Geogracom 5W – an expert system for sustainable urban and regional transport development

Abstract

This article sets out a basic approach to the problem of strategic territorial planning through transport network re-organisation. We describe the family of the expert systems designed by a company known as Geogracom. These systems begin by observing the experience of highly developed and developing countries and then formulating a set of indicators on the basis of ecological, economic, social and geopolitical parameters of territorial development. Such indices are used to derive Minimal Transport Standard (MTS), whose aim is more effective transportation. Therefore, the primary task set for the expert is to increase quality of life. Effectiveness of transport functioning is of second priority. At the same time that the degree of achieved closeness to MTS parameters is being defined, alternative well-substantiated programmes of effective development are also being suggested. Then, the volumes of investments necessary for the development of the transportation network are calculated, and the sources of programme financing are identified.

1. Introduction

Since 1986 when P.Bonsall and H.Kirby (1986) drew attention to the problem of using expert systems in the transport sector, no serious attempt has been made to create an expert system for strategic transport planning. There were many systems for tackling specific sector problems such as optimisation of flows (e.g., EMME/2, MVA), effectiveness of road projects (HDM-IV), road safety and preservation of the environment (Fergusson & Ross, 1992). Moreover, quite a few systems have been designed for dealing with concrete problems of traffic organisation on transport routes, especially in a city. These are Intelligent Transport Systems, but they solve mainly technical problems of traffic organisation.

Hence there is a need for strategic planning expert systems that address the re-estimation of a fundamental understanding of the role of transport in market economy. A lot of traditional views, approaches, methods, and criteria for measuring transport effectiveness changed drastically with the advent of market economy. Before then, transport was based on determination of volumes – the achievement of transport activity. Certain systems for transport development planning were oriented not so much towards the transport service users, but towards the providers of transport services. Moreover, it often happened that interests of different modes of transport were in conflict. That is why the problem of risk reduction due to transport network improvement appears to be of extreme importance. Its solution defines the economic and social development level and ecological situation in a region. Thus, the achievement of a desirable level of social and economic conditions in a region becomes the task of the strategic planning of any transport network development.

What are the indices of end transport service consumption? We believe they should reflect living conditions and the economic situation in a given region. Collectively, they can be called a Minimal Transport Standard (MTS).

Table 1. Indices for a Minimal Transport Standard (MTS)

№	Indicator	The best current rates in some countries	Assumed value
1	Transport share in total pollution, %	20	20-40
1	Share of motor transport in total transport pollution, %	60	60-80
2	Transport network reliability (level of transport accessibility), %	95	80-90
2	Share of roads, %	80-95	80-90[1]
3	Level of transport discrimination of population, %	0	0-10
4	Free time lost, (hours per person-week)	0	0-2
5	Accident level due to bad roads, (per 100000 trips)	0.5	0.8-1.5
6	Freight capacity of economy, thousands (km/1 USD of GDP)	0.1	0.5-2
7	Annual population movement for social and cultural purposes, (person-km)	0.9 of diameter of inhabited territory[2]	0.5-0.8
8	Ratio between infrastructure and transport costs ( all modes), %
	a) infrastructure of regional transport	60/40	65/35
	b) infrastructure of urban passenger transportation	50/50	50/50
9	Share of public transport in passenger transportation, %	30	50-70
10	Level of muscular transport (e.g. bicycles) development in urban and suburban traffic, %	35	1-10
11	"State" profitability of transport modes (% net contribution to GDP)
	a) road economy	2	0.3-0.7

The first two indices, as well as the tenth, describe the ecological situation in a region. Our wide range of indices indicates different transport shares in environmental pollution as observed in a number of countries. At the stage of network formation these indicators are taken into account before making decisions about improvement of road pavement, alternative sources of energy, mileage reduction and creation of car-competitive transport facilities.

The index of transport discrimination of population (indicator 3) shows what percentage of the population lives outside the zone of normative accessibility, which means they are not provided with the acceptable volume of transport services.

