Download presentation
Presentation is loading. Please wait.
Published byValentine Barrett Modified over 6 years ago
1
Constraint-based heuristics for amphibious embarkation planning
UNCLASSIFIED – Approved For Public Release Constraint-based heuristics for amphibious embarkation planning Paul A. Chircop & Timothy J. Surendonk D1. Mathematical modelling for defence applications MODSIM 2015
2
UNCLASSIFIED – Approved For Public Release
Introduction Amphibious operations: Transportation of army units (e.g. vehicles) from the sea to the land. Optimally embark units onto a ship’s deck: Amphibious embarkation problem: Represented as a packing problem. Units modelled by rectangular items with width, length and mass. A CH-47D Chinook conducts load lifting trials with HMAS Canberra in Jervis Bay. images.navy.gov.au
3
UNCLASSIFIED – Approved For Public Release
Packing Problem Input: A rectangular deck (with/without obstacles). A list of rectangular items (dimensions, mass). Objective: Maximize the total area of packed items. Constraints: Items cannot overlap each other or the deck boundary. No stacking and no rotation of items. Items cannot overlap obstacles (if present). Priority ordering (optional). Mass balance + threshold (optional).
4
UNCLASSIFIED – Approved For Public Release
Illustration
5
Modelling / Solution Approach
UNCLASSIFIED – Approved For Public Release Modelling / Solution Approach Constraint Programming (CP): Expressive modelling power. Domain filtering, logical constraint propagation, branch-and-bound search. Constraint Optimization Packing Tool: Java-based software package. ILOG CP Solver (IBM) as CP engine.
6
UNCLASSIFIED – Approved For Public Release
Why COmPacT? Flexibility: Choice of CP engine (commercial/open source). Customized heuristics. User Interaction for manual control: Automated and manual packing. Time-out functionality for iterative packing Run from a warmstart.
7
GUI / Test Rig CP search engine information.
Builds a tree of feasible packing solutions. Allows user to click-drag-and-lock items (manual mode). Allows for buffers (borders) on individual items. Sets movement constraints on individual items. Implements sequential packing heuristics.
8
Binary Decision Variables
UNCLASSIFIED – Approved For Public Release Binary Decision Variables
9
UNCLASSIFIED – Approved For Public Release
Relative positions (x1, y1) w1 (x3, y3) w3 (x2, y2) l1 w2 l3 l2
10
UNCLASSIFIED – Approved For Public Release
Constraint Model Objective Deck Boundary Bounds Packing Feasibility Symmetry Breaking Obstacles Variables Mass Balance
11
Example – Solver Only (I)
UNCLASSIFIED – Approved For Public Release Example – Solver Only (I) Obstacle Obstacle 15.8 m Obstacle 93.5 m
12
Example – Solver Only (II)
UNCLASSIFIED – Approved For Public Release Example – Solver Only (II) 3,003 variables. 17,774 constraints. Optimal solution found in 0.68 seconds.
13
UNCLASSIFIED – Approved For Public Release
Heuristics Aggregation: Joining (adjacent) like items together before throwing to the solver. Limited by priority ordering. Sequential Packing: Iterates through a series of warmstart solutions. Local movement (constrain position). Pack by priority or by size.
14
Aggregation Illustration (I)
UNCLASSIFIED – Approved For Public Release Aggregation Illustration (I) Combine identical items adjacent in priority list. Carried out along the length axis or width axis.
15
Aggregation Illustration (II)
UNCLASSIFIED – Approved For Public Release Aggregation Illustration (II)
16
Example – Aggregation + Solver
UNCLASSIFIED – Approved For Public Release Example – Aggregation + Solver 21.0 m Obstacles 94.5 m
17
Solution – Aggregation x 3
UNCLASSIFIED – Approved For Public Release Solution – Aggregation x 3
18
Aggregation Heuristic Performance
UNCLASSIFIED – Approved For Public Release Aggregation Heuristic Performance
19
Sequential Packing Illustration (I)
UNCLASSIFIED – Approved For Public Release Sequential Packing Illustration (I)
20
Sequential Packing Illustration (II)
UNCLASSIFIED – Approved For Public Release Sequential Packing Illustration (II)
21
Sequential Packing Illustration (III)
UNCLASSIFIED – Approved For Public Release Sequential Packing Illustration (III)
22
Sequential Packing Illustration (IV)
UNCLASSIFIED – Approved For Public Release Sequential Packing Illustration (IV)
23
Sequential Packing Illustration (V)
UNCLASSIFIED – Approved For Public Release Sequential Packing Illustration (V)
24
Example – Sequential Packing by Size
UNCLASSIFIED – Approved For Public Release Example – Sequential Packing by Size 15.8 m Obstacles 93.5 m
25
UNCLASSIFIED – Approved For Public Release
Solution
26
UNCLASSIFIED – Approved For Public Release
Solution
27
UNCLASSIFIED – Approved For Public Release
Solution
28
UNCLASSIFIED – Approved For Public Release
Solution
29
UNCLASSIFIED – Approved For Public Release
Solution
30
UNCLASSIFIED – Approved For Public Release
Solution
31
UNCLASSIFIED – Approved For Public Release
Solution
32
UNCLASSIFIED – Approved For Public Release
Solution
33
Example – Sequential Packing by Priority
UNCLASSIFIED – Approved For Public Release Example – Sequential Packing by Priority Priority Ordering 15.8 m 93.5 m
34
UNCLASSIFIED – Approved For Public Release
Solution
35
UNCLASSIFIED – Approved For Public Release
Solution
36
UNCLASSIFIED – Approved For Public Release
Solution
37
UNCLASSIFIED – Approved For Public Release
Solution
38
UNCLASSIFIED – Approved For Public Release
Solution
39
UNCLASSIFIED – Approved For Public Release
Solution
40
UNCLASSIFIED – Approved For Public Release
Solution
41
UNCLASSIFIED – Approved For Public Release
Solution
42
UNCLASSIFIED – Approved For Public Release
Summary / Future Work Constraint programming approach: CP solver + customized heuristics. Improvement on extant in-house techniques: Simulated annealing + compaction heuristics. Future work: Benchmarking performance: Broader spectrum of test problems.
43
Questions UNCLASSIFIED – Approved For Public Release
Paul Chircop & Timothy Surendonk Joint and Operations Analysis Division Defence Science & Technology Group
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.