Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Standard view
Full view
of .
Look up keyword
Like this
0 of .
Results for:
No results containing your search query
P. 1
49 Heinitz Meincke-systematizing Routing Options Air Cargo

49 Heinitz Meincke-systematizing Routing Options Air Cargo

Ratings: (0)|Views: 185 |Likes:
Published by W.J. Zondag

More info:

Published by: W.J. Zondag on Jan 06, 2013
Copyright:Attribution Non-commercial


Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less





e-zine edition 49
Systematizing Routing Options ina Global Air Cargo Network Model
This article reports on advances in building air cargo network routing software. This soft-ware module is an integral component of a multi-level air cargo supply-demand interactionmodel. The model is aimed at analyzing and forecasting airborne commodity flows on a glo-bal scale. Having a comprehensive overview of the routing options is essential for assign-ing air freight in networks in real-time as much as possible. Our modeling deals with cargo“alliances” and sub-networks defined by interlining agreements. In the absence of publiclyavailable data, we develop a route typology, as well as a methodology for subsequent choiceset formation. Itinerary level observations and preference data act as yardsticks for thisexercise. We demonstrate how to address the relevant spatial-temporal routing optionsfor cargo within a maximum range of adjustment strategies, while keeping computationalcomplexity manageable.
The conduction of quantitative research on air-freight-relatedtopics presupposes an operational model that describes cargoairline operations by a calculation of the relevant supply-and-de-mand pattern. Several large-scale freight models – e.g. WorldNet(CEC-DG VII 2009) – describe air cargo as part of overviewedaggregated commodity
. For air-cargo speci
c investiga-tion purposes, it is essential for analysts to have a more detailed,all-embracing and machine-readable capturing of the suppliernetwork structures, and their actual utilization by airborne com-modity
ows when centering e.g. around the
optimization of cargo airport infrastructure and curfew reg-ulations; 
market concentration analysis and forecasts, both globallyand regionally; 
selection and dimensioning of pre-
ight security measures;and 
internalization schemes of external effects.
by: Florian M. Heinitz and Peter A. Meincke
KLM Boeing 747-406F/ER/SCD being loaded, photo courtesy of Capital Photo (for KLM)
2The question of route choice in particular, i.e., the path actuallytaken by shipments, raises a central modeling issue in the
eldof air cargo. Common methods of ground transport planning,including optimal path search and network assignmentalgorithms, are increasingly used. In order to keep the overallmodel size manageable, one needs to
nd a reasonablecompromise between explanatory power, the granularly of observation data, and the desired level of detail.Re
ecting the routing options is not a straightforward task forvarious reasons. First, the connections between the O&D’s aremanifold and differentiated: Main-deck freighter operationsare complemented by a multitude of belly freight transports onboard of scheduled
ights – mainly within the FSNC alliances,but also with non-aligned carriers. Second, cargo charters are of-fered in addition, especially during peak capacity periods and forexceptional goods or
ight routes. Last but not least, general airfreight transport within the integrators’ networks, as a by-prod-uct of mail and express services, has been gaining importance.To connect smaller stations or provide door-to-door services,the main leg (or city pair connection) covered by air transportis typically framed by earthbound pre-carriage and on-carriage.This leads to a variety of conceivable transport chains and thusa complex choice situation faced by forwarders, which accountfor some 80% of the total transported airfreight. Unfortunately,the underlying expert knowledge of decisionmakers is seldomrevealed and is only to a minor extent included in textbooks or journal contributions.Based on this insight, a self-synthesized proxy might reproducethe choice sets at a suf 
cient quality, giving direction to the ac-tual capacity utilization and macroscopic visible airline networkload pattern. This systematization approach in the context of implementing a network routing software aims to get closer tothe alternatives that are available to airline decisionmakers, andillustrate policy sensitivity.
This approach is based on overviews such as GRANDJOT(2007) and PETERSEN (2007), which give considerableinsight into the activities within air cargo supply chains, marketstructures and the nature of air freight itineraries
. Besides thetechnological aspects, BUTTON and STOUGH (2000) explainthe (local) market concentration level and the formation of aircargo networks also as an outcome of the interplay of regulationand deregulation processes
. Further useful sources of information are case studies such as GENSEL (2005), discussinga forwarder’s perspective for the air transport chain
. Theliterature also provides well-established methods of coveringthe heterogeneity of decision-making, leading to quite differentspeci
cations of transport models. The main approaches are:
An a-priori demand segmentation (e.g. ARIMOND andELFESSI, 2001
A procedure for determining the optimal perceived choiceset sizes of business buyers (e.g. KAUFMANN and POP-KOWSKY LESZCZYC, 2004
The “narrowing down” to a shortlist of a few remainingoptions through a constraint-based choice set formationprocess (e.g. BEN-AKIVA and BOCCARA, 1995
); and
The search for the adequate functional form, in terms of ex-planatory variables entering the utility function and the useof statistical transformations (e. g. GAUDRY and WILLS,1979