The ratio of capital investments (indicator 8) reflects tendencies in the world transport policy to give priority to investments in infrastructure. Such an approach is more effective from the point of view of sustainable development since derivative indices resulting from investments in infrastructure, like connectedness and accessibility, have a long-term character and will be used by future generations.

The freight capacity of the economy (indicator 6) has a tendency to decrease in developed countries and share of public transport (indicator 9) depends on density of population. For example, in developing countries with low population density, public transport share should be considerably bigger than the share of individual motor transport.

The notion of Integral Transport Accessibility (ITA) goes together with the rest of the indices. ITA is defined as time required for reaching a certain place from any other place within a given territory, taking into account technical and topological variability. Technical reliability of transport communications allows transport system users to travel along road sections with a desired speed. Topological reliability is understood as the possibility to keep the network in service when some of its sections are out of order. Bougromenko (1996) explains that ITA is a key notion because it characterises in general the ability of a transport network to change the social and economic environment. ITA also helps to estimate the size of population living outside the zones of standardised transport accessibility (degree of transport discrimination of population), as well as free time loss.

Because our indices differ from conventional ones ([4], 1996), we believe that MTS reflects orientations of society to values that can be achieved through a long-term outlook (10-15 years) with due regard to economic, social, political, ethnic and other peculiarities of a region. Range of values for long-term indices is defined with regard to GDP per capita and its growth rate.

In order to calculate MTS indicators it is necessary to keep in mind certain factors that influence the choice of a strategy:

1. Present level of regional development (for example, it can reflect transport network backwardness);

2. Potential of regional development (defined by total prospected for reserves of power, timber and mineral resources) or percentage of population with higher education;

3. Strategic planning period set for regional development.

Figure 1. Input of the initial data for calculation of MTS indices

Fig. 1. is an interface fragment of the first part of the expert system "Geogracom 5W". It illustrates input of initial data needed for calculation of perspective MTS indices.

2. Functional structure of the expert system for strategic transport planning

Systems for strategic planning of regional development "Geogracom 5W" and of urban space "Geograd 1W" are file-server multi-user systems. The functional effectiveness of the system was greatly increased with the introduction of up-to-date GIS technologies and unification mechanisms of data actualisation that make it possible to represent current and predicted spatial and transport infrastructure condition on a cartographic background. The system consists of the following components (Fig. 2):

- Geoinformational system (GIS);

- Database management system (DBMS);

- Knowledge base and output mechanism;

- Functional core;

- Report generator.

Figure 2. Generalised structure of expert system and data flows

Integrated GIS allows downloading, uploading, viewing of vector and raster maps; creation and editing of transport network objects; search and visual selection of map objects; formation of queries to subject tables; visualisation of resulting information using GIS tools. There is also a special GIS-overlay function in the system. It simplifies the decision-making process at the stage when we receive comparative characteristics of the existing transport network and its transformed variant.

The relational database is accessible through a SQL Server. It provides an effective access to various formats of informational object presentation. It includes the following units:

- Database containing cartographic information with spatial geometrical objects (points, lines, polygons);

- Databases that can be queried by users;

- Database containing construction and exploitation costs and norms for calculation of investment programme effectiveness.

DELPHI was chosen as the software for system realisation. This choice was made because of the combination of high-productive communications of applications with a SQL Server and high-speed calculations, easy adjustment and programme testing, availability of a wide range of components and possibility to support objects created with the help of other programme languages. The expert system requires a PC running Windows 95/NT. The following is the recommended configuration:

- Pentium I-II processor;

- 32-64 MB RAM;

- 200 MB min. hard disk (for the server).

The time needed to obtain end results depends on the network size that is determined by the number of vertices. Complete analysis time (PII-400 processor) for 500 vertices is 4 min, for 1000 vertices – 25 min.

3. Network Development

It should be explained that the indicators mentioned in the introductory part are not attributes of the network configuration itself. We focus only on what proposals can bring the existing network to a given level of development that meets target indices of strategic planning? More exactly, systematisation of the empirical data obtained from transport networks in a number of counties has led us to the conclusion that there exist certain spatial and attributive parameters that are characteristic of point and linear elements of the regional transport system. Besides, analysis of the algorithms for transport network transformation has revealed regularities found in the network elements. Thus, network optimisation comes down to a set of flexible rules executed in certain sequence in order to avoid over-planning which can cause return to the starting point where the choice of wrong series of rules was made.

