As a result of extensive modelling changes and more detailed data in SIDRA INTERSECTION 6.0, as well as new and improved models introduced in Version 6.1, the capacity, performance and level of service results for a particular case may have significant changes compared with Version 5.1 (and older versions).

The effects of new and improved models introduced in Version 6.1 can cause significant differences in results between Versions 6.1 and 6.0 as well as between Versions 6.1 and 5.1 (and older versions).

Before these are discussed below, two general aspects of model changes should be emphasised. These are:

• model changes leading to changes in signal timings, and
• level of sensitivity to model changes in relation to non-linear characteristic of the fundamental performance - degree of saturation relationship.

Changes in signal timings

For signalised intersections, changes in saturation flow rates resulting from various model changes may lead to changes in signal timings (cycle time and phase times). In turn, the changes in signal timings may contribute to significant differences in delay, LOS, queue length, stop rate, and so on. The changes in performance results may be well beyond the direct effect of changes in underlying capacity models.

Non-linear characteristic of the fundamental performance - degree of saturation relationship

The level of sensitivity of performance estimates such as delay, LOS, queue length, stop rate, and so on depends on the resulting degree of saturation (v / c ratio). This means that a big change in performance results does not necessarily indicate a big change in the underlying model.

When the intersection is operating near and above capacity (high degrees of saturation), the changes in performance estimates are large as a result of small changes in the degree of saturation, i.e. the sensitivity to capacity changes is high, due to the non-linear characteristic of the fundamental performance - degree of saturation relationship. For example, if the degree of saturation, x is increased slightly from 0.95 to 1.05 (11% increase) as a result of a small change in the underlying model, delay is likely to increase substantially, e.g. from 60 s to 110 s (83%).

On the other hand, when the intersection is operating well below capacity (low degrees of saturation), the changes in performance estimates will be low even if the underlying model changes result in large changes in capacity. For example, if the degree of saturation, x is doubled from 0.30 to 0.60 (100% increase) as a result of a big change in the underlying model, delay may increase only slightly, e.g. from 30 s to 32 s (7% increase).

Use of the Sensitivity Analysis facility of SIDRA INTERSECTION will indicate the level of sensitivity for specific cases.

Reasons for differences due to model changes introduced in Version 6.0

The reasons for differences between Version 5.1 and Version 6.0 results for single Site analysis using the same input parameters include the following (changes due to Network analysis are additional to these):

• new short lane model,
• new Two-Way Sign Control method,
• changes in handling of Heavy Vehicle effects,
• changes in travel distance, travel time and geometric delay definitions,
• changes in saturation flow factors for turning vehicles,
• changes in the shared lane model,
• changes in pedestrian performance equations.

These are discussed below.

New short lane model

The short lane model in SIDRA INTERSECTION 6.1 and 6.0 differs from Version 5.1 and older versions significantly. The basis of the new model is described in the User Guide. In addition to the general aspects of model changes related to signal timings and the non-linear characteristic of the fundamental performance - degree of saturation relationship discussed above, the following should be considered when comparing the results given by SIDRA INTERSECTION Versions 5.1 (and older versions) and Version 6.1 in relation to the different short lane models used in these versions.

Definition of short lane capacity

In Versions 6.1 and 6.0, the short lane model includes the effect of short lanes on adjacent full-length lanes. This is deliberate and overcomes a shortcoming of the Version 5.1 short lane model.

The definition of short lane capacity was changed in Version 6.0. In previous versions, capacity was based on the use of short lane space, and the short lane saturation flow rate was derived from the capacity value (it was not a saturation flow rate measured at the stop line unlike full-length lanes). When the short lane queue exceeded the available short lane space, a degree of saturation of exactly 1.000 was given based on a flow rate which gave an average back of queue that fitted in the short lane space. When this occurred, an “excess flow” was added to the adjacent lane. This flow rate represented the queue overflowing into the adjacent full-length (Through or other) lane. However, this short lane queue did not have any effect on the saturation flow rate of the adjacent lane.

In Versions 6.1 and 6.0, the saturation flow rates both for short lanes and adjacent lanes represent counts of vehicles departing from the queueat the signal stop line during the green period. The basis of the short lane saturation flow model is to determine the reduced saturation flow rates at the stop-lines of the short lane and the adjacent lane due to vehicles coming from the common upstream queue with increased saturation headways after the short lane queues are discharged at the full saturation flow rate (while the vehicles queued within the short lane section are cleared). Reduction in the stop line saturation flow rates may or may not occur according to queuing conditions.

Definition of delay and queue length for short lanes

In Version 5.1 (and older versions), the definition of delay and queue length for short lanes differed from full-length lanes (they were defined as values corresponding to queues that fitted the restricted short lane space). This resulted in underestimation of delay and queue length values for short lanes.

