Calibrating VISSIM for the

German Highway Capacity Manual

The German Highway Capacity Manual provides analytical methods to use the rightmost lane that matches its actual speed. If the speed
assess the level of service of freeway segments. If these segments do not of the vehicle in the right lane is lower than the actual or desired
comply with design guidelines, microscopic traffic simulation is often speed of the trailing vehicle, the latter may change lanes and stay
used to perform quality analysis. To provide guidance and support for in that lane only as long as vehicles in the right lane are moving
these simulation studies, a research project was performed to develop at a lower speed. If the faster vehicle has overtaken all slower
a methodology for the assessment of levels of service and to find stan- vehicles in the right lane, it is supposed to return to that lane. This
dard parameter sets calibrated for the simulation of German freeways. system indicates that heavy vehicles in particular are not supposed
In this paper, the results of the microscopic simulation tool VISSIM, to use the left or leftmost lanes. In the case of two lanes and heavy-
including a short description of the characteristics of German freeways, vehicle passing, allowed trucks might change to the left lane while
model settings, and specific parameter sets for several types of freeway overtaking.
segments, are presented. In the planning process of German freeway projects, several guide-
lines and manuals are to be considered. The design of a specific freeway
facility depends on the connecting function (e.g., connecting several
For evaluation of the quality of infrastructure elements, a document higher-order centers), the actual or the forecasted traffic demand, and
equivalent to the U.S. Highway Capacity Manual 2010 (HCM 2010) the geographic situation. The details for freeway planning—such as
(1), the German Highway Capacity Manual (HBS), provides detailed freeway cross sections, alignments, or the design of merges, diverges,
analytical methods (2). Through the use of design volumes and free- weaving segments, or entire freeway interchanges—are found in the
way segment characteristics, the HBS allows levels of service (LOS) German Guidelines for the Freeway Facility Design (4). Nevertheless,
to be determined. Because of deviations from standard design patterns deviations from these guidelines can occur, mainly because of spatial
or more complex spatial conditions, traffic simulation is applied in conditions.
addition to or even instead of the computational approach. In these The HBS provides analytical methods for quality analysis of
cases, further advice is needed, especially when results differ from freeway planning (2). Every new or rebuilt infrastructure element has
one method to another. to be examined, whether the intended minimal LOS can be achieved
Therefore, the German Federal Highway Research Institute (BASt) to correspond to anticipated traffic flow rates and other circumstances.
called for proposals for a research project to develop a methodology For freeway segments matching the design guidelines and showing
and to derive parameter sets for the most common simulation software characteristic traffic conditions, the LOS can be easily computed by
tools to ensure agreement between the microscopic traffic simulation using the HBS.
and the guidelines. Geistefeldt et al. present the main contents of that However, one or more elements of a particular freeway facility,
project; this paper gives detailed information of the use of VISSIM such as a steep road gradient or a short merge, are often not com-
within the project (3). pliant with the guidelines. In addition, special demand profiles may
lead to noncompliance. In such cases, the assessment of LOS can be
derived only by traffic simulation. A guideline to a valid and reliable
Special Characteristics simulation study of traffic flow is provided in Advice for Microscopic
of German Freeways Traffic Flow Simulation (5). Nevertheless, the simulation results must
be consistent with results derived by the analytical approaches of the
German freeways differ in several respects from freeways in most
HBS. The forthcoming edition of the HBS will give some support in
other countries, including the United States. Many parts of the
simulation studies by determining the values that may be compared
German freeway network have speed limits to harmonize traffic
with the analytical results (6).
flow and increase safety. Nevertheless, they have no general speed
To assist in practical applications of microscopic traffic flow sim-
limit but rather a recommended speed of 130 km/h (about 81 mph).
ulation in a way consistent with the guidelines, a research project
Another characteristic, significant for all basic freeway segments,
assigned by the BASt was conducted. As one of the outcomes of that
is the right-side rule. This rule means that a vehicle is supposed to
project, a manual is provided that includes all calibrated parameter
sets for the simulators of the analysis.
Institute for Transport Studies, Karlsruhe Institute of Technology, Kaiserstrasse 12, The methodology and general results of the analysis are explained
76131 Karlsruhe, Germany. Corresponding author: U. Leyn, ulrike.leyn@kit.edu. in detail by Geistefeldt et al. (3). Therefore, this paper gives detailed
results of the calibration process by using the microscopic simulation
Transportation Research Record: Journal of the Transportation Research Board,
No. 2483, Transportation Research Board, Washington, D.C., 2015, pp. 74–79. tool VISSIM, including modeling details and the specific parameters
DOI: 10.3141/2483-09 for different freeway segments.

