Calibration of Capacity Models to Predict Queues and Delays at Roundabouts in South Africa Operating Under Saturated Conditions
Date
2024-02
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Stellenbosch : Stellenbosch University
Abstract
ENGLISH ABSTRACT: The investigation of operational conditions at roundabouts in South Africa by traffic engineers is complicated by a lack of local research and inconsistent geometric design of roundabouts. Particularly, there is little information about roundabout operations at or near capacity. Furthermore, no research has specifically considered methods to calibrate widely used international traffic models for South African traffic conditions. The research addressed this gap by leveraging data collected in association with the South African National Road Agency (SANRAL), employing advanced data collection methods using Unmanned Aerial Vehicles (UAVs) and models like HCM6 and VISSIM. This research focused on seven roundabouts in the City of Cape Town, Stellenbosch, Paarl, and George, which were identified to operate near capacity. The research design incorporated data collection using UAVs at roundabouts operating under saturated or near-saturated conditions. Data processing involved analysis using DataFromSky TrafficSurvey and software specifically developed for the South African National Road Agency (SANRAL) roundabout research panel. The Highway Capacity Manual (HCM) 2016 served as a significant reference, focusing on regression analysis and gap-acceptance models for roundabout capacity evaluation. The HCM6 uncalibrated and calibrated capacity models were compared with the field data. At single-lane roundabouts, the uncalibrated HCM6 capacity model presented closer to the field data, where the calibrated HCM6 capacity model overestimated the capacity indicating that the standard calibration factors proposed by the HCM2016 are not suitable for South African traffic conditions.. At double-lane roundabouts, both the uncalibrated and calibrated capacity models compared well with the field data. Microscopic simulation tools like VISSIM were used for assessing roundabout performance, considering parameters like driver behaviour, gap acceptance, and traffic assignment. Single-lane and double-lane roundabouts were calibrated and validated in VISSIM by using peak-hour entry volumes collected during separate peak-hour periods. After the calibration of the HCM6 models and the VISSIM models, the delay and queue lengths were determined using the respective models provided by each method. These results were compared to evaluate the similarities and differences in the output of the two types of roundabout analysis methods. Comparing the delay results from the HCM6 and VISSIM reveals discrepancies between analytical models and microsimulation. For instance, VISSIM includes elements that the HCM6 equations exclude, which could lead to more accurate estimations. The HCM6 model employs standardised equations and formulas to estimate delay and queue length, whereas VISSIM simulates actual vehicle interactions and traffic dynamics in a more realistic manner. However, the HCM6 models might rely on simplified assumptions and methods commonly used in practice. Examining queue length results across the models highlighted differences in how they handle lane utilisation and vehicle queues. VISSIM’s ability to simulate lane changes and dynamic traffic patterns might lead to more accurate queue length estimations compared to the other models that are hard to capture with static formulas. The HCM6 models, while convenient, might not fully account for complex interactions and lane-changing behaviours. In all of these cases, the comparison between HCM6 and VISSIM results shows how each model’s approach impacts the prediction of roundabout performance. VISSIM’s results are often considered to be closer to real-world conditions due to its ability to simulate complex traffic scenarios. Still, it requires accurate input data and proper calibration for reliable outcomes. The HCM6 offers theoretical estimations based on well-established formulas and is a quick estimation method but might lack the precision of simulation-based models.
