Did you physically install 4 instances (all with different names)? or did you install an instance on node A then add the instance on Node B (for both instances) which = 2 instances.

A cluster isn't a mirror and it cannot load balanced.

The SQL Services for the instances can only be in a "started" state on 1 node at a time (per instance). If you FAILOVER the instance to the second node then the services are shut down on the hosting node and started on the other node.

When you patch you will be patching 2 instances only assuming you installed 2 physical instances and added them to the second node.