In the last two decades, roundabouts have become the most common and efficient solution for at-grade intersections [
1,
2,
3,
4,
5]. However, many papers have evinced that standard multi-lane roundabouts offer lower capacity compared to estimates and a high rate of crashes [
6,
7]. High speeds and many potential conflicts are probably the main causes of these problems: drivers accelerate when crossing the roundabout [
8,
9,
10,
11,
12]. For these reasons, a new type of multi-lane roundabout, called the turbo roundabout, was introduced in the Netherlands [
6] in 1996. This scheme was characterized by a spiraling circular carriageway with lanes separated by curbs. The main benefits of these new roundabout layouts are: a limited number of potential conflict points, a slower speed along the circulatory lane, and a low risk of side impacts [
13,
14,
15,
16,
17,
18]. The main negative aspects are due to the presence of through conflict points in left-turning maneuvers, higher values of the critical gaps along the road, and generally lower capacity than conventional double-lane roundabouts [
19,
20,
21]. In this study, we analyzed a virtual egg turbo roundabout as a possible alternative scheme to an existing multi-lane roundabout in Cosenza (Italy). The main goal of this conversion is to evaluate the potential improvement of performance and safety [
11,
22]. We used data acquired throughout the several recordings on the investigated roundabout (traffic distribution, approaching and circulating speeds, queue lengths, approaching delays, time of service, and critical gaps) as input for the calibration procedures of VISSIM [
23,
24]. Then, we employed the calibrated VISSIM parameters obtained from the existing roundabout (minimizing the difference between simulated and field-observed queue lengths) to estimate the spatial distribution of the potential traffic conflicts at the virtual turbo roundabout by using Surrogate Safety Assessment Model (SSAM) [
25] and to compare the results of the two different roundabout solutions.
The advantage of converting traditional roundabouts into turbo roundabouts, both in terms of performance and safety, has been confirmed by several studies. Fortuijn [
6] presented the first safety analysis on turbo roundabouts. The results of before-and-after studies, carried out in the Netherlands on 70 turbo roundabouts, demonstrated an 80% reduction in the risk of crashes associated with injury with respect to conventional single-lane roundabouts. As reported by Mauro and Cattani [
26], turbo roundabouts can considerably reduce the accident rate with respect to standard roundabouts, because they eliminate conflicts between circulating and exiting flows. In this paper, they developed an accident rate model based on the concept of potential conflicts, and demonstrated that turbo roundabouts abate total car accidents by about 40–50% and injury crashes by 20–30%. The same authors [
27] confirmed these outcomes, also underlining the usefulness of turbo roundabouts in an urban area: they used the accident rate model to compare four-leg turbo roundabouts to standard roundabouts in the presence of significant two-wheeler and pedestrian traffic. Silva et al. [
19] studied, in terms both of capacity and pollutant emissions, the performance of the turbo roundabout solution, when applied in corridor, compared to a conventional double-lane roundabout. The results showed a level of performance of the turbo roundabouts corridor lower than that of standard roundabouts, with traffic conditions close to saturation; however, the turbo roundabouts ensure higher capacities when traffic levels are lower than 70% (unsaturated conditions). For the same reason, in under-saturation conditions, turbo roundabouts guarantee a potential reduction of CO, CO
2 and particulate matter (PM): high percentages of right turns increase this tendency. Bulla and Castro [
28] compared instead two-lane roundabouts and conventional turbo roundabouts in terms of capacity, level of service, and road safety. The authors used the VISSIM micro-simulation tool to determine the capacity of a standard roundabout, then converted it into a virtual basic turbo roundabout: they found a capacity increase of 7% between roundabout and turbo roundabout. Afterwards, a road safety audit methodology was proposed. This process demonstrated the road safety benefits of turbo roundabouts: a 22% reduction in overall risk estimation was found by the comparison of the average risk of both solutions. After adjusting the simulation parameters, Yperman and Immers [
29] accurately assessed the capacity of a turbo roundabout by using Paramics microsimulation model. They found a capacity increase of between 12 and 20% by turning a three-lane classical roundabout into a two-lane turbo roundabout with traffic equally distributed among the four approaches. Moreover, Engelsman and Ukel [
7] estimated a capacity increase of about 30% by using a quick-scan model developed in the Netherlands. They observed significant efficiency improvements with a turbo roundabout in comparison with single-lane roundabouts, two-lane roundabouts, traffic lights, or yield intersections, especially for approaching traffic volumes that do not exceed 3000 to 3500 veh/h. Huang et al. [
30] presented a two-stage calibration process for the VISSIM simulation model considering 10 signalized intersections, in order to identify relationships between potential conflicts obtained by SSAM and recorded conflicts. The main results of this study supplied credible estimates for rear-end and total conflicts in relatively simple driving environments, as the SSAM approach does not take into account traffic conflicts deriving from unexpected drivers’ maneuvers. Additionally, Vasconcelos et al. [
31] used SSAM to access the relative performance of three roundabout schemes: single-lane, two-lane, and turbo roundabout. In this research, the authors demonstrated that the safety level of a turbo roundabout is comparable to a single-lane (conflicts are more severe than those of the single-lane type), but with the benefit of higher capacity levels. In conclusion, they evaluated a good qualitative agreement between conventional accident prediction models and SSAM approach. Essa and Sayed [
32] investigated instead the transferability of VISSIM calibrated parameters to evaluate safety measures at different sites. They used an automated video-based survey to obtain vehicle trajectories and to identify potential conflicts for two urban signalized intersections. Then, after a two-step calibration process of the VISSIM model, they used SSAM to compare observed and simulated rear-end traffic conflicts considering Time To Collision (TTC) measures. In the conclusions, the authors demonstrated how the calibrated model matches existing traffic conditions and how safety analysis without proper calibration should be avoided.