Network LP Models
Transportation Problem Transshipment Problem Assignment Problem Network Flow Problem Min Cost Flow model Spanning Trees Shortest Path
1
Transportation Problem Example
High Plains Electronics manufactures MP3 players at three separate overseas plants. It ships finished products to three US distribution centers. Shipping costs are shown below, as are the production capacities of each plant and the demand from each distribution center (units per week):
Distribution Center LA Boston Denver 4 10 6 8 16 6 14 18 10 200 300 200
Supply = Demand
Capacity 100 300 300 700
2
Plant Singapore Spain Argentina Demand
Transportation Problem Defined
Minimize total shipping cost of products from source to destinations Supply equals demand Product measured in discrete units Integrality assured Linear costs assumed
General LP Form
min cij xij s.t.
Optimize some objective
Subject to
ij
x
j 1 m
si dj
Supply constraints
x
i 1
ij
Demand constraints
xij 0
Non-negativity constraints
4
General LP Form Schematic
Objective function
Supply constraints
Demand constraints
Prohibited Routes
What if some routes are infeasible or prohibited? Create allocation cost that is so large that it will be immediately forced from the basis Big M Example
Supply Demand
What if supply does not equal demand? Feasible solutions exist only if total supply equals total demand
s d
i 1 i j 1
Create dummy supplies or demands
7
Catering Example
The University is planning six catered events for homecoming day. Caterers Bon Apetit, Custom, and University can each handle two events, the others only one. MSU wants to minimize costs. (Costs shown in $1000s)
Alumni Brunch 12.60 14.50 13.00 11.50 10.80 13.50 12.50 Parent's Brunch 10.30 13.00 14.00 12.60 11.90 13.50 14.30 Booster Lunch 14.00 16.50 17.60 13.00 12.90 15.50 16.00 Postgame Letterman's Sponsor's Party Dinner Dinner 19.50 25.00 30.00 17.00 22.50 32.00 21.50 23.00 35.00 18.70 26.20 33.50 17.50 21.90 28.50 22.30 24.50 36.00 22.00 26.70 34.00
Al's Bon Apetit Custom Divine Epicurean Fouches University
Production Scheduling Example
Binford Access manufacturers telephone switches for private telephone carriers using inventory, overtime, and subcontracting to absorb demand fluctuations. Regular production cost is $200 per unit, overtime production is $250, subcontracting costs $270, and inventory carrying costs are $20 per unit per quarter. Binford currently has 300 units in inventory.
Regular Overtime Subcontract Capacity Capacity Capacity 900 100 300 1,000 150 300 1,200 200 300 1,200 200 300
9
Quarter 1 2 3 4
Demand 900 1,200 1,600 1,900
Transshipment Problem
Extension of transportation problem Allocate and route units
From supply centers Through transshipment points To receiving centers
Can be represented as a linear program
(or as a network flow problem)
Example
10
Transshipment Example
Upon investigation, High Plains Electronics discovers that it can ship its products to intermediate warehouses in either New Orleans or Seattle. Either location can handle up to 500 units per week. HPE wants to determine if shipping costs can be reduced by using these transshipment points.
COSTS Plant Singapore Spain Argentina Miami Seattle Demand Distribution Center Transship LA Boston Denver Miami Seattle Capacity 4 10 6 15 3 100 8 16 6 3 12 300 14 18 10 2 16 300 8 4 9 12 500 2 7 4 12 500 200 300 200 500 500 1700
11
Assignment Problem
Assign discrete resources (people, machines, vehicles, ) to tasks (jobs, routes, ) Number of resources and tasks are equal Each resource is assigned to one task Each task is assigned to one resource Integrality assured Linear costs assumed Highly degenerate (many alternative solutions)
12
Assignment Problem
Eldorado Distributors has four sales territories, each of which must be assigned a sales representative. From past experience, the firms sales manager knows which reps do best in which territories and has scored each of four reps in the four territories (low scores mean better performance). How should reps be assigned?
Rep Wilson Benson Fredrick Hodson Arvada 44 60 36 52 Territory Boulder Golden 80 52 56 40 60 48 76 36 Denver 60 72 48 40
13
Assignment Solutions
Could try complete enumeration
n! assignments possible n =10 means 10!=3.6 million possible
assignments
Linear Programming solution Dedicated Algorithm: the Hungarian method
14
General LP Form
min cij xij
i 1 j 1
Optimize some objective
s.t.
Subject to
Supply constraints
x
j 1 m i 1
ij
xij 1
0 xij 1
Demand constraints
If i is not assigned to j
If i is assigned to j
15
Assignment Problem Extensions
Supply unequal to demand
Use dummy sources or sinks
Prohibited assignments
Use large allocation cost Big M
Maximization Problems
Create opportunity loss matrix by finding difference between largest element in each column and other column elements Solve as minimization problem Example
16
Crew Assignment Problem
Western Slope Airlines, a small commuter airline in Colorado, wants to assign its six flight attendants to six monthly flight schedules so that their nights away from home are minimized.
Attendant Harry Sue Al Mary Fred Martha A 7 4 9 11 5 10 B 4 5 9 6 8 12 Flight Schedule C D 6 10 5 12 11 7 8 5 6 10 11 9 E 5 7 10 9 7 9 F 8 6 8 10 6 10
Next suppose the company has a policy that no attendant should spend more than 10 days away from home. Modifications?