Accordingly we will use the terminology of graph theory to study spatial and attribute parameters of graph vertices (settlements, crossroads, etc.) and graph edges (sections of motorways, railways, waterways, airlines, etc.). The main parameter for the graph vertices is vertex rank, which is in turn dependent on administrative and economic functions of a locality identified as a vertex and its weight (population size or volume of freight dispatched). Each vertex rank, or vertex step, is characterised by a number of out-coming edges, their normative codes, technical speed of freight and passenger traffic along these edged, ITA. The graph edges are also marked by special code that characterises network sections from the point of view of their technical reliability. For example, the set of parameters for motor roads includes the type of pavement and its condition, gradient, width of carriageway, dangerous sections, edge load, frequency of passenger communications.

Thus a prospective network for a given region is designed in the following way:

1. Vertex ranks are defined according to the information about current population size or volumes of freight transportation, forecasts of their growth and setting up of new settlements;

2. A skeleton -- a regional network at the minimal spanning tree -- is found;

3. The graph edges connecting the vertices receive normative codes in accordance with their ranks and with due regard to their degree of spatial influence;

4. Parallel or additional chains are added to the minimal spanning tree following the rule of "Creation of Additional Ways" (introduction of toll roads can be considered);

5. For edges representing railways, dangerous sections and speed limits are eliminated, new directions are added, railways are transferred to electric traction, and an automatic block system is set up;

6. For edges representing waterways, dangerous sections are eliminated and dock characteristics are changed;

7. For edges representing airways, airport classes and runways are changed;

8. The number of out-coming edges is checked. Additional ways are joined according to the rules expressed by the "IF-THEN" condition;

9. After intermediate calculations have been made for re-organised transport network, the degree of disparity between the groups of MTS indices is determined. Accents are placed then, and chains of rules with changed factor thresholds are chosen. Development of alternative transport modes is envisaged in order to ensure ecological safety. Elimination of peripheral points is performed for the transport provision indicators. Construction of capital bridges, overpasses, as well as elimination of so-called "Steiner Bridges" (topologically unreliable network sections), is considered.

10. Indices for an ideal network are calculated. Usually these will exceed MTS indices. In order to bring MTS indices to the standard an all-round algorithm is applied to re-code some of the already transformed edges and those remaining.

The size of this article does not allow a more detailed description of our network transformation algorithm, but elimination of "Steiner Bridge" performed at step 9 of the algorithm will be adduced here as an illustration of the complex decision-making processes used. A "Steiner Bridge" is an edge whose load coefficient is the ratio of the number of the shortest ways to the total number of routes from each network vertex to every other vertex, and whose bypass exceeds the thresholds. This edge is considered potentially dangerous, as its failure will produce global catastrophic consequences. Elimination of this danger will increase network design (topology) quality, as well as it will guarantee reduction of possible losses.

The use of empirically obtained rules for determination of the priority order for network transformation proposals will spare you the above-mentioned procedure of planning the order of the proposals for network transformation. The user is supplied with the table reflecting the priority of the proposals, which is defined on the basis of GDP per capita and the realisation time of the transport network development programme.

Figure 3. Fragment of Geogracom 5W expert system

Fig. 3. is a fragment of the expert system interface. It shows two stages, the first being the bottleneck diagnosis and the second being the generation of proposals. Not only the target region (county, province) development programme is formed following the above-mentioned algorithm, but also programmes for all of its territorial subdivisions are outlined. This approach allows us to define the interests of inferior administrative units, to compare them with those of the upper level in order to strike a balance between them. As a rule, proposals for regional and sub-regional network development are the same for 20-50 %, and it seems impossible to satisfy the interests of all sides involved. One of the possible solutions is to delegate the administrative functions to those vertices that meet the optimal network topology. At the end the system will output a number of alternative programmes for network development in a region and in each of its subdivisions.

