Modeling for Urban Goods Movement – a case study of Indian Cities Dr. S. L. Dhingra Adjunct Professor Ex. Institute Chair Professor & Emeritus Fellow Transportation Systems Engineering Civil Engineering Department IIT Bombay, India by On April 9-10,2014 Workshop on Urban Freight Transport : A Global Perspective By TSE/CE/IIT Bombay and Center of Excellence for Sustainable Urban Freight Systems, RPI, Troy,NY
62
Embed
Modeling for Urban Goods Movement - Indian Institute of · PDF fileModeling for Urban Goods Movement –a case study of Indian Cities Dr. S. L. Dhingra Adjunct Professor Ex. Institute
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Modeling for Urban Goods Movement – a case study of
Indian Cities
Dr. S. L. Dhingra Adjunct Professor
Ex. Institute Chair Professor & Emeritus FellowTransportation Systems Engineering
Civil Engineering DepartmentIIT Bombay, India
by
On April 9-10,2014
Workshop on
Urban Freight Transport : A Global PerspectiveBy
TSE/CE/IIT Bombay and Center of Excellence for SustainableUrban Freight Systems, RPI, Troy,NY
What makes it important?
• Traffic congestion
• Environmental impacts
• Traffic accidents
• Terminal facilities
2
Goods movement pattern
Intra-city flows – Flowswhose origin anddestination are within thecity
Inter-city flows - Flowswhose one end (origin ordestination) is within thecity and other outside thecity
Regional flows - Flowswhose both ends (originand destination) areoutside the city
3
Possible patterns of urban goods flows
4
Intra urban freight movement
• Goods movement is directly related to populationand to understand that one must know thephysical, economic, and social make up of the city
• Urban goods may be classified depending on itsphysical state, handling needs, modes of vehiclesused, direction of movement etc., for analyzingthe demands
• Whole problem of goods movement would notbe solved all at once, but modeling frameworkcan be flexibly adopted to make progress in smallsteps
5
Modeling frame work
6
• Aggregate analysis of total establishments employingthe aggregated parameters to yield trip rates may beadequate and useful for planning process
• Urban goods movement forecasting techniques mustbe developed in terms of fairly simple measures ofeconomic activities
• Any modeling efforts should begin with the datacollection relating to urban goods movementsthrough primary surveys of consignment movementsand supplementing them by secondary sources
7
8
Inter urban freight movement
• Situation in the case of inter urban freight movements has bright patches
• Consignment size and distance of haul are the most significant parameters in choosing the own transport, hired transport or railways for goods movement
• Firms owning transport generally utilized their own transport for medium and short hauls and preferred hired mode for long distance trips
9
Selection of cities for study
• Cities of varying sizes with respect todemography and economic activities
• Federation of Indian Chamber of Commerceand Industry (FICCI) proposal that classifyingcities on the grounds of economy is anappropriate one as the urban economystructure has direct influence on the urbangoods flows
10
11
India
12
Case Study Cities
Data collection
• Truck operator surveys
• Traffic counts at selected points in the city
• Outer cordon surveys
• Focal point surveys
Owing to the complexity of the goodsmovement, no single method of datacollection could cover complete goodsmovement and its characteristics
13
Traffic survey stations in selected cities
14
Outer cordon surveys
Sample size between 6.8 to 100% dependingupon the city was taken
The following particulars of the sampled goodsvehicle were collected
1. Type of vehicle
2. Origin of trip
3. Destination of trip
4. Land use at destination
5. Type of commodity carried
6. Quantity of commodity carried
15
16
Goods focal point surveys
• Major goods focal points are industries, whole sale trade, ware houses, freight terminals
• Goods focal point surveys are involved in identifying the extent of market in space and drawing a cordon line around these spaces
• In most of the cities whole sale markets were concentrated at one place
• Separate surveys were organized for each market in Delhi as different markets are located at different points