74
Leyn and Vortisch 75

Calibration Results in the parameter sets for a certain segment type and (b) leaving the
results most flexible for other kinds of the same segment type. A
As noted earlier, on basic freeway segments in Germany (and some more detailed description appears in Geistefeldt et al. (3).
other European countries), vehicles are supposed to drive according The different freeway segments are modeled with the microscopic
to the right-side rule. For this reason, the driving behavior for basic simulation tool VISSIM, Version 5.40, and in accordance with the
freeway segments is obviously specific for these conditions. German Guidelines for the Freeway Facility Design (4). The share of
However, in merge and diverge areas, other aspects of driving heavy vehicles is set to the German default value of 10%. Figure 1
behavior become much more important than otherwise. Vehicles are shows the different freeway segment types calibrated in VISSIM.
not allowed to move right to the merging lanes in merging areas but
rather tend to change left in situations in which many vehicles are
entering the mainline. Likewise, in diverging areas, absolute adher- Basic Model Settings for All Types
ence to the right-side rule would lead to vehicles changing to the exit of Freeway Segments
lanes and then changing back to the main lanes instead of exiting. In
When a new network model is being built from scratch, differentiation
the case of weaving segments, no such rule can be followed either.
between passenger cars and heavy vehicles can be found only in a
Therefore, the right-side rule is not characteristic of driving few predefined settings in VISSIM. The acceleration and deceleration
behavior on road–freeway junction areas, especially for passenger functions, for example, are vehicle-class specific, unlike the desired
cars. In the case of heavy vehicles, lane change settings follow the speed distributions. Therefore, adaption to more realistic assumptions
principle of the right-side rule—but mainly to prevent those vehicles is an essential step.
from engaging in cooperative lane changing in merge and diverge In relation to driving behavior especially, a separate parameter
areas. Thus, links and connectors have to be modeled carefully with set for every behavior type and vehicle class (e.g., passenger cars
lane-closure and no-lane-change settings. and heavy vehicles) is crucial, as certain passenger car driving
Hence, the parameter sets for calibrated driving behavior devel- characteristics are inappropriate for heavy vehicles.
oped for merging, diverging, and weaving segments in Germany The standard desired speed distributions contain a speed range
can contribute to freeway facility modeling elsewhere, too. One can representing passenger cars rather than heavy vehicles. So, for heavy
assume that driving behavior differs in relation to the driving situ- vehicles, the speed range should be reduced. Figure 2 shows the
ation (e.g., merging) rather than on the specific construction design predefined speed distribution for 90 km/h (Figure 2a) compared
of the corresponding segment. with the adaption of the desired speed distribution for 90 km/h for
The calibration of the simulation models is based on a simulation heavy vehicles (Figure 2b). In the software screenshots, the abscissa
setup that includes different demand scenarios. The simulation results represents the speed in kilometers per hour, and the ordinate shows
are used to determine the average breakdown volumes when the the corresponding cumulative share.
average speed drops below a certain threshold speed. The quality of The standard truck for the model of vehicle types has a predefined
a calibration step is assessed by the deviation of these values from length of 10.2 m (33 ft). It is replaced by a more realistic distribution
the HBS curves between the LOS E and F. of vehicle models that contains standard trucks, with and without
The calibration is performed nonautomatically but always considers additional trailer, and semitrailer trucks.
realistic driving maneuvers and reasonable parameter adjustments. For all freeway segment types discussed in the following
Furthermore, it is aimed at (a) getting as few differences as possible subsections—and for every specific speed described there—the use

Merge segment
Type E1.
One lane on-ramp, 3 main lanes

Diverge segments
Type A1.
One lane off-ramp, 2 main lanes

Type A2.
Two lane off-ramp, 3 main lanes

Type A4.
Two lane off-ramp, 3 to 2 main lanes
lane reduction

Weaving segment
Type VR1.
One lane to one lane

FIGURE 1   Freeway segment types in the analysis according to guidelines (2, 4).
76 Transportation Research Record 2483

(a) (b)

