I think what is BEHIND the stainless steel/ gal steel/ titanium firewall is as important as the firewall.
First comment - People do not understand FIRE. Let's get into the details.
My comments here pertain to composite airframes and firewalls. Between the firewall and humans is a bad place to put the insulation. Composite airframes are easily compromised by temperatures that are quite modest compared to typical combustion products. We have fire at somewhere around 1800F to 2000F.
Long held and well demonstrated practice is Fiberfrax between the fire and firewall plus a thin layer of metal to protect the Fiberfrax until the fire comes. The composite resins usually go soft around 200F. That is a long ways below 2000F. A typical composite firewall is either plastic foam or plywood cored, and will quickly be charred and fall apart when exposed to flame gases at 2000F. Beside the awful thought of the fire intruding into the cabin, there is the issue of the engine and mount departing the airframe, resulting in a tumbling airframe and crash at whatever terminal velocity impact orientation that produces.
Put simply, the epoxy must be kept cool to keep the airframe together. 200F is probably a good steady state target. Having done the heat transfer calculations, to keep the 2000F flame from raising the composite firewall above 200F in a steady state case, four plies - 1/2" - of Fiberfrax is appropriate. Three plies was not enough. Why not do the transient case? First, your recognition of the failure may take more time than we would all like, then you have to do an emergency descent and landing. IIRC, the FAA standard is 2000F for 15 minutes, and that is well along the transient case and looking pretty close to steady state to this engineer.
Fiberfrax is fragile in day to day living, and needs a layer of something to protect it from being abraded or wiped off during service work as well as to keep it from being loaded with liquids common in aviation that would put the flame much closer to the structural firewall. Yes, a nice 0.015 stainless or titanium sheet will do and keep any flame out of the Fiberfrax. A 0.015 aluminum sheet will protect the Fiberfrax from day to day, will keep flammable liquids out of the Fiberfrax, and will melt local to a 2000F fire, but its job when installed this way is done once the fire starts, and you just saved some weight. There are other folks coming to the same conclusion. But if you feel better with a 0.015 ferrous sheet (~4-1/2 pounds) instead of aluminum (~2 pounds), go for it.
Our friends using pusher prop designs have historically used 1/8" to 1/4" of Fiberfrax then a thin metal covering. They also have the advantage of airflow running away from the firewall, not toward it. With a substantially less severe heat dose on the insulation, this may indeed be completely adequate.
Get into a metal airframe structure with a ferrous firewall, and all of this may change...
(a) Don't make anything critical in the engine bay out of aluminum, including the prop- control or cowl flap hinge.
Hmm, this might be difficult to achieve. Things like cases for prop governors, carburators/throttle bodies, and other components are usually made of low melting point metals. VW cases are magnesium alloy. This must scare the poop out of some folks, but I have yet to hear of major issues with them.
(b) Don't use foam/ fiberglass sandwich behind the firewall, I only use plywood now. Much stronger for longer.
Tests have been done with plywood cored laminates. While you may get more time out of plywood cores, the composite resins loose strength at only a little above 200F and the wood chars quickly and loses all structural integrity at temperatures far closer to 200F than the flame at 2000F. Protecting and structures from fire requires significant insulation. Once we have that, I doubt that the firewall will be easily compromised.
(c) Put a fire resistant and heat insulation blanket or material on the pilot side of the firewall and use extra cover over the engine bolts and intrusions.
I do recognize that the fire on the steel engine mount will tend to transfer heat to the firewall itself. I think most of us are counting on usually only burning the firewall in one place with connections holding (because we applied suitable fireproof insulation to the firewall). A layer or two of refractory material over the engine side of the mount bolts and tubes may have real benefit here.
(d) use a stainless steel strip in the cowling layup to support the cowl fasteners.
Harump. You think the cowling will stay put if it sees these kinds of temperatures? I suspect that a ply of Fiberfrax, other commercial heat shield, or even tumescent paint on the inside of the cowling will do a lot more for keeping the flame products going out the intended cowl outlets.
Billski