AFRIKAANSE OPSOMMING: ’n Gebrek aan plaaslike studies en onkonsekwente ontwerpe skep uitdagings vir verkeersingenieurs wat met verkeersirkels in Suid-Afrika werk. Hierdie navorsing pak hierdie leemte aan deur gebruik te maak van data van die Suid-Afrikaanse Nasionale Padagentskap (SANRAL) en gevorderde metodes soos onbemande lugvaartuie of (UAVs of hommeltuig) en modelle soos HCM6 en VISSIM. Deur te fokus op sewe verkeersirkels in Kaapstad, Stellenbosch en Paarl wat naby kapasiteit funksioneer, integreer die studie hommeltuig-data-insameling in versadigde toestande, deur gebruik te maak van DataFromSky TrafficSurvey en spesifieke sagteware wat ontwikkeld was vir die SANRAL navorsings projek. Die studie evalueer die HCM 2016. Vergelykings tussen ongekalibreerde en gekalibreerde HCM6-modelle en velddata onthul nuanses. Enkellaan verkeersirkels toon dat die ongekalibreerde model naby aan die velddata is, terwyl die gekalibreerde model kapasiteit oorskat. Dubbellaan verkeersirkels vertoon ’n goeie ooreenkoms tussen beide modelle en velddata. VISSIM, ’n mikroskopiese simulasie-pakket, evalueer verkeersirkels deur bestuurdersgedrag en gapingaanvaarding in ag te neem. Kalibrasie en validering in VISSIM, met behulp van spits-uur inkomendevolumes vir enkel- en dubbellane verkeersirkels, vind plaas. Na kalibrasie word vertragings en toulengtes bepaal, wat ’n vergelykende analise met HCM6 moontlik maak. Die vergelyking onthul verskille tussen HCM6 se analitiese benadering en VISSIM se simulasie. VISSIM se insluiting van dinamiese elemente lei tot potensieel meer akkurate raminge, in teenstelling met HCM6 se gestandaardiseerde vergelykings. HCM6-modelle mag op vereenvoudigde aanames en metodes staatmaak, terwyl VISSIM komplekse interaksies en baanveranderinge vasvang. By die analisering van vertragingsresultate illustreer VISSIM se simulasie van werklike dinamika die teenstelling tussen HCM6 se teoretiese raminge. VISSIM se presisie in komplekse gevalle, afhanklik van akkurate insette en kalibrasie, oorskadu HCM6 se doeltreffendheid maar moontlike gebrek aan presisie. Verskille in toulengtes beklemtoon VISSIM se voordeel in die simulasie van dinamiese verkeerspatrone en baanveranderinge oor HCM6 se statiese formules. Die studie beklemtoon die impak van modelleringsbenaderings op die voorspelling van verkeersirkels se operasionele aspekte, waar VISSIM ’n meer geverfynde, simulasiegebaseerde perspektief bied terwyl HCM6 vinnige, teoretiese raminge verskaf.
AFRIKAANSE OPSOMMING: ’n Gebrek aan plaaslike studies en onkonsekwente ontwerpe skep uitdagings vir verkeersingenieurs wat met verkeersirkels in Suid-Afrika werk. Hierdie navorsing pak hierdie leemte aan deur gebruik te maak van data van die Suid-Afrikaanse Nasionale Padagentskap (SANRAL) en gevorderde metodes soos onbemande lugvaartuie of (UAVs of hommeltuig) en modelle soos HCM6 en VISSIM. Deur te fokus op sewe verkeersirkels in Kaapstad, Stellenbosch en Paarl wat naby kapasiteit funksioneer, integreer die studie hommeltuig-data-insameling in versadigde toestande, deur gebruik te maak van DataFromSky TrafficSurvey en spesifieke sagteware wat ontwikkeld was vir die SANRAL navorsings projek. Die studie evalueer die HCM 2016. Vergelykings tussen ongekalibreerde en gekalibreerde HCM6-modelle en velddata onthul nuanses. Enkellaan verkeersirkels toon dat die ongekalibreerde model naby aan die velddata is, terwyl die gekalibreerde model kapasiteit oorskat. Dubbellaan verkeersirkels vertoon ’n goeie ooreenkoms tussen beide modelle en velddata. VISSIM, ’n mikroskopiese simulasie-pakket, evalueer verkeersirkels deur bestuurdersgedrag en gapingaanvaarding in ag te neem. Kalibrasie en validering in VISSIM, met behulp van spits-uur inkomendevolumes vir enkel- en dubbellane verkeersirkels, vind plaas. Na kalibrasie word vertragings en toulengtes bepaal, wat ’n vergelykende analise met HCM6 moontlik maak. Die vergelyking onthul verskille tussen HCM6 se analitiese benadering en VISSIM se simulasie. VISSIM se insluiting van dinamiese elemente lei tot potensieel meer akkurate raminge, in teenstelling met HCM6 se gestandaardiseerde vergelykings. HCM6-modelle mag op vereenvoudigde aanames en metodes staatmaak, terwyl VISSIM komplekse interaksies en baanveranderinge vasvang. By die analisering van vertragingsresultate illustreer VISSIM se simulasie van werklike dinamika die teenstelling tussen HCM6 se teoretiese raminge. VISSIM se presisie in komplekse gevalle, afhanklik van akkurate insette en kalibrasie, oorskadu HCM6 se doeltreffendheid maar moontlike gebrek aan presisie. Verskille in toulengtes beklemtoon VISSIM se voordeel in die simulasie van dinamiese verkeerspatrone en baanveranderinge oor HCM6 se statiese formules. Die studie beklemtoon die impak van modelleringsbenaderings op die voorspelling van verkeersirkels se operasionele aspekte, waar VISSIM ’n meer geverfynde, simulasiegebaseerde perspektief bied terwyl HCM6 vinnige, teoretiese raminge verskaf.
Description
Thesis (MEng)--Stellenbosch University, 2024.