Concept Development and Evaluation of Order Assignment

Strategies in a Highly-dynamic, Hybrid Pallet Storage and Retrieval

System

Giulia Siciliano

and Johannes Fottner

Chair of Materials Handling, Material Flow, Logistics, Technical University of Munich, Boltzmannstraße 15,

Garching bei München, Germany

Keywords: Stacker-Crane-based Warehouse, Shuttle System, Discrete Event Simulation, Order Assignment Strategies,

Control System.

Abstract: In this paper, we propose and evaluate various order assignment strategies for a new, highly-dynamic hybrid

pallet storage and retrieval system or dynamic hybrid pallet warehouse (DHPW). The research gap we fill is

the identification of the order assignment strategies, which guarantees the highest performance for this new

warehouse system in different operational conditions. First, we present a brief description of the system under

consideration. We then go on to present a brief literature review on order assignment strategies for stacker-

crane-based AS/RSs and AVS/RSs. Next, we develop our concept of order assignment strategies for a DHPW

with sequenced retrieval and storage, followed by a discrete event simulation and an evaluation of the concepts

according to the level of performance they enable. Finally, we identify the operations of the process whose

order assignment strategy has the greatest impact on performance, and suggest an optimum combination of

order assignment strategies.

1 INTRODUCTION

In this contribution, we propose a number of order

assignment strategies for a dynamic hybrid pallet

warehouse (DHPW) and evaluate these strategies and

their combinations using discrete event simulation to

determine which strategies provide the best

performance for this newly developed warehouse

system. The goal for this new system is to combine

the advantages of stacker-crane-based warehouses

and shuttle-based ones (Eder et al., 2019; Siciliano et

al., 2020). A shuttle system forms the base tier, with

a multiple-depth channel storage system constructed

above it (see Fig. 1). A transfer buffer, served by

stacker cranes with satellites, connects the base tier

and the channel storage system. Fig. 2 shows the

various elements of the base tier, including the

transfer buffers. For retrieval, a stacker crane takes a

pallet from the multiple-depth channel storage and

brings it to the transfer buffer on the base tier. A

shuttle then brings the pallet to the output location

https://orcid.org/0000-0002-8438-9409

https://orcid.org/0000-0001-6392-0371

Figure 1: DHPW (image courtesy of Gebhardt

Fördertechnik GmbH).

(O location). The moment a shuttle receives an order

with a certain location on the transfer buffer as the

start position and the O location as the destination, it

reserves that start location on the transfer buffer. For

storage, a shuttle takes a pallet from the input location

(I location) and brings it to the transfer buffer. A

stacker crane then takes the pallet and places it in the

multiple-depth channel storage system. The moment

a shuttle is preparing to leave the I location with a

pallet to store, it selects and reserves the destination

location on the transfer buffer.

360

Siciliano, G. and Fottner, J.

Concept Development and Evaluation of Order Assignment Strategies in a Highly-dynamic, Hybrid Pallet Storage and Retrieval System.

DOI: 10.5220/0010565903600368

In Proceedings of the 11th International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH 2021), pages 360-368

ISBN: 978-989-758-528-9; ISSN: 2184-2841

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

Figure 2: Shuttle base tier in the simulation model

(Siciliano et al., 2020).

In this paper, order assignment strategies determine

which job order should be assigned to which stacker

crane or shuttle. They are also employed to determine

which idle shuttle to activate. We therefore present a

brief literature review of those order assignment

strategies for stacker-crane-based and shuttle-based

warehouses that form the basis of our concept.

1.1 Order Assignment Strategies for

Stacker-Crane-based Warehouses

The order assignment decision taken by a stacker-

crane-based AS/RS concerns the I/O location from

which the stacker crane should take a pallet and the

I/O location to which it should deliver the pallet. In a

stacker-crane-based warehouse with more than one

I/O location, the simplest strategy is to randomly

select the I/O location to which the stacker crane

delivers the pallet (Arantes and Kompella, 1993).

However, other strategies may enable a higher

throughput. In fact, (Lantschner, 2015) proposes two

additional strategies. The first selects the I/O point

nearest to the stacker crane’s current position. The

second selects the I/O location nearest to the next job.

