Regional LandUse and Transportation Planning Using a Genetic Algorithm

1
Regional Land-Use and Transportation Planning
Using a Genetic Algorithm

• Brigham Young University
• Richard Balling, Ph.D., P.E.
• Michael Lowry
• Mitsuru Saito, Ph.D., P.E.

funded by the National Science Foundation
2
Outline

• Problem Formulation
• Genetic Algorithm
• Results
• Conclusions and Future Work

3
Problem FormulationWasatch Front Region

• Divide region into 343 districts.
• Find optimum scenario assignment for each
  district from set of defined scenarios.

Status Quo Scenario Assignment
4
Problem FormulationWasatch Front Region

• Identify 260 inter-district streets.
• Find optimum street type assignment for each
  street.

C2 two-lane collector C3 three-lane
collector C4 four-lane collector C5 five-lane
collector A2 two-lane arterial A3 three-lane
arterial A4 four-lane arterial A5 five-lane
arterial A6 six-lane arterial A7 seven-lane
arterial F1 freeway
Status Quo Street Assignment
5
Feasible Plans
Wasatch Front Region 10420 possible plans
housing capacity gt 2,401,000 residents (2020
Forecast) employment capacity gt 1,210,000
jobs (2020 Forecast) open space gt 165,000
acres (20 of developable land)
6
Objectives
Minimize Travel Time of all trips in a 24 hour day
Minimize Land-Use and Street Change from Status
Quo

• link-node network
• peak commute period, off-peak period
• home-based work trips, home-based non-work
  trips, non-home-based trips
• trip production and attraction rates for each
  scenario
• gravity model
• Dial's multipath assignment model
• congestion delays for peak commute period

• measured in terms of status quo people affected
• multiply people affected by degree of change
  factor
• summed over streets and over districts

7
Genetic Algorithm

• Represent plans as chromosomes
• 1) Random starting generation
• 2) Calculate feasibility and fitness of each plan
• 3) Create child generation from parent generation
• a) tournament selection
• b) single-point crossover
• c) gene-wise mutation
• d) maturation (elitism)

343 District Genes
260 Street Genes
...
...
...
A2
C4
A3
F
C2
8
Genetic AlgorithmWasatch Front Region
Start Generation
9
Genetic AlgorithmWasatch Front Region
2nd Generation
10
Genetic AlgorithmWasatch Front Region
4th Generation
11
Genetic AlgorithmWasatch Front Region
6th Generation
12
Genetic AlgorithmWasatch Front Region
12th Generation
13
Genetic AlgorithmWasatch Front Region
30th Generation
14
Genetic AlgorithmWasatch Front Region
100th Generation
15
ResultsWasatch Front Region
16
ResultsWasatch Front Region
17
ResultsWasatch Front Region
18
ResultsWasatch Front Region
19
ResultsWasatch Front Region
20
ResultsWasatch Front Region
21
ResultsWasatch Front Region
22
Conclusions

• Genetic algorithms can be used to search over
  thousands of plans to find an optimum trade-off
  set of plans for regions.
• 2) Minimizing change converted open space land to
  residential land sprawl. This seems to be what
  has occurred in the Wasatch Front Region over
  the past two decades.
• Minimizing travel time favored mixed usage land
  and upgraded street capacity. Total travel time
  was less than half the travel time of the min
  change plan.

23
The Next Step City Planning
Region specifies scenario for a particular
district
City determines zone land uses that match the
scenario percentages
Region specifies inter-district street types
City determines intra-district street types