieval system”. Computers

& Industrial Engineering, Volume 20, Issue 4, 401–

410.

ICORES2014-InternationalConferenceonOperationsResearchandEnterpriseSystems

340

Randhawa, S. U., Shroff, R., 1995. “Simulation-based

design evaluation of unit load automated

storage/retrieval systems”. Computers & Industrial

Engineering 28 (1), 71–79.

Van den Berg, J. P., Gademann, A. J. R. M., 2000.

“Simulation study of an automated storage/retrieval

system”. International Journal of Production

Research 38 (6), 1339–1356.

SimulationbasedPerformanceAnalysisofanEnd-of-AisleAutomatedStorageandRetrievalSystem

341