Behavioral Micro-Simulation Jose Holguin-Veras, Ph.D., P.E. William - - PowerPoint PPT Presentation

Behavioral Micro-Simulation

1

Jose Holguin-Veras, Ph.D., P.E.

William H. Hart Professor VREF’s Center of Excellence for Sustainable Urban Freight Systems Center for Infrastructure, Transportation, and the Environment Rensselaer Polytechnic Institute jhv@rpi.edu

Main goals

To produce a reasonable guess of freight traffic in metropolitan areas using:

Freight trip generation estimates (using NCFRP 25 models) Known delivery patterns, such as tour length distributions by industry sectors (obtained from data collected by RPI from carriers and receivers) Observed traffic counts at key corridors

The BMS was originally developed to assess the impacts of alternative policies to foster off-hour deliveries (7PM to 6AM)

2

Key components

Freight trip generation (FTG): estimated using the NCFRP 25 models and Zip Code Business Pattern data Synthetic population of carriers (and receivers, if needed) is created

Using the data collected by RPI, the sample data is used to create the population of carriers needed to make all deliveries in the metro area The origin of the deliveries are set to be the locations were warehouses and distribution centers are located

Delivery tours are created:

Match the tour length (number of stops) by industry sector Match the number of deliveries by ZIP code (or any other level of geography used)

3

Graphically: Freight Trip Generation

4

Graphically: Synthetic population of carriers

Different industry sectors have different tour lengths NYC and NJ (Holguin-Veras et al. 2012):

Average: 8.0 stops/tour; 12.6% do 1 stop/tour; 54.9% do < 6 stops/tour; 8.7% do > 20 stops

Synthetic population match observed traffic and FTG

5

Tour simulations

 Select a truck in an industry sector

 Number of stops is randomly assigned  Select receivers at random from the group of receivers in that sector  Compute optimal tour and store it

 Repeat until delivery tours satisfy the FTG for the entire area

6

1) Origin of a truck that carries food products to five restaurants 2) Five receivers

Example: Use of the BMS in the OHD project

BMS use in the off-hour delivery project

8

Carrier/receiver synthetic generation  Randomly select industry segment

• Generate/locate carrier
• Generate/locate receivers to serve

Receiver behavioral simulation  Model receiver’s decision to accept OHD Carrier behavioral simulation  Compute costs for base case and mixed operation  Model carrier’s decision Repeat for another carrier-receivers set End Change incentives, reset participation counts Define range of incentives to receivers for OHD Ordinal logit model (Holguin-Veras et al 2013) Regular-hour receiver Off-hour receiver a) Base case (no OHD) b) Mixed operation Carrier depot Legend:

Output: Joint Market Share (JMS) of OHD Receivers Market Share (RMS) at TAZ level

Ordered logit model with random effects

This model reproduces receivers’ response to incentives

9

Model Independent variables Parameter t-stat Parameter t-stat Constant 0.61 (2.78) 0.22 (1.00) Number of deliveries

• 0.07

(-9.17)

• 0.08

(-11.66) Incentives One time incentive in $1000 (OTI) 0.18 (6.95) 0.17 (6.76) Carrier discount in percent (CDR*100) 3.00 (6.81) 3.10 (7.12) Business Support (BS) 0.55 (3.82) 0.51 (3.52) Public Recognition (PR) 0.34 (2.24) 0.38 (2.48) Trusted Vendor (TV) 0.94 (4.29) NAICS Clothing stores, binary variable

• 2.73

(-4.57)

• 2.46

(-4.32) Performing arts, binary variable

• 1.96

(-5.69)

• 4.80

(-12.38) Interaction terms: OTI and NAICS OTI for food and beverage stores 0.12 (2.56) 0.20 (4.24) OTI for apparel manufacture stores 0.23 (1.72) 0.11 (1.88) OTI for clothing stores 0.24 (3.18) 0.25 (3.40) OTI for nondurable wholesalers 0.33 ( 6.83) 0.37 (7.62) Interaction terms: CDR and NAICS CDR for personal laundry