17
Maximum Flow Problem
Western Pipeline wishes to determine the maximum flow of gas that can be delivered to Denver from its Texas wells through its network of gas pipelines. 3 6 Texas (source) 2 3 2 5
2 6
1 2
5 Denver (sink)
18
Max Z = Xdt subject to: Xta 6 Xab 3 capacity constraints Xac 2 Xta Xab Xac = 0 Xab Xbd Xbc = 0 conservation constraints
a
Texas (source) t [6] Xta
Xab [2]
[3] [2]
b [5]
Xac [3] e
c [6] [1]
[2] [2]
Xdt
f [5] d Denver (sink)
19
Min Spanning Tree
A cable company is laying new fiber-optic cable in a one of its client cities. The number of feet of cable (in 10,000s) required to connect various switching stations is shown below. 25 12 16 14 19
35
15 14
22
20
Min Spanning Tree
A new Colorado state park is being planned with roads required to proposed park facilities. Minimize total road mileage (and thus environmental impact).
4 2 2 4 2 5 5 3
3
2
3 7
21
Shortest Path Problem
Mountain Construction maintains a central equipment warehouse and must equip each of its 6 construction sites with supplies each day. The travel times between its warehouses and sites are given in the following network. 25
30
Warehouse
12 6 8
15
11
18
16 5
22
Minimum Cost Flow Problem
Generalized network problem
Transportation, assignment, max flow, min spanning, shortest path,
Formulated as linear program
Wide utility, flexible, general
Solved using Network Simplex Method General LP formulation
23
Shortest Path Problem
Undirected and connected network Distance (cost) same in both directions between two nodes Find the shortest (distance, cost) route from one to a number of other points given a network with know arc penalties (distance, cost) Solve using shortest path algorithm
24
Shortest Path Example
Mountain Construction maintains a central equipment warehouse and must equip each of its 6 construction sites with supplies each day. The travel times between its warehouses and sites are given in the following network. 25
30
Warehouse
12 6 8
15
11
18
16 5
25
Min Spanning Tree Problem
Undirected and connected network Distance (cost) same in both directions between two nodes Minimize (cost, distance) of connecting all nodes in the network Connection costs are independent Solve using Minimum Spanning Tree Algorithm
26
Min Spanning Tree Example 1
A new Colorado state park is being planned with roads required to proposed park facilities. Minimize total road mileage (and thus environmental impact).
4 2 2 4 2 5 5 3
3
2
3 7
27
Min Spanning Tree Example 2
A cable company is laying new fiber-optic cable in a one of its client cities. The number of feet of cable (in 10,000s) required to connect various switching stations is shown below.
25 12 16 14 19 8
35
15 14
22
28
Maximum Flow Problem
Directed and connected network Find the maximum flow between a source and sink (e.g., supply and demand) through a network of branches with known flow capacities Linear costs assumed Augmenting Path Algorithm
29
Maximum Flow Example 1
Western Pipeline wishes to determine the maximum flow of gas that can be delivered to Denver from its Texas wells through its network of gas pipelines. 3 6 Texas (source) 2 3 2 5
2 6
1 2
5 Denver (sink)
30
Maximum Flow Example 2
Midwest Grain ships railcars of grain from its St. Louis barge terminal to its ship terminal in LA. During the fall harvest, shipping capacity is tight. How much can Midwest ship? (Amounts shown in 10s of railcars daily.)
Salt Lake 8
0 6 0 3 0 4 6 5 6
Des Moines
4 0 0 0
St. Louis
2 Denver
LA
Phoenix
2 6
Dallas
31
Transportation Problem
Shipping products from production plants to consumer centers Minimize shipping cost Deal with unbalance between production and demand when it exists
32
Transportation Problem
High Plains Electronics manufactures MP3 players at three separate overseas plants. It ships finished products to three US distribution centers. Shipping costs are shown below, as are the production capacities of each plant and the demand from each distribution center (units per week):
COSTS Plant Singapore Spain Argentina Demand LA 4 8 14 200
Distribution Center Boston 10 16 18 300 Denver 6 6 10 200 Capacity 100 300 300 700
33
Minimum Cost Flow Problem
Generalized network problem
Transportation, assignment, max flow, min spanning, shortest path,
Formulated as linear program
Wide utility, flexible, general
Solved using Network Simplex Method
We will not review in this class
General LP formulation
34
Minimum Cost Flow Problem
[-30] D [9] [50] A [10] 2 4 3 B [10]
All of the following are particular cases of the MinimumCostFlow problem: Transshipment (and consequently Transport, Assignment) Shortest path Maximum Flow Prove it
E 1 [80] [-60]
3
C
Network Flow Optimization for Load Balancing Used (or suggested) in core networks, as superior to shortest-path optimization for network traffic engineering
The Single Commodity problem:
Similar to MinimumCostFlow problem with only one source and one sink. Same constraints but different metric: Minimize Z = Max(xij / cij) Easily extensible to cases when we have many sources and one sink or many sinks and one source
The Multi-Commodity problem (this is the general case)
Conservation constraints apply to each flow separately, therefore we have a decision variable per link per flow Capacity constraints and economic function apply to the total flow over each link
Flow optimization Single Commodity example
Given many paths between points 1 and 2, how to distribute the traffic so that to minimize the maximal utilization on all links. Minimize F=max(x13/c13, x33/c32, x14/c14, x42/c42) subject to x13 + x14= h12 x32 = x13 x42 = x14 x13 c13 x13 c32 x14 c14 x14 c42
N.B.The general case is the multicommodity problem
Single commodity example