(Lantschner, 2015) demonstrates analytically that

these two strategies provide a shorter mean driveway

for a stacker-crane-based AS/RS with one stacker

crane, in which the number of I/O locations is varied

between two and five. (Gagliardi et al., 2014)

provides an extensive literature review on operational

decisions that can be taken in AS/RSs.

1.2 Order Assignment Strategies for

Shuttle-based Warehouses

An essential decision to be taken in a shuttle-based

AS/RS concerns which shuttle should take which

order, which requires selecting a dispatching strategy.

The majority of authors investigated dispatching

strategies for automated guided vehicles (AGVs) but

not directly for shuttles. According to (Grunow et al.,

2006), the two dispatching strategies adopted the most

frequently in the literature are assigning the next

available job to the vehicle nearest to the pick-up

location or assigning it to the vehicle that has

completed the least transportation orders. More

recently, (Habl et al., 2020) proposed order assignment

strategies in a single-tier, double-deep shuttle level and

evaluated them by means of discrete event simulation.

The next section describes the operations that take

place in the DHPW and explains when a decision is

required during those operations. We then suggest

several strategies by which to address this decision.

2 CONCEPT DEVELOPMENT:

RETRIEVAL IN SEQUENCE

The operations for retrieving in sequence are as

follows:

I. A stacker crane retrieves a pallet from the channel

storage and brings it to an available position on

the transfer buffer;

II. The stacker crane activates an idle shuttle;

III. The activated shuttle retrieves a pallet from the

transfer buffer and brings it to the I/O locations.

We now describe various strategies for addressing the

decision to be taken in each operation. For the sake of

clarity, the strategies are numbered.

2.1 Retrieval Strategies, Group I

In the first group of strategies relating to retrieval in

sequence, the stacker crane selects an available

transfer buffer position to which the pallet can be

brought. The following strategies are available:

1. Random Position

The stacker crane randomly chooses the position from

those available on the transfer buffer.

2. Nearest Position to O Location

The stacker crane chooses the available transfer

buffer position at the smallest distance from the O

location. The result of this is that the positions on the

transfer buffer nearest to the O location are constantly

occupied by retrieved pallets. This reduces the

distance the shuttles travel from the O location to the

transfer buffer to retrieve pallets, which minimizes

their travel time. However, concentrating all shuttles

Concept Development and Evaluation of Order Assignment Strategies in a Highly-dynamic, Hybrid Pallet Storage and Retrieval System

361

in a small area of the shuttle base tier causes increased

interference between the shuttles as they move. They

thus spend more time waiting to give way to shuttles

that have priority on certain routes.

3. Shortest Path for the Stacker Crane

The stacker crane chooses the available transfer

buffer position that provides the shortest path for the

stacker crane. When retrieving, both the stacker

crane’s idle position (IP) or dwell point and the

position P2, where the pallet to be retrieved in the

channel warehouse is stored, are fixed. Therefore, the

position in the transfer buffer ensuring the shortest

path for a single stacker crane cycle has the

coordinates (P2x, 0), as shown in Fig. 3:

Figure 3: Determination of PT, the retrieval location of the

transfer buffer ensuring the shortest path for the stacker

crane.

4. Position Available for the Longest Time

The stacker crane chooses the transfer buffer position

that has been available for the longest time. A shuttle

deletes its reservation of a transfer buffer location

once its chassis has fully exited it. From that moment

on, the position is therefore available again. This

strategy aims to balance the distribution of orders

among the different transfer buffer locations. As a

result, the routes of the shuttles are distributed over a

larger area. This increases their travel distances but

reduces interference between shuttles on their routes.

2.2 Retrieval Strategies, Group II

In the second group of strategies relating to retrieval in

sequence, the stacker crane selects an idle shuttle to

activate and to which to assign a new retrieval order.

For retrieval in sequence, idle shuttles wait at storage

locations. The following strategies are available:

5. Random Shuttle

The stacker crane randomly activates a shuttle among

those available at the storage locations on the shuttle

base tier.

6. Longest Shuttle Idle Time

The stacker crane awakens the shuttle that has been

waiting at a storage location for the longest time.

