What does "thermal runs" mean?
If you are using taq polymerase, it can usually handle this, but some specific sequences are more difficult than others, and perhaps a different polymerase will work better. You could try a polymerase that specifies that it can PCR up to 5 or 8 kb. Proofreading polymerases, Phusion, Pfu, etc, or a taq polymerase "long pcr" mix might be worth trying.
For using the polymerase you already have, you can try the usual optimizations:
First, a couple of new primer sets can be tried, or used nested if they fail individually.
The anneal temp can be adjusted up (probably) and/or down, depending on the primer Tm.
The extension time should be ample.
Added cosolvent helps for some templates. Betaine, or 3-10% DMSO can be tried, also glycerol, and formamide work in some cases.
Various of these variations can all be run at once (particularly if you have two machines, or a gradient machine) and run on a gel.
In terms of buffer itself, the only thing I do is use the alternate "GC Buffer" that my Phusion polymerase provides (contains a cosolvent, I assume), or I used to try changing the Mg concentration, but I stopped messing with that.