Aimsun Saturation Flow Rate Calibration for Hybrid SimulationChallenges and Recommendations

John Bennett and Brian Betts

August, 2017

Traffic modelling levels

Three broad levels

Macro:

- Large scale, strategic, no

detailed representation of

congestion.

Meso:

- Medium scale, models

intersections in detail, capable

of macro and dynamic

assignment.

Micro:

- Finest level of detail, complex

signal operation, models

individual vehicle movements.

Aimsun capable of all three

levels and “Hybrid”

simulation

Page 2

Source: Barcelo, J. Casas J, Garcia D & Perarnau J

(2005)

Aimsun meso assignment

Discrete event simulation:

- Simulation time changes when

an event occurs.

- Vehicle generation, vehicle

node movement, traffic signal

change.

- Simplified car-following and gap

acceptance model

Vehicle considered only as

it enters and exits a node

section.

Does not consider

acceleration/ deceleration

or details lane changing

behaviour.

Page 3

Source: TSS Aimsun 8.1.4 User’s Manual 2016

Challenges for Hybrid simulation

Differences in vehicle

movement, car-following

and gap acceptance.

No acceleration/

deceleration or detailed lane

changing.

Requirement for a level of

comparability between

meso and micro area

operation.

Intersection throughput/

saturation flow

comparability is key factor.

Page 4

Source: TSS Aimsun 8.1.4 User’s Manual 2016

What is saturation flow?

“The maximum uniform

discharge rate across a stop

line” (Webster F.V. and Cobbe

B.M. 1963).

Can be measured on-site

(ideal) or predicted using

TRRL and HCM2010

formulas.

Research indicates that

saturation flow is primarily

impacted by geometrical

factors:

- Lane widths

- Gradient

- Turn radii

Page 5

Source: HCM2010

Saturation flow impact variables

Page 6

Source: TRRL and HCM2010

Saturation flow adjustment by location

Page 7

Site designation

Description of characteristics Adjustment to TRRL RR67

Good Dual carriageway intersections.

No noticeable interference from pedestrians, parked vehicles, right –turning traffic (either owing to their absence of because special provision is made for them).

Good visibility and adequately large turning radii.

Exit of adequate width and alignment.

Good quality road surface.

+10%

Average Some characteristics of good sites and poor sites. -5% to 10%

Poor Average speeds low.

Some interference from standing vehicle, pedestrians and right turning traffic.

Poor visibility and or poor alignment of intersection.

Busy shopping street with pedestrian activity.

Poor road surface.

Traffic calming measures on either/both entry and exit.

Congestion or downstream queueing discouraging drivers from pulling away cleanly.

15% to 25%

Source: TRRL Food for Thought Article 97 (2005)

Saturation flow calibration in Aimsun

Vehicle reaction times (Reaction Time and Reaction Time at Stop

parameters).

Vehicle kinematics in micro simulation only (acceleration/ deceleration

profiles, etc.)

Speeds on turns – common approach used to adjust saturation flows in

VISSIM.

Section Jam Density in meso simulation only:

- Governs the number of vehicles that can stay at the same time in a section

(model link).

TRRL and HCM2010 variables such as lane width and gradient have no

significant impact on saturation flow.

Page 8

Analysis aim and objectives

Challenges for Hybrid simulation:

- Calibration of saturation flow rate using non-traditional impact variables.

- Disconnect between micro and meso vehicle movement.

Aim:

- To calibrate saturation flow and achieve operational comparability between meso and

micro simulation within the same network.

Objectives:

- Identify parameters that have most impact on saturation flow in micro and meso.

- Determine parameters that result in typical saturation flow rates in micro and meso

simulation, which could be used as a starting point in a model calibration process.

- Recommend parameters that result in comparable intersection operation in micro and

meso, thereby providing a level of comparability for hybrid simulation.

Page 9

Analysis methodology

Development of a dummy signalised approach using Aimsun version 8.1.4.

Assessment steps:

1. Calculate saturation flow rate in micro using software default reaction time and

turn speed values.

2. Refine reaction time and turn speed parameters in micro to achieve typical

saturation flow of around 2,000 vehicles per hour.

3. Calculate average delay for the intersection approach using preferred reaction

time and turn speed parameters.