FIGURE 2   Desired speed distributions for 90 km/h: (a) VISSIM predefined and (b) adaption for heavy vehicles.

of different desired speed distributions for passenger cars and for Basic Freeway Segments
heavy vehicles can be assumed. Furthermore, for basic freeway seg-
ments without a speed limit, an appropriate desired speed distribu- To reach or even exceed the HBS capacity limits, the object basic
tion has to be chosen. For these cases, the predefined 140 km/h for freeway segment is modeled by (a) a lane reduction for the case of
passenger cars and the customized 90 km/h for heavy vehicles speed two lanes and (b) a section with different positive gradients for the
distributions fit best. case of three lanes. In both cases, links and connectors must be mod-
The parameter combinations provided in the following sections eled carefully under by consideration of their lengths, sequence, and
do not represent the sole solution. During the calibration pro- the use of the no-lane-change and lane-closure settings. The latter
cess, a certain number of similar parameter combinations could settings are both crucial in obtaining a realistic driving behavior,
be found that fail to reach the quality requirements by just a small especially for heavy vehicles, in relation to the right-side rule. To
margin. In addition to these basic settings, which depend on the model a speed limit of 120 km/h for passenger cars, the desired speed
different kinds of freeway segments, additional parameters are distribution of 130 km/h is chosen.
to be changed. These adjustments are described in the following Table 1 shows the calibrated parameters for basic freeway segments.
sections. For both passenger cars and heavy vehicles, only those parameters

TABLE 1   Calibrated VISSIM Driving Behavior for Basic Freeway Segments

Parameter for Basic Freeway Segment Passenger Cars Heavy Vehicles

Following
  Car-following model Wiedemann 99 Wiedemann 99
   CC1 (headway time) [s] 1.05 1.05
   CC4 (negative “following” threshold) [−] 20.30 20.30
   CC8 (standstill acceleration) [m/s2] 3.50 2.50
   CC9 (acceleration at 80 km/h) [m/s2] 1.50 1.00
Lane change
  General behavior Right-side rule Right-side rule
   Cooperative lane change yes no
   Within: maximum speed difference [km/h] 10.8 NA
   Within: maximum collision time [s] 10.0 NA
  −1 m/s2 per distance, own [m] 300.00 200.00
   Accepted deceleration of trailing vehicle [m/s2] 20.75 −0.50
   To slower lane if collision time above [s] 15.00 1.00
   Advanced merging yes no

Note: NA = not available.

differing from VISSIM’s default values for the predefined driving Merge Segments
behavior of right-side rule (motorized) are given in bold.
The specific driving behavior for merge areas begins for the modeled
main lanes 50 m upstream of the gore point; for the on-ramp this
Diverge Segments distance is 125 m. It lasts until the end of the merging lane. The lane
change distance for the connectors upstream of the end of the merging
Three kinds of diverge areas were chosen for examination: one with lane is set to 250 m, with an emergency stop value of 5 m.
a one-lane exit ramp, one with a two-lane exit ramp, and one with a To model a realistic merging behavior, settings for lane closure
two-lane exit ramp and lane reduction (Figure 1). and no lane change for the links and connectors in the merge area
In diverge areas, the specific driving behavior for the diverging depend on the number of ramp lanes and the speed limit. In the case
section is modeled from 250 m upstream of the beginning of the exit of a one-lane ramp (Type E1), the left lane or lanes are modeled
lane. The remaining sections are characterized by the basic freeway with lanes closures. In addition, no lane change has to be set for
segment’s driving behavior. the rightmost main lane to prevent vehicles from changing to the
The lane change distance for the connectors in the diverge area merging lane. Heavy vehicles should not change to the left lanes
downstream of the gore point has to be made sufficiently large, in throughout the merge area. If a speed limit exists, no lane change
these cases, 700 m with an emergency stop of 5 m for the main lanes has to be set for passenger cars to avoid changing to the right lane
connector and 700 m with an emergency stop of 50 m for the off-ramp or lanes.
connector. If a lane reduction is included, these values decrease To reproduce speed profiles of entering and merging vehicles,
to 500 m for the main lanes connector and 250 m for the off-ramp several desired speed decisions must be made on the on-ramp,
connector. For these values, the routing decisions are to be set far starting with 50 km/h for passenger cars and 40 km/h for heavy
enough upstream of the diverge area. vehicles just before the beginning of the circular curve. About 15 m
In the diverge area, lane closures and no lane changes are to be before the gore point, passenger cars are assigned desired speeds of
modeled in such a way that vehicles in the exit lane are not allowed 100 km/h and heavy vehicles 90 km/h (specific for heavy vehicles).
to return to the main lanes. In addition, heavy vehicles should not Immediately downstream of the end of the merging lane, addi-
change from a main lane to the next left lane. This setting should tional desired speed decisions are crucial to allowing vehicles
begin about 50 m upstream of the start of the exit lane and remain until entering the mainline to accelerate to the prevailing speed of the
some meters after the gore point. Of course, in the case of the lane section.
reduction, lane changes attributable to pursuit of the routes must The calibrated parameters that are different from the predefined
remain possible. VISSIM driving behavior types for the merging segments are shown
To achieve a speed reduction in the exit ramp early enough, in bold type in Table 3.
reduced speed areas and desired speed decisions with the desired
speed distribution of 50 km/h for both passenger cars and heavy
vehicles are modeled. Passenger cars should decelerate by 3 m/s2 and Weaving Segments
heavy vehicles by 2 m/s2.
The detectors to determine capacity are placed 500 m upstream of For weaving segments, a reduction of the desired speed is essential
the gore point. Table 2 shows that the parameters resulting from cali- if the passenger cars’ speed upstream of the weaving area exceeds
bration of the three examined diverge types differing from VISSIM’s 120 km/h. The lane change distance for the downstream connectors
default values of the predefined driving-behavior type called freeway is set to 200 m and, with an emergency stop value, to 250 m. The
(free lane selection) for passenger cars and right-side rule (motorized) driving behavior in the weaving section shown in Table 4 is modeled
for heavy vehicles in bold type. Most of the parameters are the same from 50 m upstream of the first gore point to 50 m downstream of
for all types; in those cases, only one value is given. the rear gore point. As for the parameter sets presented earlier, only

