VRP with pickup quantities to be distributed across several delivery locations with limited capacity each

[biswanathbasak09]

I am trying to develop a VRP during an evacuation scenario where there is a set of waiting (pickup) locations (around 100) and another set of shelter locations (around 1000). The number of people waiting at each pickup location varies from each other. Also, the shelter locations have limited capacities (but not equal to each other). There is a fleet of 85 buses, each with a fixed (all same) capacity, stationed across 9 depots (the number of buses at each depot is known), to carry these people to the shelters.

My objective is to pick up ALL waiting people from the pickup points and dispatch them across the shelters, with the minimum possible distance traveled. In the end, it does not matter if any shelter is left unoccupied or partially occupied, but no person should be left behind at the pickup locations, and no shelter can be overloaded. The buses need not return to their depots at the end of their trip. It is obvious that the sum of shelter capacities is much larger than the total number of people awaiting pickup, so reaching a solution is guaranteed.

I have checked with several VRP algorithms, and the current problem seems to be something between CVRP and VRPPD with special constraints on pickup and shelter capacities. It would be nice if anyone could provide any insight on how I could proceed.