In recent years the use of traffic simulation models for conducting conflict safety evaluation has been
proposed as a surrogate safety measure instead of the traditional traffic conflict technique. The method
makes it possible to analyze the simulated vehicle trajectories and from safety indicators find conflicts
at a given location. This method has the ability to make a safety assessment of different road designs
in project planning before making the changes. Also, the method can estimate simulated conflicts
without the need for observing traffic at the location. The software Surrogate Safety Assessment Model
(SSAM) can estimate traffic conflicts with the use of trajectories from microsimulation models, such
as the microscopic traffic simulation program VISSIM.

It is however questionable whether traffic simulation has the ability to simulate the unsafe interactions
between vehicles leading to traffic conflicts and collisions and thereby recreate the real traffic behaviour
at the given location. Different studies investigate the validity of this method in the conlict safety
evaluation and the amount of calibration needed to realistically represent the behaviour. Most studies
investigate the use of microsimulation models for safety assessment at signalized intersections in city
areas focusing on back-end conflicts. Studies regarding this method at unsignalized intersections in rural
areas with focus on all types of conflicts is however still limited and need to be further investigated.

The objective of this study is to validate the use of VISSIM and SSAM by comparing field-measured
conflictdata, from a previous study, with simulated conflicts both at an unsignalized intersection in a
rural area. Two models in VISSIM was generated to investigate the effect of calibrating the model. The
first model uses parameters from Vejdirektoratet, and in the other model certain behavioral parameters
was calibrated to match the behaviour at the project location. Video data from the existing safety
analysis from the project’s location were used to calibrate the model. Vehicle trajectories from VISSIM
were extracted and used in SSAM. The safety indicators TTC and PET used to register conflicts in
SSAM were found by making a literature study.

When comparing the conflicts between the two models, it was found that the calibration did not have
a significant impact on conflicts. However, when comparing the results from the simulations with the
results from the existing study, it was found that there was a significant difference between the conflicts.
The results make it questionable whether VISSIM and SSAM are able to correctly identify conflicts
at unsignalized intersections in rural areas, and how significantly the uncertainties in the calibration
effected the results. It is concluded that this subject needs to be further investigated.