INTERSTATE FREIGHT ON STATES’ ROAD David Gargett Afzal Hossain David Cosgrove 29 September 2006

Background From 1971–1995, the SMVU conducted every 3–5 years, and annually since 1998 Not specifically designed to measure the change between years due to no overlap Moreover, the methodological adjustment in 1998 complicates the use of the data in computing growth rates in road freight

Background (cont.) The ABS warns: ‘Caution must be used when using the SMVU to measure change’ Thus BTRE adjusted past SMVU freight data to be comparable to the current SMVU methodology (BTRE Freight Measurement and Modelling in Australia, Report 112) In addition, a methodology for estimating road freight (1971–2003) for each State and Territory was proposed earlier, based on adjusted national aggregates from the ABS SMVU

Background (cont.) But the SMVU definition of IS freight has a flaw: The amount of tonne-kilometres done by other States’ trucks on a State’s road (Source: ABS) i.e., limited to ‘freight carried by trucks registered in other states on a state’s roads’

Background (cont.) i.e. Includes intrastate freight carried mostly within, say NSW, by trucks registered in Victoria. Excludes portion of interstate tkm performed by NSW trucks within NSW as they head to, say Victoria On both counts, estimates of interstate freight for NSW do not measure what road authorities wish measured

Objectives of Current Research To estimate IS road freight tasks within States [based on O–D matrices] To present a methodology for continually updating the estimates from future SMVUs To provide a time-series of the IS road freight task on States’ roads (1971–2004)

Four Past O-D Road Freight Matrices Published in: (1) BTE Estimates of Australian Interregional Freight Movements (1971–72) (2) ABS Interstate Freight Movement Survey (1980–81) (3) ABS Experimental Estimates of Freight Movements (1994–95) (4) ABS Freight Movements Survey (2000–01)

BTE 1972 Estimates of Australian Interregional Freight Movements IS freight movements undertaken by the various transport modes for 1971–72 First of their kind ever compiled on an Australia-wide basis Road freight estimates were subject to more error than for other modes (i.e. rail, sea & air) May overestimate road freight flows, but the data not adjusted

ABS 1981 IS Freight Movement Survey ‘Business-based’ Collected by a census of approx enterprises ( tonnes or more of interstate road freight movements in a year) Sub-contract arrangements excluded Tonnage data for capital cities and some specific areas by O–D IFMS best and most reliable ABS surveys – no adjustment for IS component

ABS 1995 Experimental Estimates of Freight Movements Carried out quarterly, collecting freight movements by commodity group, mode (including road, rail, sea, and air), weight and O–D Rigid and articulated trucks (3.5 tonnes or more) included Adjusted downwards (factor of 0.87) for overestimation due to double-counting of sub-contractor loads

ABS 2001 Freight Movements Survey Road component based on articulated vehicles Rigid trucks and other commercial vehicles were excluded Adjusted upwards (factor of 1.15) for underestimation due to “missed” trips and no rigids

Estimated IS road freight task (million tkm) (2001) Origin DestinationIS NSWVICQldSAWATASNTACTTotal NSW VIC QLD SA WA TAS NT ACT IS Total

Adapting ABS SMVU Data to an O–D Matrix Basis The protocol for calculating O–D matrices of ABS road freight task for 1998–2004 is: Take a three year average of SMVU 2000, 2001 and 2002 (Table 18, SMVU) Calculate a cell factor for scaling, equal to the 2001 adjusted FMS cell tkm divided by the average 2000–02 SMVU cell tkm Use this scaling factor on that O–D cell in each of the SMVU matrices of 1998 to 2004 A similar operation gives the 1982 and 1985 matrices by using the 1981 O–D matrix in the scaling factor

Estimated IS road freight task (million tkm) (2004) Origin DestinationIS NSWVICQldSAWATASNTACTTotal NSW VIC QLD SA WA TAS NT ACT IS Total

Cell-by-Cell O–D Matrix Modelling Done by interpolating individual O–D road freight task data using a series on tonnages through Marulan (the number of trucks passing through Marulan, multiplied by a load per truck series) Example – actual and predicted road freight tasks (million tkm) between 1972 and 2005 for QLD–NSW.

Actual and predicted road freight task, 1970–2005, Qld–NSW

Data Analysis For the regressions ‘Log’ transformation was used In many cases, a time trend was included, depending on data set Some of the years’ data for some routes were omitted - large variability

Coefficients of regression (selected routes) Route Coefficients InterceptLog MarulanTime NSW-VIC VIC-NSW NSW-QLD QLD-NSW VIC-QLD QLD-VIC

Example of the results of regression interpolation for O–D pairs (million tkm)

Splitting O–D Flows by State To split the tkm for an O–D pair by the state in which it is performed, we use a “fractions by state” table This roughly allocates the total tkm for a specific O–D by the states in which it is performed

Splitting O–D Flows (Contd.) For example, for the SA to QLD, the task should be split by the fractions 0.10 from the origin in SA to the VIC border 0.05 through the north-west corner of VIC 0.58 through NSW to the QLD border 0.27 within QLD

Fractions by States from Origin to Destination SA to Task split by States (fractions) NSWVICQLDSAWATASNTACTTotal NSW VIC QLD WA TAS 0.00 NT ACT 0.00

Next Step The fractions are then multiplied by the O–D cell’s total freight task SA = 0.10 * 840 = 84 VIC = 0.05 * 840 = 42 NSW = 0.58 * 840 = 487, and QLD = 0.27 * 840 = 227 Thus total gives 840.

Road freight task for 2005 by States, calculated using fractions SA to Task split by States (million tkm) NSWVICQLDSAWATASNTACTTotal NSW VIC QLD WA TAS NT ACT

Next Step (contd.) Then the O–D task components are characterised as either ‘from’, ‘through’, or ‘to’. Thus in our example, SA ’from’ = 84 VIC ‘through’ = 42 NSW ‘through’ = 487, and QLD ’to’ = 227.

Final O-D based IS Freight Flow Estimates Summing the O–D component tasks gives an estimate of the amount of different types (‘from’, ‘through’ and ‘to’) of interstate freight being carried on each state’s roads The last column gives the new O–D based estimate of total IS freight on each State’s roads

Final estimates of IS freight (million tkm) on States’ roads (5-yr interval) NSWVIC FromToThruTotalFromToThruTotal

Comparison of current and previous estimates Current estimates compare quite well to estimates of total interstate freight published in BTRE Report 112 This previous estimate was based on total Australia interstate freight on the ABS definition, times 1.4 to account for the portion of interstate trips done within a state by that state’s trucks Individual State estimates calculated in this paper are now different, being based on a true ‘state of task performance’ basis, rather than a ‘state of registration by main area of operation outside of the state’ basis

Comparison of current vs previous estimates of total IS road freight

Comparison of current and previous estimates (contd.) This, then, was the solution we sought to the problem posed at the beginning of the paper The new data for each state tells state authorities the growth rates of interstate freight flowing across their roads In addition, it allows them to understand the growth rates of particular O–D combinations This allows a focus on growth along the probable routes the trucks will be taking

Summary O–D matrices derived from 1971 to 2004 Allow for logical control over the definition of interstate freight Analysis method generates levels of total IS freight similar to previous BTRE estimates, but the state split differs

Summary (contd.) First estimates of the concept ‘the interstate freight task performed on each state’s roads’ Also, rough annual updates of the O–D matrix can be derived from each new SMVU Time series estimates can also be used for forecasts of the volume of interstate freight on states’ roads.

Thank you Any question ?