Shuttles wait at storage locations as long as no

retrieval order is available. The aim of this strategy is

to distribute the orders among the shuttles and avoid

some shuttles being overloaded while others are

underused. This makes maintenance more easily

predictable and more delayed in time, thus reducing

maintenance costs.

7. Nearest Shuttle to the Job

The stacker crane awakens the shuttle that is waiting

at the storage location with the shortest distance to the

pallet to be retrieved. The aim of this strategy is to

reduce shuttle travel distances and, in turn, their travel

times.

8. Least Utilized Shuttle

The stacker crane awakens the shuttle that has

executed the fewest orders since the commencement

of the retrieval process. Balancing the distribution of

jobs among shuttles in this way avoids some shuttles

being used more than others.

2.3 Retrieval Strategies, Group III

In the third group of strategies relating to retrieval in

sequence, an active shuttle selects an available pallet

on the transfer buffer for retrieval. The following

strategies are available:

9. Random Pallet

The shuttle randomly chooses a pallet from those

available on the transfer buffer.

10. Nearest Position to O Location

The shuttle chooses the available transfer buffer pallet

with the smallest distance to the O location. The aim

is to reduce the travel distance of the shuttles. If

multiple stacker cranes are in a single aisle, the

stacker crane serving the transfer buffer area nearest

to the O location is used more than the others.

11. Smallest Sequence Number

The shuttle chooses the pallet available on the transfer

buffer with the smallest sequence number. A loaded

shuttle is only permitted to depart from the transfer

buffer when the shuttle containing the pallet with the

preceding sequence number has left the transfer

buffer. The aim of this strategy is to reduce the time

a loaded shuttle has to wait on the transfer buffer

before being permitted to depart. When there are only

a small number of shuttles in the base tier, this

strategy is the only one that avoids deadlocks.

(IPx, IPy) fixed

(P2x, P2y) fixed

(P2x, 0)

(0,0)

SIMULTECH 2021 - 11th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

362

However, with a large number of shuttles, other

strategies may provide shorter shuttle travel times

and, in turn, a higher throughput than the one

achievable with this strategy.

3 CONCEPT DEVELOPMENT:

STORAGE

The operations for storage are as follows:

I. An available shuttle receives a storage order. It

chooses an updated transfer buffer destination

location. It takes the pallet from the I location and

delivers it to the chosen transfer buffer location;

II. A stacker crane chooses a pallet from the transfer

buffer that is available for storing. It takes it and

brings it to the channel of the warehouse in which

this type of product is stored.

We now describe various strategies for addressing the

decision to be taken in each operation.

3.1 Storage Strategies, Group I

In the first group of strategies relating to storage, a

shuttle selects a position available on the transfer

buffer to which to bring a pallet for storing. The

following strategies are available:

12. Random Position

The shuttle randomly chooses the position from

those available on the transfer buffer.

13. Nearest position to I location

The shuttle chooses the available transfer buffer

position with the smallest distance to the I location.

14. Position Available for the Longest Time

The shuttle chooses the available transfer buffer

position that has not been occupied by a pallet for the

longest time. An empty location on a transfer buffer

is available from the moment a satellite of a stacker

crane has exited it with its whole chassis until a

shuttle reserves it.

3.2 Storage Strategies, Group II

In the second group of strategies relating to storage,

the stacker crane selects an available pallet on the

transfer buffer. The following strategies are available:

15. Random Pallet

A stacker crane chooses a pallet randomly from the

ones available on the transfer buffer.

16. Nearest Pallet to I Location

The stacker crane chooses the available transfer

buffer pallet that is at the smallest distance to the I

location.

17. Shortest Path for the Stacker Crane

The stacker crane chooses the available transfer

buffer pallet whose position provides the shortest

cycle path for the stacker crane. In the case of storing,

IP and P1 (the position of the channel warehouse

where the pallet will be stored by the stacker crane)

are fixed. Therefore, the position (PT) on the transfer

buffer, where it is necessary to pick up the pallet in

order for the stacker crane to have the shortest path,

can be calculated using the optical law of reflection.

