I've been doing some reading & experimenting with modules in endgame furnace designs. What follows is a whole lot of thinking out loud. Inevitably I've overlooked something or don't have a complete understanding or perspective - please feel free to discuss - I'm by no means an expert.
I've managed to figure out a layout where most furnaces can be affected by as many as 8 beacons. Naturally, I've been placing productivity L3 modules in the Furnaces and Speed L3 modules in the beacons.
The wiki has some neat information on this kind of setup, and if you scroll halfway down the page there is a really cool table: https://wiki.factorio.com/index.php?title=Module
I've replicated this table for my Electric Furnace setup rather than the table in the Wiki for the Assembling Machine 3. It's pretty easy to sort out the maths and since the table is based on percentages, it extends very nicely. With my setup, the max rates are:
Electric Furnace + 2 Productivity L3 with 8 Beacons + 2 Speed L3 (each):
Production: 120%
Electricity: 820%
Speed: 470%
Energy/Unit: 145%
Units/Time: 564%
So, for about 50% more energy in the furnaces, I can more than quintuple my effective production speed (and remember 20% of it is bonus products from the production modules). This sounds fantastic (I actually need 2 fast inserters pulling plates out of the Furnaces to keep from backing up), but I also have to consider the cost to power the beacons, which is 3.84 MW for 8 beacons. In larger designs, each beacon can affect up to 8 Furnaces and each Furnaces would be best served by 8 beacons - a 1:1 ratio - but we'll generally fall far short of that (I can't figure a setup with better than 1.5 beacons for every 1 furnace, but I haven't tried too much). At a 1.5:1 (B:F ratio), this would bring the Energy/Unit to 215%, overall.
It's interesting to look at the table on the wiki and see that the Energy/Unit of the build actually decreases as we add more Beacons, but given the non-negligible energy cost of Beacons, I wanted to see if that's actually true in practice, after accounting for that cost. In fact, it does, albeit at a slower rate, and it seems difficult to get lower than 200%. The theoretical minimum - using a 1:1 Beacon-to-Furnace Ratio, would be 191%.
Those are the numbers, at least. Myself, I feel like this starts to seem a bit costly. Sure it's a huge space saver, but considering the solar needed to run it, I'm not certain it's a net gain - at least in terms of space (and, of course, discounting the materials needed to build out the solar).
On the flip side, production & speed modules seem like an all-or-nothing type of deal. Either you save the energy costs and stick with stock / Efficiency, or you really have to hit your Furnaces that use Productivity modules with as many Speed Beacons as possible to get the energy / unit cost down.
My next natural question, is - well, is this layout more compact than a Stock / Efficiency Furnace? In large designs, I can only fit about half as many actual Furnaces into a given area, because I have to make room for the Beacons. Given that I'm producing at 5.64 times the normal rate, even at double the area taken, we're still doing better than 2.5x the production rate for a given area. But that's not including the power. An optimal 1:1 F-B setup would require 33 solar panels per Furnace/Beacon combo (just during the day) to power it. A vanilla Furnace would require just 3. An Efficiency Furnace would require 0.6 solar panels.
So here's where we are now: An optimal Production/Speed Furnace produces at 5.64x the normal rate for ~double the energy cost per unit with a 20% bonus to yield. The energy requirement requires (33/3)/5.64 = ~double the amount of space, overall, of a stock Furnace with the same production output per second when powering with solar, and about (33/0.6)/ 5.64 ten times the amount of space, overall of a max efficiency Furnace.
We can see that the space requirements at best will be greater with a Productivity/Speed Furnace. Relaxing the Efficiency Furnace to using only Level 1 Modules, we use double the energy of a Max Efficiency setup. This requires ~1/5th the overall space of a P/S Furnace. Why are we talking about space when space is infinite? Well, solar doesn't just cost space, it costs materials.
I've more questions, but I feel like I've thought out loud for a bit too long here. Ultimately, I feel like P/S Furnace design is situational. Most Furnace layouts scale fairly well. We can't say "Space in Factorio is infinite" as a defense of Solar, and then conveniently forget that same fact when it comes to the idea of simply expanding the furnace setup. The logistics will vary by player, but with good design (indeed, scalable smelting setups are something of a bedrock for proper play), it feels like Efficiency Furnaces come out ahead.
The modules are cheaper, the solar cost is less, and scaling to L2 modules happens much more naturally than L3 modules and beacons, freeing up resources for other aspects of the base (L1 & L2 modules have a place in the mid/endgame, while L3 feels mostly like an endgame / post-endgame item). I don't want to discount P/S setups entirely - while furnaces can be expanded, later production chains can be much more painful to scale and expand. The compounding effect of an P/S setup is also very interesting for long production chains, and is worth its own discussion - especially when compared to efficiency modules (which also compound).
Ultimately, I feel like Efficiency modules are compelling (an Efficiency 1 chip seems to cost less than a solar panel) and P/S optimal designs are strictly situational, given the scarcity of resources, with a high cost.
