Safety performance indicators for roads: Pilots in the Netherlands, Greece, Israel and Portugal

Deliverable D3.10c of the EU FP6 project SafetyNet
Weijermars (ed.), W.A.M.

This report discusses the results of pilot projects in which the road network SPI was applied. The road network SPI assesses whether the 'right road' is in the 'right place', i.e. it investigates whether the actual road category meets the road category that should be present given the sizes of the urban areas that it connects. The SPI can be calculated following a number of steps. First of all, the urban and rural centres in the area under study are listed and classified into groups on the basis of the number of inhabitants. Second, it is determined which centres should be connected to each other and by what type of road (road category). These theoretically desired connections are compared to the actual connections. The SPI is defined as the percentage of appropriate current road category length per road category.

In order to obtain more insight into the limitations and the applicability of the SPI, it was applied in the following pilot countries: Greece, Israel, Portugal, and the Netherlands. From these pilot projects we conclude that it is possible to calculate the SPI scores. The data necessary for calculating the SPI scores was available in the pilot countries and the different steps of the method could be completed. Moreover, we defined some limitations of the method and some directions for further research in order to further develop the SPI.

First of all, the definition and classification of the centres need attention. Ideally, each urban area that is surrounded by rural area is defined as a centre. In practice, data is not always available on this level. In some cases, a centre consists of separate sub-centres with rural area in between. In other cases, two centres are so close to each other that there is no rural area in between them. Moreover, the centres were classified using the number of inhabitants, whereas in reality other facilities may also generate traffic. From the pilot projects it was concluded that some important existing centres may be excluded or classified to a lower category than they should be in case only the number of inhabitants is taken into account. We recommend to deal with these problems in a pragmatic way; centres can be classified on the basis of inhabitants, using the available data and the resulting classification should be evaluated and adapted by somebody with knowledge of the local situation. All changes should be reported and explained. Another issue is that the limits of the classes are chosen quite arbitrarily. Further research is necessary in order to obtain a theoretically sound classification system.

In general, the lists of connections that were determined for the pilot countries appeared to be realistic. In Portugal however, some connections were not found by the method due to small search areas. In other cases, theoretically desired connections were determined that are not found in practice, due to natural barriers.

Regarding the identification and matching of actual connections, the guidelines were very brief. Each pilot country executed this step in its own way, using a route planner or a GIS tool. In general, it appeared to be possible to select one actual connection and to categorise the roads on the basis of the SafetyNet road categorization. The detour factor of 1.6 that was suggested in the Manual (Hakkert & Gitelman, 2007) was concluded not to be appropriate for all pilot areas, yet it was found to be difficult to define one detour factor that is appropriate in all cases. Therefore, we propose to deal with it in a pragmatic way and to examine connection with a detour factor higher than 1.6 on a case-specific basis. In the pilot projects, we did not take into account the suggestion from the manual to upgrade the theoretically desired road category in certain cases that a road is used by several connections. This issue should be analysed further when improving the method.

The calculation of the SPI is quite straightforward if the preceding steps are applied. In general, the resulting scores seem quite realistic, although the scores of Greece seem low compared to the other pilot areas. One overall score may give a biased view, since it depends on the theoretically desired road network that differs greatly between the case study areas. Therefore, we recommend not to combine the results into one overall score. When the results are analysed in more detail, there appear to be some road categories that have lower scores. The SN requirements for the different road categories possibly should be interpreted less strictly and the theoretically desired road categories between different types of centres possibly need revision. More research is needed concerning this issue. Moreover, in some cases relatively short sections of connections (connecting freeways to urban areas) failed. Maybe one should allow a certain percentage of the route to be of a lower category and only assess the primary part of the connection.

The experiences of the pilot countries were used to update the guidelines for calculating the SPI scores. In our opinion, the current method provides insight into the strengths and weaknesses in a road network. However, especially the classification of urban centres and the SafetyNet road categorization need further research to improve the method for calculating the SPI. Moreover, the relationship between the SPI scores and traffic safety should be analyzed in order to obtain more insight into the validity of the road network SPI.

Gepubliceerd door
European Commission, Brussels


Dit is een publicatie van SWOV, of waar SWOV een bijdrage aan heeft geleverd.