The stacker crane takes the available pallet that is as

close as possible to the calculated position. As shown

in Fig. 4, we locate PT graphically by considering the

x axis as a symmetry axis and projecting P1 onto the

other side of it. Then we join the projection of P1 and

the idle position with a straight-line s. The

intersection between the s and x axes is PT. We

calculate the abscissae of PT analytically as the point

belonging to s whose ordinate is zero:

−(−P1

)

PTx

−

P1x

−(−P1

)

IPx

−

P1x

(1)

𝑃𝑇𝑥 = P1

IPx − P1x

−P1

+𝑃1𝑥

(2)

Figure 4: Determination of PT, the storage location of the

transfer buffer that ensures the shortest path for the stacker

crane.

18. Pallet Available for the Longest Time

The stacker crane chooses the pallet that has been

available on the transfer buffer for the longest time. A

pallet delivered by a shuttle is available from the

moment when the shuttle has exited a delivery

location with its entire chassis.

(IPx, IPy) fixed

(P1x, P1y) fixed

(?, 0)

Projection of P1

(P1x, -P1y)

(0,0)

Concept Development and Evaluation of Order Assignment Strategies in a Highly-dynamic, Hybrid Pallet Storage and Retrieval System

363

4 SIMULATION STUDY

In this section, we analyze the results of the

simulation study and evaluate the order assignment

strategies described above as well as their

combinations. The performance evaluation criteria

are warehouse throughput, shuttle utilization ratio,

and length of time spent by shuttles on different

operations (Lienert et al., 2018).

4.1 Model Description

The model is implemented using the Plant Simulation

discrete event simulation environment. (Siciliano et

al., 2020) provides a detailed description of the

system’s modelling and implementation. We

validated the model comparing the analytically

calculated travel time of individual shuttles with the

simulation. Moreover, we compared the simulated

travel times of shuttles and stacker crane with the

real-world subsystems.

4.2 Parameters

In this paper, we consider a system with two stacker

cranes in a single aisle, serving 56 transfer buffers

arranged in two rows alongside the stacker cranes

(see Fig. 2). 512 storage locations for intermediate

buffering are in the base tier. Each half of the base tier

has three storage aisles. Four cross aisles are

positioned within the storage area. Two I/O areas are

located on either side of the system. Two bidirectional

lanes connect them. One I/O area is dedicated to

incoming pallets and the other to outgoing pallets.

Furthermore, within each I/O area there are two I/O

locations (see Fig. 2). In the channel storage, 56

storage channels are on each of the eight tiers. Each

channel has a capacity of nine pallets.

In the following experiments, five replications are

performed with the parameters in Tab. 1 and Tab. 2.

Each replication lasts for 24 hours simulated time.

4.3 Evaluation: Retrieval in Sequence

We evaluate by simulation the throughput of the

warehouse for each of the 48 combinations of

retrieval in sequence strategies (see Fig. 5).

Combination {3, 8, 11} achieves the highest

throughput. However, this combination only displays

a higher throughput than random ‘sequenced’

combination {1, 5, 11} when the stacker cranes

become the bottleneck of the system, i.e. for 10 or

more shuttles.

Table 1: Stacker crane parameters.

Parameter Value

Travel speed x

4.0 𝑚/𝑠

Travel acceleration x

0.5 𝑚/𝑠



Lifting speed y

1.0 𝑚/𝑠

Lifting acceleration y

1.0 𝑚/𝑠



Satellite speed z

1.2 𝑚/𝑠

Satellite acceleration loaded z

0.5 𝑚/𝑠



Satellite acceleration unloaded z

1.0 𝑚/𝑠



Time of pallet handover

2.0 𝑠

Time of satellite handover

6.0 𝑠

Time for positioning in channel

1.0 𝑠

Time for positioning before channel

1.0 𝑠

Table 2: Shuttle parameters.

Parameter Value

Speed (loaded)

0.6 𝑚/𝑠

Speed (empty)

1.0 𝑚/𝑠

Acceleration (loaded)

0.3 𝑚/𝑠



Acceleration (empty)

0.6 𝑚/𝑠



Turning time

6.6 𝑠

Handover time

10.0 𝑠

Combination {2, 8, 11} has the highest throughput for

eight shuttles, which is when the shuttles are still the

bottleneck of the system.

