I'm retrofitting staple-up radiant into the floors of a ~2000' house with 4 zones (one of which will have low-temp baseboard).  Design-day temp is 5F, the total design-day heat load is ~28KBTU/H,  so the smaller zones (or even the bigger ones) are well below the output of even the tiniest mod-con boilers on DESIGN DAY, let alone the bulk of the hourly-binned loads for the heating season.

So I'm thinkin'...

A. With staple up radiant much of the time it would need 120F+ water, and on design-day maybe even 140, in which case much of the condensation using a mod-con would be in the flue, not the heat exchanger.  (You really need 110F or lower return water to really get a mod con to hit it's rated-efficiency stride, no?)

B. I'll need to add mass with a buffer tank in any case to limit cycling losses.

C. Gee, 130-140F is a decent temp for a DHW indirect...  

...hmm...

D. Modulating HW heaters are a helluva lot cheaper than mod-cons...

E. HW heaters are sure tolerant of lower temp return water than cast iron...

The solution I'm looking at is using a heat-exchanger type reverse-indirect (Ergomax E23) as both the DHW and a buffer tank, kept at 120-140F, with a tempering valve at the DHW output, and no mixing valves to the radiant zones.  For a heat source a Takagi  KD20 (same as the T-K2, but sealed combustion) in a short loop to the tank, with the circulation pump controlled by the tank's aquastat.

Sure, it's only 82% efficient, but if I model it against a Peerless Pinnacle T50 mod-con and the binned hourly loads over the ~6800 degree day heating season the difference in annual fuel use is only about 100-110 therms.  (For the difference in the installed PRICE I could add an evacuated tube solar collector in a boiler loop that could more than make up the difference.  The collector could be freeze-protected with controls sipping off the tank at minimal loss rather than dealing with glycol in the heating system or adding yet another heat exchanger, but I'll treat any solar inputs as a seperate project.)  We're talking just under 900 therms vs. a little under 800 for the season, including hot water.  If standby losses become an issue, it's pretty easy to retro-insulate the Ergomax to an arbirarily high level, but it'll be running much cooler than most indirects. A hacked-together spreadsheet model isn't perfect, but am I way off base here?

Are there any code issues with using Takagi HW heaters for space heating, or Ergomax indirects in MA?  (The Ergomax stores boiler water, with DHW heated as it passes through the heat transfer coils.  I don't THINK it needs to be ASME-H rated in MA for this application, nor doubly isolated from the boiler water. Or does it?)

I don't see how it's possible that a takagi and an ergomax are so much cheaper than a t50 that the payback isn't there, first of all, in a reasonable (if not immediate, of course, on a small heat load, you are in a low but not tiny load situation) timeframe. If it is, I suspect you are missing something in your cost equation. I really don't believe it will pay for an evacuated tube addition done to any code that would cover solar, and if it did, the evac tube would be nearly useless in a system running this hot for heating anyway and so I would consider that pretty much wasted money. You'd likely be far better off with the boiler upgrade.

Second, the takagi will not achieve its rated efficiency in a heating application. What you are doing is a far cry from a 70 degree single pass dt to a 120 outlet temp. 82% is generous.

Third, if you actually let the mod/con run reset instead of crippling its ability to condense by strapping it to a DHW temp tank as your buffer, it's likely to exceed its rated efficiency most, if not all of the winter. you can use a small tank of nearly anything for a buffer if you must (cheap electric 10 gallon heater, for example), or a boiler buddy, or use a prestige solo which is much less susceptible to short cycling problems due to its water content. if your DHW load is low, the excellence even removes the need for a separate tank.

Basically, if you do not have payback with a mod/con, use a tank water heater, dedicated or separated with a heat exchanger, and call it a day. You'll achieve similar efficiencies as the takagi, much more cheaply.

I do not know your code issues, but technically you're kind of shooting at the wrong target here.

In the sketched out system the Takagi would be running something like a 30-45F delta T with 155F (programmed- could be raised if need be) at it's outlet- 82% steady-state isn't approachable? (Methinks it probably is, if the heater output, boiler-loop pump size, and tank setpoint temps are set up reasonably.) Temps at middle of the tank will oscillate around a 120-130F setpoint with something like a 7F hysteresis (nominal, built into the aquastat that comes with it, but that could be swapped for something else.) There is considerable temperature slope to the tank when running in heating-buffer mode- the boiler return water is close to the radiation return temp, since the heating water return feeds the bottom of the tank about 6" from the boiler return pipe.

If evacuated tube solar can't deliver a significant quantity of 120F water with reasonable efficiency during most of the year I'd be stunned (it's what they're designed for, after all!) It's mid-winter output would be low, but in the shoulder seasons not bad, and in the summers it'd likely provide a 75%+ solar fraction for DHW. It shouldn't be tough to get a hundred delivered-therms out of it annually at 120F. Hybrid solar-boost systems using boiler water without heat exchangers (coupled with a mod-con and a small tank) are now commercially available in Germany. But I consider any solar additions to the Rube Goldberg contraption a separate design project, to be evaluated on it's own NPV.

In the comparative cost equation, with the mod-con we'd still be adding an indirect (be it Ergomax or something else) so the cost of the Ergomax roughly drops out. Added into the comparison on the mod-con side are the more sophisticated controls and the additional annual maintenance. (The annual service would end up being a double-digit percentage of the mod-con's operational cost at current NG pricing.) The installed-price delta with all controls & hardware included is over $2500, best-case (probably over $3K), for a net annual operational savings of under $200 (maybe even under $50.) Of course, were it propane-fueled (or were NG pricing to double/triple) the dollar savings would be much more significant, tipping in favor of a mod-con solution.

I like the concept of heating with tank HW heater also, but haven't explored it fully. In MA, by code you're not allowed to run DHW through your heating system, so there would need to be additional heat exchangers at a minimum. But it may prove to be a better-value option. (The Takagi/Ergomax comes in under $2K + installation.)

