I think the below should answer your questions. You basically get charged per instance for per second of usage.
You're charged an hourly rate based on the pricing tier of your App Service plan, prorated to the second.
The charge is applied to each scaled-out instance in your plan, based on the amount of time that the VM instance is allocated.