Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust...
Transcript of Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust...
![Page 2: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/2.jpg)
10-Jun-14
Warehouse Operations
Flow Path Schematic (FFN)
Sharp et al. 1991 Receiving
Cross Docking
Storage
Case PickingFull Pallet
PickingItem Picking
Sortation
Packaging
Shipping
![Page 3: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/3.jpg)
Design framework for storage systems
Unit loads
• Single and dual command
Direct access
• Single-deep rack and single-load high floor stacks
Comprehensive
• Rich set of facility configurations and storage policies
Robust: efficiency and risk (stochastic)
Component of design methodology for
warehousing systems
Research Goal
3
![Page 4: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/4.jpg)
Sainsbury’s Grocery
Distribution Center
4
![Page 5: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/5.jpg)
Empty Single-Deep Pallet Rack
with Four Levels
8
![Page 6: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/6.jpg)
ASRS Pallet Unit Load High-Rise
Storage
9
![Page 7: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/7.jpg)
Wine Barrels in a Cantilever
Rack
11
![Page 8: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/8.jpg)
Definitions
Storage Policy
Set of rules that determine where to store arriving
SKUs in a warehousing system
Unit Load
A collection of materials that can be transported,
stored, and controlled (managed) as a single unit
• Examples
• Vast majority of discrete goods
13
![Page 9: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/9.jpg)
Warehousing Storage Objectives:
Back to Basics
Minimize the cost of expected travel time
for given input-output operations
Minimize MH equipment and personnel
Variable (marginal) costs
Minimize the cost of required storage
space for given stored inventory
Minimize capital investment
Fixed costs
19
![Page 10: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/10.jpg)
Very few configuration decisions
Most compared with complete
enumeration (user driven comparison)
Technology, type of material handling equipment,
aisles have ladder structure or not, aisle
orientation, location of the input/output points,
storage policy
Many combinations
• Need computational support to evaluate designs
quickly
Main Design Observation
20
![Page 11: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/11.jpg)
Main design decision variables
Number of aisles, number of levels (rack height),
number of columns (aisle length)
Secondary decisions
Load locations, number of personnel and MH
equipment
Decomposition
Pareto optimal comparison of efficiency
versus risk
Design Decision Variables
21
![Page 12: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/12.jpg)
Pareto Risk versus Efficiency
Comparison
22
![Page 13: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/13.jpg)
Long research history and still active area
Heskett (COI) 1963,…to Ang et al. 2012
Most recent reviews Gu et al. 2007 + 2010
Contemporary blogs
Industry norms FEM,VDI
Results and algorithms are strongly
assumption driven
Integration and unified assumptions are the
challenge
Prior Research on Storage Systems
Design and Storage Policies
23
![Page 14: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/14.jpg)
Storage Policies Classification
10-Jun-14
Storage Policies
Unit Load
Non Unit Load
Non Direct Access
Direct Access
No Information
Product Based
Load Based
Factoring
Non-Factoring
Demand RankedInventory RankedFrequency of Access Ranked#-Class
Demand RankedInventory RankedFrequency of Access Ranked#-Class
#-Zone
#-Zone
Perfectly Balanced
Assignment Formulation
Assignment Formulation
Duration of Stay
Duration of Stay
RandomClosest Open Location
RandomClosest Open Location
Non Perfectly Balanced
Factoring
Non-Factoring
Vector Assignment Formulation
![Page 15: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/15.jpg)
Stationary or not warehousing operations
Repetitive, seasonal, build-up (single use),
random events
Storage Policy Classification:
Additional Considerations
25
![Page 16: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/16.jpg)
One user-specified design
E.g. ASRS, random storage
Master problem: determine NA, NL, NC
Sub problem:
Split by scenario
Compute assignment costs (parameters)
Optimize scenario variables and (objective) cost
Return EV and SD of scenario costs
Decomposition Algorithm
31
![Page 17: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/17.jpg)
General load-based assignment (VAP)
Most general, very large MIPs, most
computationally demanding
Ultimate verification algorithm
Technology comparison with random
storage
Using FEM travel time norms
Different risk measures
Two Examples
32
![Page 18: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/18.jpg)
Occupancy Gantt Chart:
Rack Based Direct Access
Time Periods
J1
J2
J3
J4
J5 I
6
I1
I9
I3
I7
I4
I8
I2
I10
I5
33
![Page 19: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/19.jpg)
VAP Conclusions
Very large integer optimization problem
Very tight LP relaxation
Efficient sub problem and problem size
indicate decomposition
Very small gap for Lagrangean relaxation
upper bound
Highly primal and dual degenerate
Acceptable penalty for primal heuristic
44
![Page 20: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/20.jpg)
Automated storage and retrieval system
(ASRS) versus person-controlled narrow
aisle reach truck (NAT)
System and construction, operations, and
maintenance costs
ASRS
Simultaneous travel, aisle-captive crane
NAT
Sequential travel in the aisle, non aisle-captive
Technology Comparison Example
45
![Page 21: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/21.jpg)
Model characteristics
Cubic space constraint (master), volume and area
cost terms (sub) become parameters, quadratic
sub objective (risk = variance), efficiency versus
risk tradeoff weight
Algorithm
Finite ranges for NA, NL, NC
Solved by complete enumeration in master
Technology Comparison Example
46
![Page 22: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/22.jpg)
Technology Comparison Example:
Standard Deviation Risk
47
![Page 23: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/23.jpg)
Technology Comparison Example:
Downside Risk (Semi-Deviation)
48
![Page 24: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/24.jpg)
Unit Load Storage Policy
Conclusions
Unit load systems are very common
Single or dual command cycles
Two main objectives:
• Cost of storage space,
• Cost of total travel time
Three planning problems
• Strategic configuration and sizing
• Tactical storage policy
• Operational storage & retrieval sequence
49
![Page 25: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/25.jpg)
Operator-controlled systems are less
expensive, but have larger cost variability
Above is true regardless of the risk
measure (standard deviation or downside
risk)
On an equal low-risk basis automated
systems are less expensive
Unit Load Storage Policy
Conclusions Continued
50
![Page 26: Robust Storage Systems Design - Georgia Institute of ...mgoetsch/cali/conferences/iserc...Robust Storage Systems Design Marc Goetschalckx marc.goetschalckx@isye.gatech.edu ISERC 2014](https://reader031.fdocuments.us/reader031/viewer/2022011802/5b0b5d867f8b9a45518dded4/html5/thumbnails/26.jpg)
May I answer any questions?
51