About the low vs. tiny load situation: Were this a single zone it would be small, but the largest zone is a little more than half the load, and the smaller zones are not just tiny, they're indeed MICRO-loads (each with very low thermal mass.) Switching them in & out from an un-buffered mod/con is bound to generate cycling efficiencies. (When it's 35F out with no passive solar gain, the whole-house load is still less than the lowest-modulated output of the T50.) Ten gallons (a mere 83lbs) will help, but more is better (maybe even 50-100 gallons?). A buffer that small would rise ~30F in a 10 minute burn at the lowest output of either the T50 or KD20. Burns a lot shorter than that probably don't achieve anything like steady-state efficiencies, but 30F is a huge swing. With the E23 you have 215lbs of water, with 7F hysteresis that's a minimum 6 minutes of burn in the micro-load condition (not great, but not terrible- most burns will be much longer). With bigger buffers whole burn cycles may be skipped when servicing micro-load zones during October-November & April-May. Small staple-up zones have very minimal thermal mass, and the lowest modulated outputs of any of these boilers are several times any single zone load for much of the year.

You have no need for anything more than a small buffer. a mod con on low fire doesn't need much: 2.5 gallons of water content in the prestige solo gives almost a full minute of burn with NO load or flow... more like 1.5 minutes to 2 minutes for the TT Solo 60. That's no load.. non existant in your case as even a modest dt of floor temperature adds some mass too. Remember these are low mass boilers: you don't need a ten minute burn to get to "steady state". That said, wear and tear on the boiler is of concern. and if you are microzoning a buffer can make sense, but there is no need to go crazy. the cost upcharge of a larger vs smaller tank would likely never pay for itself.

Also, the Ergomax is at least double the cost of most regular indirects of its volume based on a quick internet search. If you need one that small, again, I'd look at the excellence boiler which has 14 gallons of indirect built in and 3GPM continuous production. otherwise, about half of the ergomax cost is added cost for the on demand setup you don't need.

The TT also does not need yearly servicing. Note I don't sell boilers. it's just that the TT makes an awful lot of sense. and i'm not sure what you're saying about sophisticated controls, they all come with outdoor reset built in, that's not an added cost.

You are definitely wasting significant money if you don't have payback on a mod/con and choose to instead go with an on demand heater and buffer, instead of a tank water heater and heat exchanger. If not significant enough to make you blink during initial install, you still have electrical use to consider and the on demand's required pumping power is at least 4 times more than the TT's single pump design would need, and something similar compared to a tank water heater which would likely have a lot less draw of its own as well.

You are also missing a key element of your solar idea. If your panel gives 75% of your domestic in the summer, great. It won't help on your heating load at all though, as whatever it generated in the summer will be less in the winter, and you still have a domestic load. So you'll offset less than 75% of domestic in the winter and still need to meet your heat demand. So while it may be nice to have solar, it's not a heating solution in your case unless you dramatically increase the size of your array.

and if you think you'll achieve the same efficiency at 155 degree outlet on an on demand as it's rated for at 115 to 120, I'm not sure what to say. I doubt it. With the buffer tank you're ahead of where you'd be if you were running a 20 dt, but this is a water heater, not a boiler. it's not really optimized for heating. now, I'm not a boiler engineer, so perhaps I'm taking that a bit on faith, but I have not yet been pleasantly surprised by an on-demand's performance in the field yet.

Basically: on demands make no sense for heating unless you are doing low zone count high mass systems with very low heat loads. If you spend $3k extra and get a 15 year payback, ok, you have a low load. But there is payback; and if that's not good enough, then save real money and use the tank water heater. it'll run at very similar efficiency, for a lot less money and operating cost compared to the mod/con. Otherwise, you can spend the extra money and jump through the hoops you need to, in order to get the on demand to do something it was never built to do at acceptable performance... but you're not far ahead of where you could have been with a tank heater in the first place.

Hope it helps anyway; you're obviously doing some serious homework, so I'm not trying to be discouraging, you're just barking up the wrong tree is all. there is a whole forest here ;)

All good feedback- just the type I was looking for! Thanks!

A couple of notes:

*Internet pricing on the Ergomax is higher than through distribution (~$1200+ shipping over the internet, vs. $800 f.o.b. the local distributor's loading dock.) It's similar for the KD20, only more so ($550 f.o.b. the local distributor as overstock vs. $1K + shipping internet stores.) That adds up to the cost of the smallest low-end cast iron boilers, but it'll run more efficiently (no matter WHAT the ill-considered AFUE test results might indicate.) The cheapest mod-cons start at $2500 and the costs of the controls add up quickly.

**Takagi units are indeed designed & specified for use as hydronic boilers (unlike most other on-demand HW heaters) and are installed as such all over Japan. Some have been run through (somewhat bogus IMHO) AFUE testing for the North American market, but most have not. The documentation that comes with them contains basic hydronic system-architecture schematics for using them with baseboards &/or radiant floors, and how to use them in mixed DHW & space heating applications, etc. None of the schematic show buffer tanks, but documentation for the T-K3 (the follow-on product that replaces the KD20) even has a specific heatin-schematic page showing how to make a combined DHW/hydronic system Massachusetts code-compliant without resorting to heat-exchangers- I don't think they just made that up! :-) (I think I can improve on that with a buffer though...)

The KD20's ~82% steady-state combustion efficiency is rated at some delta-T (20C?) in a heating environment (I'd have to dig it up- they may or may not have specified the exact temps in the bit that I read, but I'm pretty sure it isn't 122F out, the standard DHW preset), which is nearly identical to the steady-state numbers of the T-K3. While I believe MOST on-demand HW heaters fall way off the efficiency charts in heating environments, I suspect (or hope? :-) ) that the Takagis will all run ~80% or better (even when programmed for 176F output, which is where they max out) in a reasonably designed hydronic system, and will approach the steady-state numbers when buffered or with a sufficiently high thermal mass to work against.

I know of one installation in UT where a Navien condensing HW heater is the backup source for a solar space heating installation, but I doubt it ever actually runs in condensing mode (maybe it condenses in the flue though. ;-) ) It's running both the coil in an air handler as well as maintaining the solar store to DHW temps, but I suspect it needs sub 80F to get any condensation in the heat exchangers. It's efficiency is probably not terrible- probably has a 25-30F delta-T on the AH coil, but it probably wasn't worth the extra $1500 to go with a condensing version there. The design day heat load at that house is probably lower than mine, and there's 150-200' of flat panel providing most of the heat. We'll see what the total NG use is this winter, but I suspect it'll be well under 500 therms, at half the NG pricing we pay in New England.

Using the Ergomax as both a heat exchanger & buffer tank was recommended by this guy ( www.heatpro.us/) on another forum, but I thought it worth running up the flagpole here to get a more critical feedback before diving in.

