Material Handling & Transportation Issues

Shared Transportation Capacity

Large shipments reduce transportation costs but increase inventory costs EOQ trades off these two costs Reduce shipment size without increasing transportation costs? Combine shipments on one vehicle

Inventory Inventory Transport Transport

TL vs LTL

Inventory Inventory

?
Transport Transport?

Shared Loads

Issues
Design Routes that
Minimize the transportation cost Respect the capacity of the vehicle
This may require several routes

Consider inventory holding costs

This may require more frequent visits

Vehicle Routing Problem

Classical VRP

n customers must be served from a single depot utilizing vehicles with capacity Q for delivering goods Each customer requires a quantity qi Q of goods Customer orders cannot be split

Graphical Representation

Depot

Additional Features
Depots
Multiple locations

Routes
Maximum time Link costs

Vehicles
Multiple vehicle types and capacities Release, maximum and down times

Objective Functions
Minimize total traveled distance Minimize total traveled time Minimize number of vehicles Maximize quality of service Multiple objective functions

Customers
Time windows (soft or hard) Accessibility restrictions Priority Pickup and delivery

Vehicle Routing & Scheduling

Model Problem Variety
Pure Pickup or Delivery Problems Mixed pickups and deliveries Pickup-Delivery Problems

Backhauls

Complications
Simplest Model: TSP

Vehicle Routing
Find best vehicle route(s) to serve a set of orders from customers. Best route may be
minimum cost, minimum distance, or minimum travel time.

Orders may be
Delivery from depot to customer. Pickup at customer and return to depot. Pickup at one place and deliver to another place.

General Setup
Assign customer orders to vehicle routes (designing routes).

Assign vehicles to routes.

Assigned vehicle must be compatible with customers and orders on a route.

Assign drivers to vehicles.

Assigned driver must be compatible with vehicle.

Assign tractors to trailers.

Tractors must be compatible with trailers.

Model
Nodes: physical locations
Depot. Customers.
1 4 5 6 3 6 8 8 4 4

Arcs or Links
Transportation links.

depot

Number on each arc represents cost, distance, or travel time.

Pure Pickup or Delivery

Delivery: Load vehicle at depot. Design route to deliver to many customers (destinations). Pickup: Design route to pickup orders from many customers and deliver to depot. Examples:
UPS, FedEx, etc. Manufacturers & carriers. Carpools, school buses, etc.
depot

Pure Pickup or Delivery

depot

Which route is best????

depot

depot

TSP & VRP

TSP: Travelling Salesman Problem
One vehicle can deliver all orders.

VRP: Vehicle Routing Problem

More than one vehicle is required to serve all orders.

TSP

VRP

depot

depot

Mixed Pickup & Delivery

Pickup Delivery

depot

Can pickups and deliveries be made on same trip? Can they be interspersed?

Mixed Pickup & Delivery

Interspersed
Pickup Delivery

Not Interspersed

depot

Separate routes

depot

depot

Interspersed Routes
F For clockwise trip: Load at depot ACDFIJK D E A G H

Pickup Delivery

I
K J

Stop 1: Deliver A
Stop 2: Pickup B Stop 3: Deliver C Stop4: Deliver D etc.

CDFIJK
BCDFIJK BDFIJK BFIJK

C
B

depot

Delivering C requires moving B Delivering D requires moving B

Pickup-Delivery Problems
Pickup at one or more origin and delivery to one or more destinations. Often long haul trips.
C B

Pickup Delivery depot A

A
C

Intersperse Pickups and Deliveries?

Can pickups and deliveries be interspersed?
C
Interspersed B A B C

depot
Pickup Delivery

C Not Interspersed A B A depot C B

Backhauls
If vehicle does not end at depot, should it return empty (deadhead) or find a backhaul?
How far out of the way should it look for a backhaul?
C
Pickup Delivery depot A B A C B

Backhauls
Compare profit from deadheading and carrying backhaul.
Pickup Delivery B A depot A Empty Return D Backhaul C B

Complications
Multiple vehicle types. Multiple vehicle capacities.
Weight, Cubic feet, Floor space, Value.

