The capacity increase could come a few ways
- Perhaps there is currently a rule against running at max capacity in case of bridge issues
- Block signalling means that increasing the speed of the trains increases the throughput. If the max speed on Portal is lower than the max speed in the North River Tunnels...
- Trains stopping at Secaucus v. skipping it mean intricately-timed mergers. The throughput increase at Portal could translate to an overall throughput increase even if the North River Tunnels have always been the bottleneck component, because a better Portal bridge means trains can run closer together (in time, due to higher speeds) west of Secaucus, allowing more scheduling flexibility east of Secaucus.
Say the old schedule required 2 minutes between each train in the NRT and at Portal each, and a Secaucus stop takes 3 minutes, and every other train stops at Secaucus. Before, if you tried to have trains out of New York every two minutes arrive at Secaucus at :00 (through), :02 (stopping), :04 (through), :06 (stopping), etc... (30 tph) which then depart at :00, :05, :04, :09, etc... but the departing trains must wait a minute to have the signal clearance, holding up incoming trains, since they can't be less than 2 minutes apart. So instead you have to run trains at :00 (through), :03 (stopping), :05 (through), :08 (stopping), :10 (through), which depart at :00, then the :58 leaving at :02, then the :03 pulls in, then the :05 pulls through (it can't be at :04, two minutes after the previous departure, because then it would be too close after the :03; the :03 can't be earlier because then it would have a conflict with the :58's departure), etc... and now the :03 can't leave until :07, hand the :08 pulls in, the :10 pulls through, the :08 leaves at :12 so the :13 can pull in, etc... and we have 20 tph.
But the new Portal Bridge with higher speeds might allow for tighter clearances west of Secaucus, so now our original :00 (through), :02 (stopping), :04 (through), :06 (stopping), etc... (30 tph) which then depart at :00, :05, :04, :09, etc... are allowed, because the higher speed means that the one minute of time clearance is enough space clearance, say.
Fill in whatever the correct values of 2 minutes (max frequency through NRT / old Portal), 3 minutes (Secaucus stopping time), 1 minute (max frequency through new Portal / clearance needed at Secaucus) etc... are for the actual physical constraints - my numbers are probably too low. This is just an example of how changing Portal capacity could increase throughput even if the NRT are still the bottleneck.