Influences of developments and measures can be studied by applying VCM. Seven scenarios are generated in this work with five in land use planning and two in transport planning. The following findings from these application examples can be concluded:
• The separation of land uses, i.e. work places in the centre and residence in the periphery area, leads to more PuT trips. It confirms the findings in literature such as from LECK (2006). Compared to separation of land uses, the ideal mixture of land uses leads to more car trips.
• Locating the new development in the city area rather than in the region area induces more PuT trips due to good PuT services in the city area. However, relevant facilities such as schools in the city area should be provided accordingly for the increased population.
• Ideal improvement of capacity on roads, i.e. no congestion in the network, induces more car trips with longer distance but slightly shorter time. It also conforms to the research in the literature such as from MOKHTARIAN and CHEN (2004).
• Ideal improvement of the PuT service increases the PuT share significantly. However, this conclusion is only suitable for cities with a relative high density.
These application examples are on the basis of “other things remaining equal”, which ignores the possible interaction between the changed measure and other elements in the model. VCM is applied to model the influence of land use structure and transport supply on travel demand. However, it is not able to model the interaction between land use structure and transport supply. For example, changes in the land use structure would also cause changes in the transport infrastructure, especially the arrangement of public transport lines. Another example of disadvantage of this basis “other things remaining equal” lies in the first scenario (S1), where all work places are located in the central area of a city. In this scenario the increase of parking search time due to limited parking places is not modelled, otherwise it results in even more PuT trips. Therefore, the modelling result of VCM can be interpreted as the moment that one change happens and other changes have not yet reacted to this change.
A VCM is able to represent travel demand of the reference area with a certain validity.
However, different aspects of VCM can be improved in further studies, for example:
• The land use structure in a polycentric city can be transferred by the cross-classification of land use categories, densities and rings referring to not only the city centre, but also to sub-centres. In the current version of VCM, only rings referring to the city centre are considered.
• A database of tile structures with typical network forms and components of zones can be created for the network generator tool. Different concepts of cities can be generated with corresponding tile structures in the database.
• The region area can be modelled on a lower aggregation level with more zones, so that regional planning measures can also be studied applying VCM. For example, the
unrealistically high number of walk trips can be avoided in scenarios S4 and S5 of locating new development in the region area.
• New version of VISUM 15.0 introduces also modelling of freight transport with different structure from person transport by means of generating real tours and trip tables of freight transport. Applying this version of VISUM should be able to model freight transport in a better way than the current VCM does.
More development scenarios and measures in land use and transport planning than scenarios in chapter 5 can be modelled and analysed in VCM. As a potential further application of VCM, a combination of the PuT improvement in S7 and the limitation of car use can be tested. Besides, as it is suggested to adopt an integrative approach for transport planning and land use planning in the literature, the influence of combing measures in land use planning and transport planning can be studied. A measure that can be tested is the high density with mixed-use and restricted car use policies, as suggested by CERVERO and KOCKELMAN (1997).
Another interesting potential further research applying VCM is the comparison of VCMs with different land use structure and transport supply. The methodological processes can be applied to develop new virtual city models, such as with another network form or with a new reference model. An example is the comparison between Detroit Region in the USA and Stuttgart Region, as these two regions have similar characteristics of population and size. Land use data in Detroit Region can be transferred with the methods of cross classification into a new VCM. Similarly a new network model based on the network structure in Detroit Region can also be generated with the network generator tool. For a precise forecast, this network model should be calibrated and validated with the existing data from the Detroit Region. Assuming that all of the residents in Stuttgart Region are living in Detroit Region in the USA, i.e. behaviour data in Stuttgart Region are applied, travel demand of these travellers in the new VCM can be investigated.
Comparing travel demand of these two VCMs, although they share the same behavioural data, the land use and network structure in the U.S.A are supposed to produce more and longer car trips, as shown in differences between surveys in the USA and Germany.
This version of VCM is available for more applications, and other versions of VCM can be generated following the same methods.