List of Demographic Variables

1
Efficient Demand Responsive Transit Systems
Maged M. Dessouky Kurt Palmer Majid
Aldaihani Tamer Abdelmaguid
Department of Industrial and Systems
Engineering University of Southern
California Los Angeles, CA 90089-0193
2
Motivation

• There has been a significant increase in demand
  of DRT service providers due to the American with
  Disabilities and Act (ADA).
• Los Angeles County alone has more than 5000 vans
  and 4200 cabs providing paratransit delivery
  service generating 8 million trips per year.
• ADA has also set strict service requirements and
  standards.
• In LA county, the transit budget in the future
  can be entirely absorbed by the ADA types
  service.
• In LA, Access is incorporating a new trial that
  allows ADA eligible passengers to ride for free
  on fixed route buses.

3
Research Objective

• To evaluate a new service delivery method for DRT
  providers (hybrid system)
• Minimize passenger travel miles and time
• Minimize fleet size
• Minimize on-demand vehicle miles

4
Background of PDP
Pickup and Delivery Problem Solution methods
and problem types...
5
Previous Work

• Algorithms for curb-to-curb system

6
Previous Work

• There is little work in the literature that
  introduce methods to integrate the
  demand-responsive service with the fixed route
  service.
• Liaw, White, and Bander (1996)
• Hickman and Blume (2000)
• Our research contrasts from their work in the
  following aspects
• Mathematical formulation
• Criteria of selecting candidate requests and
  paths
• Insertion procedures
• System-wide heuristic

7
Problem Description
General Assumptions There are N passengers, M
paratransit vehicles, and O fixed bus routes For
each passenger, there is a pick-up point
(origin), drop-off point(destination), desired
pick-up time and/or drop-off time. Each vehicle
has a known capacity Each fixed route has a set
of bus stops and a schedule Requests are made in
advance (before the service day) Maximum of two
transfers are allowed (not between same
mode) The distance matrix is given and the
network is symmetric. Objective Integrating the
curb to curb system with the fixed routes system
in order to reduce the total cost and/or increase
the total productivity while not significantly
reducing the service level
8
Hybrid Scheduling Approach
PATH DISTANCES F1,F2 1-2 (8)8 1-1C-2C-2
(1,9,1)11 .72,.22 1-1C-1A-2 (1,11,2)14 .57,.2
7 1-2D-2C-2 (5,3,1)9 .88,2 1-1C-TC-2 (1,5,4)1
0 .80,1 1-1C-1B-2 (1,8,4)13 .61,.625
9
Heuristic Approach
Notations BB (r, B1, B2) Distance from bus stop
B1 to bus stop B2 on route r DD (i) Door to
door distance of request i PB(r, B1, i) Distance
from the origin point of request i to bus stop
B1 on route r DB(r, B2, i) Distance from bus
stop B2 on route r to the drop off
(destination) point of request i DBD (r, B1,
B2, i) PB (r, B1, i) DB (r, B2, i) HYB (r,
B1, B2, i) DBD (r, B1, B2, i) BB (r, B1, B2)
Phase I Screening Two conditions to candidate
paths for Hybrid system
1. Minimizing the hybrid distance
DD(i) /
HYB (r, B1, B2, i) gt F1 2. Minimizing the
distance traveled by vehicle
DBD (r, B1,
B2, i) / BB (r, B1, B2) lt F2
Phase II Scheduling Select the feasible
candidate paths that minimize the
objective function and schedule them in the
existing vehicle schedule
10
Insertion Heuristic Approach
11
Insertion Heuristic Approach
12
Illustrations
Combinations
Current schedule
P1 D1 P2 P3 D3 D2 P1 D1 P2 P3 D2 D3 P1 D1 P2 D2
P3 D3
P1 D1 P2 D2
P1 P2 P3 D3 D1 D2 P1 P2 P3 D1 D3 D2 P1 P2 P3 D1
D2 D3 P1 P2 P3 D3 D2 D1 P1 P2 P3 D2 D3 D1 P1 P2
P3 D2 D1 D3 P1 P2 D1 P3 D3 D2 P1 P2 D1 P3 D2
D3 P1 P2 D1 D2 P3 D3
P1 P2 D1 D2
13
Improvement Heuristic
14
Comparison