• 2.11

(-2.98)

• 2.08

(-3.25) Interaction terms: Trusted vendor and NAICS TV for food and beverage stores 4.35 (7.29) 2.02 (3.17) TV for performing arts 4.65 (2.56) 13.49 (11.16) TV for clothing stores 5.06 (8.28) 2.24 (4.06) TV for miscellaneous stores retailers 6.59 (13.63) 3.17 (5.86) Parameters µ(1) 1.88 ( 21.54) 1.91 (21.36) µ(2) 4.56 (34.64) 4.56 (34.14) µ(3) 7.63 (40.45) 7.55 (40.51) Sigma 4.58 (27.64) 4.74 (25.83) n Log likelihood

• 1390.89
• 1388.50

1522 Model 1 Model 2 1522

Incentives Interaction terms: OTI and NAICS NAICS code Interaction terms: TV and NAICS

BMS Results

10

OTI = $0 avg = 2.2% max = 6.2% min = 0.6% OTI = $2,000 avg = 2.7% max = 7.6% min = 1.2% OTI = $4,000 avg = 3.4% max = 7.6% min = 1.3% OTI = $6,000 avg = 4.3% max = 9.9% min = 1.9% OTI = $8,000 avg = 5.5% max = 11.9% min = 2.6% OTI = $10,00 avg = 7.0% max = 13.4% min = 3.5%

Example: Geographically Oriented Incentives

Geographically focused incentives: case of NYC

50% of establishments are located in Midtown Manhattan being responsible for 52% of the incoming freight trips to the city Two geographic distribution have been considered: (1) Lower and Midtown (2) Central Park and Upper Scenarios consider giving incentives to either the entire Manhattan

• r only to Lower and Midtown Manhattan

Lower Manhattan (LM) Midtown Manhattan (MM) Central Park (CP) Upper Manhattan (UM) + +

Results of geographically focused incentives

Ratio Budget/JMS provides an idea about the amount

• f resources required to achieve a 1% JMS

The results also show the superiority of geographically focused incentives which requires between 71% and 75% less expenditures than incentives spread out all

• ver Manhattan

13

OTI ($K) JMS (%) RMS (%) Budget ($M) JMS (%) RMS (%) Budget ($M) 1 6.5 1.7 1.4 6.7 1.8 0.4 2 7.0 1.9 3.4 7.0 1.9 0.8 3 7.4 2.1 5.5 7.0 1.9 1.2 4 7.8 2.3 8.5 7.5 2.2 2.1 5 8.0 2.4 11.2 8.2 2.4 2.8 6 8.6 2.7 15.2 8.4 2.6 3.7 7 8.9 2.7 19.2 8.6 2.7 4.5 8 9.7 3.2 26.1 9.1 2.9 5.9 9 9.6 3.3 29.7 9.7 3.3 7.3 10 10.3 3.6 36.2 9.9 3.4 8.8 Lower and Midtown Manhattan Central Park and Upper Manhattan OTI ($K) Manhattan Lower and Midtown Manhattan 1 0.31 0.22 2 0.67 0.48 3 1.05 0.75 4 1.45 1.08 5 1.89 1.40 6 2.41 1.76 7 2.85 2.16 8 3.46 2.68 9 4.07 3.10 10 4.68 3.52 Ratio Budget/JMS

Example: Self-Supported Freight Demand Management

Self supported freight demand management

A self-supported freight demand management system (SS-FDM), is one that generates the funds required for a continuing improvement towards sustainability The incentives to be handed out to the receivers are generated by a toll surcharge to the vehicles that travel in the regular hours The analyses consider tolls to only trucks (per axle) or both; trucks and cars. Finally, different levels of toll collection efficiency were also considered

15

Results: tolls to trucks (per axle)