When very heavy flows used a short lane, it was necessary to switch the short lane specification to the adjacent lane for realistic delay and queue length estimation. The Version 6 short lane model does this switching automatically.

Network model requirements

It was necessary to improve the areas discussed above, especially for the purpose of a good network model that requires (i) stop-line saturation flow rates for signal platooning and (ii) unrestricted queue lengths to model queue blockage of upstream intersection lanes by queues developing in short lanes, overflowing onto adjacent lanes and extending to the upstream intersection.

Thus, the model in Versions 6.1 and 6.0 removes various shortcoming of the short lane model used in previous versions. These shortcomings were explained in the User Guides of previous versions.

Changes during Version 6.0 updates

Short lane model improvements were also introduced in several Version 6.0 updates. Those introduced in later updates include the following.

Version 6.0.20 (released on 7 April 2014): “Short lane capacity calculations for approaches with multiple short lanes and for unsignalised cases have been improved.”.

Version 6.0.22 (released on 14 May 2014): “Improvements have been made to the short lane capacity model in relation to cases with continuous movements, very short lane lengths and some signalised cases with two green periods.”.

Lane Summary report

Model results will be better understood by inspecting the Lane Summary report (lane-by-lane results) rather than the Movement Summary report since individual movements may use several lanes with different characteristics especially when using a combination of short lanes and full-length lanes. As a result, it is not easy to see why the movement performance results change.

Shared lane model with Free Queues

The use of the shared lane model with Free Queues is recommended for slip/bypass lanes at signals where queue space for vehicles using the slip/bypass lane is very short (entry to the slip lane is very close to the signal stop line). As a rough guide, use the shared lane model with Free Queues if short lane is less than 30 m / 100 ft as discussed in the User Guide.

New Two-Way Sign Control method

There are significant changes in modelling of two-way stop and give-way / yield sign controlled intersections in SIDRA INTERSECTION 6.1 compared with Version 5.1 (and older versions). These include:

• a new method to make automatic adjustments to the base values of critical gap and follow-up headway as a function of intersection geometry, control and flow conditions,
• allowing for the effect of unequal lane utilisation on gap-acceptance capacity, and
• the ability to specify the percentage of the minor movement flow rate giving way to the nearest lane only.

Refer to the User Guide for detailed discussion.

Changes in handling of Heavy Vehicle effects

In Version 5.1 (and older versions), the parameter "HV Method for Gap Acceptance" was available (for roundabouts, sign control and filter turns and slip lane movements at signals) with two options: "Include HV Effect if Above 5%" and "Include HV Effect for All Percentages". In Version 6.1, only the "Include HV Effect for All Percentages" method is used. Since the option "Include HV Effect if Above 5%" was used in the SIDRA Standard models, the default values of Gap Acceptance Factor and the Opposing Vehicle Factor parameters for the Heavy Vehicles movement class have been set as 1.5 to compensate for the change in the method. These can be specified by the user in the Vehicle Movement Data dialog, Calibration tab.

The effect of Heavy Vehicles (and all other movement classes) on saturation flow rates of unopposed movements at signals continues to apply for all HV percentages.

The effect of Heavy Vehicles (and all other movement classes) on saturation flow rates of opposed and unopposed movements at signals is therefore handled in a consistent way in Version 6.

Changes in travel distance, travel time and geometric delay definitions

The road section and travel distance definitions used for Site OD movements differ from older versions. The new method determines travel distance and travel time values according to travel on External Approach, Internal Approach and External Exit sections. A single Site analysed separately (not as part of a Network) will have External Approach and External Exit sections only. An Internal Approach is relevant to a Site which is analysed as part of a Network with an exit section connected to a downstream Site.

Geometric delay values in SIDRA INTERSECTION Versions 6.1 and 6.0 are generally lower than those in Version 5.1 (and older versions). In the past, delay due to acceleration to the exit cruise speed on the downstream exit section was included in geometric delay (this was because only External Exits were relevant due to single intersection modelling in earlier versions). In the new definition, the exit acceleration delay is assigned to midblock travel in the downstream section after the intersection.

See the detailed discussion on this subject given in the User Guide.

Changes in saturation flow factors for turning vehicles

New values of Turning Vehicle Factors (through car equivalents) that vary by OD Movement and Movement Class are used for adjusting saturation flow rates as signals. These factors also apply to uninterrupted (unsignalised) movements. Various related changes are introduced in the Vehicle Movement Data dialog, Calibration tab. Refer to the User Guide.

Changes in the shared lane model

The modelling of different movements with different signal phasings, or opposed and unopposed movements, blocking each other in a shared lane, especially the case of more than two movements in a single lane has been much improved. This area of the code was totally rewritten in Version 6.0 as we now need to handle (as a theoretical upper limit) up to 8 movement classes with 8 destinations (i.e. potentially 64 movements) with different timing characteristics.

