Discussion
Linear relationship
A general linear relationship appears to be present when comparing SMTD and MPD for all the material types, particularly at a texture range between 0 and 1.5mm. A visual inspection of the relationship for individual surface type suggests that the data points cease to be linear at higher SMTD texture depths and they appear to curve upwards from around 1.0 to 1.5mm. Detailed analysis of the data shows there are subtle variations dependent on surface type but the relationship exists for both negative and positive textured materials.
The cumulative distribution chart in Figure 3-3 highlights that 93.2% of all the SMTD values measured on the network are less than 1.5mm. The latter explains why some of the trendlines for graphs presented in Appendix A and Appendix B appear to be flatter than the data would indicate. The reason for this is mainly due to the density of data points between 0 and 1.5mm, i.e. data points will be stacked on top of each other. The latter has an influence on how the trendline is calculated and explains why the linear trendlines look slightly flat compared to if trendlines were estimated using the human eye, i.e., the latter would be placed at a steeper angle.
Linear relationships were determined using the individual material data sets using all the data and values less than 1.5mm. A weighted average gradient was adopted to reflect the proportions of different material that exist on the Scottish network. The two weighted average gradients were then used to convert the current SMTD condition category criteria for use with MPD. Dependent on the relationship used new thresholds for MPD were calculated. Using the data from the entire data set the existing (SMTD) threshold changed from 0.7mm to 0.76mm and from 0.4mm to 0.44mm. The dataset containing SMTD that was less than 1.5mm produced new MPA thresholds of 0.75mm and 0.43mm. Based on the analyses undertaken these increases appear sensible and reflect the general observation that, although SMTD and MPD estimate texture depth in a different way, MPD values generally produce higher estimates of texture. It also seems sensible to adopt the latter thresholds, based on less than 1.5mm SMTD, as the relationship is strongly linear in this range, particularly between 0mm and 1mm.
Maintenance Comparison matrices
Comparison matrices were used to assess the impact of using MPD rather than SMTD. The main purpose of this exercise was to determine how MPD would categorise the condition of the network in terms of maintenance requirements when compared to SMTD. The performance of MPD compared to SMTD is examined in Table 3-7 and Table 3-8. Figure 4-1 summarises the lengths of surfacing identified as being in the Green, Amber and Red categories by SMTD, MPD and the lengths where they both agreed. It shows that currently SMTD places more lengths (3,990km) than MPD (3,768km) in the Green category. In contrast, the MPD places more lengths (792km) that SMTD (570km) in the Amber category. The lengths placed in the Red category by SMTD (3km) and MPD are both very small (2.9km).
It can be seen that MPD classified 3,718.5km of surfacing as being in the same Green category as SMTD, i.e. 93.2% agreement. The remaining portion, 271.6km or 6.8%, was classified as Amber. In accordance with CS 2306 this category is defined as having some to moderate deterioration. These sections would be added to Transport Scotland’s site ranking score and be considered along with friction (SCRIM) and accident data to determine whether any corrective action would be required.
The Red category which is defined in CS 230 as moderate to severe deterioration, is very small irrespective of the texture estimate used. The current SMTD method classifies 3km of surfacing in this category, whereas MPD classifies 2.2km and 0.8km in the higher Amber category.
Comparisons with other countries
The Office of Rail and Road (ORR) commissioned a study to benchmark the condition of the National Highways (NH) network against Transport Scotland (TS), the Welsh Assembly (WA) and the Netherlands. The measurements of road surface condition included texture depth. For the UK networks, NH, TS and WA, texture depth was estimated using SMTD. In the Netherlands SMTD is not used and texture depth is reported as MPD. In order to compare texture depth with the Netherlands, a relationship based on SMTD and MPD measurements collected on surfacing in England was used. The data collected on the NH network produced a weighted gradient of 1.21 and was based on the relationship between SMTD and MPD on different surface types for SMTD values less than 1.5mm. Although the approach is similar to that adopted for this study, it should be noted that the NH sample was around five times larger (23,500km) and contained 16 different types of material, albeit Clause 942 and HRA were the dominant types.
The weighted gradient used to convert SMTD threshold values to MPD values in this study was 1.068, which is around 13% smaller than that used in the ORR study. One possible explanation for the difference in calculated gradients relates to the standard used to calculate MPD. The NH relationship utilised data that was processed in accordance with the 2019 ISO standard for the calculation of MPD3, and it was noted that this produces different estimates of texture compared to the previous 2004 ISO standard. Work carried out by WSP in New Zealand (Internal correspondence) indicates that the difference between MPD2004 and MPD2019 is not linear, i.e. the difference is greater at low textures than high textures. A check was undertaken to establish what standard the Scottish SCANNER data was processed to and it was confirmed that the MPD values were calculated using the same standard as NH, i.e. ISO, 13473-2019.
A second plausible reason for the difference in relationships could be due to Scottish materials producing lower texture levels owing to their material components and configurations. It is recognised that certain materials, such as TS2010, are denser and produce different surface characteristics from some of the materials used on the NH network. The ORR report also contains some visible differences in histogram distributions for the UK networks. Nonetheless, it would be expected that texture levels for individual material types such as HRA would provide a similar relationship. For example, the average weighted gradient for HRA on the NH network was 1.24 and this compares to 1.08 for HRA on the Scottish network, which is 15% smaller. However, on closer examination of material specifications, it became evident that the HRAs were produced to different requirements. The HRA used on the NH network relates to Clause 943, and the HRA on the Scottish network relates to Clause 911TS. The former relates to a mixture that contains a PmB to produce an enhanced resistance to deformation. Other factors that could influence the texture between the two HRAs include levels of trafficking, which are significantly higher on the NH network.
Skid policy
Texture depth forms part of Transport Scotland’s skid policy and any decision to transition to MPD requires to be considered carefully. Accident studies show that the contribution of texture depth is complex and appears to be an important factor in some circumstances but not in others10. As such, it is recommended that texture depth should not be used in isolation as a surrogate for risk of skidding accidents. However, as a general rule, it is widely acknowledged that maintaining adequate levels of texture is essential where the measured skid resistance is low. The current study has not considered in detail how the application of MPD could affect the use of texture within the current skid policy. Yet, early indications suggest that the adoption of MPD may be more conservative as it categorised more of the network as Amber when compared to SMTD.