BTW: 1-2 minute no-load burn is a short cycle, and most stuff I've read on boiler efficiencies show an asymptotic approach to steady state efficiency with somewhere between 5-8 minutes required to reach 90% of steady state-efficiency. (The lower the boiler mass, the steeper the asymptote.) Jacket & flue losses go up dramatically with duty cycle too, so a 16 minute burn followed by a couple of hours of off time ends up being significantly more efficient than four 4-minute burns separated by 26 minutes of off time (a single jacket & flue loss off-cycle instead of four), whether it's a condensing boiler or not. A 2% steady-state jacket & flue loss is multiplied by duty-cycle in the shorter off-durations, but decays over longer off cycles. (Which is where a first-order estimate spreadsheet model can give pretty-good relative numbers based on the sizing of the buffer and the anticipated duty cycles.) Buffer tank standby losses are comparatively quite low (and lower with larger buffers due to the simple geometry of surface area vs. volume.) Buffer standby can also be made arbitrarily small at low cost via that newfangled stuff, what do they call it?? oh yeah, INSULATION! :-)

The notion of adding solar inputs to the buffer isn't intended to be anything like a major space-heating solution here, but if the buffer is already acting as the DHW heat exchanger it can provide a path to a lower-cost more integrated & simplified DHW boost than going with a full-on separate solar water heating system, using messier controls, more heat exchangers, etc., and may provide a small amount of space heat in the shoulder seasons. More solar could of course be added for a bigger boost, but that would require a bigger tank and a heat-dump circuit- I'm looking for Keep-It-Simple-Stupid and low cost, with optimal annual return. I'd size & orient the array to stay under the max operating temp of the tank in summer with the anticipated DHW load, installing the collectors vertically (not on a roof pitch), so that it's peak output days occur during October & April, when dumping the heat indoors via the heating system isn't a disaster, even if the thermostats aren't calling for heat. Like I said, it's a separate design project, with it's own technical & financial analysis, but it doesn't seem like it would be rocket-science to get a useful amount of heat out of it without breaking the bank (even without subsidized solar.) I'm thinkin' it'll be somewhere between 15-25 tubes and maybe $1K in pumps & controls to make that happen. The fact that such systems are being marketed in Europe gets us past the "can it work" hurdle, but the design details & cost will of course be important. Whether or not it could be implemented the same way using a tank HW heater remains to be seen, but it probably could.

you keep talking about the costs of controls for mod/cons; they come with controls built in, I'm not sure what you're talking about. Whatever external controls you are talking about adding, you would likely need to add with your alternate configuration.

simple because a mfg has a detail for how to shoehorn their unit into a heating system doesn't make it a good choice for a boiler. likewise, a detail for being "code compliant" without a heat exchanger simply illustrates how great those diagrams are ;) A heat exchanger is, in my humble opinion, a very necessary evil if you want a dual use system. Code compliant or not.

If they really were so good, I, among others, would love to know how they are so much cheaper than a mod/con. Considering the long history of on demands in heating applications, I do not think takagi has recently unlocked the magic mystery that has suddenly brought on demand heaters to the forefront of heating technology. I think they are simply looking for another avenue to sell their units.

MOST water heaters GAIN efficiency when run in a heating environment compared to the EF they advertise. Tankless do not, because their EF is not shaded by the same standby loss the tanks are; which is nearly completely negligible when stacked up to a heating demand instead of a small DHW demand like the EF is calculated on.

The Navien appears to be a special unit. But I am not enough of an expert on that to say.

Also I would not want to run a mod/con without any load ;) but with any startup load (which even in a low mass system will typically exist, it doesn't go to zero), the cycles get longer to the point where sure, a buffer tank might gain you a couple percent on average, but if you are sacrificing condensing ability to get it, you're not really ahead. You'd be much better off going for condensing ability if the mod/con were the boiler of choice. That's more like several percent than just a couple. You could do condensing with just a small dedicated buffer of 10 or 20 gallons (couple hundred bucks for a small electric water heater..) if you prefer.

If you're not going to condense, then again the tankless does practically nothing for you that a tank water heater cannot. What exactly is the benefit of modulation if every load is a buffered one?

You don't need an outdoor reset if you're not planning to run the tank less than 120F, mult-speed pumps, etc.

Takagi hasn't unlocked any mysteries, it's just a crudely modulating wall hung copper boiler, lower mass and more thermal shock & condensation resistant than cast iron boilers making it radiant-heat tolerant. They're cheaper (and less efficient) than mod/con's 'cuz the draft controls and heat exchangers are just far simpler (and less sensitive) beasts. It's not a mod-con wannabe by any means, just a cheaper more flexible & space efficient alternative to a bang-bang controlled mid-efficiency boiler, with output that can be dialed for a low-load low temp scenario that can't be efficiently met with cast iron. With the much lower return water temps, cast iron is out (even with boiler-bypass piping, etc), and these are just cheaper. But since we're micro-loading with low temp low-delta-T loads, buffering and bumping the delta-T and letting it be a slave to the average load (the buffer) rather than the micro-loads (the small zones) will allow it to run in a more efficient mode than it would otherwise.

I agree 200% that the detail about not using heat exchangers to be MA code-compliant doesn't mean following that schematic plan is a good idea. (It's basically a note about requiring that the circulation pumps in the radiation cycle at least 60 seconds out of every 6 hours so that water doesn't stagnate over the summer, and another note about the cold inlet check valve have a 1/8" hole per MA code.) I have no interest in running domestic water in my heating tubes. But to be MA compliant, a NG boiler has to test at 80% AFUE min, so it apparently passed that (admittedly low) hurdle.

It's obvious that setting up a mod-con with a 120F buffer tank would knock it's condensing efficiency way back (but larger buffers at lower temps controlled by an outdoor reset could save a lot of cycling losses in a mod-con. IIRC Peerless even sells a set of controls designed to take advantage of larger buffers. But the opposite is true for the non-condensing boiler- even with a 120F buffer it gains efficiency both with reduced cycling losses and guarantees longer burns with a somewhat modulated output (the modulation depending on how much return water is entering the tank, and at what temp.) Whether it's ultimately more (or less) efficient than a tank HW is dependent on a lot of particulars, but the tank HW heater path is one I'll be looking further into at your prompting. (Again, thanks!)