4. Investment programme and financial provision forecast

The economic expediency of proposals was not taken into account when the procedure of spatial development of transport network was discussed earlier. Road construction and rehabilitation were planned in order to achieve MTS indices without consideration of financial limitations. These limitations are used at the next stage when investment programmes are drawn up.

The following initial data are used for elaboration of investment programmes: tables that contain work cost of road, bridge and overpass construction, their rehabilitation and maintenance; coefficients of work cost change necessitated by climate conditions, altitude above sea level, seismic activity in the region, influence of big cities; regional unit prices; per-unit costs. The amount of returns received in the region (or in each district taken separately) due to change of ITA index is introduced or given from a reference book. On the basis of prediction of spatial and economic transformation of the region the proposals are put in the necessary order within the frames of the following alternative programmes: purely economical, social, safety programmes and agreement programmes. In economical programmes priority is given to the proposals whose outlay will be covered in a short period of time. In social programmes proposals are arranged on the basis of free time increase and improvement of living condition indicators achieved as a result of cut of non-production time spent on obligatory trips. In safety programmes proposals are arranged on the basis of reduction of the factors affecting traffic safety. Alternative programmes are presented in the form of reports with priority order of proposals, their costs and values of decision-making criteria. In agreement programmes multitarget assessment is made with the help of hierarchical a decision-making method. The most rational decision about the order of proposals directed towards regional improvement is chosen on the basis of aggregated weights. On the level of programme realisation experts are free to determine weights themselves.

On the basis of the agreement programme, the system builds regional transport network development scenarios that help to estimate annual financing volumes, volumes of network maintenance, rehabilitation and development, current project costs as for returns and outlay, forecast of fuel consumption by all users of the network for each transport mode, unemployment reduction, and other macro-economical parameters.

The profitability index and net costs are used to compare different variants of the project in order to select the optimal one. Histograms help to assess investment attraction of the project with different financing volumes and associated risks.

Figure 4. Dependence of annual effect upon financing level

Figure 4. illustrates how the annual effect depends on the financing level (the so-called "road curve"). With the help of the method of spline approximation points marking the bends of the curve, or critical values (e.g. the point where the curve crosses the abscissa), are found. A reasonable financing level for road network development is determined with the help of this chart (see fig. 4):

- up to point 1 there is practically no change in the losses caused by road network underdevelopment;

- in the interval lying between points 1 and 2 losses per year are bigger than annual effect, and creation of an optimal transport network is not feasible with the given financing;

- in the interval 2-3 a net economic effect will be obtained;

- beyond point 3 there is no sense in further increase of financing from economic point of view.

5. Features of the approach to urban strategic planning

Our approach to strategic planning directed at improvement of life quality in a city by means of transport infrastructure development has many distinctive features. Some of the MTS indices are changed and supplemented. Transport mobility of the population is measured in passenger-kilometres per one inhabitant, and it reflects the number and length of trips within a city. Values of muscular transport index are changed since there are more pedestrian and bicycle communications. The index of transport share in the environmental pollution is also recalculated after the noise level has been included into the ecological parameter. A new index - level of transportation convenience and comfort (passengers per square metre in the public transport)- is added. The ITA index, which remains the basis for calculation of transport discrimination of population, free time lost fund, and road reliability, is considered with and without time needed to get to a stop, travel time, change time, and time spent at a stop.

The procedure of defining target urban MTS indices has a specific character. Contingent of passengers, their purposes, and trip structure can be defined provided we know real income per capita and its alteration rate during implementation of the urban transport development programme with due regard to social and economic peculiarities of the population, its size and demographic structure (sex, age, social groups). Functions of gravity, functions of preference of transport to pedestrian communications, as well as preference of individual vehicles to public transport, are formulated on the basis of the obtained data. Requirements of each user group for transport service quality (travel time, time needed to get to a stop, waiting for transport, number of changes, traffic regularity, aggregated travel time with regard to biorhythm and comfort) are defined. Then the obtained indices are corrected according to the planning and territorial characteristics and the transport network configuration of a city. The results are registered in MTS as indicators.

