Results of the evaluation of pilot tests. Deliverable D4.4 of the SAFER-LC project

Kurzfassung

This deliverable collects the main results obtained from evaluations of the piloted safety measures selected in earlier phases of the SAFER-LC project. This deliverable reports the descriptions of the piloted measures, method and data to evaluate the safety effects of the selected measures, as well as the results of evaluations together with their discussion. More detailed information about the implementation of the measures and execution of pilots can be found from deliverable D4.3 of the SAFER-LC project (Carrese et al., 2019). In some cases, deliverable D4.3 also reports details on the development of the measure. The main inputs for this deliverable from other SAFER-LC activities originate from Work Package 2 (WP2), Work Package 3 (WP3) and earlier tasks of WP4. The earlier deliverables of WP4 produced implementation guidelines for the pilots (D4.1; SAFER-LC Consortium, 2018a) by providing an overview of the major testing environments that were available for piloting in the SAFER-LC project. The available pilot test environments ranged from simulation environments to real (or close to real) traffic circumstances. Deliverable D4.2 (SAFER-LC Consortium, 2018b) describes the proposed evaluation framework including a list of parameters from which the partners could select the most appropriate ones for the evaluation of their pilot. The identified Key Performance Indicators (KPIs) were arranged into five categories: ‘Safety’, ‘Traffic’, ‘Human behaviour’, ‘Technical’, and ‘Business’. Finally, the deliverable D4.3 (Carrese et al., 2019) describes the pilot activities carried out in WP4 by documenting the implementation and execution of pilots in various level crossing environments in different countries. This deliverable reports the evaluation results of 21 safety measures that were piloted at eight pilot sites during the SAFER-LC project. The number of piloted safety measures varied by pilot site and the pilot test sites varied from simulation studies to controlled conditions and real railway environments. In some cases, the selected measures were not suitable for piloting in a real world experimental context and/or the implementation in real railway environment was not feasible, for example, due to financial resources, timing of our piloting period and/or lack of suitable pilot site(s). Therefore, pilot test sites in the SAFER-LC projects varied from simulation studies to controlled conditions and real railway environments. Some of the measures (‘In-vehicle warnings to driver’, and ‘Additional lights to train front’) were tested in two different environments to collect complementary information on their safety effects via two types of installation. Due to the nature of the conducted pilots (small-scale pilot tests), it was hardly possible to calculate any quantitative estimates for safety effects of the measures in terms of annual reductions in the number of LC fatalities and/or accidents based on the results of the pilots. However, since numerical estimates of safety effects are needed for cost-benefit calculations (WP5 of the SAFER-LC project), the authors made an attempt to draw these estimates based on the applicability of safety measures to different LC types, road users and behaviours leading to LC accidents based on pre-existing information on the effects of LC safety measures. The authors acknowledge that many uncertainties are related to these estimates. However, the assumptions used in the calculations are clearly documented and hence the estimates can be easily updated if more detailed statistics or more information on safety effects become available. Therefore, a detailed documentation of LC accident data (information on additional variables and details) is highly recommended to enable drawing of these estimates. Based on the safety potential calculations presented in chapter 5 the piloted measures that were estimated to have the highest safety benefits are: − Additional lights at the train front, covering measures ‘Additional warning light system at front of the locomotive (6.0–12.0%)’ and ‘Improved train visibility using lights (6.0–30.0%)’. This measure was estimated to have rather high effectiveness (prevention of 15–30% of relevant LC accidents) and target rather large share of LC accidents (19.9−96.3% depending on the approach). − In-vehicle train and LC proximity warning (4.4–15.0%). It is important to be noted that the effectiveness of this measure depends on the usage of the in-vehicle devices. In practice, the car driver needs to install the application on a smart mobile device, and location tracking should be enabled on this device while driving. Furthermore, the driver needs to allow the application to run seamlessly on the background and also notice the visual or auditory warning in order to perform the required action on time (e.g. stop before the LC). However, these latter requirements are valid for all LC safety measures. − Speed bumps and flashing posts (2.0–8.0%). This accident reduction estimate concerns the situation where the measure is implemented to passive LCs (where the highest safety effects were expected in Dressler et al. 2018). − Blinking lights drawing driver attention (Perilight) (2.0–8.0%). This measure is targeted to passive LCs. Some concerns on applicability of piloted safety measures in different railway environments are listed below: − Written letters on ground and coloured road marking: Any road marking can only be applied on a paved road with an even surface. Thus, the message written on the road does not hold for road environments such as gravel roads, cobblestone, tracks etc. Furthermore, these measures are not perfectly suitable to countries with snow and long winter with darkness. − Noise-producing pavement and speed bumps: These measures are not well suited to gravel roads. In addition, these measures are not effective in case of snow. − Blinking amber light with train symbol and blinking lights drawing driver attention (Perilight): It is important to note that these measures are targeted to passive LCs and require power. However, in practice many of passive LCs no mains power is available and thus other alternative power sources need to be investigated. The effectiveness of these measures was estimated somewhat lower than active LCs with sound and/or light warning since the warning in these measures is linked to LC approach and not to actual arrival of train. − In-vehicle train and LC proximity warning: This system may not operate satisfactory for LCs surrounded by roads on which Global Navigation Satellite System (GNSS) reception is poor. Overall, the safety effect results of the piloted measures are promising. Therefore, it is recommended that some of most promising measures will be tested in larger scale real world experiments with well-planned research designs to obtain more information on their effects (also on long term) on road user behaviour and thus on road safety. This would also support the more exact numerical estimation of safety effects of the piloted measures. The results of this deliverable will serve as input for WP5 that deals with cost-benefit analyses. The estimates of safety effects of each measure will be used in cost-benefit or cost-effectiveness calculations and the experiences collected during the piloting will support the drawing of final recommendations for the SAFER-LC project.