Looking at it another way, the Takagi/buffer combination allows you to "right size" the boiler (treating it much as you would a bang-bang controlled cast-iron boiler)- the modulation level can be fairly well controlled to be at or below the design-day rating for the house/heat load in question simply by the tank setpoint, the pump speed, and the Takagi output temp. It's not hard math to figure out how many gpm you need at what approximate delta-T to deliver the design-day BTUs to the tank, and if you're off by a little bit, the modulation will correct for it some. My design day load is about 1/5 the maximum fire on the KD20, but twice the low-end modulation- it can be dialed in for my load to operate much more load-matched than any cast iron boiler out there. It's a bit of a Rube Goldberg contraption compared to a straight-ahead HW tank, but the latter may have a more limited output range. Got any models to recommend for a 5K-30KBTU/H whole house load? (I know one person in MA heating her 4000' house with a tank HW heater. Her average heat load is quite a bit bigger than mine, but that probably doesn't matter TOO much.)

the takagi/buffer is no better than a tank. the buffer capacity practically eliminates any modulation benefit.... practically. you might gain a couple percent but again, nothing that would remotely look like payback... especially not after your first service call/breakdown on the takagi.

I would not advocate for cast iron (though really, that's just its own buffer tank if you control it adequately), or anything else. Quite simply: Mod/Con running reset, or tank water heater. I stand by the TT excellence being your best/simplest choice if longer term payback is acceptable (or, if you use gas prices from six months ago, or six months from now, in your calcs ;)). If not, the tank water heater is the sweet spot of efficiency/economy.

Your acquaintance, however, made the wrong choice. even a 25% increase over your numbers starts to eat into that payback period quickly, and that assumes a superinsulated shell. Under any circumstance that really requires heat, she'd have been better off with a mod/con all around.

I'm not too into particular tank water heaters though. 30k upper range should be fairly easy to find. with a HE, reset control to the radiant can be had for about $250 as well... a nice, inexpensive comfort upgrade in that case more than an efficiency upgrade of course (though the HE will run cooler).

I'm looking at the tt site, seems the smallest they're currently offering is the 110, and it's LOWEST modulated input is 30K (delivering almost exactly my design-day requirements)?? Given the significant cycling losses, I think I'll pass on that model- too big. (A boiler size would also offer little benefit from modulation, but the condensing will still give it SOME boost.)

The Solo 60 seems to be on web-store shelves (for a bit under $3K, shipped, and I'd still have to buy a water heater or indirect), and it's not exactly confidence inspiring that it's been discontinued. (Or is it so new that it's not on their website yet, even though it's for sale?) But a 15+ year payback on something with a 10 year warranty is less than confidence inspiring, eh? ;-) But OK it's on my consideration list (at $1500 close-out pricing, I'm there!) Most mod-con manufacturers specify a maintenance schedule for tune up by factory-qualified techs to stay in warranty, and TT is no exception. The Solo 110 manual (available online) has a significant maintenance checklist (including the hydronic water chemistry "at least annually" to head off corrosion in the more-sensitive heat exchanger, presumably.)

Don't know what people charge for that service in your neighborhood, but around here its a couple or three hundred 'merican shekels, which definitely adds to cost of operation. If the price of NG doubles there will be some payback, but reviewing the history of the early '80s energy price shock it's unlikely that it'll double before the warranty is up. (In the '80s oil shock it took 9 years for consumption to resume to pre-shock levels, and 25 years for the price to rise that high again. It'll likely be faster this time around, but absent a heavy carbon-tax, I don't see it doubling over current pricing during the lifespan of the boiler.)

The woman who heats her zoned radiant house with a HW heater is a heating contractor- if it's the wrong choice for her I figure it's HER problem. :-) (She even runs a driveway snow-melting zone off it.) Hopefully she did the math- I don't know much detail about her system, but I view it as an existence proof that it can work in my zip code.

My original hope had been to run the radiant off a micro-cogen (see www.freewatt.com), since they're available in my zip code (I even had them come out and size me up.) The 11KBTU heat output, moderately buffered would be a decent baseline for my needs most of the heating system- that's my heat load on a non-sunny mid-30s F outdoor temp, well within the sweet spot of the binned hourly weather data. But the hydronic boiler they coupled it with is grotequely oversized for me (70KBTU/H out at lowest fire, on top of the 11K out of the Honda.) They claim to be working on marrying it to a smaller boiler, but don't give out any details (and certainly no release date.) Since I live in one of the highest-priced electricity markets in the lower 48 the payback there would be certain for me.

Part of my reluctance to go spendy on a high-efficiency boiler in the short term is the hope that a reasonably sized cogenerator will become available in the not-too-distant future. (If not Climate-Energy's Honda-based Rube Goldberg contraption, something else.) But just about anything will be more efficient than the 4x oversized cast-iron beast currently taking up space in the basement, (and retrofitting it to run radiant would be just silly.) There are several under development in Europe, where the heating loads in dwellings tends to be half or less than the US average, but a 50hertz 240V inverter might make for some interesting smoke if I bootlegged one in greymarket. :-)

the 60 is new, not discontinued.

the solo 110 is a stainless steel heat exchanger (it's not "sensitive" at all, which is one of the reasons I like it. it's not flow sensitive either). warranty info on the website is out of date; I believe the more current requirements are available through suppliers and are much less stringent.

perhaps your contractor friend is using a modulating water heater. the Phoenix was just released by HTP recently. You're still losing the super low end condensing that way, but it's a solid idea otherwise. That's not cheap either though. on a larger system, it would be much better than a typical water heater at least.

Hey Bob!

Regarding combustion efficiencies of tankless HW heaters in heating applications, I'm somewhat iinclined to believe folks who actually measure stuff:

http://dsp-psd.pwgsc.gc.ca/collection_2007/cmhc-schl/nh18-22/NH18-22-106-108E.pdf

Seems like 80%+ combustion efficiency is pretty likely over a large operating range!

In this case it was a Rinnai (not Takagi) running a coil in an air handler, but the return temps were most likely not below 100F, since most peops find exit-air cooler than 100F uncomfortable. The Rinnai's rated EF was 82%, the combustion effciency measured between 82.9% in space heating mode and 83.4-85.9% during DHW loads.