Many Costs:
Fixed charge. Variable costs per loaded mile & per empty mile. Waiting time; Layover time. Cost per stop (handling). Loading and unloading cost.

Priorities for customers or orders.

More Complications
Time windows for pickup and delivery.
Hard vs. soft

Compatibility
Vehicles and customers. Vehicles and orders. Order types. Drivers and vehicles.

Driver rules (DOT)

Max drive duration = 10 hrs. before 8 hr. break. Max work duration = 15 hrs. before 8 hr break. Max trip duration = 144 hrs.

Simple Models
Homogeneous vehicles. One capacity (weight or volume). Minimize distance. No time windows or one time window per customer.

No compatibility constraints.
No DOT rules.

Simplest Model: TSP

Given a depot and a set of n customers, find a tour (route) starting and ending at the depot, that visits each customer once and is of minimum length.

One vehicle. No capacities. Minimize distance. No time windows. No compatibility constraints. No DOT rules.

Multiple Routes
Capacitated VRP: vehicles have capacities.
Weight, Cubic feet, Floor space, Value.

Deadlines force short routes.

Pickup at end of day. Deliver in early morning.

Time windows
Pickup. Delivery. Hard or Soft.

Multiple Route Solution Strategies

Find feasible routes.
Cluster first, route second.
Cluster orders to satisfy capacities. Create one route per cluster. (TSP for each cluster)

Route first, cluster second.

Create one route (TSP). Break route into pieces satisfying capacities.

Build multiple routes simultaneously.

Improve routes iteratively.

Construction Heuristics
Heuristics that Grow Fragments
Strip Heuristic Nearest neighbor Double-ended nearest neighbor

Heuristics Based on Trees

Minimum spanning tree Christofides heuristic Fast recursive partitioning

Heuristics that Grow Tours

Nearest addition Farthest addition Random addition

The Strip Heuristic

The Strip Heuristic Partition the region into narrow strips Routing in each strip is easy ~ 1-Dimensional Paste the routes together x x x x x x x x x x x

x
x x x

Sweep Algorithm
Draw a ray starting from the depot. Sweep clockwise (or counter-clockwise) and add customers to the route as encountered.

Start a new route when vehicle is full.

Re-optimize each route (solve a TSP for customers in each route).

Sweep Algorithm
Suppose each vehicle capacity = 4 customers

depot

depot

Nearest Neighbor
1. 2. 3. Randomly select a starting node Add to the last node the closest node until no more nodes are available Connect the last node with the first node O(n2) running time

Double-ended Nearest Neighbor

Conceptually the same as nearest neighbor heuristic The fragment is allowed to grow from both ends

Nearest Neighbor
Add nearest customer to end of the route.

depot

depot

depot

Nearest Neighbor
Add nearest customer to end of the route.

depot

depot

depot

Nearest Neighbor
x x x x x x x x

x x

x
x x

Nearest Neighbor Algorithm

Suppose each vehicle capacity = 4 customers

depot

First route

depot

Nearest Neighbor Algorithm

Suppose each vehicle capacity = 4 customers Second route

depot

Third route

depot

Nearest Insertion
1. Select a node and the node nearest to it and build a two-node tour 2. Insert in the tour the closest node y until no more nodes are available
O(n2) running time

y x z Minimize d(x,y) + d(y,z) d(x,z)

Nearest Insertion
Insert customer closest to the route in the best sequence. 1 2 3

depot

depot

depot

Nearest Insertion
Insert customer closest to the route in the best sequence. 4 5 6

depot

depot

depot

Nearest Insertion
x x x x x x x x x x x x x x

Nearest Insertion
x x x x x x x x x x x x x x

Nearest Insertion
x x x x x x x x x x x x x x

Nearest Insertion
x x x x x x x x x x x x x x

Nearest Insertion
x x x x x x x x x x x x x x

Nearest Insertion
x x x x x x x x x x x x x x

Nearest Insertion
x x x x x x x x x x x x x x

Nearest Insertion
x x x x x x x x x x x x x x

Nearest Insertion
x x x x x x x x x x x x x x

Nearest Insertion
x x x x x x x x x x x x x x

Nearest Insertion
x x x x x x x x x x x x x x

Nearest Insertion
x x x x x x x x x x x x x x

Nearest Insertion
x x x x x x x x x x x x x x

Nearest Insertion
x x x x x x x x x x x x x x

Nearest Insertion
x x x x x x x x x x x x x x

Farthest Insertion
1. Select a node and the node farthest to it and build a two-node tour 2. Insert in the tour the farthest node y until no more nodes are available
O(n2) running time

