Tying tankless to tank version

[Bill Hamberg]

(short lead in... had an addition build a couple years back and a Rheem ECO200 9.5 GPM tankless unit installed - addition is 1 bth, kitchen and laundry - with the anticipation of tying it into the main section at some point)

Question: what is the best practice to do so? Would it be best to run the hot side of the tankless directly into the cold side of the tank heater - thus feeding hot water into that unit or tee it into the hot side of the tank heater?

If you tee it in, does there need to be an isolator or check valve in line?

Thanks in adavance! -Bill

 

[Cacher_Chick]

Why would you want to connect a tankless to a tank heater?

 

[Sponsor]

 

[Dana]

What is the intended goal here?

For the tankless to fire and control temperature properly it needs a minimum flow, which it won't necessarily get if teed in the hot side of the tank.

 

[Phog]

Hello Bill, there's nothing inherently wrong with doing what you describe, however it's a bit of a Rube Goldberg situation. Typically you would just eliminate the tank, for the sake of cost efficiency.

The reason being that there are standby energy losses through any tank, aka heat that leaks out to the room through the insulation & is lost. The tankless unit will also capture 90+% of the energy out of the burning gas flame, while the tank unit will only recover 50-70% (and more likely closer to 50%).

With tankless systems there is something called "cold water sandwich" effect that afflicts many installations. You can google it if you're unfamiliar and find a hundred different explanations of what it is. To mitigate it, some people have been known to pair a tankless with a very small electric tank just downstream, think 2-6 gal; others will install a recirculation loop. Many people just live with the issue.

Finally, if you do decide to remove the tank, pay attention to how the tankless was installed. Check for a T&P relief valve on the hot outlet & an expansion tank on the cold inlet. You probably have them there, however it is possible that a lazy/unscrupulous installer could have skipped installing them & cross-connected the hot branches. (Thereby relying on your existing tank water heater system & saving a few bucks). Not likely but needs to be checked.

 

[Dana]

The tankless unit will also capture 90+% of the energy out of the burning gas flame, while the tank unit will only recover 50-70% (and more likely closer to 50%).

A typical tank water heater captures about 80% of the source fuel heat, not 50-70%.

The EF or UEF number being lower is an artifact of the standby losses, based on a presumed duty cycle. If the burner is running more than 25% of the time it's "as used EF" will be about 80%. If it's burning at less a 2% duty cycle (a half hour per day) it won't come anywhere near it's labeled efficiency numbers.

Feeding a tank type water heater with at tankless does not remove the standby losses.

The standby losses of an electric water heater are considerably lower than those of a gas water heater, since it doesn't have a center flue heat exchanger or burner area creating large gaps in the insulation. Convective air loops through the heat exchanger of an atmospheric drafted water heater moves heat from the tanke into the room air and up the flue.

Teeing into the hot side effectively plumbs the water heaters in parallel with each other. The higher resistance to flow of the heat exchanger of the tankless means that the lion's share of the water would be coming from the tank, and only at very high flow would the tankless fire at all. It will work fine if the tankless output feeds the cold input of the tank, but it's not clear what the purpose of that would be. It's less efficient, but allows for somewhat faster tub-fills. Still, running a tankless at high flow a lot also reduces it's lifecycle. The large pressure drop across any tankless at 9.5-10 gpm is large enough to interfere with anti-scald valves on tub mixers too.

So, what performance goals are you shooting for?

 

[Bill Hamberg]

Wow, you guys are putting a whole of thought and science behind my feeble-minded question (haha).

Here was my thought process. I'm in the Philly area (go EAGLES!) and I think the TWH is large enough to supply a 2 1/2 bath home (in my area) of which we have 3 & 2 half baths (including the addition). (like a lot of baby boomers) I also have 3 adult children living at home and it is (very) common for the main section to run out of hot water - 50 gallon DV water heater is 4-5 years old.