17
18
• Data collected in goods focal point surveys was1. Type of activity2. Type of vehicle3. Origin4. Destination5. Destination of land use6. Type of commodity carried7. Quantity of commodity carried8. Average distance travelled in a day
19
Goods transport flows
• Magnitude of goods transport flows in each ofthe cities was determined by analyzing datacollected through cordon surveys and focalpoint surveys
• Volume of incoming vehicles and quantum ofincoming goods increased with city size
• Outgoing goods traffic was also found to beincreasing with city size
20
21
Urban goods flow characteristics
Commodity Classification is important to makethe analysis manageable. Eight broad categoriesare listed1. Perishable food products2. Non-Perishable food products3. Beverages4. Industrial Inputs5. Industrial Outputs (Consumer Products)6. Building materials7. Industrial Outputs (Intermediate Products)8. Other Categories
22
Intercity Inbound flow characteristics Trucks and mini trucks are the major carriers of intercity
inbound flows with more than 91% of goods moving bythese vehicles
Building materials (28.9%), Industrial inputs (18.1%)and food products (16.7%) are the major constituentsof the intercity inbound flows
Whole sale markets (35.8%) and retail markets (21.3%)are the major attractors of the inbound goods flows
The intercity inbound flows are dominated by heavyconsignments with more than 55% consignmentsweighing more than 4 tonnes
23
Mode split of intercity inbound goods flows
24
Commodity wise composition of intercity inbound goods flows
25
Intercity inbound goods flows destined to different land uses
26
Intercity inbound goods flows as per consignment size
27
Intercity Outbound flow characteristics The dominant carriers of Intercity Outbound flows are
trucks (70%) and mini trucks (18%) The major constituents of outbound flows are food
products, industrial outputs and industrial raw materials Whole sale trade (52.4%), industries (23.1%), transport
terminals (17.6%) are the major generators of outboundflows from the cities
Intercity outbound flows are also dominated by heavier consignments with more than 4 tonnes accounting for 44.4% of the total consignments
28
Mode split of intercity outbound goods flows
29
Commodity wise composition of intercity outbound goods flows
30
Intercity outbound goods flows destined to different land uses
31
Intercity outbound goods flows as per consignment size
32
Regional flow characteristics
• The proportion of through traffic in a city is found tobe dependent more on its locations with respect totrunk routes
• The industrial raw materials (24.4%), buildingmaterials (16.9%) and food products (16.1%) arefound to be major constituents of the through flows
33
Intra-city flow characteristics The major generators of intracity goods flows are whole sale
markets and warehouses with a contribution of 62% of totalintracity flows
Retail trade is found to be the major attractor (39.4%) of theintracity flows
Trucks carry 45% and mini trucks carry 16% of the goodstransported within the city
Slow moving vehicles constitute 70% of the intracity goodsvehicle trips and carry about 40% of goods transported in cities
Non-Perishable food products (21.3%), industrial raw materials(20.6%), building materials (17.4%) and intermediate industrialoutputs (16.5%) are the major contributors of Intracity flows
34
Mode split of intracity goods flows
35
Intracity goods flows originating from different land uses
36
Commodity wise composition of intracity goods flows
37
• The average consignment size of intracity flows ranged from 0.5 tonnesto 2 tonnes and also the average consignment size of differentcommodities varied widely
• The average distance of haul varied with city size and it is found toincrease with the city size
• As expected the average trip length of trucks was higher and they alsocarried heavier consignments
• Trucks accounted for 40.8% of the tonne kilometers made in the citiesand is followed by LCVs with 18.8%
• Fast moving vehicles contributed to about 30% of the vehicle km whilethe slow moving vehicles contributed to more than 70% of the vehiclekm while the slow moving vehicles contributed more than 70% of thevehicle kilometers made in urban areas
