presentation at graduate student day, 2012 inter-block yard crane scheduling
DESCRIPTION
Presentation at Graduate Student Day, 2012 Inter-Block Yard Crane Scheduling a t a Marine Container Terminal by Omor Sharif, Nathan Huynh, Mashrur Chowdhury , Jose Vidal Paper submitted to International Journal of Transportation Science and Technology. - PowerPoint PPT PresentationTRANSCRIPT
1
PRESENTATION AT GRADUATE STUDENT DAY, 2012
INTER-BLOCK YARD CRANE SCHEDULING
AT A MARINE CONTAINER TERMINAL
byOmor Sharif, Nathan Huynh, Mashrur Chowdhury, Jose
Vidal
Paper submitted to International Journal of Transportation Science and
Technology
2
A VERY BRIEF REVIEW OF A CONTAINER TERMINAL!
• An interface between ocean and land transport• Containers are loaded/unloaded to/from a ship• Operation involves a large number of decisions• Capacity Constraints, Environmental Concerns etc
3
RESEARCH PROBLEM AND MOTIVATION
• Container Yard and Yard Crane (YC)• A container yard is made up of several blocks of
containers• Workload varies among blocks during operational
hours• Efficiency of YCs impacts truck waiting/ship waiting
time• Assign/relocate YCs among blocks to finish most work • Known as Interblock YC Scheduling Problem
4
MODEL ASSUMPTIONS
• Total operational hours is divided in several planning periods
• A workload forecast for blocks is known at start (time-units)
• At most two cranes work at a block at the same time• At most one transfer per planning period• Other operations will not introduce delay• 10 minutes transfer time for each longitudinal block
traveled
5
INITIAL ASSIGNMENT OF YARD CRANES
High to low work volume• Sort blocks in decreasing order of initial work volume (IWb)• Assign NCb
max cranes to each block starting from topmost item
Crane at each block • If nc = nb, assign one crane per block.• If nc < nb, sort blocks in decreasing IWb , assign cranes.• If nc > nb, first assign one crane per block, then sort in
decreasing IWb . Reduce transfers • Assign cranes to reduce future transfers.• Based on pass/fail of a series of iequalites.• A four step approach
6
THREE PRINCIPAL PARAMETERS
Extra capacity of a crane
• E (c) = Tc × NCbinitial − IWb
Amount of help needed by a block
• H (b) = IWb − Tc × NCbcurrent
Transfer time
• TTcod
7
PREFERENCE FUNCTIONS FOR BLOCK AND CRANE AGENTS
Minimum transfer time • Transfer time of a crane from origin to destination block
Positive difference • Extra Capacity of a crane – Help needed by a block –
Transfer time
Absolute difference • abs (Extra Capacity of a crane – Help needed by a block
– Transfer time)
Absolute difference squared distance• abs (Extra Capacity of a crane – Help needed by a block
– Transfer time2)
8
AN ALGORITHM TO ASSIGN CRANES TO BLOCKS
Crane proposing version
Block proposing version • …
1. Each crane j proposes to first block i from its preference list2. Each block i receiving more than qi proposals, ‘holds’ the most preferred qi cranes and rejects all others.n. Each crane j rejected at step n − 1 removes the block i rejecting the crane from its preference list. Then the rejected crane j makes a new proposal to its next most preferred block i who hasn’t yet rejected it. Go to step n − 1.
9
IMPLEMENTATION
• Multi-agent simulation GUI• Implemented in NETLOGO• Stationary block and Mobile crane agents
10
DESIGN OF EXPERIMENTS
Parameter Value UnitNumber of blocks
10, 20, 30 Nos
Number of cranes
1 or 1.5 × Number of blocks Nos
Length of planning period
240 Minutes
Work volume (1) Moderate (2) Heavy (3) Above capacity
Minutes
Initial assignment
(1) Crane at each block (2) High to low work volume (3) Reduce transfers
-
Preference strategy
(1) Minimum transfer time (2) Absolute difference (3) Positive difference (4) Absolute inverse squared distance
-
AlgorithmVersion
(1) Blocks proposing (2) Cranes proposing
-
11
RESULTS
Percentage incomplete work volume: Case I - average number of cranes per block = 1.0
Number of Blocks 10
20
30
Number of Cranes 10
20
30
Work Volume M H AC M H AC M H AC
Minimum Transfer Time 0 0.13 11.34
0 0.10 11.14
0 0.12 11.28
Absolute difference 0 0.16 12.16
0 0.13 12.09
0 0.11 12.16
Positive difference 0 0 13.55
0 0.05 13.08
0 0.04 12.86
Absolute difference squared distance
0 0.13 11.80
0 0.08 11.69
0 0.10 11.81
Mathematical Program 0 0 10.57
0 0 10.57
0 0 10.60
12
RESULTS (CONTD..)
Percentage incomplete work volume: Case II - average number of cranes per block = 1.5
Number of Blocks 10 20 30
Number of Cranes 15 30 45
Work Volume M H AC M H AC M H AC
Minimum Transfer Time 0 0.02 9.79 0 0.20 10.00
0 0.48 10.04
Absolute difference 0 0.07 10.19
0 0.18 10.45
0 0.43 10.48
Positive difference 0 0.14 10.40
0 0.53 10.73
0 1.00 10.71
Absolute difference squared distance
0 0.01 10.06
0 0 10.31
0 0.01 10.38
Mathematical Program 0 0 9.69 0 0 9.77 0 0 9.79
13
CONCLUDING REMARKS
• In ‘medium’ condition all work can be finished • In ‘heavy’ condition the percentage incomplete is 1% or
less• In ‘above capacity’ condition the percentage remaining is
within 3% of the lower bound.• Scalable; a test case with 30 blocks can be solved within 3
seconds
• Consider relocating cranes multiple time• Include forecasts for multiple planning periods • Solve integrated problems involving other related
decisions
Summary of Findings
Future Work
14
Thank You
Questions ?
15
A SAMPLE EXAMPLE