Random Insertion
This heuristic adds points to the tour in a random order

Minimum Spanning Tree

Construct a minimum spanning tree Traverse the tree to build a tour by eliminating edges from vertices with degree three and adding edges to vertices with degree one

Minimum Spanning Tree

Build a minimum spanning tree on the edges between customers Double the tree to get an Eulerian Tour (visits everyone perhaps several times and returns to the start) Short cut the Eulerian Tour to get a Hamilton Tour (Traveling Salesman Tour)

Minimum Spanning Tree

Is Easy to construct
Use the Greedy Algorithm
Add edges in increasing order of length Discard any that create a cycle

Is a Lower bound on the TSP

Drop one edge from the TSP and you have a spanning tree It must be at least as long as the minimum spanning tree

Minimum Spanning Tree

x x

x
x x x x x

Minimum Spanning Tree

Double the Spanning Tree
Duplicate each edge in the Spanning Tree The resulting graph is connected The degree at every node must be even Thats an Eulerian Graph (you can start at a city, walk on each edge exactly once and return to where you started) Its no more than twice the length of the shortest TSP

Minimum Spanning Tree

14 15

x
16

18 17

x
19 20

13

12

11

x
10 9

x
23

x
21 4

x
6

x
7

8x

x
26 25

22

x
3
28

x
24

27

x x
29

x
30

Minimum Spanning Tree

Short Cut the Eulerian Tour
14 15

x
16

18 17

x
19 20

13

12

11

x
10 9

x
23

x
21 4

x
6

x
7

8x

x
26 25

22

x
3
28

x
24

27

x x
29

x
30

Minimum Spanning Tree

Short Cut the Eulerian Tour
14 15

x
16

18 17

x
19 20

13

12

11

x
10 9

x
23

x
21 4

x
6

x
7

8x

x
26 25

22

x
3
28

x
24

27

x x
29

x
30

Christofides Heuristic
Construct an MST[Min. Spanning Tree] Construct a matching of all vertices with odd degree Combine the matching edges with the MST edges, so that now all vertices have an even degree Compute an Eulerian tour through the graph

Savings Method
Start with an initial solution where each customer is serviced individually from the depot
1 3

Clarke G. and J. W. Wright (1964) Scheduling of vehicles from a central depot to a number of delivery points, Operations Research, vol. 12, pp. 568-581.

Savings Method

Savings Calculation
i

cij

c0i
0

cj0

sij = ci0 + c0j - cij

Savings Method

Procedure
1. Compute savings sij for all pairs 2. Choose the pair with the largest savings and join customers in a new route if feasible. 3. If all positive savings have been examined, stop. Otherwise go to 2.

Savings Method
Start with separate one stop routes from depot to each customer. Calculate all savings for joining two customers and eliminating a trip back to the depot.
Sij = Ci0 + C0j - Cij

Order savings from largest to smallest. Form route by linking customers according to savings.

Savings Method

Remove

Add

Savings = S12 = C10+C02 -C12

3

depot

depot

Savings Method
3

S12
1
5

S13
1
5

3 4

depot

depot

depot

S14
1 2 4

S15
1 5 2 4

3 5

S16
1 2 4

3 5

depot

depot

depot

Small savings

Savings Method
S23
1 2
4 3

S24
1
5

S25
1
5

3 4

depot

depot

depot

S26
1 2 4

S34
1 5 2 4

3 5

S35
1 2 4

3 5

depot

depot Large savings

depot

Large savings

Savings Method
S36
1 2
4 3

Large savings

S45
1
5

S46
1
5

3 4

depot

depot

depot

S56
1 2 4

depot

In general, with n customers there are n(n-1)/2 savings to calculate.

