lower operating time of servers

Hi, in my model, it has a checking logic to check if the servers breakdown or not every 8 hour. An the breakdown percentage will change every 24 hours. I use year 2011 as the time-slot and divide it into monthly and weekly.


When the simulation run for 8 hours, a checking logic in MTBF MTTR table is used. A random number is generated by the formula dnuiform(0,1000) , and then I used this random number compare with the operating percentage I set to decide if the server breakdown or not. However, I tried to do simulation for 20 time and found the simulated total operating hour of each server is normally lower then the estimated hour calculated by excel as input data around 100-300 hours (total simulation time is 8760 hour) . I think this factor quiet affect the simulation result. Is that any problem of random number generation ? Is that the random number is not random enough ? Or there are some error in my checking logic??

Also, during simulation, I found that there is a server state called "setup" occurred, it occupies about 50- 600 hours, what does this state means ? Is it a processing state ? If not, how to cannel it in my model as I just want the server in one of [processing, idle, breakdown] state.

Thanks in advice.