So if I were to feed the cold side of the tank heater from the hot side of the TWH it, in (my mind) theory, would eliminate the "depleted" hot water problem in the main section.

From your replies above it sounds like it would - next question would be; is there any type of isolation or back-feed valve needed?

 

[Bill Hamberg]

Just found this too - exactly what I was thinking...

 

[Dana]

No check valves required if plumbed in series, with the tankless feeding the cold side of the tank.

Simply raising the storage temperature of the tank and installing (the now almost universally required) tempering valve or thermostatic mixing valve on the output of the tank might fix the depletion problem for less work.

If you have five adults showering daily a decent sized drainwater heat exchanger would turn a 50 gallon tank into an "endless shower" situation, and would lower the gas burned for showering by quite a bit (more than half if it's a big enough heat exchanger.

It won't do anything for tub-filling, but for showers it's a huge benefit. If you're interested in pursuing that I can go into more detail. A decent one will run about a grand- they're not super-cheap.

 

[LLigetfa]

Simply raising the storage temperature of the tank and installing (the now almost universally required) tempering valve or thermostatic mixing valve on the output of the tank might fix the depletion problem for less work.

I agree. I've not looked at the specs of the tankless, but don't many of them restrict the GPM if/when they cannot keep up with demand? I suppose if they do that the setpoint would be a factor.

 

[Dana]

Older tankless heaters with smaller burners would restrict flow to maintain output temperature, but most 199KBTU units like that Rheem don't.

Having designed heating systems using a tankless as a boiler I can say with some certainty that at 10gpm ANY of them will be dropping more than 20 psi, many would be dropping over 10 psi, no matter what the setpoint or incoming water temps is. Just because a tankless is "rated" at 9.5gpm at some stated temperature rise doesn't mean your house has sufficient water pressure to actually deliver 9.5 gpm of flow through a tankless. Almost nobody on a private well runs that kind of pressure.

 

[Phog]

The expansion tank shown in the diagram you posted is isolated from the hot water tank, by the check valve located between the tankless & tank.

The check valve is unnecessary, and if you copy this diagram exactly, is even counterproductive from the standpoint of thermal expansion. (The incoming water to the tank will already be preheated by the tankless, so this may not be a big deal in practice. But why put a check valve in, if it doesn't help and might hurt?).

Your tankless unit is 199k btu, which is about the highest powered unit you typically find in residential. It has enough heating capacity to run 2 showers simultaneously in the dead of winter with 34F incoming water.

As Dana alluded to earlier, if you're trying to upgrade the system to handle all your showers running simultaneously, you're going to have to start paying attention to pressure drop. You might not have enough pressure for 3 showers after plumbing the tankless into the tank.

If you think your family can live with being limited to 2 simultaneous showers at any given time, your tankless unit alone should be sufficient without even needing the tank. .. assuming no one tries to use the clothes washer while 2 people are showering ;-)

 

[Jadnashua]

There will be some pressure drop across the tankless system. That may be somewhat overcome if you've got a tempering valve on the tank heater and the storage temperature is high enough to require some tempering.

A tankless is rated with a specific incoming water temperature, often 50-degrees. In the wintertime in many parts of PA, your water temperature can be considerably lower than that, meaning the 9.1gpm just won't happen, at least at the same operating temps as in the summer.

 

[Dana]

A tempering valve has a substantial pressure drop too a high flow, but doesn't go to practically infinity at 10 gpm. With a tankless you can't just crank the temp to 140F and temper down or you may not get enough flow through the tankless to fire it up or keep it firing at modest flows of 105-110F hand washing water out of a faucet.

To get the benefits of high flow really takes a tank. To get the fast recovery time or "endless shower" experience takes a big burner and/or a drainwater heat recovery heat exchanger (or both). There are commercial water heaters with 199K modulating burners, but it sounds like you're trying to re-invent that with the Rheem and some unknown tank heater. You can sort of get there by slaving a pump to the aquastat on the tank and cranking the output temp of the tankless to the max, but don't try to pump 9gpm or even 5 gpm through the tankless all the time or you'll wear it out.

