You are on page 1of 25

Road Network Design Problems with

Spatial Harvest Scheduling

Evelyn W. Richards
Forest Engineering
Faculty of Forestry and Environmental Management
University of New Brunswick

Road Network Design Problems with


Spatial Harvest Scheduling

Adjacency Constraints

Road Network Planning

What is the real definition?


Modelled adjacency conditions: a mis-match
Results from a Stand-Centred Model
Importance
Integer models that can work
Integrating Access and Harvest Planning

Spatial Datasets: where are they?


Summary

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
2

Adjacency

What is the real definition?


How have we modeled it?

A mis-match?

Does this matter?


Exact Solutions
Stand Centred Model
Examples and Results

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
3

Adjacency Definition
Given an existing clearcut and a
definition of the maximum
clearcut size M = 50.
1)Adjacent areas not to be
harvested until area is
regenerated
2)If harvesting in an adjacent
area would increase the size of
the clearcut beyond M, then it
is forbidden.
1) or 2) ?

35

20

30

20

40
30

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
4

There is a significant difference


35
20

30

30

35

35
20

30

20

40
30

20

40

Harvest now

Harvest during green-up period

30

20

30

20

40

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
5

What has been modeled?


Closer to regulation

Trickier to model

1 Pre-blocked prior to
modeling.

35
30

2 Not pre-blocked: the model


groups stands dynamically.
3 Not pre-blocked: openings
can grow to maximum size
over several periods.

20
30

40

35
30

20

30

We developed a Stand Centred Model that


gets near-optimal solutions

20

20
40

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
6

Does Adjacency Matter?


Percentage Reduction in Objective Function: NPR
Forest

1 period

2 period

3 period

Eldorado

0.31%

3.25%

7.00%

nbcl9_01

3.40%

5.12%

7.83%

nbcl9_01b

0.77%

3.12%

6.60%

nbcl9_01c

0.88%

3.08%

6.99%

nbcl9_01d

1.14%

3.73%

7.68%

This is the percentage reduction in AAC when there are greenup


constraints.
______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
7

Harvest & Road Choices


Given a set of stands or blocks, and a potential network of road segments that access all
stands:
Choose the best set of harvest and road construction projects

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
8

Harvest and Road Planning


Integrated Model
Maximize Net Revenue - Road Construction Costs Transportation Costs
Subject to:
1.
Average ending age >= 75
2.
Non-Declining Yield (Maximum increment 10%)
3.
Adjacency Constraints
4.
All timber transported to exit node
5.
Road Network Connectivity

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
9

Linked Models
1. Maximize Net Revenue
Subject to
Average ending age >= 75
Non-Declining Yield (Maximum increment 10%)
Adjacency Constraints

2. Minimize Road Costs + Transportation Costs


Subject to
Harvest Solution Found in 1.
______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
10

Results: Benefits of Integrated Model


Road Construction Cost
(Discounted $'000)

1,200

19.38%

1,000
800
600
400

8.36%

200
5x5
Linked Models

10x10
Integrated

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
11

Comparing Solutions

Linked Process

Integrated Model

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
12

Impact of Greenup on Roads


Minimize NPV (Transport + Build)
Subject to: Volume >= Minimum
10x10

Minimize NPV (Transport + Build)


Subject to: Volume >= Minimum
6x6
2,500,000

500,000

2,000,000

400,000
300,000

1,500,000

200,000

1,000,000

100,000

500,000

# Greenup Periods

# Greenup Periods

Clearly, longer green-up increases road costs when volume is fixed


______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
13

Impact of Greenup on Roads


Road Construction Cost

140,000
120,000
100,000
80,000
60,000
40,000
20,000
1

Planning Period
0

Road costs increase and roads are built earlier

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
14

Spatial Datasets: Do we have any?

Crown License 5, NB
131029 polygons
Mean area: 0.4699
Maximum: 133.6287
Minimum: 0.0000
______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
15

Spatial Datasets
Photo Interpretation
Tile Matching
Slivers and Fragments from
buffering
...

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
16

The Result?
... no polite way to
express it

This "stand" has 29 adjacent stands!


______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
17

Aggregation heuristics

Match important features


Merge slivers
Search for wild shapes (shape index)
...
131029 polygons
Mean area = 0.4699

5881 Polygons Mean


area = 10.27

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
18

ifmlab.for.unb.ca/fmos

1. Existing "real" forests


2. Research Forests
3. Forest Generator
4. Forest Generator with roads

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
19

nb crown licence 5

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
20

Forest Landscape Generator

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
21

Forests and Roads

Simulates stands and potential road networks.


Users enter parameters for distributions
Creates data structures necessary for optimization models.

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
22

Summary
Spatial acess and production models

Integer programming optimization brings clarity


yield interesting insights
are tractable
require better datasets

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
23

Thanks!

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
24

Hierarchical FM Re-visited
Strategic

What outputs from


higher levels need to be
enforced at lower
levels?
Strategic Tactical

AAC?
Spatial AAC?
Exact Aspatial Schedule?

Tactical

Tactical/Operational

Annual

______________________________________________________________
Evelyn Richards, Forest Engineering, UNB
25

You might also like