What you have build doesn't make much sense to be honest.
The "product" label on the tokens is not updated, so nothing will happen because the second production grid table doesn't use those product names. Even if the label was updated, you would run into the issue that the number of products (and thus the needed number of tokens) differs between the two tables.
And even if you worked around that somehow the whole logic doesn't really make sense with how the model is presumably supposed to work. Currently every combiner queue would first work to complete the first demand table and move on to the second one. As I understand it from how the objects are arranged, one group of combiners should work on the first demand table and the second group uses those products and further combines them based on the second table.
To achieve this you do not need make many changes to the Process Flow at all. All you have to is to allow for the tables which the instance object/combiner uses to be variable.
A simple implementation would be to write the respective table names into labels on the queues. Then read the name from those labels whenever the Process Flow previously used a fixed name.

This way you can control which parts a combiner works on.
One more step you will have to do is to update the "PartsNum" label of a combined item to reflect the new product type (Assign Labels activity) before it is moved out of the combiner.
And you used the wrong trigger on "Queue2". It should be "On Entry".
And the "ProductPartQuantities2" doesn't have the correct column headers.
Oh, and FlexSim is likely going to brick itself if you try to run this with 150 instances and over 130.000 items.
And finally: Please don't post additional questions as answers to your own question. Either continue the commend chain, or, if the topic has moved on from the original question, post a new question, possibly linking to the previous one to provide context if you deem that necessary.