They don't specifiy the output temp, but it was most likely a relatively low delta-T given they were also using it for DHW. It was beating the combustion efficiency of a combi-tank heater using the identical air handler by several percent in the nearby similar apt unit. But of course, system-efficiency isn't the same as combustion efficiency (but it's surely LIMITED by combustion efficiency.)

Which leads me to believe that the buffer/tankless configuration isn't the efficiency disaster that you imply (even if it may not be the sweet spot in the $/purchased-efficiency, reliability, or optimal NPV point of view.) The efficiency of the tankless burner system starts higher and falls more slowly under partial load than the combi-tank, but a buffered-tankless will probably be closer to the combi than the air-handler/tankless setup. But standby losses on a buffer tank can be brought down to arbitrarily low levels at low cost.

FWIW, the design day heat load in the apartments tested are ~30% higher than my house. Whether mod-con or tankless, modulation advantages don't exist for me for >75% of the heating season even it were a single zone. The fact that it's multi-zone, with the largest zone not-quite reaching the lowest modulated output of the TT Solo on design day, without buffering it's little better than a buffered bang-bang control scheme. I've read elsewhere (unconfirmed) that TT's controls are set up for a 10 minute minimum burn cycle, which will require significant buffering on my smaller zones, in which case it's about the same as running all zones on a central tank. The outdoor reset won't do squat, since it's delivering several X the total heat required when a small zone calls for heat, independent of outdoor temp. An outdoor reset controlling a central tank's temp is probably more appropriate, if the efficiency gains pretty limited, even were it decoupled from DHW.

Still working the design problem...

Please don't put words in my mouth: I never called it an "efficiency disaster". I said, I doubt the on demand would achieve its efficiency rating, and by the time you jump through all the hoops you are trying to jump though you end up spending a lot more money to do what you could probably do with a tank water heater just as efficiently.

Which this study largely agrees with, by the way, note that the tank water heater is wildly outperforming its 0.58 rated EF, acheiving very close to the same efficiency as the on demand; though you're right, it's not measuring the efficiency drop I would expect in the Rinnai configuration, we know nothing about how this air handler was operated. For all we know it could have been run at a 50 degree drop which would not be the first time I'd seen that with air handlers. But, perhaps I'm wrong about on-demands. Historically, I'd say the great weight of anecdotal evidence I've seen would say not, but I'm not one to insist that anecdotal evidence is proof either.

Whether I am or not though still doesn't really change the base facts:

a mod/con will condense during all of these demands, and you're wrong about reset not doing squat even when cycling (for one, it reduces cycling in the first place by not just automatically firing on a demand, whenever you are below the floating target). a $250 twenty gallon tank dedicated for bufferring (NOT maintained) takes care of pretty much all of your cycling concerns in any case, but even without it, short of a condensing on demand, the mod/con will outperform the on demands in most cases, assuming reset is used.

you do not need a ten minute burn cycle with the TT either, incidentally, and i'm not sure where you are getting that from.

the mod/con will still run more efficiently, but I'm not assuming payback there. I'm saying IF fuel costs based on a higher efficiency target come out to give you payback with a mod/con, great; but the mod/con will do better than a tank or an on demand running a tank, even if the payback equation doesn't work out. In your case, you could downrate your mod/con efficiency expectation slightly, but they are rated at 92% under part load conditions without condensing, and in your case you could always condense, so I doubt you'll drop below that even with additional cycling factored in.

IF that calc does NOT give you payback; then use a tank water heater. You are just wasting money considering the tankless and ergomax, which is basically the same as a tank and heat exchanger, but more expensive.

That is the long, and the short. Whether the on demand is below 80% or slightly above doesn't affect it much... which is basically the extent of my unsurety.

The words from your mouth (fingers?) were

" What you are doing is a far cry from a 70 degree single pass dt to a 120 outlet temp. 82% is generous."

...and...

"if you think you'll achieve the same efficiency at 155 degree outlet on an on demand as it's rated for at 115 to 120, I'm not sure what to say. I doubt it."

...and...

"Considering the long history of on demands in heating applications, I do not think takagi has recently unlocked the magic mystery that has suddenly brought on demand heaters to the forefront of heating technology. "

Which all implied "woeful shortcomings" or "efficiency disaster" to me. Was I misconstruing you?

I was only counting on low-80s combustion efficiency range, and it seems it probably WOULD run right about where I had assumed. If you hadn't challenged the assumption I wouldn't have bothered looking it up. (There isn't a lot of easily available data online, but many examples of using on-demands with air-handlers. Unless it's something special most AH coils run ~30F deltas @160-180F water, less when operated at lower temps.)

The short-sheet for the air handler they used with the Rinnai lives here:

http://www.loraxec.com/media/Enerboss_Spec_Sheet_500700_Series.pdf

It looks like it can deliver the design-day heat load at 130F (no big delta-T required), but they may have run it at higher temps. Either way, it wouldn't be tough to set up a similarly low-head boiler loop to a tank (if you wanted to throw money in the air instead of going with a tank HW heater, expense be damned! :-) ) And were it significantly more efficient to run it into the tank with a higher delta-T, lower gallons-per minute, that's easily configured via pump selection & programming the on-demand's output temp. Max output on Takagis is ~180F, so a 70F delta isn't out of the question, if that's what it took, but I'm betting it runs about as efficiently with half that delta-T and lower output temp. (Cood be rong, offen am, eh? ;-) )

I never doubted that tank heaters far outstrip their EF numbers when operated in space-heating mode, but I was a bit surprised that the one tested didn't beat 80% combustion efficiency, which kinda puts a hard upper-limit on it. (I'm suspect condensing versions would do better, even at non-condensing temps.) I'm definitely leaning toward a tank HW heater solution (at your prompting) at this point, but the Solo 60 and some level of buffering isn't completely out of the question either. Cycling it 10x/day on zone calls (buffered) seems abusive, and perhaps the least-best use of the technology, but condensing vs. non-condensing isn't arguable. Even with cycling losses my crude model shows a hundred therms/year or so improvement. Think anybody will ever make a 30K mod-con with 5/1 turndown? My load isn't atypical, and current available mod-cons are basically NO-MOD/cons at my load.

Can you 'splain me better how the outdoor reset makes a significant difference at low load? I'm not sure what you mean by "by not just automatically firing on a demand, whenever you are below the floating target."