At 4 gpm you're pumping 2000lbs/hr. With a 199K burner that would support 199000 BTU /2000lbs= ~100F temperature rise. The max output temp on the Rheem ECO200 is 140F, so obviously you'll never get the burner to deliver the full 199K into a 130F tank at 4 gpm.

A 130F tank is 10F cooler than the 140F output. 4gpm x 10F is 20,000 BTU/hr, which is about twice the min-fire output, so you should be able to set the tank to something like 130-135F without causing error conditions or flame-out if it's pumping at 4 gpm. With a typical 1/25 hp pump you might only be able to muster 3 gpm, which is fine, but for now assume it's 4 gpm. A stainless "smart" circulation pump could be useful if you're really going to run with this, but they're not cheap.

When the tank temp drops to 110F (tub filling temp) it's a 30F temperature rise, x 2000lbs/hr = 60,000 BTU/hr. When it hits 105F (showering temp) it's a 35F temperature rise, x 2000lbs/hr= 70,000 BTU/hr, which is enough output to support a single endless 2 gpm shower at wintertime incoming water temps, and about twice the burner output of a typical standalone tank, and comparable performance to a 76K condensing tank type water heater.

When the entering water at the tankless is above 125F there's effectively zero condensing going on, so it'll take an efficiency hit, but it won't be short cycling on small water draws either.

You could put a monster sized pump on it and get more BTU/hr out of it, but the tankless will usually be toast in less than 5 years if (ab)using it that way. If running it at 2 gpm or 3 gpm delivers sufficient hot water performance that would even kinder to the tankless than the 4 gpm paradigm.

 

[Tim Ross]

I’m sorry to resurrect a dead thread, but this is great information, I’m looking to do a tankless into a holding tank myself, the tankless will be on its own loop and will not have any impact on flow in the home, it will pull water from the bottom of the tank, circulated by an aqua stat controlled pump, through the tankless and back into the tank. The issue of really variable inlet temps for the tankless concerns me, so I’m wondering if I could utilize a thermostatic valve that mixes in cold water from the line coming into the house with water from the holding tank with hopes it can provide a narrower range of water temps for the tankless to adjust to.


The reason I want to do this is because I have a shower that flows 9gpm and while I love my old faithful AO smith Vertex 100 from back in the day, the replacement costs are too high. Our incoming water temps are 7deg C (ahhh, that is 45deg F I believe), so a tankless will never supply the volume I need on the hot side. The idea of going with a hybrid setup is reasonable IMHO, I’m just wondering if since this thread was created there has been more science done on how to make this work well. The thermostatic valve idea I understand is imprecise and suboptimal but may provide a better range of temp conditions for the tankless to operate with and maybe extend its life span a bit. My goal is to provide a solid recovery rate in the 6GPM range by feeding the tankless warmer water from the holding tank mixed with some cold water from the mains.

if I am way off base, please let me know. 95% chance I’m an idiot. I’m certainly not a plumber but I’m trying to develop this design so the plumber is presented with a well thought out plan. This is not a common configuration.

Thanks all.

Tim

 

[Dana]

I’m sorry to resurrect a dead thread, but this is great information, I’m looking to do a tankless into a holding tank myself, the tankless will be on its own loop and will not have any impact on flow in the home, it will pull water from the bottom of the tank, circulated by an aqua stat controlled pump, through the tankless and back into the tank. The issue of really variable inlet temps for the tankless concerns me, so I’m wondering if I could utilize a thermostatic valve that mixes in cold water from the line coming into the house with water from the holding tank with hopes it can provide a narrower range of water temps for the tankless to adjust to.