TABLE 2   Calibrated VISSIM Driving Behavior for Diverge Segments

Parameter for Diverge Segments, Type A1/A2/A4 Passenger Cars Heavy Vehicles

Following
  Car-following model Wiedemann 99 Wiedemann 99
   CC0 (standstill distance) [m] 2.50 2.50
   CC1 (headway time) [s] 1.15/1.25/1.15 1.15/1.25/1.15
   CC2 (“following” variation) [m] 5.00/4.00/4.00 5.00/4.00/4.00
   CC8 (standstill acceleration) [m/s2] 3.50 2.50
   CC9 (acceleration at 80 km/h) [m/s2] 1.50 1.00
Lane change
  General behavior Free lane selection Right-side rule
   Cooperative lane change yes yes
   Within: maximum speed difference [km/h] 10.8 10.8
   Within: maximum collision time [s] 10.0 10.0
   Accepted deceleration of trailing vehicle [m/s2] 21.00 21.00
   To slower lane if collision time above [s] NA 1.00
   Safety distance reduction factor [−] 0.85/0.80/0.80 0.85/0.80/0.80
   Maximum deceleration for cooperative braking [m/s2] 29.00 29.00
   Advanced merging yes yes
78 Transportation Research Record 2483

TABLE 3   Calibrated VISSIM Driving Behavior for Merge Segments

Parameter for Merge Segments, Type E1 Passenger Cars Heavy Vehicles

Following
  Car-following model Wiedemann 99 Wiedemann 99
   CC0 (standstill distance) [m] 2.50 2.50
   CC1 (headway time) [s] 1.25 1.25
   CC2 (“Following” variation) [m] 4.00 4.00
   CC3 (threshold for entering “following”) [−] −8.00 −8.00
   CC4 (negative “Following” threshold) [−] −0.35 −0.35
   CC5 (positive “Following” threshold) [−] 0.35 0.35
   CC8 (standstill acceleration) [m/s2] 3.50 2.50
   CC9 (acceleration at 80 km/h) [m/s2] 1.50 1.50
Lane change
  General behavior Free lane selection Right-side rule
   Cooperative lane change yes no
   Within: maximum speed difference [km/h] 10.8 NA
   Within: maximum collision time [s] 10.0 NA
  −1 m/s2 per distance, own [m] 300.00 200.00
   Accepted deceleration, own [m/s2] 21.50 −1.00
   Accepted deceleration of trailing vehicle [m/s2] 21.00 −0.50
   To slower lane if collision time above [s] NA 1.00
   Safety distance reduction factor [−] 0.80 0.80
   Maximum deceleration for cooperative braking [m/s2] 26.00 26.00
   Advanced merging yes yes