• Compare the performance of curb-to-curb to hybrid
  system
• Vehicle productivity
• Passenger travel miles and time
• Clearly curb-to-curb minimizes passenger travel
  miles
• The question is can a hybrid system provide near
  the service level of curb-to-curb at a cheaper
  cost
• Comparison will be based on real data from
  Antelope Valley Transit Authority (AVTA)

15
Why AVTA?

• Distances are large enough to justify transfer
  between modes.
• Most of the passengers travel to a central
  location.
• AVTA is a small to mid-size agency (ideal to
  integrate the two systems).

16
Snapshot of Arcview for the AVTA Data
17
Miles Traveled per Request
18
Actual Pick-up Time
19
Actual Drop-off Time
20
Occupancy Rate - Percentage Time
21
Number of Passengers per Request
22
Ride Sharing
23
Miles Traveled per Driver
24
Computational Experiments
Number of Candidate Requests
25
Computational Experiments
Number of Candidate Paths
26
Computational Experiments
Sensitivity Analysis (Distance)
27
Computational Experiments
Sensitivity Analysis (Trip Time)
28
Daily F1 and F2 Values
29
Daily Vehicle Distance and Customer Time
30
Total Vehicle Distance and Customer Time
31
Components of Customer Trip Time for Improvement
Heuristic
32
Sensitivity Analysis on Number of Vehicles
33
DRT Benchmarking Study

• Surveyed
• 180 agencies listed in 1999 NTD as serving
  populations larger than 200,000
• 25 agencies from California serving
  populations smaller than 200,000
• Responses
• 62 large national agencies
• 13 small California agencies

34
Demographic Variables

• Population Density
• Passenger Trips per Capita
• Passenger Trips per Vehicle

35
Demographic Clusters
36
Demographic Segmentation
Surveyed Responses
Cluster 1 41 15
Cluster 2 29 9
Cluster 3 48 20
Cluster 5 29 9
Cluster 4 Others 33 9
Small CA Agencies 25 13
37
Performance Variables

• Cost Efficiency
• Operating Expense per Passenger Trip
• Operating Expense per Revenue Mile
• Operating Expense per Passenger Mile
• System Productivity
• Revenue Miles per Vehicle
• Revenue Miles per Total Vehicle Mile
• Passenger Miles per Revenue Mile
• Passenger Trips per Revenue Mile
• System Effectiveness
• Passenger Trips per Capita

38
Results of Correlation Principal Components
Analysis

• Cost Efficiency
• Average Expense for Service
• System Productivity
• Average Mileage Productivity
• Average People Loading Productivity
• System Effectiveness
• Passenger Trips per Capita

39
Preliminary Performance Segmentation
Surveyed Responses
All 4 Measures Good 15 8
3 of 4 Good 45 15
2 Good, 2 Bad 58 19
3 of 4 Bad 39 12
All 4 Measures Bad 21 6
Good value is on desirable side of median for
surveyed agencies Bad value is on undesirable
side of median for surveyed agencies
40
Conclusion

• A service delivery method (hybrid system) is
  identified to handle the increased demands and
  strict service standards
• A heuristic approach is developed to solve the
  hybrid DARP.
• The heuristic succeeded in identifying a
  significant number of candidate requests that can
  be transferred to the fixed route service.
• We compared the curb-to-curb system with the
  hybrid system based on real data from Antelope
  Valley Transit Authority (AVTA).
• The hybrid approach can significantly reduce the
  vehicle travel miles, however, at the expense of
  increased hybrid passenger trip time. Although,
  the total passenger trip time does not increase.
• Future research will explore a system-wide
  heuristic versus the current insertion procedure.