With the thermostatic mixing valve in the tankless loop it cold water would/could only be mixed in when hot water was being drawn from the tank. When the cold water flow stopped the thermostatic mixer would cut back the flow to zero, limiting the temperature in the tank to whatever temp the mixer was set to.

As the temperature of the tank gets close to the temperature of the mixer output the flow will drop below the minimum flow required to keep the flame going on the tankless and it'll be prone to short cycling as the temp falls to the point where the tankless will re-light, but the pump will run continuously (and likely cavitate at the high back pressure of the restricted low flow) , never actually satisfying the aquastat if the aquastat isn't set several degrees below the mixer temp (which would render the mixer a bit useless.)

Setting the output of the tankless to it's maximum (typically ~55C for condensing tankless) and setting the temperature on the tank to ~50C would likely get you there without complicating things with the thermostatic mixing valve in the tankless loop. Install the thermostatic mixing valve between the output of the tank and the hot water distribution plumbing of the house, and be sure to select one that is good for the whole 9gpm.

The reason I want to do this is because I have a shower that flows 9gpm and while I love my old faithful AO smith Vertex 100 from back in the day, the replacement costs are too high. Our incoming water temps are 7deg C (ahhh, that is 45deg F I believe), so a tankless will never supply the volume I need on the hot side. The idea of going with a hybrid setup is reasonable IMHO, I’m just wondering if since this thread was created there has been more science done on how to make this work well. The thermostatic valve idea I understand is imprecise and suboptimal but may provide a better range of temp conditions for the tankless to operate with and maybe extend its life span a bit. My goal is to provide a solid recovery rate in the 6GPM range by feeding the tankless warmer water from the holding tank mixed with some cold water from the mains.

if I am way off base, please let me know. 95% chance I’m an idiot. I’m certainly not a plumber but I’m trying to develop this design so the plumber is presented with a well thought out plan. This is not a common configuration.

Thanks all.

Tim

You need to do some math to know what will/won't work.

A (45C-7C=)38C /100F temperature rise at 9gpm (or ~4500lbs/hr) is a heat rate of 450,000 BTU/hr more than 2x (getting on to 3x) the output of a 199K-in tankless. With a sufficiently large drainwater heat recovery heat exchanger feeding it's output to...

1 ...the cold input of the tank...

2 ... the cold feed to the thermostatic mixing valve, and...

3... the cold feed to the shower...

...can cut the heat rate needed to be supplied by the tank + tankless to about half of the 450KBTU/hr. (Again make sure the drainwater heat exchanger can handle the full 9gpm, which may require a commercial rather than a residential version.)

When being filled with ~25C water from the heat exchanger the recovery time on the tank is reduced by half, even when drawn down to near the comfort-limit of a shower (<<40C, a typical showering temp).

Do NOT set the flow on the tankless to 6gpm or you'll shorten it's life substantially. At a 5C rise (50C storage temp, when then tank is nearly fully heated, 55C output temp) and a more temperate 4 gpm (=~2000lbs/hr)t the tankless would be running it'll be burning at a comfortable ~18,000 BTU/hr, (=the minimum firing rate of some tankelss units) but when the bottom of the tank is 25C (roughly the output of a bigger drainwater heat exchanger) that's a 30C (54F) delta-T, and a burn output rate of (54F x 2000lbs/hr =) 108,000 BTU/hr, which is more than the output of a Vertex 100.

Even 4 gpm is a high enough flow rate to wreck the flow sensors on some smaller tankless units in short years, so even though you're never using it at it's max burn rate, a 180KBTU/hr - in (or bigger) tankless would be far better choice than a 150K-in (or smaller) unit.

If a drainwater heat recovery unit isn't in the budget you'll need a (roughly 2x) bigger tank in order to get reasonable showering performance out of it.

You may have to play around with the flow rate on the tankless and the storage temps to optimize hot water delivery on the equipment you choose. Raising the storage temp much higher than 5C below the output temp will cause many tankless units to flame out, or even shut down & spit out an error code.