38
Intracity veh-km made by different vehicles
39
Intracity tonne-km made by different vehicles
40
Urban Goods Transport Demand Modeling
• Input – Output model
Goods demanded by each sector of economy from all other sectors of economy can be determined but non-availability of Input – Output tables in terms of commodities makes the use of model difficult
• Sequential model
Similar to urban passenger transport planning modeling with certain variations in the specifics of the models
41
42
• Variables Selection
Urban goods flows = f(Population, Industrial Workers, Workers in Trade and Commerce)
43
Proposed modeling approach
44
Sequential Flow Models
• Intercity Inbound Vehicle Trips Model
Vehicle trips = -233+0.00102(Pop)+59.6(PIW)
where,
Pop = Population of city
PIW = Industrial workers as % of total workers
• Intercity Inbound Goods Flows Model
Flow in tonnes = -556+0.00736(Pop)+281.1(PIW)
45
Intercity Outbound Vehicle TripsVehicle trips = -417+.00139 Pop
Intercity Outbound Goods FlowsFlow in tonnes = -3242+.00913 Pop
Intracity Outbound Goods FlowsFlow in tonnes = -3167+.0077 PopTonne Kilometers = -46109+.0846 Pop
46
Commodity Flow modelsIntercity Inbound flows
47
Outbound flows
48
Intracity flows
49
Mode Split Models
• Regression analysis was conducted withpercentage of consignments of a givencategory, going by a designated mode, asdependent variable and the size ofconsignment and length of haul asindependent variable
• Linear and exponential forms of functions areestablished
50
Linear ModelsPCS 1 = 96.33 - 2.83 LPCS 2 = 85.15 - 4.96 LPCS 3 = 36.42 - 3.94 L
Where,L = Haulage length in kmPCS 1 = % of consignment < .5 tonnes by slow moving VehiclesPCS 1 = % of consignment <1 & >.5 tonnes by slow moving VehiclesPCS 1 = % of consignment <2 & >1 tonnes by slow moving Vehicles
51
• Validation of Demand Models
Data generated in Hyderabad is used to validate the models. Estimated and observed trips are matching closely
52
Strategy for goods transport facility planning
• Quantum of goods categorized commodity wise must be known for planning terminal facilities more rationally and efficiently
• Standardize the total requirements of various types of commodities and flows for given population and percentage of industrial workers in city as it is cumbersome to use large number of models practically
53
Facilities Required In Goods Terminals
• Docking and loading/unloading area
• Fuel and servicing stations
• Office spaces
• Parking spaces
• Covered storage space for handling goods
• Lodges and dormitories
• Shops selling motor spares
54
Spacing Required
• Assuming that loading / unloading takes placefrom 8.00 am to 8.00 pm and loading /unloading time per intercity truck is 2 hoursand for intracity truck is one hour
Number of spaces = (Number of trips/Working hours )*Timerequired for loading
55
Time series techniques for forecasting truck traffic
• Spectral analysis which include line spectrum and power spectrum are required to check whether time series models are suitable for the data sets
• Two types of models – ARMA and ARIMA
1.Auto Regressive Moving Average
2.Auto Regressive Integrated Moving Average
56
• Auto Regressive Moving Average Model
2
1
1
1
)()()(m
j
j
m
j
jt twCjtwjtyu
57
• ARIMA model is given by
2
1
1
1
)()()(m
j
j
m
j
jdttt twCjtwqjtyfyyu
58
Where,
Y(t), t=1,2,…n is the series being modeled
M1 is the number of AR parameters
Fj is the jth AR parameter
M2 is the number of MA parameters
Qj is the jth MA parameter
C is a constant and
{w(t)=1,2,…n} is the residual series
59
• Model selection is based on two criteria
1.Maximum likelihood rule
2.Minimum mean square error
• Model is validated for the data available in Mumbai Metropolitan region
• Utilizing time series models weekly truck traffic on national highways 3,4 and 8 and L.B.S road at Vashi, Mulund(E), Dahisar and Mulund(W) in Mumbai Metropolitan region were modeled
60
• Complexities associated with the urban goodstransport analysis and modeling do not permitcomprehensive planning of the transportsystem
• Development of planning methods is inhibitedby the absence of urban goods movementdata and the inadequate knowledge of theurban goods transport needs