Microsimulation of Intra-Urban Commercial Vehicle and Person Movements 11th National Transportation Planning Applications Conference Session 11: May 8, 2007, Daytona Beach, Florida * Contact Information: | Ofir Cohen, PB, San Francisco* John Gliebe, Portland State University Doug Hunt, University of Calgary
Agenda Motivation- why? D isaggregate CO mmercial M odel Scope Survey and Segmentation of Establishments Model Components Calibration results
Motivation Commercial travel comprises a large share of weekday urban traffic, but has received scant attention from modelers (Regan and Garrido, 2002) –~11% of overall vehicle trips in the state. –Emphasize on Tour rather than trip –Standard freight models miss short-hauls and multi- stop deliveries within urban areas –Freight models don’t represent service provision, sales calls and travel for meetings –Large variation in firm operations Practical yet realistic approach needed
Scope: What is a Commercial Trip? Intra-urban trips only – up to 50 miles* –ACOM is an econometric model that simulates Inter-urban trips. Weekday simulation of a typical 24 hrs* All trip purposes combinations are available Includes goods pickup and delivery, meetings, business supply acquisition, service provision, sales, driver’s lunch, etc.
Establishment Types Industrial: 4 sub establishments categories »Agriculture »Construction »Heavy Industry »Other ( Mines, Metal, Light Industrial, etc.) Wholesale: warehousing and distribution Retail: stores and restaurants Transport: for-hire trucking and delivery Service: 5 sub-establishments types »Hotel »Health »Government »Education »Other – e.g., banking, consulting
Ohio Establishment Survey Surveys: –Data on the firm: employees, number who travel for job, commodities, occupations –One-day activity/travel diaries –Shipment data corresponding to travel diary Sample: –561 public and private establishments –1,640 workers who traveled –1,951 work-based tours –9,588 activity/trip records
Ohio Establishment Survey, Cont. Limitations / Simplifications –No data on intra-establishment relationships –One vehicle per day per employee –Occupations of individuals not identified –No observations for Non-Motorized or Transit trips –No data on delivery company such as FedEx, DHL, or UPS
Traveler Generation Model Number of employees segmented by establishment type is defined per TAZ Binary Logit function- an employee did a Commercial Tour or not A traveler will do at least 2 trips (First trip+ return to his establishment) EstablishmentIndustrialWholesaleRetailTransportServicesAll Total Employees2,057,520386,4601,471,444264,8664,121,8538,302,143 % Who Travel9%15%7%14%9% Total Travelers180,57056,810103,00138,141379, ,749
Traveler Worker TAZ 1457 = 17 Construction Workers
D.C Log Sum Industrial EstablishmentsService Establishments Time coef= Time coef=
Vehicle Type Model Assign to each traveling employee a vehicle type for the entire day MediumHeavy Industrial Wholesale Retail Transport Service Resid_LU Ind_Mix_LU Indus_LU Office_LU Retail_LU CBD__LU Rural_LU
Vehicle Use by Establishment Type
Start Time Model
Day patterns formed through dynamic choice approach Not a pattern based model Any number of tours and trips is possible Sensitive to accumulated time at multiple levels: - activity, tour and workday duration Previous decisions affect future decisions
Trip Purpose Model Multinomial Logit function with 6 alternatives: 1. Good - Distribution/pickup of goods 2. Service - Providing Service 3. Meeting - Limited to Light / Medium vehicle –Available only between 07:30-21:30 4. Other- Personal needs (Food, Gas) –Available only between 06:00-22:45 5. Back To Establishment - ends this tour 6. Stay in Current Activity - increment times by 5 minutes, simulates the trip duration
08:05 AM T.P 06:20 AM06:52 AM06:57 AM07:20 AM09:20 AM08:00 AM T.P 12:00 AM
Trip Purpose Model SERVICE TRIP Establishment=Wholesale GoodServiceOtherMeetingReturn current- Good current- Service current- Other current- Meeting current- Back to Estab Constant Time Hour 08:00-09: Time Hour 17:00-18: Stay Duration when current= LN (Stay Duration) when current= Wholesale Stay Effect when current= Overall Tour Duration -1.1E-34.5E-3 Total Activity Duration - current tour Vehicle Light GOODS TRIPOTHER TRIPMEETING TRIPRETURNSTAY U (purpose) = c1+c2*EstablishmentType +c3*currentPurpose + StayEffectConstant + timeWindowConstant*time+ c4*tourDuration+ c5*DayDuration+ c6*stayDuration+ c7*ln (stayDuration) +c8*VehicleType
Next Stop Location U(TAZ)=f( Chosen Purpose, Establishment, Vehicle, eTime, tTime, Jobs(14 categories), HH, LU type)
Next Stop Location Results Industrial Establishment DestinationsWholesale Establishment Destinations
Establishment & Destination Establishment locations Destination locations Columbus Area
Destination Choice Distance Calibration
Lesson learned “Worth the effort” – shouldn’t be neglected. Capture “real-time” decisions Huge variation in patterns Estimation shouldn’t be over-segmented. More vehicle types. Can be applied for Weekend HH activity model Easily calibrated
Acknowledgements Ohio Department of Transportation –Greg Giaimo –Rebekah Anderson –Sam Granato
Questions?