The common way of analyzing roadway safety is to model and run your scenario using VISSIM. Afterwards, you can use VISSIM output, trajectory file, as an input for SSAM to get the number of conflicts with their characteristics including lat and long.
It is true that classic simulation models do not consider safety. The reason for this is that nobody considered the introduction of human error in traffic simulation. We introduced human error in microscopic simulation. You can find details in our paper: "From traffic conflict simulation to traffic crash simulation ...." Morever anyone interested in experimenting our software will find it more user friendly than SSAM (in that case please contact us).
A key parameter for road safety is operating speed. If you can lower that, you will improve road safety outcomes. Not something that you can "model", but the speed-safety relationship is reasonably well-known. Here's a recent paper by one of my colleagues on that topic: http://bit.ly/2ZggbNn