**1. Introduction**

The Travel Time Information System (TTIS) was implemented in Poland in 2012 with the aim to solve the problem of seasonal congestion of the road network in the recreation area of the Malopolska region. This complex Intelligent Transport System (ITS) covers both national and regional rural and suburban roads. The aim of the TTIS is the e ffective exploitation of capacity reserves existing in the road network by providing information to road users about alternative routes with a lower tra ffic density and shorter travel time.

The tra ffic redistribution on the road network not only impacts the environment (fuel consumption and an increase in pollution) but it can a ffect in a not negligible way the safety of the road users. The common problem in the implementation of this kind of complex system is that the travel time (measured as tra ffic volume in relation to the infrastructure capacity) is the only factor in the decision about rerouting tra ffic instead of considering multiple criteria [1]. The present research aims to analyze the e ffects of the TTIS system on road safety and travel time by verifying the overall safety and tra ffic operation performance of the road network, including national and regional roads and their intersections where tra ffic is rerouted. The road safety assessment was carried out by calibrating Safety Performance Functions (SPFs) for national and regional roads (with a greater density of access points), for both sections and intersections. The use of SPFs calibrated on the basis of empirical data o ffered the advantage to predict the safety conditions of the whole road network included in the TTIS for di fferent values of rerouted tra ffic. Furthermore, by assessing the safety level of each road section and intersection as a function of the Annual Average Daily Tra ffic (AADT) and geometric parameters, it was possible to simulate in real time the network performance in terms of road safety. This was possible since the e ffects of the TTIS system did not change the risk related to the di fferent road categories and

intersection types (the estimated crash modification factor for the TTIS was close to 1, as shown in [2]). In fact, by rerouting tra ffic, the TTIS a ffects crash frequency in non-homogenous road categories and intersection designs. A comparison of the tra ffic operation and road safety was also assessed, looking for the optimum of two measures of system operation, changes in crashes and travel time.

This paper is a further step to develop a preliminary analysis on the safety performance of the network when tra ffic is rerouted between paths with di fferent road segmen<sup>t</sup> categories and characteristics [2]. The mentioned study has high reliability in the methodology and data but was conducted only in terms of road safety for road sections, while intersections were not included in the analysis.

## **2. Literature Review**

The TTIS, recommending alternative routes on the basis of travel time, is commonly used. It improves the level of service and may have controversial e ffects on road safety on the main regional and national road networks based on the magnitude of the rerouted tra ffic [2]. This preliminary study [2] considers the e ffects of the system by developing simulated scenarios of tra ffic redistribution on the network. The main problem of those applications is that the TTIS may also influence the change in tra ffic at intersections. It may lead to a greater number of crashes than those related only to road sections, based on the di fferent risks related to the specific crossed intersections [3].

The added value of including intersections in the overall analysis of the real-time safety conditions of the network is due to the fact that drivers may more frequently change their travel routes, preferring routes with low priority and a shorter travel time. This may result in a greater number of dangerous maneuvers at intersections and consequently a greater probability of multiple vehicle road crashes.

The TTIS is more often implemented in urban areas than on rural roads. For example, such a system operates in the Norwegian city of Trondheim [3] and Hong Kong [4]. A similar experience was carried out in London, and showed an increase in the number of crashes in connection with the increase in the proportion of vehicles equipped with connected on-board tools for rerouting [4]. Assuming a total share of vehicles equipped with on-board tools to be 100%, the costs of road crashes will increase by 1.5% [5]. Other studies sugges<sup>t</sup> that the distribution of tra ffic in the suburban road network, relying on the shortest travel time while still maintaining an acceptable level of service, led to an increase in the risk of crashes (considering a non-linear relationship between crashes and tra ffic volume) [6,7]. In all of those systems, the basic principle is to reroute tra ffic in the road network to minimize the delays (travel time) of users. Based on the real-time tra ffic volume, the system calculates the tra ffic performance and gives information about alternative routes to users, often without considering the impact of rerouting on future tra ffic conditions [8]. Research [9] found that driver route decisions depend not only on travel time information, but also on route scenery, the number of intersections, and tra ffic signals along the alternative route. The mentioned approach also lacks knowledge about the safety conditions of the network, which require a specific analysis based on the risk level related to road characteristics and intersection types.

The analysis of the e ffects of route recommendations on accident risk in urban networks [8] indicated that accident reductions, resulting from a more e fficient distribution of tra ffic in congested networks, are small. The use of minor roads can reduce travel time, but at the same time can increase the accident frequency.

The variation of network-wide accidents caused by tra ffic redistribution, subject to various levels of dynamic route guidance, market penetration, and the potential of a new safety-enhanced route guidance system based on di fferent levels and pattern simulation [10], showed approximately a 10% increase in accidents.

The available worldwide experience and research sugges<sup>t</sup> the need to take into account not only the tra ffic performance but also an assessment of the safety performance on the alternative routes in the road networks covered by TTIS [11–13]. Furthermore, a reliable balance or comparison between tra ffic operation and road safety was never carried out for the existing TTIS for rural two-lane roads.

The aim of the paper is an assessment of the road network covered by the TTIS in terms of road safety and tra ffic operation, considering a dynamic tra ffic distribution. The paper proposes the next stage of the study of road safety on roads included in the TTIS [2] with consideration for road safety and the impact of travel time at intersections as part of the network. Therefore, for road networks included in the TTIS: 1) a detailed systematic road safety and tra ffic operation analysis has been done, considering the predicted number of crashes on the basis of ad hoc calibrated SPFs, 2) tra ffic data and their monthly variability were considered by getting data from the measuring devices implemented in the system, and 3) the relation between tra ffic volume and speed was developed by the authors based on empirical data from the TTIS.

#### **3. Travel Time Information System for Rural Roads and Data**

The aim of the Travel Time Information System (TTIS) implementation was to improve tra ffic performance (reduction of travel time), by rerouting tra ffic in the road network covered by the system between two tourist sites, Zakopane (Z) and Rabka (R), in a recreational region in Poland (Figure 1). In Figure 1, the possible routes in the TTIS and the abbreviations of town names are presented.

**Figure 1.** National and regional road network included in the analysis and alternative route identification.

#### *3.1. Structure of the TTIS*

To collect tra ffic data (tra ffic volume as well as travel and spot speed), the TTIS consists of a series of devices and sensors, i.e.,:


The data collected by the sensors are used to estimate travel times from one location to another for di fferent sections of roads (RMTSs) and routes (ANPR). The driver can select a route based on data on travel times provided via a VMS (Figure 2), an internet website, or a mobile app. As a result, the tra ffic

volume distribution may vary depending on the system recommendations and the drivers' decisions to change route between the cities of Zakopane (Z) and Rabka (R) (Figure 1).

**Figure 2.** Variable message signs for travel time of the Travel Time Information System (TTIS).
