About AS/RS (1/7)
About AS/RS (2/7)

Research areas
 Storage
locations assignment
 Storage
and retrieval policy
 Physical
design
About AS/RS (3/7)

Storage locations assignment
 Randomized


Uniformly
Closest open location
 Class-based

Turnover ratio
 Dedicated
About AS/RS (4/7)

Storage and retrieval policy
 Single

command (SC)
First-come-first-serve
 Dual
command (DC)
First-come-first-serve
 Dynamic TSP problem

Assignment problem (static & dynamic)
 Nearest neighbor

About AS/RS (5/7)

Physical design

Configurations of rack


Velocity



Height & Length
Acceleration & Deceleration
Dwell point
Bozer & White(1984)

E(SC)=[ 1  1 b 2] T
3

E(DC)=[ 4  1 b 2  1 b 3 ] T
3 2
30
About AS/RS (6/7)

Performance measure
 Travel
time
 Throughout
About AS/RS (7/7)

Load types
 Mini-load
 Unit-load
Effects of relocation to AS/RS
storage location policy with
production quantity variation
Introduction
 Experiment design
 Simulation
 Result
 Conclusion

Introduction
The performance of will not be good if
production quantity variation and current
storage policy is kept
 Find the tolerance limits and appropriate
strategies for storage policy when
production plan change
 Random,2 and 3 class-based will be
examined

Experiment design (1/3)

There are 20 items in the production plan
Production quantities vary between 1080%
 Items are assumed as end products


For 2 class-based storage
 Top
four items: first class
Experiment design (2/3)

For 3 class-based storage
 Top
four items: first class
 Next six items: second class

For random storage
 1500,

5000 and 95000 unit storage time
Relocation process will be done whenever the
crane idles
Experiment design (3/3)
Simulation (1/4)

GPSS

Simulated for 900,000 unit time

Warn up 300,000

300,000~600,000 unit time: original

Repeat 10 times
Simulation (2/4)
Simulation (3/4)
Simulation (4/4)
Result (1/3)

Random storage
Result (2/3)

2 class-based
Result (3/3)

3 class-based
Result(4/4)
Conclusion (1/2)
Conclusion (2/2)

Class-based storage policies perform better but
have a lower tolerance

Class-based would be better if performance is
the most critical issue

If performance is not critical and frequent
production plan changes are expected, random
storage would be better