Rob
Ansys Employee

That's the same collision twice.  You need one pair for particle & wall and a second pair for particle & particle. There may be a little trial and error, or my test case is corrupt due to being built in Fluent v6

To be honest, DEM in Fluent will cost you an awful lot of time because of the time steps needed to resolve the collisions. I'd favour Ansys Rocky coupled to Fluent, pure DPM or Eulerian over Fluent's DEM depending on need, particle loading and software availability. 

The other properties are based on how the collisions behave, so may require experimental data.