question

The first thing I would do is get working runs in the experimenter with the variable you want to change. Then, I would create a variable that represents the number you want to optimize. In your case, that would be "Harvester Capability" (I am not sure what that is). You will likely need to create global variables that represent these things. Global variables are created in the Toolbox, and those are added to the Optimizer/Experimenter in the Scenarios Tab. Make sure when running your experiments that changing the number of the variable you created impacts the Harvester Capability Variable.

Once you have running experiments, You can go to the Optimizer Design Tab. Since you are changing the number of Task Executors, you will select that Variable and make its type Integer. You will add an upper and lower bound, likely defined by the maximum number of AGV's your real world network can support, or the business can afford. Likely if given no upper limit, the Optimizer would choose a very high number that would not be feasible to purchase in real life.

Next you choose your objective, which in your case is "Harvester Capability". You can choose to maximize or minimize this variable. If this is the only variable you want to optimize, you will choose single for search mode.

After this, your Optimizer should be ready to run. If your model does not have statistical distributions, you can see when your model ends and put that for your run time. If it has statistical distributions, I would check the run multiple replications per solution box, and add a big cushion for the amount of run time.

I hope this clarifies what the tutorials have said.