| Title |
Modeling Length-Dependent Driving Speed Reduction in Highway Tunnels with Structural Type Classification |
| Authors |
주윤호(Ju, Yunho) ; 김태준(Kim, Taejun) ; 김정화(Kim, Junghwa) |
| DOI |
https://doi.org/10.12652/Ksce.2026.46.2.0149 |
| Keywords |
고속도로 터널; 구조 변수 기반 유형화; 군집분석; 속도비; 비선형 회귀분석 Highway tunnel; Structural typology; Hierarchical clustering; Speed ratio; Nonlinear regression |
| Abstract |
This study investigates how structural characteristics of highway tunnels affect traffic performance deterioration and proposes a differentiated correction framework for volume?delay functions (VDFs) based on tunnel typology. Using nationwide data from 204 highway tunnels in South Korea, a three-year dataset of 5-minute vehicle detection system (VDS) observations was constructed by matching each tunnel section with an adjacent general road section under identical traffic conditions. Traffic performance was quantified using the speed ratio, defined as the ratio of average tunnel speed to that of the corresponding general section, thereby isolating structural effects from demand and temporal variations. To account for structural heterogeneity, cluster analysis was conducted using key geometric and design variables. Three structurally distinct tunnel clusters were identified, and statistical tests confirmed significant inter-cluster differences, particularly in tunnel length, curvature radius, and longitudinal gradient. Cluster-specific univariate and multivariate nonlinear regression analyses reveal that the effect of tunnel length on speed reduction is not universal. A statistically
significant nonlinear length effect was observed only in one structurally constrained cluster, where speed ratio decreased with increasing tunnel length following a quadratic relationship. In contrast, structural variables failed to explain speed ratio variations in the remaining clusters, suggesting the dominance of non-structural factors. These findings demonstrate that traffic performance degradation mechanisms in tunnels are heterogeneous and depend on structural typology rather than individual geometric variables alone. Accordingly, applying uniform VDF correction factors may introduce systematic bias. This study provides empirical support for a selective, type-based VDF correction strategy and enhances the realism of macroscopic traffic assignment and tunnel performance evaluation. |