One interesting finding is that with less than 12

shuttles, random combination {1, 5, 9} results in a

very low throughput.

We also investigate the utilization ratio and time

components of the shuttles (see Fig. 6). We define the

utilization ratio as the sum of column components in

Fig. 6, which are dark-green, light green and yellow.

We note that combination {3, 8, 11} has the highest

utilization ratio. In fact, in combination {3, 8, 11} the

shuttles spend less time waiting for available transfer

buffer pallets than in combinations {1, 5, 11} and {2,

8, 11}.

Figure 5: Throughput provided by 48 combinations of order

assignment strategies for retrieval in sequence.

100

110

120

130

0246810121416

als pe

Hou

of Shuttles

{1,5,9}

{1,5,11}

{2,8,11}

{3,8,11}

SIMULTECH 2021 - 11th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

364

100

2 4 6 8 10 12 14 16

Amount of time [%]

Number of huttlesS

{2,8,11}

100

2 4 6 8 10 12 14 16

Amount of time [%]

Number of huttlesS

{3,8,11}

100

2 4 6 8 10 12 14 16

Amount of time [%]

Number of huttlesS

{1,5,11}

100

2 4 6 8 10121416

Amount of time [%]

Number of huttlesS

{1,5,9}

waiting for previous sequence number

waiting for available transfer buffer pallet

blocked

loading and unloading

Figure 6: Time components of significant combinations for

retrieval in sequence.

With random combination {1, 5, 9}, loaded

shuttles spend most of their time waiting on the

transfer buffer for the pallet with the previous

sequence number to be taken.

Next, we investigate which operation for retrieval

in sequence has the greatest impact on throughput.

For this purpose, we filter the throughput results of

the simulation, considering one operation at a time

(see Fig.7).

On one hand, the shuttle chooses an available

pallet on the transfer buffer operation has the biggest

effect on throughput when the shuttles are the

bottleneck of the system.

On the other hand, the stacker crane chooses an

available position on the transfer buffer operation has

the biggest effect on throughput when the stacker

cranes are the bottleneck of the system.

The stacker crane chooses a shuttle to awaken

operation has virtually no influence on throughput.

Figure 7: (from top) Influence of operations stacker crane

chooses an available position on the transfer buffer, shuttle

chooses an available pallet on the transfer buffer and

stacker crane chooses a shuttle to awake on throughput.

4.4 Evaluation: Storage

We evaluate by simulation the throughput of the

warehouse for each of the eleven combinations of

strategies for storing (see Fig. 8).

Combination {13, 18} achieves the highest

throughput when the stacker cranes are the bottleneck

of the system, i.e. for eight or more shuttles.

100

110

120

130

0 2 4 6 8 10121416

Retrievals per Hour

Number of Shuttles

{1,5,10}

{1,5,9}

{1,5,11}

100

110

120

130

0 2 4 6 8 10121416

Retrievals per Hour

Number of Shuttles

{1,5,9}

{2,5,9}

{3,5,9}

{4,5,9}

100

110

120

130

0 2 4 6 8 10121416

Retrievals per Hour

Number of Shuttles

{1,5,9}

{1,6,9}

{1,7,9}

{1,8,9}

Concept Development and Evaluation of Order Assignment Strategies in a Highly-dynamic, Hybrid Pallet Storage and Retrieval System

365

Combination {13, 15}, provides the highest

throughput when the shuttles are the bottleneck of the

system, i.e. for six or fewer shuttles. Note that random

combination {12, 15} achieves a low throughput.

Figure 8: Throughput achieved by eleven combinations of

order assignment strategies for storing.

In addition, we investigate the utilization ratio and

time components of the shuttles (see Fig. 9).

When the shuttles are the bottleneck,

combinations {13, 18} and {13, 15} achieve a higher

utilization ratio than random combination {12, 15}.

In fact, in combinations {13, 18} and {13, 15}, the

shuttles spend less time being blocked due to mutual

route interferences than in random combination {12,

15}. When the stacker cranes are the bottleneck,

combination {13, 18} ensures the highest utilization

ratio.

