I tend to believe numbers far more when it's actually measured rather than inferred- it's quite easy to fool oneself (I've fallen into that trap innumerabley times.) You are likely be underestimating by quite a bit:
Doubtful as there is more calculating and measuring than "inferred" implies. I've accounted for and measured most of the key differences (ambient room temp, seaononal incoming water temp), plus I've made conservative assumptions (relying on the higher demand of the winter mix.) I don't have external envelop lines, tank, or recirculation, and my tank is located in the basement center of the structure. I've insulated about 75% of the linear feet of my hot water lines.
If I had a spare gas meter about I would install it for a mid-winter measurement. I do, however, intend to do some retests where I can during mid-late winter when the ground is at its coldest. If I'm appreciably off, the results will show it.
In other words, winter use showed an average increase of over 100% of the summer average, not merely "...over 50%..."
Reality check: the basis of their midpoint is unlikely to be anything resembling my own from what I can see in their numbers and my own usages over various years. In fact, it wouldn't even hold for the electric water heater in my previous home. If it did the winter electric use would have spiked like summer and exceeded my total actual electric use each winter month. They are so far off of my usage profile it's not even funny.
That isn't only true for this home and the last, but also the
three prior to that. If I used their 300% factor then it appears I didn't use more than a tiny amount of gas for the furnace in those homes in the coldest months--actually most months the furnace would have had to be producing gas to run the water heater.
As it is, I can do actual measurement during some milder weather stretches during mid-winter when the furnace won't cycle, so I'll see. It won't effect the supply water temp much as the ground temperature and reservoirs don't change temp that rapidly. Should make for a nice averaging effect.
AFUE & EF are horribly flawed tests- no better than educated guesses when applied to actual as-used efficiencies & measured fuel use, and would be an even lousier method for estimating. With AFUE it's a matter of how well matched the burner is to the actual load and whether short-cycles can be prevented. With EF it's all about volume of use when it comes to tanks- if you use more than the ~62 gallons/day you beat the EF numbers, if you use less, you get less (often FAR less) efficiency. With a tankless it's more about the volume per draw- short-cycles kill efficiency pretty fast, but shower-length draws meet/beat the EF numbers. Even a 0.66EF induced-draft electronic ignition tank drops below 50% efficiency when the daily draw drops to under 30 gallons. A 0.90EF condensing tankless drops to under 80% average efficiency under predominantly short-draw use profiles. An average use scenario for a tankless roughly equals a best-case high-volume scenario for a standard tank, but the worst case low-volume use scenario for a tank is off an efficiency cliff.
Compared to those California numbers they appear extremely accurate.
I'm not seeing a burn efficiency cliff in my summer gas use. Instead it is tracking with hot water use reduction and the fixed storage loss that I measured and already factored in. Actually, if you look again at the water heater testing, you can see that it agrees with what I'm seeing. His EF conclusion leaves me with a big, "Duh! You didn't have to do any testing to determine that, it's first principles." Storage losses are essentially fixed, and don't scale with use. That's why you subtract them out and use some approximation of the burner system AFUE...essentially what I'm doing.
One of the things I've learned over the years is that estimating with reasonably thought out engineering models works even when I expect poor results. I've not had trouble matching up my water use, gas use, or electric use within about 10%. When I'm off by more than that, I try to determine why and adjust.
The problem with the highlighted part above is that it would counteract the very effect you are claiming I am undercounting. Winter will demand longer, harder burns with a better (colder) initial profile. Thermal efficiency would increase rather than decrease. This would dampen the other weather impacts. (Raising summer use, reducing winter compared to fixed efficiency.) No doubt the efficiency impact is real, but it is not a major factor compared to things I am already accounting for.
Continuing with the AFUE analysis you gave above, one wouldn't expect a tankless unit sized for multi-point operation to be efficient with my typical shower draws. Half of them are near the limit of where the tankless even turns on.