Diagnosis and evaluation are performed before modelling of a long-term development programme of an urban transport network, as well as before the already mentioned territorial development strategy. Urban spatial zones with borders of transport districts are traced on a city map with the street-road network graph. This stage is similar to traditional approaches, but the method is different ([5], 1997). The quality of urban space is then analysed. For this purpose it is necessary to define urban spatial non-uniformity, convenience of consumption of services provided by a certain object (e.g. by a shop or chemist's) within a certain territorial unit, level of transport service, spatial organisation of business activities. Already at this stage on the basis of the results of the analysis some recommendations concerning relocation, re-profilasation, and setting up of more service-providing objects are given to system users. A model of mutual correspondences between transport districts is used to build isochrones, as well as to define traffic flows, congestion probability, urban transport service quality. Design values of average section speed, traffic flows, ITA, supply and demand, and provision level as ratio of supply to demand are outputted for each object of urban transport network (UTN).

Network modification is performed after a possible re-location of urban objects. It starts with an independent algorithm of the existing network rationalisation that includes traffic flow re-organisation on the urban road network, changes of public traffic flow intervals due to optimal route scheduling, re-location of stops, vehicle re-stocking and increase in transport pool capacity. UTN diagnosis is made, and if we have not yet achieved the level of desired indices, then we pass on to the algorithm of an ideal transport network modelling. The programme of road network transformation is carried out. It includes construction, rehabilitation and introduction of new routes of urban passenger transportation. Then the algorithm of UTN rationalisation and diagnosis is started. As a result, users receive end network indices and visual presentations of a rationalised ideal network. The expert system has an interactive mode of UTN transformation that allows an operator to change spatial network objects himself.

Development of an investment programme with evaluation of financial effectiveness is the finishing stage of the process.

Conclusions

The transition to a market economy in Russia caused some new issues to appear. These issues in their turn happened to be pre-requisitions for the creation of methods of their solution with the help of up-to-date technical equipment. Under the new conditions a number of social, economic, demographic, ecological factors should be taken into consideration before approaching the task of strategic planning of a transport network. The Minimal Transport Standard together with the values of its indices calculated on the basis of GDP per capita and GDP growth provides, as we believe, the most reliable combination of indices for working out a strategy of territorial development in the areas dependent on traffic functioning.

Strategic planning consists of several stages:

1. Diagnosis of the existing transport network, calculation of MTS indices for the current period;

2. Introduction of GDP values, its growth and calculation of indicators for the given period;

3. Network development through application of the rule sequence that will result in network transformation;

4. Generation of alternative programmes for regional development and on their basis elaboration of agreement programmes;

5. Forecast of financial provision of investment programmes and effectiveness of capital investments.

As a result, users can interfere with the decision-making process at any of the above-enumerated stages. They can even transform the transport network by varying initial data all by themselves. Installation of the expert system on widely used PCs made the expert systems designed by Geogracom a useful tool of regional administrators and strategic managers. People in 24 regions of Russia and the CIS employ the expert systems of the scientific and consulting company Geogracom to achieve the goals worthy of the 21st century.

References

1. Bonsall, P. and Kirby, H.,(1986) The Role of Expert System in Transport. // Information Technology Applications in Transport: Leeds University, p. 353-382.

2. Fergusson, E. and Ross, C., PC Software for Urban Transportation Planning. // Journal of American Planning Association – vol. 58, N2, p. 238 (1992).

3. Bougromenko, V., Economic Equivalent of Road Accessibility: case study of Russia. // The Role of Transportation in Economic Development. Proceedings of IRF Asia-Pacific Regional Meeting – vol. 4, Taipei, p. 284-293 (1996).

4. Urban transportation indicators in Eight Canadian Urban Areas. – TAС, 31 p. (1996).

5. Strengthening Ashagabat's Urban Transport System. // Social Assessments for Better Development. Case studies in Russia and Central Asia. – Washington, World Bank, p. 140-164 (1997).