Hey @Jingyong P
One way I could think to do this without going to extreme would be to have Queues at locations throughout a city, and they send requests for packages with frequency based on population density in a particular area. You could also change the amount of time it takes to unload packages based on the spread of the population in a neighborhood, for example in a suburb it takes longer to deliver house house vs an apartment building.
You could incorporate traffic by changing the speed a transporter could travel on a network node. This could get quite complicated with tables changing time based on simulation time. It could also get very complicated programming logic for the path the transporter should follow.
I don't think population density would be the best indicator of how many packages would be delivered to an area, although it may not be possible to get a heatmap of deliveries from a carrier such as UPS or FEDEX so population density may be your best bet.
Without an idea of the format of the data you will be receiving, it is hard to think of a way to accomplish this for your specific case, but I hope this idea helps you.
If you need any more help with this please tag me @Braydn T.