Next, we investigate which storage operation has

the biggest impact on throughput. For this purpose,

we filter the throughput results of the simulation,

considering one operation at a time (see Fig 10).

On one hand, the shuttle chooses a free position

on the transfer buffer operation has the greatest

impact on throughput when the shuttles are the

bottleneck of the system.

On the other hand, the stacker crane chooses an

available pallet on the transfer buffer operation has

the greatest influence on throughput when the stacker

cranes are the bottleneck of the system.

4.5 Evaluation: Double Cycle

We define a double cycle as the combination of a

storage and a retrieval in sequence. Specifically, in a

double cycle, each shuttle takes a pallet to store from

the I location, brings it to the transfer buffer, takes a

retrieval pallet on the transfer buffer, brings it to the

O location – while respecting the correct sequence –

and returns to the I location to restart the double cycle.

In the meantime, each stacker crane alternates a

storage cycle with a retrieval cycle.

Figure 9: Time components of significant combinations for

storage.

We evaluate by simulation the throughput of the

warehouse for each of the 48 combinations of the

strategies for storing and retrieving in sequence, as

applied to double cycles (see Fig. 11). We consider

only 48 combinations, because we always choose the

pallet with the smallest sequence number for the

retrieval operation shuttle chooses an available pallet

on the transfer buffer and we always choose a random

idle shuttle for the retrieval operation stacker crane

chooses a shuttle to awake.

The results show that all combinations achieve a

very similar throughput when the shuttles are the

bottleneck. When the stacker cranes are the

bottleneck, the combination {3, 5, 11, 13, 17}

achieves the highest throughput. Combination {3, 5,

11, 13, 17} means that each shuttle takes a pallet to

store from the I location and brings it to the nearest

available position on the transfer buffer. Next, the

100

110

120

0 2 4 6 8 10 12 14 16

Hou

of Shuttles

{12,15}

{13,15 }

{13,18}

100

2 4 6 8 10 12 14 16

Amount of time [%]

Number of huttlesS

{12,15}

100

2 4 6 8 10 12 14 16

Amount of time [%]

Number of huttlesS

{13,15}

100

2 4 6 8 10 12 14 16

Amount of time [%]

Number of huttlesS

{13,18}

waiting for free place in trans fer buffer

blocked

loading and unloading

ivin

SIMULTECH 2021 - 11th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

366

Figure 10: (from up to down) Influence of operations

shuttle chooses a free position on the transfer buffer and

stacker crane chooses an available pallet on the transfer

buffer on throughput.

shuttle takes the available retrieval pallet with the

smallest sequence number on the transfer buffer and

brings it to the O location. Each stacker crane

alternately takes the storage and the retrieval pallets

whose locations on the transfer buffer ensure the

shortest path for the stacker crane.

Note that random ‘sequenced’ combination {1, 5, 11,

12, 15} also achieves a high throughput compared to

the majority of the other combinations.

We also

investigate the utilization ratio and time components

of the shuttles (see Fig. 12). Combination {3, 5, 11,

13, 17} achieves the highest utilization ratio.

Specifically, in combination {3, 5, 11, 13, 17}, the

shuttles wait for a shorter time for an available

transfer buffer pallet than in random ‘sequenced’

combination {1, 5, 11, 12, 15}.

0246810121416

Double cycles per Hour

Number of Shuttles

{1, 5, 11, 12, 15}

{3, 5, 11, 13, 17}

Figure 11: Throughput achieved by 48 combinations of

order assignment strategies for double cycles.

100

2 4 6 8 10121416

Amount of time [%]

Number of huttlesS

{3,5,11,13,17}

100

2 4 6 8 10121416

Amount of time [%]

Number of huttlesS

{1,5,11,12,15}

waiting for free place in transfer buffer

waiting for available transfer buffer pallet

waiting for previous sequence number

blocked

loading and unloading

driving

Figure 12: Time components of significant combinations

for double cycles.