4. Calculate average delay for the intersection approach using meso simulation with

software default reaction time and turn speed parameters.

5. Refine reaction time, turn speed and section jam density parameters produce

approach average delay comparable to levels achieved in micro.

Page 10

Analysis assumptions

Single lane intersection approach to eliminate impacts of vehicle lane changing.

Assignment of car vehicle type only.

Consistent signal timings and operation.

Consistent approach volume and vehicle arrival profile.

Calibration of saturation flow rates were assessed separately for each of these

adjustment methods, not in combination.

Page 11

Analysis results

Page 12

Micro simulation parametersMethod Turn

speed (kph)

Time step (s)

Reaction time (s)

Reaction time at stop (s)

Saturation flow rate

(vph)

Average delay (s)

Software default parameters

70 0.8 0.8 1.2 2,571 26.21

Preferred turn speed

25 0.8 0.8 1.2 2,057 27.84

Preferred global reaction times

70 0.5 1.0 1.6 2,057 30.50

Preferred section reaction times

70 0.5 1.0 1.6 2,000 31.61

Meso simulation parametersMethod Turn

speed (kph)

Jam density

(veh/km)

Reaction time (s)

Reaction time at stop (s)

Saturation flow rate

(vph)

Average delay (s)

Software default parameters

70 200 1.2 1.6 N/A 14.59

Preferred turn speed

25 200 1.2 1.6 N/A 14.59

Preferred section jam density

70 40 1.2 1.6 N/A 31.41

Preferred global reaction times

70 200 2.1 1.6 N/A 32.16

Preferred section reaction times

70 200 2.04 1.6 N/A 32.53

Other calibration considerations

Vehicle lane changing on approach to stop line:

- Site specific, can be managed using Look Ahead Distance parameters and solid lines.

Impacts of calibration of angled turn movements

Lane utilisation:

- Appears balanced in micro by default, but appears to favour kerb-side lanes in meso.

Impact of posted speeds, lane widths and gradient

Page 13

Conclusions

Saturation flow rate is not directly influenced in Aimsun by the same parameters

outlined by TRRL and HCM2010.

Software default parameters (version 8.1.4) don’t appear to produce typical saturation

flow rates.

Software default parameters (version 8.1.4) don’t appear to produce a level of

comparability between meso and micro assignment in terms of saturation flow.

Reaction times and turn speeds can be used to influence saturation flow in micro:

- Recommended values to be used as a starting point for calibration.

Turn speed does not seem to impact saturation flow in meso.

Reaction times and jam density can be used to influence saturation flow in meso and

calibrate to micro delay output:

- Recommended approach and parameters values for starting point in calibration.

Review other factors such as lane changing and lane utilisation:

- Meso appears to weight lane utilisation to kerb side lanes.

Page 14

References

Akcelik R. (1981), Traffic Signals: Capacity and Timing Analysis: ARRB Research

Record 123, Australian Road Research Board.

Barcelo, J, Casas J, Garcia D & Perarnau J (2005), Methodological Notes on

Combining Macr, Meso and Micro Models for Transportation Analysis.

Gipps, P.G. (1981), A behavioural car-following model for computer simulation.

Transportation Research Board Part B, 15, 105-111.

Gipps, P.G. (1986a), A model for the structure of lane-changing decisions.

Transportation Research - B. Vol. 20-B, No. 5, pp. 403-414.

Gipps, P.G. (1986b). MULTSIM: A Model for Simulating Vehicular Traffic on Multi-Lane

Arterial Roads. Mathematics and Computers in Simulation, 28. 291-295.

Highway Capacity Manual 2010: Volume 3 – Interrupted Flow: Chapter 18 Signalized

Intersections, Transport Research Board.

Kimber RM, Mcdonald M, & Hounsell NB (1986), The prediction of saturation flows for

single road junctions controlled by traffic signals, Research Report 67, Transport and

Road Research Laboratory.

Law, Averill M. and Kelton W. David, (1991) Simulation Modeling and Analysis.

McGraw-Hill International Editions. Second Edition.

Transport System Solutions (2016) Aimsun version 8.1.4 User’s Manual.

Webster F.V., Cobbe B.M., (1966) Traffic Signals, Note 34, Transport and Road

Research Laboratory.

Page 15