Seems to me it's firing at lowest mod (or not) and the boiler's output temp is somewhat irrelevant: All of the heat from a firing is either dumped in the buffer or delivered to the zone. Standby losses on the buffer are low, and the heat delivered to the zone is whatever it's it's zone t-stat demands. For a given heat load the duty cycle requirements of the burner remains about the same, but without ever larger buffers, cycle numbers INCREASE (not decrease) as output temps get very low. To be sure there's a crossover point as return water temps rise and condensing drops off, eventually exceeding cycling losses, but even 0.5% cycling losses add up (and I'll bet they're higher than that on most mod-cons.) Feel free to show me the math of how that all stacks up, but it seems the only time you get SIGNIFICANT benefit out of outdoor reset is when your load reaches or exceeds a major fraction of the lowest modulated output of the boiler. Marginal gains of a few percent from using the O.R. are significantly offset by cycling losses before then, and the gains don't really take off until the boiler is fully modulating.

No, you are correct that I was saying it would not achieve rated efficiency; and I may be wrong at how much to de-rate it, but I'm not quite ready to just assume so based on a report using a system with no details on its operation. Even so, I did not say or imply that it would be an "efficiency disaster" is all: just that it doesn't suddenly make sense to use an on demand. It still does not. It does nothing for you that a tank heater can't do, far more cheaply. except higher volume outputs, perhaps.

It may be that the unit holds to 80% under higher temps as well. However, we still don't know. If that air handler were run with a 120 out 70 back or something like that, it's very close to domestic operation. I have no idea how it was run though, so I can't say. I can say, that you would not be likely to replicate that scenario in a buffer tank, IF.. IF... IF... that were what was required to run the on demand at the efficiencies calculated. It could be done with an air handler in a room temperature space, however.

output temp is not irrelevant when the boiler is condensing. You are still cycling on a differential. You can run that differential at higher or lower temps, and at lower temps you are condensing more. you dont' just "start at 70" with every demand. Cycling around 140 is not the same as cyling around 95 or 105. There is no difference in cycling numbers unless, of course, you change your differential, other than the difference in load itself. raising a buffer or a zone's water temp 10 degrees is the same amount of heat either way, it's just a question of how efficiently you are operating to get that heat into the water.

WITHOUT reset, OR a maintained storage tank, a heat source generally turns on whenever there is a heat demand, unless/until it hits a fixed high limit amount. So you cycle with every single heat demand, no matter what.

With Reset, a heat demand ONLY fires the boiler if the water is too cold to meet the demand. So it's october 31st, 40 degrees outside, you only need, say, 100 degree water. if your water were still 105 degrees from a demand a few minutes ago, you won't cycle again; you'll just run the pump and draw out the heat that is left. This is obviously more effective with buffer capacity but does not require a maintained storage tank. But this ability to actually determine whether water temps are hot enough or not does reduce cycling compared to a system without reset or maintained storage tank.

This just brings the system to "necessary cycles" and minimizes operating temps while it is at is. Otherwise, imagine a string of demands on an otherwise uncontrolled heat source with a fixed output temp; each demand raises its temperature further and further, completely unnecessarily.. and incidentally, satisfies each demand faster and faster to boot.

I am not a boiler engineer though, so I don't have the math to show this. I'm just passing along the best info I have thus far. I don't really think cycling efficiency is that big of a deal with low mass heat sources that get up to temp in about 30 seconds anyway... especially not with reset with which residual heat is not "wasted". I use buffers for heat source longevity more than anything, as whatever small percentage is lost in cycling with low mass heat sources (especially with reset, so they can purge themselves as well) would have quite a long payback period for any buffer tank .

Show me the hot air heating system that would return 70F water without FREEZING the occupants with the chilly breeze, eh!?! :-) 100F water returned from an AH coil is something of a lower limit for comfort, (and that's without standing in the exit air draft.) 100F exit air is the lowest for standing in the draft, and heat transfer across the coil isn't all that efficient. It would have to be a very special coil indeed to deliver both 100F+ air & 70F return water (and 95F average temp in the coil.)

By far the greatest (the only first order factor) determinate of condensing operation efficiency is the temperature of the return water. Whether the water flow is higher, producing lower output temps or lower flow to produce higher temps. If you have a fixed firing rate (lowest modulation) and a load significantly lower than that, the duty cycle is known. If the return water is 90F, it doesn't matter whether the output is 100F or 120F, the boiler runs at about the same efficiency (since the flue-gas temp is determined by the return water temp, not the output temp.) But at a fixed buffer capacity, if you're buffering 120F it'll require fewer cycle numbers than if producing 120F. And in a scenario with multiple micro-loads that are far below the boiler's lowest fire, servicing them directly from the boiler output is a recipe for maximizing the cycle numbers. Buffer the output- more mass=longer burns=higher operating efficiency. (With duty cycles below 25% the efficiency difference approaches double-digits.) Buffers can get you some of the best features of both high mass & low mass boilers, but only by decoupling the boiler output from the load, serving the loads from the buffer.

I was led to the buffer-isolation concept via a white paper written by Jack McKeegan of Patterson Kelley Boilers. (I can't find it on the web in a quick google, but it's out there somewhere, and well worth reading.) While written from a large-facility point of view, the math (and the solution) is the same for the micro-loading situation.

The crude first-order effects cycling loss model I'm using was clipped from Raypak's site: http://www.raypak.com/module.htm Scroll down to graphs 3 & 4. The shape of the curve is the same whether the top line is 85% (non-condensing model) or 95% (condensing model- you ain't gettin' better than about 95% without return water under 90F). Assume a 0.5-1.0% jacket & flue loss for a mod-con, maybe 1-1.5% for a tankless, 2-4% for a cast-iron. If the total load is less than a 25% duty cycle at lowest-fire, you can see what that does to your efficiency. (An 83% steady-state efficiency boiler becomes a 75% efficiency boiler, a 95% efficiency boiler drops to 87%.) The only way to game the system is with a buffer big enough to completely skip cycles. Buffer standby losses may be as much as a percent, but that's hugely better than 8-10%. Properly configured & insulated, a cheap 83% boiler + a buffer can beat an oversized mod-con with a lowest-fire output 2x the design day load, 5-10x the typical shoulder-season load. At 20% duty cycles & less you're well into double-digit losses, no matter what you're assuming for jacket & flue loss. If you're cycling more than once or twice per hour (see graph 5), you're giving up another few percent in startup losses, well in excess of large-buffer standby losses. The bottom line always seems to come down to, cycling=bad, more cycling=worse, no matter what the operating temps are. The flip side of that is the fewer cycles per-DAY (hourly isn't good enough) the better.

