Hi,

Our file server is getting filled with .hst files whenever a couple of our engineers run Sim Traffic. I am trying to convince them that .hst files are just artificial simulated data, and should be temporary files. Their argument is that some jurisdictions will someday ask for the simulation data (.hst files). I was under the impression that you only need to save the model, and use the same seed numbers, and you will get an identical .hst file so there is no need to keep a .hst file on our network and on our server backups. So is there any randomization other than the seed number? When I was programming C in college I know the random function always had the same output if you used the same seed! There's no black magic involved and thats the point I was hoping to confirm with this posting. =) Thanks...

*.hst files can be replicated with the seed numbers as long as NOTHING has changed about your file. I've learned from experience that the slightest little change, even a street name will change the results. However, if you are not keeping the hst files, then make sure you keep the installation files of the version you are using. Any new version or build will change your results.
As for your server.... it is best to run it on your hard drive anyway. It will run faster and not clog up the server. I believe trafficware will tell you it is better to run it on the hard drive.

At the very least, save your files - including the hst files - to CD, and include a notepad file with the file date and build version.

For Synchro purposes, it's best to run the program on your hard drive. But for business purposes, in my experience it is always best to save files to a network. If your hard drive crashes and you can't recover your files, that is not good.

I’m trying to generate a few .hst files under different conditions but with the same seed numbers to check if .hst files are consistent when using the same model and same seed numbers. I thought of 5 different conditions so far that we might encounter, (no change, filename change, file path change, file date change, and non-network changes to the model) and if you can think of any other conditions to test go ahead and make note of them.

When I reference H:\ drive it is a network drive... the drive that is being filled with these .hst files

These are the tests we'll try:

1. Make a Synchro model (or we can use an existing one). Save it under H:\SynchroTest
2. Run Sim Traffic once and save the .hst file as test1.hst under H:\SynchroTest
3. Run the model again with the same seed numbers and save the .hst file as test2.hst under H:\SynchroTest
4. Copy, and paste the model so that there is a “Copy of whatever” in the directory and run SimTraffic again on the copy with the same seed numbers and save the .hst file as test3.hst under H:\SynchroTest
5. Copy the original model to the local drive and run SimTraffic with the same seed numbers, save the .hst file as test4.hst under H:\SynchroTest
6. Delete the original copy on H:\ drive, then copy it back to the H:\ from the local drive. (changes the File Created date). Save the .hst file under H:\SynchroTest as test5.hst
7. Change something in the model such as the street names, or something else that you wouldn’t expect to affect the simulation and run SimTraffic again with the same seed numbers and save the .hst file as test6.hst under H:\SynchroTest


After that we will compare the .hst files to see if there are any differences. If there are none then we should be safe to delete .hst files from H:\ drive each week. All we would need to save would be the model, the seed numbers, and the build number of Synchro/SimTraffic, because different builds can produce different results.

Then I was thinking if there ever is a situation where we absolutely need to keep the .hst files in the projfile, we can ask users to rename the files as *.hst.save so that the script will not detect them as being .hst files and will not delete them.