Toll collection 100% Toll collection 75%

16

$1 $2 $5 $8 $10 $0 7.1% 8,467 $0.00 $0.00 $58.08 $116.16 $290.40 $464.63 $580.79 $1,000 7.6% 9,031 $5.65 $1.88 $57.78 $115.56 $288.91 $462.26 $577.82 $2,000 8.2% 9,783 $26.32 $8.77 $57.39 $114.77 $286.93 $459.10 $573.87 $3,000 8.8% 10,463 $59.89 $19.96 $57.03 $114.06 $285.15 $456.23 $570.29 $4,000 9.5% 11,265 $111.95 $37.32 $56.61 $113.21 $283.04 $452.86 $566.07 $5,000 10.3% 12,209 $187.14 $62.38 $56.11 $112.22 $280.55 $448.89 $561.11 $6,000 11.0% 13,058 $275.51 $91.84 $55.66 $111.33 $278.32 $445.31 $556.64 $7,000 11.9% 14,175 $399.58 $133.19 $55.08 $110.15 $275.38 $440.62 $550.77 $8,000 12.8% 15,200 $538.65 $179.55 $54.54 $109.08 $272.69 $436.30 $545.38 $9,000 13.7% 16,279 $703.14 $234.38 $53.97 $107.94 $269.85 $431.76 $539.70 $10,000 14.9% 17,754 $928.70 $309.57 $53.19 $106.39 $265.97 $425.56 $531.95 Freight vehicle surcharge per axle: One-time- incentive % OHD OHD tours (year) Total incentive budget Yearly incentive budget Yearly toll revenues (car surcharge = $0) $1 $2 $5 $8 $10 $0 7.1% 8,467 $0.00 $0.00 $77.44 $154.88 $387.19 $619.51 $774.39 $1,000 7.6% 9,031 $5.65 $1.88 $77.04 $154.09 $385.21 $616.34 $770.43 $2,000 8.2% 9,783 $26.32 $8.77 $76.52 $153.03 $382.58 $612.13 $765.16 $3,000 8.8% 10,463 $59.89 $19.96 $76.04 $152.08 $380.19 $608.31 $760.39 $4,000 9.5% 11,265 $111.95 $37.32 $75.48 $150.95 $377.38 $603.81 $754.76 $5,000 10.3% 12,209 $187.14 $62.38 $74.81 $149.63 $374.07 $598.51 $748.14 $6,000 11.0% 13,058 $275.51 $91.84 $74.22 $148.44 $371.09 $593.75 $742.19 $7,000 11.9% 14,175 $399.58 $133.19 $73.44 $146.87 $367.18 $587.49 $734.36 $8,000 12.8% 15,200 $538.65 $179.55 $72.72 $145.43 $363.59 $581.74 $727.17 $9,000 13.7% 16,279 $703.14 $234.38 $71.96 $143.92 $359.80 $575.68 $719.60 $10,000 14.9% 17,754 $928.70 $309.57 $70.93 $141.85 $354.63 $567.41 $709.26 Yearly toll revenues (car surcharge = $0) Freight vehicle surcharge per axle: One-time- incentive % OHD OHD tours (year) Total incentive budget Yearly incentive budget

Note: The shaded cells represent non-feasible combinations of financial incentives to receivers and tolls.