).In what ways can these methods, to some degree, be adopted inpractice?Starting with a broad, assumption-based set of options availableto the decisionmaker, the task is to devise numerous routing op-tions and narrow down the result list to the ultimately relevantchoice set while keeping the information retrieval costs withinrealistic bounds.Taking a look at the real world supply-side information sources forrouting and schedule enquiries, one ob-tains a
rst set of criteria for identifyingthe offer most suitable to the customerneeds. Table 1 summarizes the requisitedata queried by the web user interfacesfor some of the leading cargo airlines.These speci
cations always include thecity pair codes and preset time frames,that is, latest acceptance time (LAT) and/ or time of availability (TOA). In somecases, a selection of the actual operatingcarrier, the necessary equipment, and thechoice of a branded solution are queried,as they restrict the set of eligible itiner-aries. Except for a minority of carriers,integration with the route search enginesseems to be missing. In other words –the abovementioned speci
cation doesnot affect the resulting list of itinerarieslater put on display. As demonstrated bythe web user interfaces, timetables andlevels of service may be evaluated forthe single airline, for “not-aligned” car-rier, or at a “cargo alliance” level, e.g.for
, pillared by the AF/ KL joint venture.
Table 1 :
Air Cargo Routing Options Queried by A/L Web Portals
3The four previously mentioned methods are now arranged asfour successive stages of a choice process: Demand segmen-tation, choice sets’ size determination, choice set formation/ quali
cation, and choice. Note that the full set of these stagesis needed to fully cover the complex supply side, the needs of various air cargo consignment types, and the resulting multi-dimensional space of conceivable itineraries.The ways to handle this problem at the air network topologyside include the intentional creation of a hierarchy of trunk lines,feeder lines, and ground access lines, making sure that the sys-tem relevant paths are taken within that synthetic timetable net-works.This paper focuses on the scope of model designs beyond theaspects of the network topology. We presuppose real-worldtimetables such as the OAG database
. Table 2 overviews themodeler’s degrees of freedom when executing the four stages. Inthe following, the possible model speci
cations will be exploredand their explanatory content will be analyzed – process stage byprocess stage– in more detail.
Demand Segmentation
In contrast to the relative uniformity of passengersconveyed by air, the nature of goods varies widely inair freight. It makes sense to differentiate between thetypes of commodities that need to be transported inso-far it is of any relevance to the route selection. This canbe achieved by a breakdown of the origin-destinationtransport volume by commodity groups and in terms of time criticality. The segmentation should be carried outat the beginning of the modeled routing process in or-der to submit the speci
c requirements to the succeed-ing process stages. Once this classi
cation is applied,constraints will be imposed on the search for potentialitineraries, ensuring the respective deadlines or the ex-clusion of hazardous goods.
Choice Set Size Determination
According to ORTÚZAR and WILLUMSEN (2001),the delimitation of options in large choice sets is a cru-cial point in the model speci
cation, leading to a searchfor a trade-off between relevance and complexity
.There are two main dimensions constituting the choiceset: City pair options (a) on the one hand and the vari-ety of suppliers (b) serving that origin-destination pairon the other.a) City Pair OptionsIn contrast to passenger airlines, in air cargo one maysuppose a high geographical variability of the actualdeparture, transfer, and arrival airport. Also, unusualitineraries should be considered as long as the servicelevel agreement to the customer can be met and the air-line’s income statement is still positive. The extensionof catchment areas for both origin and destination isan arbitrary decision taken by the modeler. For smallerstates by land area, harboring major hub airports suchas Luxemburg (LUX), the cross-national assignmentsto traf 
c zones may far outnumber domestic traf 
c.The geographical mapping of transports originating orterminating at a traf 
c cell to “neighboring” airportsthus opens a choice dimension. Particularly for con-nections between areas of a high airport population,this may yield a multitude of possible itineraries. Toillustrate the city pair options generated by the two in-dependent sub-dimensions for origin and destination, one canimagine plane grids, as depicted in Figure 1.b) Supplier OptionsA second choice dimension arises from the existence of oligopo-listic origin-destination markets. The expected choice set size isroughly given by the inverse of supply side market concentra-tion from origin
destination cells’ point of view. Networkalliances, grouping several cargo airlines, as well as notable“not-aligned” carriers of local presence are regarded as the pri-mary market players, accounting for most of the route capac-ity. According to a brief analysis of popular origin-destinationairfreight markets, this should keep the number of options inthe single digit range, if the residual capacity, represented by aassumption of marginal market actors, is neglected.Note that the reproduction of global network alliances is not justa methodological means of choice set delimitation. Althoughwe see cargo alliances expand and decrease over time, the as-sociated advantages such as market coverage, extended network
Table 2 :
Systematization of Air Cargo Route Choice
Figure 1:
 Route Choice Set Size Determination Assuming Oligopolistic Markets(Source: Own Representation)

You're Reading a Free Preview

/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->