Shilpa,
Not sure what is going on...let me understand the workflow, you create a CF layer, add a field to the Incidents called ADD_PT_ID and then do a load locations and map ADD_PT_ID and load fails?
Does the type of the ADD_PT_ID field agree with the contents of the ADD_PT_ID field on your data?
If you plan to re-use these addresses, perhaps you can run the GP tool call Add Locations to add the location fields once and then when you do the Load Locations, choose the "Use Network Location Fields" to make it load much faster. This will make it possible to "debug" what is going on you will not have to wait for so long each time you load the points while trying to make this process of mapping the ADD_PT_ID field work.
And I will suggest couple of other things since you are loading a large number of locations:
If you only want the shortest path distances and not the actual route geometry to the four closest, then use the OD Cost Matrix solver as it will go quicker. Here the CF facilities would mean destinations and the CF Incidents would be Origins.
Also, if the four closest locations are reasonably close, say 30 minutes, then turn off hierarchy AND put a cutoff of 45 minutes and solve. It will solve much faster as hierarchy works great for long distances and has a larger overhead for short distances (I am assuming you have a hierarchy built into your network dataset).
Regards,
Jay Sandhu