I just completed a simulation model of a wheigingbridge. Where two streams pass the same wheiging bridge (incoming and outgoing). The arrival lists are based on the distribution of arrivals en departures, these are fitted by adding an excelconnection and then used the autofit option.
Afterwhich I added a table with a growfactor I eventualy could use. Now it is just based on the current situation (100%). Hour labels where added and an empirical distribution was added to determine how much of arrivals and depatures cumulative entered the system per hour. Afterwhich a queue has been added for incoming and outgoing traffic with a capacity of both 30 units.
The server(weighingbridge) has been linked to Excel and has been given a distribution via autofit.
The most important thing about this model are the queues. However after some test runs the results are very fluctuating even though I haven't changed anything.
I attached my model + excel maybe if anyone would like to help that would be highly appreciated.
As on request of Mr. jongboer some extra information regarding issues en resulst I struggle with.
The meaning of this model is to measure the average waiting time + average content of trucks in both queues. incoming and outgoing that pass the weighingbridge in a round robin procedure.
The model is built so trucks go round robin on the server - 1 in - 1 out.
The problems are that in reality during daytime lets say between 06:00 and 16:00 the average waitingline incoming would not exceed around 9 trucks and outgoing not more than 6 trucks. That would be very maximum on a average day. There are soms busy days in the year but according the data it would not influence the average patern very much.
When I do some tests it says average 14 trucks incoming and around 9 outgoing.
Also I would like to know if its possible to measure the waiting time + number of trucks labeled to hours of the day?
Attached my latest model + excel file if extra data is needed please let me know.
If seen that you have made some more progress with your model. If I understand correctly your models gives you different waiting time result and different number of waiting trucks then you see in your data. I have not checked if you are using the right distribution based on the data you have. But in general there is no problem in the Empirical Distribution atom or the 4DScript of how you use it. What can be a problem is that you are using the Round Robin input strategy of your Wheigingbridge process. This strategy will choose an incoming truck and then a outgoing truck and will repeat this in this same order. That means that if in a specific hour 5 more incoming trucks arrive compared to outgoing. These will all have to wait until another incoming truck arrives in the next hour. Even if the Wheigingbridge is not doing anything for a long time. During that time you will have a long queue of incoming trucks which get a lot of waiting time. Maybe you should reconsider if this is really the correct input strategy. For example longest waiting will give you less waiting times and queues then round robin.
Regarding your output question. I see that you have used the experiment wizard with method “separate runs” method and you are doing 20 runs with an observation period of 20*hr(24) (so also 20 days) without warmup period. To get results by the hour there are several options. You could change your Experiment definition and use the subrun method. The easiest would be to do runs of 1 day, but I do not know if this is suitable for you current problem. Another option is to use the Data Recorder atom and write all the waiting times of each truck to this atom and also record the time that a truck arrived in the waiting line and when it left. You can easily export data recorded by the Data recorder to Excel and make all the graphs you want. The data in this recorder can again also be used in the Experiment wizard. Important is that you first think exactly what kind of an experiment you would like to do.
1. How exactly do you apply the settings to write all the waiting times of each truck and record the time that a truck arrived and left ? The datarecording atom does not have the fixed option in the list of the settings. Already checked the help file but you don't get any wiser from that.
2. Currently I've done runs with queues of max. capacity of 15 but if I change my queues to lets say 50 trucks. The maximum in queue are also 50 trucks. actualy that isn't possible because I have a normaldistribution with 109 trucks average with a deviation of 20 trucks and this empirical distributed. Than its not possible to have 50 trucks in queues average seems to me ?
3. If i change the growfactor with normal queue capacity (15) to 30% growth my average in queue grows with like 0.40 pieces ? little bit strange if I have a total of 27865 arrivals.
4. What is now actually the best way to set up the experiment (times, number of runs etc). My data is fron 1 year. ?
In the Trigger on entry and Trigger on exit of the queue you can store the Time in a label on the product. For this you need the following code:
Trigger on entry: Label([entrytime], i) := Time
Trigger on exit: Label([exittime], i) := Time
This way the current entry time is stored in the label Entrytime on the product and the current exit time in label Exittime.
In the Data Recorder you can create two variables, one for each of the labels.
Regarding your second question, have you tried to change the Input strategy of the server and did you still have same problem? Before we look at the third question, let's first get this to work.
Regarding the fourth question, for example the number of runs is related to the required width of the confidence interval. I suggest you discuss this further with your supervisor.
outgoing + wheiging bridge data.xlsx