If you had a buffer that could store an entire day's worth of BTUs at low temps you can clearly ignore cycling losses, just feed the buffer and be happy! :-)

With a tank HW heater the jacket losses can be controlled, even if the raw combustion efficiency is lower. It's lower upfront cost and simplicity of installation makes it hard to rationalize the Rube Goldberg approaches of gee-whiz control schemes. Just getting the burner sized right for the load is more than half the efficiency battle- the rest is gravy.

Dana is precisely correct. We often used buffer tanks in the 90's for our fixed output Glowcore boilers, decoupling flow, stretching cycle times and allowing for smaller loads.

Unfortunately most of the standard water heater's standby losses result from an open chimney. A ModCon is still the best buy when multi-temp or outdoor reset is considered. And then there is the exclusion of the drafty and energy sucking combustion air.

Dana; air handlers have been run with 100 degree SUPPLY temps routinely in geothermal system for about 20 years. You can't blow the air right on the occupants, and they run lower delta-T to maximize output, but you are not limited in the way you think you are just because they put a fan on some fins. they are, of course, oversized by regular standards. and I'd rather see other solutions, but I can't say it doesn't work. It does.

Again, I"m not saying that IS what they did.. I'm saying we DON'T KNOW what they did, and the range of possibilities is quite large. Again, regardless of what they did, it doesn't change my basic premise; an on demand on buffer gets you practically nothing a tank water heater can't give you, and if you want more efficiency than that, a mod/con is a better choice.

I'm also not arguing AGAINST buffers. I'm arguing against maintaining a storage tank at a fixed temperature that is much higher than you need it to be most of the winter. A properly sized and controlled buffer tank is not maintained at any temperature.

I've seen that raypak page before (it's very good) but you do note that the vast majority of your loads are within a few percent of ideal efficiency with a mod/con cycling at low fire, right? If you run this like it's a 30k burner (or another mod con with a lower min mod, if the prestige doesn't do it for you), then your "on-off" boiler is within 3% or so of "peak dynamic efficiency" down to half that number. or so. graph 3.

on graph 4, it shows a pretty minor drop until the very bottom of the load equation.

and all of this is for cast iron boilers, I might note: none of it takes low mass boilers into consideration which I have to think changes things a lot, since you don't lose the embodied energy in the mass of the boiler on a cycle. Well, you might, but it's a lot less energy. They don't discuss boiler mass at all, even though it does actually matter.

so again: I like buffers. I think they make sense. I use them a lot when I'm not using other techniques with "fancy controls" like zone syncronization. but they don't make so much sense that it would make sense to run a condenser like an cast iron boiler on a maintained storage tank at peak seasonal temperature, not when you can add a small, low cost reservoir as a dedicated buffer tank and get BOTH buffering AND low temp, reset operation.

I would never limit my condenser to DHW instantaneous indirect storage temperatures a minimum, in other words, if I had any other choice.

Again: payback might not be there for you. But there is no real doubt that a mod/con with reset will run this more efficiently... significantly... than a tank water heater or an on demand.

Rob- Raypak is a modulating copper-tube boiler company- the primary difference in the cycling-loss model is the constant you select for jacket & flue losses. On-demands are effectively mini copper-tube boilers, more similar in mass to mod-cons (far more similar than cast iron is to either.) If you're generous to the mod-con and assign it 0.5% and assign the on-demand 1% the picture doesn't change much- duty-cycle & total cycles per day eats up efficiency much more quickly than buffer-tank standby losses.

The vast majority of my loads are micro-load zones, each a tiny fraction of the whole-house design-day load or the lowest modulated output of most boilers. (Indeed the largest zone almost but doesn't quite reach the lowest output of the Solo 60 on design day.) Without buffering this would plain kill the efficiency of the mod-con, and I suspect (without doin' th' math) that 10 gallons of whatever-temp the outdoor-reset sets up just isn't gonna be enough to really fix it. OTOH 25 gallons of 120-130F water with 7F of hysteresis (which I HAVE done the math on- it's more than half the load of my smaller zones at 40F outdoor temps, a very large annual hourly-bin) would be a huge improvement-it'll skip LOTS of cycles, but 50 gallons would probably be better. (Inducing a higher hysterisis is probably cheaper & just as effective though.) It's true that maintaining a 120-130F buffer with a mod-con will cut into it's condensing effectiveness, but it probably won't much affect the efficiency of a low-mass copper tube boiler like an on-demand.

You seem convinced that the on-demand needs significant derating when run at higher return water temps, but I've yet to find published evidence that it does- most of what I find online implies the opposite, that they're pretty much like any other copper tube boiler. These folks evaluated systems similar to the ekoComfort experiement using Takagi KD20s & T-K2s and air-handlers with about a 20F delta-T (120F return water):

http://www.toolbase.org/pdf/fieldevaluations/Tankless_Hot_Water_Heater_EvaluationSWD.pdf

Unfortunately they didn't publish combustion-efficiency or system efficiency measurements (but I think/hope they woulda noticed had the efficiency fallen off a cliff. :-) ) Their biggest design issues were related to pump sizing to get it to modulate up sufficiently to keep air-output temps comfy under sustained DHW loads, and output temps that resulted in 133F+ return water inducing short cycles (an artifact of the internal controls of the on-demand), neither of which would be an issue in the KD20/buffer-tank configuration, not that I'm goin' there.

There's a rule of thumb out there that a 40F increase in flue temp is about a 1% loss in combustion efficiency (40F is pretty close to the difference you'd get in flue gas temps between operation with 70F return water you cite and the 110F return water I'd expect.) If the on-demand runs 83% combustion efficiency in DHW mode using groundwater temps, maybe it gets derated to what, 81-82%? Unlike mod-cons, non-condensing appliances don't have the knee in the condensing curve to treat with kid-gloves- it's pretty linear. They probably all run higher than 85% combustion efficiency steady-state with 50F water, and the mere 80-82% typical EF is an artifact of the cycling losses from the average DHW draw (a very different test condition from heating loads, where steady state efficiency is probably achieved in no more than 10 minutes, given the low mass.) But I wish there were actual test data available, eh?