Savings Method
Order savings from largest to smallest.
S35 S34 S45 S36 S56 S23 S46 S24 S25 S12 S26 S13 etc.

Savings Method
Form route by linking customers according to savings.

S35:link 3&5
3
1 2 4

S34:link 3&4 (keep 3-5)

3
1 5 2 4

depot

depot

0-3-5-0

0-4-3-5-0

Savings Method
Form route by linking customers according to savings.
S35 S34 S45 S36 S56 S23 S46 S24 S25 S12 S26 S13 etc. 0-3-5-0 0-4-3-5-0

Savings Method

S45: skip
3 1 2

S36: skip
3 1 5 2

S56:link 5&6
3 1 5 2

depot

depot

depot

Savings Method

S23: skip
3 1 2

S46: skip
3 1 5 2

S24:link 2&4
3 1 5 2

depot

depot

depot

Savings Method
S25: skip
1 2 4 3 5

S12: link 1&2

1 2 4

depot

depot

Final route: 0-1-2-4-3-5-6-0 Optimal?

Savings Method
Form route by linking customers according to savings.
S35 S34 S45 S36 S56 S23 S46 S24 S25 S12 0-3-5-0 0-4-3-5-0 skip skip 0-4-3-5-6-0 skip skip 0-2-4-3-5-6-0 skip 0-1-2-4-3-5-6-0

Different Approaches

Route First - Cluster Second

Build a TSP tour Partition it to meet capacity

Cluster First - Route Second

Decide who gets served by each route Then build the routes

Route First
Vehicle Cap: 15
4 6

x2 x3

x6

x
2
5

x
x5
5

x
3

x
4

x
6

x
5

x5

Cluster First
Sweep Heuristic
4

Vehicle Cap: 15
6

x2 x3

x6

x
2
5

x
x5
5

x
4

x
6

x
5

x5

Cluster Algorithms
Select certain customers as seed points for routes.
Farthest from depot. Highest priority. Equally spaced.

Grow routes starting at seeds. Add customers:

Based on nearest neighbor or nearest insertion Based on savings. Based on minimum angle.

Re-optimize each route (solve a TSP for customers in each route).

Cluster with Nearest Neighbor

Suppose each vehicle capacity = 4 customers

depot

depot

Select 3 seeds

Add nearest neighbor

Cluster with Nearest Neighbor

Suppose each vehicle capacity = 4 customers

depot

depot

Add nearest neighbor

Add nearest neighbor

Cluster with Minimum Angle

Suppose each vehicle capacity = 4 customers

depot

depot

Select 3 seeds

Add customers with minimum angle

Cluster with Minimum Angle

Suppose each vehicle capacity = 4 customers

depot

depot

Add customers with minimum angle

Add customers with minimum angle

Improvement Heuristics
Intra-route improvement

Inter-route improvement

depot

depot

Starting routes

Improved routes

VRP Route Improvement Heuristics

Start with a feasible route. Make changes to improve route.
Exchange heuristics within a route
Switch position of one customer in the route. Switch 2 arcs in a route. Switch 3 arcs in a route.

Exchange heuristics between routes.

Move a customer from one route to another. Switch two customers between routes.

Local search methods.

Simulated Annealing. Tabu Search. Genetic Algorithms.

K-opt Exchange
Replace k arcs in a given route by k new arcs so the result is a route with lower cost. 2-opt: Replace 4-5 and 3-6 by 4-3 and 5-6.
Original route
3 1 2

Improved route
3 5

2
4 5

depot

depot

2-opt Exchange
2-Opt
x x x x x x x x x x x

x
x

3-opt Exchange
3-opt: Replace 2-3, 5-4 and 4-6 by 2-4, 4-3 and 5-6.

Original route
3 1 2 4 5 1

Improved route
3 2 4 6 5

depot

depot

Improvement Heuristics
Cluster with Nearest Neighbor

depot

depot

Starting routes

Optimized routes