5 CONCLUSION AND OUTLOOK

In this paper, we develop and evaluate different order

assignment strategies for a DHPW by discrete event

simulation, considering retrieval in sequence, storage

and double cycles. As evaluation criteria, we analyze

warehouse throughput and the utilization ratio and

time components of the shuttles. We also investigate

which retrieval in sequence and storage operations

have the greatest impact on warehouse performance.

Finally, we suggest an optimum combination of order

assignment strategies, which guarantees the highest

performance in the different operational conditions.

Our main results are as follows:

 To achieve the best performance for retrieval

in sequence, each shuttle takes the pallet of the

transfer buffer with the smallest sequence

number. Each stacker crane brings its pallet to

the transfer buffer location that ensures the

shortest path for the stacker crane. Stacker

cranes awaken those idle shuttles first that

have executed the fewest orders since the

commencement of retrieval.

Concept Development and Evaluation of Order Assignment Strategies in a Highly-dynamic, Hybrid Pallet Storage and Retrieval System

367

 To achieve the best storage performance, each

shuttle brings its pallet to the available transfer

buffer location nearest to the I location. For

stacker cranes, the following two cases are

possible: when the shuttles are the bottleneck,

each stacker crane takes a random pallet from

the transfer buffer; when the stacker cranes are

the bottleneck, each stacker crane takes the

pallet from the transfer buffer that has been

available for the longest time.

 To achieve the best double cycle performance,

each shuttle brings storage pallets to the

available position of the transfer buffer that is

nearest to the I location. Each shuttle takes

available retrieval pallets with the smallest

sequence number on the transfer buffer. Each

stacker crane alternately takes storage and

retrieval pallets whose locations on the

transfer buffer ensure the shortest path for the

stacker crane.

For future research, we suggest investigating the

influence of layout on performance. Examples of

characteristic layout parameters that can be varied are

the dimensions of the shuttle base tier, the

arrangement of the retrieval and storage positions in

the transfer buffer, and the zoning strategy used.

ACKNOWLEDGEMENTS

We would like to thank Joerg Eder and Thomas

Klopfenstein from the firm Gebhardt Fördertechnik

GmbH for the fruitful collaboration.

REFERENCES

Arantes, J. C., Kompella, S., 1993, Travel-time models for

AS/RS with multiple I/O stations, 2nd Industrial

Engineering Research Conference Proceedings, IIE,

Norcross, GA (USA), 405-409.

Eder, J., Klopfenstein, T., & Gebhardt, M., 2019, Patent:

Lagersystem zur Speicherung und Abgabe von

Ladungsträgern, DE102019211804, German Patent and

Trade Mark Office (DPMA).

Gagliardi, J. P., Renaud, J., Ruitz, A., 2014, A simulation

modeling framework for multiple-aisle automated

storage and retrieval systems, Journal of Intelligent

Manufacturing, 25, 193-207.

Grunow, M., Günther, H.-O., Lehmann, M., 2006,

Strategies for dispatching AGVs at automated seaport

container terminals, Strategies for dispatching AGVs at

automated seaport container terminals, OR Spectrum,

28, 4, 587–610.

Habl, A., Rautenberg, A., Fottner, J., 2020, Dynamic

control of multiple vehicles moving along the same rail

in automated vehicle storage and retrieval systems, 19th

International Conference on Modeling and Applied

Simulation, Athen (Greece).

Lantschner, D., 2015, Eine algorithmenbasierte Spielzeit

automatischer Lagersysteme mit mehreren

Übergabepunkten, Dissertation, Technische

Universität München.

Lienert, T., Staab, T., Ludwig, C., Fottner, J., 2018,

Simulation-based Performance Analysis in Robotic

Mobile Fulfilment Systems – Analyzing the

Throughput of Different Layout Configurations, 8

International Conference on Simulation an Modeling

Methodologies, Technologies and Applications

(SIMULTECH 2018), 383-390.

Siciliano, G., Lienert, T., Fottner, J., 2020, Design,

Simulation and Performance of a Highly-Dynamic,

Hybrid Pallet Storage and Retrieval System, 19th

International Conference on Modeling and Applied

Simulation, Athen (Greece).

SIMULTECH 2021 - 11th International Conference on Simulation and Modeling Methodologies, Technologies and Applications

368