values differing from the predefined VISSIM driving behavior type weaving segments are not exclusively valid for German freeways
freeway (free lane selection) are given. but can be used for non-German freeways too. This adaptability
gives modelers a good starting point for calibration in concrete
simulation projects for both HBS-compliant and other network
Conclusion elements.
These parameter sets do not replace calibration and validation for
The analysis showed that VISSIM can be calibrated to reproduce a concrete project. The HBS is conservative in relation to capacities.
capacities and speed–flow curves given in the HBS guideline for Higher capacities and better LOS for a given facility can be proven
freeways and road–freeway junctions. The HBS-compliant parameter by simulation in concrete projects if parameter changes are made on
sets and other model settings are published in the project report the basis of sound calibration for each project.
as well as in this paper. This paper assumes that, unlike the basic The ability of microscopic simulation to demonstrate higher
road segment parameters, the parameters for merge, diverge, and capacities in specific situations than calculated by the methods in

TABLE 4   Calibrated VISSIM Driving Behavior for Weaving Segments

Parameter for Weaving Segment, Type VR 1 Passenger Cars Heavy Vehicles

Following
   Observed vehicles [−] 4 4
  Car-following model Wiedemann 99 Wiedemann 99
   CC0 (standstill distance) [m] 3.00 3.00
   CC1 (headway time) [s] 1.65 1.65
   CC8 (standstill acceleration) [m/s2] 3.50 2.50
   CC9 (acceleration at 80 km/h) [m/s2] 1.50 1.00
Lane change
  General behavior Free lane selection Free lane selection
   Cooperative lane change yes yes
   Within: maximum speed difference [km/h] 10.8 10.8
   Within: maximum collision time [s] 10.0 10.0
  −1 m/s2 per distance, own [m] 400.00 400.00
   Accepted deceleration, own [m/s2] 21.50 21.50
   Maximum deceleration of trailing vehicle [m/s2] 24.00 24.00
  −1 m/s2 per distance of trailing vehicle [m] 400.00 400.00
   Accepted deceleration of trailing vehicle [m/s2] 21.50 21.50
   Safety distance reduction factor [−] 0.30 0.30
   Maximum deceleration for cooperative braking [m/s2] 29.00 29.00
   Advanced merging yes yes
Leyn and Vortisch 79

the guidelines can save a considerable amount of investment costs 3. Geistefeldt, J., S. Giuliani, P. Vortisch, U. Leyn, R. Trapp, F. Busch, A.
and is one of the reasons that simulation is applied at all. Rascher, and N. Celikkaya. Assessment of Level of Service on Freeways
by Microscopic Traffic Simulation. In Transportation Research Record:
Journal of the Transportation Research Board, No. 2461, Transportation
Research Board of the National Academies, Washington, D.C., 2014,
Acknowledgments pp. 41–49.
4. Richtlinien fuer die Anlage von Autobahnen (RAA): (German Guidelines
The paper is based on research sponsored by the German Federal for the Freeway Facility Design). Forschungsgesellschaft fuer Strassen- und
Ministry of Transport and Digital Infrastructure, represented by the Verkehrswesen, Cologne, Germany, 2008.
5. Hinweise zur mikroskopischen Verkehrsflusssimulation (Advice for Micro-
Federal Highway Research Institute. scopic Traffic Flow Simulation). Forschungsgesellschaft fuer Strassen- und
Verkehrswesen, Cologne, Germany, 2006.
6. Handbuch fuer die Bemessung von Strassenverkehrsanlagen (HBS),
References Entwurf: (German Highway Capacity Manual, draft). Forschungsgesell-
schaft fuer Strassen- und Verkehrswesen, Cologne, Germany, 2013.
1. Highway Capacity Manual 2010. Transportation Research Board of the
National Academies, Washington, D.C., 2010. The contents of this paper solely reflect the views of the authors.
2. Handbuch fuer die Bemessung von Strassenverkehrsanlagen (HBS):
(German Highway Capacity Manual). Forschungsgesellschaft fuer The Standing Committee on Highway Capacity and Quality of Service peer-reviewed
Strassen- und Verkehrswesen, Cologne, Germany, 2001. this paper.