Help on analysis of simulation results

#1
Hello. I am performing simulations on a project (changing durations of activities randomly from lognormal distribution). I perform 100 simulation runs with different acitivity durations on the same project. Afterwards, I insert additional activities in the project schedule and run 100 simulations on the altered project schedule. I am interested in the total project duration, i.e. want to find out if inserting additional activities significantly changes the average project duration.

Started analysing the results and have 2 problems:
1. Some of the 100 run output data does not follow any distribution. I am using EasyFit software and trying to find which distribution fits best by looking at the graph shape, Kolmogorov-Smirnov, Anderson Darling and Chi Squared test. Unfortunately, for some of the data one or another test says "reject". How to choose the distribution in that case?
2. I want to check if the average project duration of the initial and the altered projects is significantly different, not just a random thing. I was reading a lot about the statistical significance tests. As I understand, t test would be the best to use in my case. But the output data does not follow normal distribution. What else could I use instead of t test for this problem?

Thank you all very much for the help.