Badger- Forced-draft tank HW heaters (like the one use in the ekoComfort experiment) don't have the open-flue problem. I suspect their lower combustion efficiency is related to a lack of turbulence on the water side the heat exchanger resulting in less-efficient heat exchange. The tank designers probably maximized flue-gas turbulence with the forced-draft to compensate to the extent they could, but it takes both sides. Forced draft boiler & on-demand heat exchangers are easily designed to have far fewer laminar-flow issues. But low-mass boilers (condensing or otherwise) seem to have more issues with micro-bubbles forming on the water side cutting into heat transfer efficiency, but at lowest fire mod-cons have been shown to lose efficiency due to laminar flow on the flue-gas side as well. (Lowest fire lowest temp isn't necessarily the most-efficient mode of operation- depends on the specific design.)

The on-demand HW heaters I've considered are all sealed-combustion. I'm not sure if there are any cheap tank heaters that comes configured that way. Hmmm... (It's a good thing I'm not in a hurry, eh? ;-) )

I understand what you are doing and that you are micro-zoned. But in practice, zones don't patiently wait for each other to stop firing to pick up, and with reset curves your heat demands are fairly constant. On a 20 degree dt, 10 gallons gives you almost 1700 stored BTUs when a demand is triggered. That would seem to buffer out quite a few tiny demands.

You're making some very good points, but a 50 gallon buffer on a tiny system like this is ridiculous. You're getting into full on geothermal system buffer tank sizes at that point. While I can't pick your math apart here, that just goes very far outside of the realm of normal operating characteristics even for a very small system. I suspect you are looking at lot at percentages and not a lot at volumes of heat, and I also suspect you're not crediting your distribution system with any mass or heat retention (which may not be much, but does help). But I am speculating, admittedly, grasping at straws because what you're saying to a large degree just doesn't make any sense.

so that's making me wonder how close to reality your perceptions are here, is all. I hope it doesn't sound like I'm just digging in my heels to call you wrong, just please understand, that's a pretty wild claim for situation you're in from where I'm sitting, and I'm not sure what you're basing it on so I can't really address it.

anyway: I am accepting your idea that perhaps the on demand doesn't lose that much at higher temps. Likewise, I still demand that a mod/con will see significant efficiency benefits; one more time, maybe not to payback, but not equivalent either. and I would rather run a mod/con with reset than without. Without a mod/con, I would run a tank. I am not sure how many times I need to restate that to focus on it, but I am not really focusing on the on demand, because it doesn't really make any sense. Why spend that much more to do almost exactly what the tank heater is doing, by the numbers of the article you posted? Obviously you can't use reset with an on demand because if it doesn't know what the setpoint is, it can't adjust any modulation to achieve it, and it still uses the buffer tank to on demand setpoint measurement to determine its modulation. You can reset the tank, but not the unit itself (until they come out with that feature, of course).

I'm curious about your ten minute statement though. These exchangers heat up way faster than that, I believe, why do you think it would take so long to hit "steady state' efficiency?

Thanks for the nudge on Raypak's boiler types as well, I had missed that. Still, I have to wonder if they are really equivalent, given a nearly 10% differential in AFUE to a mod/con (which is typically tested under the same test conditions, not condensing, IIRC) there must be a significant different in construction that might affect those numbers.. right?

Finally, I"m not sure what you're pointing at with the toolbase article. it says nothing at all about the units' efficiency, only its cycling characteristics. Are you just pointing out that some people design to a 20 degree drop across air handlers? You'll get no arguement from me there. They could also have upsized their air handler and run a lower flow and higher delta T, raising the outgoing water temp of the unit (or not, depending on their ventilation diffuser design). but yes, most people run 20 degree drops across most everything.

Another dissenting opinion (from a former-skeptic), with some real world heating system experience behind it:

http://www.profitableplumbing.com/_wsn/page5.html (You'll have to scroll down a bit to hit heating bit.)

So, where I'd started out, I need warmer water to be able to deliver the heat to the zones via the lower-transfer-efficiency radiation that a staple up on wood provides, which would already reduce the efficiency of a mod-con for half the season (the coldest half). And the zone loads are low enough mass (under 10 gallons) to be needing buffering anyway. It's maybe not the cheapest solution, but non-terrible system-efficiencies, even by mod-con standards, and measurably better than any cast iron beastie. Sure, you'll get 10% better performance with the smallest mod-cons without much forethought, but you'd have to design it VERY carefully to get as much as 15% better. (The crude models estimate around 12-13% lower fuel economy.) If mod-cons were 50% smaller then you'd be talkin'!

The 10 minutes is an upper limit for hitting 99% of steady-state thermal efficiency for the unit, not just the combustion-efficiency. I suspect the 90% level is hit within 3 minutes in a boiler that small. The heat-exchanger is more than half the equation, but heat losses to the rest of the unit go down over time as the other internal mechanicals & jacket temps rise. It's a secondary effect, but still significant when estimating the true efficiency costs of shorter & more numerous cycles.

AFUE is a bizarre test- very low delta-T, at return water temps that would ruin (and violate the warranties) of most non-condensing boilers. Most residential boilers NEVER run to full steady state before serving a zone call, and the specified duty cycles seem arbitrary. All boilers will have higher efficiencies at lower return water temps, and longer/fewer burn cycles- system designers are better off looking at the raw combustion efficiency and ways to ensure the lowest tolerable return water temps, and fewest burn cycles. There's no magic to AFUE- it probably cheats most mod-con numbers (particularly in slab-radiant operation) while overstating most real-world cast iron boiler efficiencies.

I only pointed to the toolbase article as another sample system being touted as "high efficiency" (albeit, without efficiency data, what with that? :-) ) with low delta-Ts and high return water temps using an on-demand as a boiler. Whether it actually achieved any efficiency we'll never know, but had it been crazy-bad I'm sure they would have (or at least SHOULD have) noticed. It makes me want to set one up in a test lab and pull out thermocouples and flue-gas analyzers- you'da thunk SOMEBODY in the heating biz might have done that, given the number of people using them as slab-radiant boilers.

But the real-world experience of using them successfully & reasonably-efficiently in non-radiant hydronic systems cited in the "Tankless 101" article makes me believe they really ARE just tiny forced-draft copper-tube boilers, with similar combustion efficiencies (at lower cost & higher operational efficiency than cast-iron blobs, yet cool return-water tolerant to boot.)