Intersection Performance in the Kathmandu Valley: A Systematic Review of Capacity, Level of Service, and Intervention Effectiveness

Abstract

Urban intersections in rapidly motorising South Asian cities routinely operate beyond design capacity, yet the severity of failure and the comparative effectiveness of remedial interventions remain poorly characterised for heterogeneous, motorcycle-dominated traffic streams. This systematic review synthesises quantitative findings from 32 peer-reviewed intersection performance studies and six contextual studies examining capacity, level of service (LOS), and intervention effectiveness across 25 intersections in the Kathmandu Valley, Nepal, spanning 2014–2026. Five principal findings emerge. LOS F prevails at most of studied signalised intersections while DOS among LOS F intersections ranges from approximately 1.3 to 17.37. The absence of a unified motorcycle PCU standard produces a 22–48% divergence in estimated traffic demand, shifting LOS classifications by up to two tiers from the same field data; a parallel inconsistency in simulation calibration — with VISSIM and SIDRA parameters routinely applied without reference to prior site-specific derivations — compounds cross-study incomparability. Intervention effectiveness is strongly DOS-stratified: signal optimisation yields meaningful delay reduction only below DOS ~2.5; grade separation achieves LOS C with 87–90% delay reductions in all modelled flyover scenarios; and a 50% modal shift from motorcycles to public transport produces a 74% corridor delay reduction without infrastructure investment. Signal coordination is spacing-dependent: beneficial at 180 m (Shital Nivas–Kanti, −75% and −50% delay reductions) and counterproductive at 1,130 m (Satdobato–Gwarko, +4.7% network queue increase). Across all 32 studies, only one provides genuine before-after field measurement of a recommended intervention; five independent studies at New Baneshwor remain unvalidated following implementation, confirming that the primary barrier to performance improvement is institutional rather than technical. These findings collectively identify PCU and calibration parameter standardisation, simulation-to-field validation protocols, and cumulative knowledge production as the three priorities for advancing intersection management in the Valley.