The current code is considerably more sophisticated in this respect. By comparison, the older (Version 5.1 and previous) code was essentially written for two movements in one lane and the third movement was a later “add-on” handled in a simplistic way.

Pedestrian performance

Performance (delay and queue length) equations for pedestrians at signals were improved in SIDRA INTERSECTION 6.0 by including the effect of high pedestrian volumes through the use of the flow ratio parameter. The model for pedestrian movements with two green periods was also improved. Coupled with the Walk Time extension parameter, this produces much improved performance estimation for pedestrians compared with previous versions.

Reasons for differences due to model changes introduced in Version 6.1

Signal platoon model using signal Offsets

A new signal platoon model using signal Offsets is introduced in Version 6.1 (Section 7.2). This replaces the simple platoon model based on the use of Arrival Types and Percent Arriving During Green specified as user input under the Signal Coordination parameter in the Vehicle Movement Data dialog, Signals tab (Section 5.11.3) in previous versions.

When the Signal Coordination parameter is specified as Program (default), the lane-based second-by-second platoon model as a function of signal Offsets will be used for internal approach lanes in a Network of signalised Sites (at-grade intersections, interchanges, pedestrian crossings) identified as Coordinated Sites in the Network Timing input dialog in Version 6.1 (Section 6.5.1). The performance results based on the more detailed new platoon model are expected to be significantly different from the results based on the simple platoon model used in previous versions.

Also importantly, when the Signal Coordination parameter is specified as Program (default), the movement will be treated as an isolated (not platooned) movement in single Site analysis under the Site tab. Therefore there will be large differences in results for a Site between single Site analysis (under the Site tab) and Network analysis (under the Network tab).

Extra Bunching

In previous versions, only a user-specified Extra Bunching parameter was used for the effect of upstream signals on gap-acceptance capacities of roundabout and two-way sign controlled intersections. In Version 6.1, when the Program (default) option is selected for this parameter in the Intersection input dialog (Section 5.2.1), the Extra Bunching value will be determined by the program automatically in Network analysis. The value determined by the program is likely to differ from a value specified as user input previously resulting in differences in capacity and performance results.

If the Extra Bunching value was zero in a Project file created with a previous version, the Program option will be selected automatically when opening the file in Version 6.1. In Network analysis, this may result in significant parameter values determined by the program, leading to significant differences in capacity and performance results.

Thus, when the Extra Bunching parameter is specified as Program (default), there may be large differences in results for a Site between single Site analysis (under the Site tab) and Network analysis (under the Network tab).

Continuous lanes in Network analysis

In Network analysis, excess back of queue will be passed on to upstream continuous lanes (e.g. Major Road approach lanes at two-way sign control and continuous lanes of a signalised seagull intersection upstream of a signalised intersection) to allow for queue blockage from a downstream intersection to apply to intersections further upstream. This new model introduced in Version 6.1 may cause differences in results between versions, and also between single Site analysis (under the Site tab) and Network analysis (under the Network tab).

Another new model introduced in Version 6.1 in relation to continuous lanes is allowing for the effect of opposed turns in Major Road short lanes at Two-Way Sign Control Sites on the capacity and performance of the adjacent through (continuous) movement when there is overflow from the back of the short lane. This is relevant to basic Site analysis. Differences between Version 6.1 and older versions may result from this change.

Pedestrian Actuation and Phase Actuation methods

The Pedestrian Actuation method for improved signal timing calculations when pedestrian volumes are low, and Phase Actuation method for improved signal timing calculations when vehicle volumes are low are used to reduce the minimum required Phase Times according to the probability of a pedestrian call or a vehicle phase call in a signal cycle. Reduced minimum times may lead to significant changes in signal timings (Cycle Time and Phase Times) resulting in large changes in performance estimates.

Delay calculations for movements in shared lanes at unsignalised intersections

In Version 6.1, delay calculations for movements in shared lanes at two-way sign controlled intersections and roundabouts including shared continuous and opposed movement lanes was improved for better reflection of the conditions of individual movements in a shared lane. This model improvement is relevant to basic Site analysis and may cause differences between Version 6.1 and older versions.

Two-segment lanes

In SIDRA INTERSECTION 6.1, significant enhancements have been introduced to the handling of two-segment lanes (Section 5.4.1). A two-segment lane can be treated as a full-length lane or a short lane depending on the Movement Classes using the upstream and downstream segments of the lane. In both cases, the probability of blockage of upstream lanes will be determined, and the probability of short lane overflow into the adjacent lane will be determined when a two-segment lane is treated as a short lane. This model improvement is relevant to basic Site analysis and may cause differences between Version 6.1 and older versions.

Cost model parameters

Cost model parameters updated regularly in each major release of SIDRA INTERSECTION. Therefore differences are expected in operating cost estimates between versions.