Results: tolls to trucks (per axle) and cars

Toll collection 100% Toll collection 75%

17

$1 $2 $5 $8 $10 $0 7.1% 8,467 $0.00 $0.00 $382.94 $460.38 $692.69 $925.01 $1,079.89 $1,000 7.6% 9,031 $5.65 $1.88 $382.54 $459.59 $690.71 $921.84 $1,075.93 $2,000 8.2% 9,783 $26.32 $8.77 $382.02 $458.53 $688.08 $917.63 $1,070.66 $3,000 8.8% 10,463 $59.89 $19.96 $381.54 $457.58 $685.69 $913.81 $1,065.89 $4,000 9.5% 11,265 $111.95 $37.32 $380.98 $456.45 $682.88 $909.31 $1,060.26 $5,000 10.3% 12,209 $187.14 $62.38 $380.31 $455.13 $679.57 $904.01 $1,053.64 $6,000 11.0% 13,058 $275.51 $91.84 $379.72 $453.94 $676.59 $899.25 $1,047.69 $7,000 11.9% 14,175 $399.58 $133.19 $378.94 $452.37 $672.68 $892.99 $1,039.86 $8,000 12.8% 15,200 $538.65 $179.55 $378.22 $450.93 $669.09 $887.24 $1,032.67 $9,000 13.7% 16,279 $703.14 $234.38 $377.46 $449.42 $665.30 $881.18 $1,025.10 $10,000 14.9% 17,754 $928.70 $309.57 $376.43 $447.35 $660.13 $872.91 $1,014.76 Yearly toll revenues (car surcharge = $1) Freight vehicle surcharge per axle: One-time- incentive % OHD OHD tours (year) Total incentive budget Yearly incentive budget $1 $2 $5 $8 $10 $0 7.1% 8,467 $0.00 $0.00 $363.58 $421.66 $595.90 $770.13 $886.29 $1,000 7.6% 9,031 $5.65 $1.88 $363.28 $421.06 $594.41 $767.76 $883.32 $2,000 8.2% 9,783 $26.32 $8.77 $362.89 $420.27 $592.43 $764.60 $879.37 $3,000 8.8% 10,463 $59.89 $19.96 $362.53 $419.56 $590.65 $761.73 $875.79 $4,000 9.5% 11,265 $111.95 $37.32 $362.11 $418.71 $588.54 $758.36 $871.57 $5,000 10.3% 12,209 $187.14 $62.38 $361.61 $417.72 $586.05 $754.39 $866.61 $6,000 11.0% 13,058 $275.51 $91.84 $361.16 $416.83 $583.82 $750.81 $862.14 $7,000 11.9% 14,175 $399.58 $133.19 $360.58 $415.65 $580.88 $746.12 $856.27 $8,000 12.8% 15,200 $538.65 $179.55 $360.04 $414.58 $578.19 $741.80 $850.88 $9,000 13.7% 16,279 $703.14 $234.38 $359.47 $413.44 $575.35 $737.26 $845.20 $10,000 14.9% 17,754 $928.70 $309.57 $358.69 $411.89 $571.47 $731.06 $837.45 Freight vehicle surcharge per axle: One-time- incentive % OHD OHD tours (year) Total incentive budget Yearly incentive budget Yearly toll revenues (car surcharge = $1)

Note: in this case all combinations of financial incentives to receivers and tolls are feasible

Potential Uses

Potential Uses

The BMS will replicate freight traffic in any metro area The BMS could be used to:

Produce realistic estimates of freight VMT Analyze the impacts of alternative logistical configurations (using a Urban Consolidation Center, transfers of cargo to environmentally friendly modes like freight bicycles) Analyze the impacts of retiming of deliveries, or receiver-led consolidation programs by receivers Analyze the impacts of policies that change operational patterns, technologies, or infrastructure used by carriers Changes in work hours, limited emission zones, etc.

19

Expected outputs of the BMS

Acceptance rate of technology/ operations/ infrastructure in response to policy measures Freight (large and small trucks) VMT by industry segment for the initiatives considered, including time

• f day for some

Freight traffic by origin-destination before/after, a key input for traffic simulation models Cost impacts on carriers and receivers

20

Limitations

Estimation of air pollution

The BMS is not a traffic simulator, it does not account for traffic behavior in networks

Potential solution:

 Use the BMS output as an input to traffic simulators  Purchase GPS data for key metro areas and post-process it with

MOVES to produce estimates, add the estimates to BMS

The BMS is very good for urban freight modeling, though it does not consider intercity freight (and things like truck stop electrification, etc.)

Potential solution: create modules that perform these

computations, add to BMS

21

Conclusions

Conclusions

The BMS is an important tool to evaluate TDM policies The application to the Manhattan case study provides insight into the potential benefits, and limitations:

Off-Hour Deliveries Geographic oriented incentives Self Supported Freight Demand Management

Other extensions of the BMS include the analysis of incentives according to industry segments

23

Questions?

24