How to use experimenter to travel number of flowitems and bring them back to qu?

[hebah_m2]

[ FlexSim 18.0.10 ]

small-port.fsm

I have a port and assume there a cyber-attack on the port. Thus, the ship gets wrong massage and travel to the separator when there is no separator available. Then, because there is no place, the ship back to the queue again to wait. Thus, I need to do an experiment on this case with the scenarios that show if that happened to a different number of ships in one month. The performance measures will be the travel time and the stay time in the queue.

I attached the FlexSim file.

Thank you

[benjamin_w2 · Accepted solution]

Hi @Hebah M2,

The first thing that you need to do is figure out how you are going to mark if a ship was cyber attacked. I would recommend setting a label on the "On Exit" trigger of the combiners. Then, you will need to send cyber attacked flow items to different ports.

Here is how I approached this problem:

1. Create a global variable called PercentCyberAttack that is set to a value between 0-100.
2. In the combiners, set the "On Exit" trigger to the following:

   1. 

3. Make sure to configure your model to send items based on these labels.
4. Open the experimenter and configure your scenarios.

   1. 

5. Figure out which statistic you would like to track, and place this into the Performance Measures tab. In the attached example, I chose the number of ships that went through the sink

   1. 

6. Click the "Experiment Run" tab, and click the "Run Experiment" button. This will then run all of your scenarios.

cyberattack.fsm

[hebah_m2]

@Benjamin W2

It does not show any differences in the results?

[benjamin_w2]

@Hebah M2,

In the simple case that I tested it did not show any difference in the results. This is because your barges travel so fast and are created on such long intervals, than a little extra travel time had almost no effect on the number of ships that traveled to the sink.

However, you can track any statistic you would like in the experimenter. Different performance measures may produce varying results.

[hebah_m2]

@Benjamin W2,

How about if I use Global Table (by putting a percentage for the defect ship) is that possible?

[benjamin_w2]

@Hebah M2, Yes, you can refer to a global table in the experimenter. I think the key to your model is figuring out exactly what statistic you are looking for.

Remember, in the model that I sent you, you set the percentage of ships that are cyber-attacked. If you try to track this same percentage as a performance measure, you are only going to get your input back.

The key to your model is figuring out 2 things:

1. How often are ships cyber attacked? This will be your control variable.
2. What performance measure do you care about?

Once you figure out these things, you can design variables and performance measures in the expiramenter to test out variations to your control variable.