HPAC Magazine

One side or the other

March 1, 2015 | By John Siegenthaler

Alternate ways to pipe up a buffer tank.

Short operating cycles continue to be one of the chronic complaints associated with modern hydronic heat sources. State-of-the-art mod/con boilers with 5:1, 8:1, or even 10:1 ratios between their maximum firing rate and minimum stable firing rate cannot always match the heating load imposed by a single small zone, such as a towel warmer radiator in the master bathroom, on a mild day. This is where the additional thermal mass provided by a buffer tank provides the “thermal elasticity” needed between heat supply and heat demand.

Over the last couple of decades, the North American hydronics industry has been learning the importance of buffer tanks as low thermal mass boilers and water-to-water heat pumps have increasingly been combined with highly zoned distribution systems.

As biomass fueled boilers make their mark on the industry, one of the principles that most system designers agree on is that a properly sized buffer tank is essential to good performance for wood gasification boilers, as well as boilers fueled by pellets and wood chips.

In North America, one of the most common arrangements for a buffer tank is to install it between the heat source and distribution system as shown in Figure 1.

“Hot” water from the heat source enters an upper side wall connection on the tank. Water headed for the distribution system exits from another upper side wall connection that is usually directly across from the “hot” water inlet.

Because the entering water is hotter than the water in the tank, it is slightly less dense and remains in the upper portion of the tank. When the flow rate into the distribution system is about equal to the flow entering the tank from the heat source, the entering hot water tends to “slide” across the upper portion and does not disturb the cooler water in lower portions of the tank.

Notice that a swing check is shown on the pipe leading into the tank from the heat source. It is there to stop reverse thermosiphoning during times when the heat source is not operating but the tank contains heated water. The upper right connection leading from the tank to the distribution system also contains a check valve.

I suggest using a spring-loaded check valve in this location to stop potential forward thermosiphoning when the distribution system is not operating. A swing check does not have sufficient forward opening resistance to stop such thermosiphoning, while most spring loaded check valves have a forward “cracking” pressure of about 0.5 psi, which is generally enough to stop this unwanted flow.

Cooler water returning from the distribution system tends to remain in the lower portion of the tank and slide across to the bottom left sidewall connection back to the heat source. These effects are both desirable because they help maintain temperature stratification in the tank.

If the flow rate from the tank to the distribution system is greater than flow rate entering the tank from the heat source, cooler water will begin to migrate upward in the tank. The tank is giving up heat to the load under this condition. If the flow rate from the heat source is greater than the flow rate to the distribution system, hot water will begin to migrate downward in the tank. The average tank temperature increases as the surplus output of the heat source is stored. The tank is taking on heat under this condition. Both of these conditions are shown in Figure 2.


Besides providing thermal storage that allows reasonably long operating cycles of the heat source, a buffer tank piped as shown in Figure 1 also provides excellent hydraulic separation between the heat source circuit and load circuit. This happens because the internal flow velocities within the tank are very low compared to the flow velocities in the piping connections to and from the tank. Due to these low velocities, there is almost zero head loss across the tank or from the top to the bottom of the tank.

The low flow velocities also allow dirt particles, which may be present in the flow returning from the distribution system, to drop to the bottom of the tank. The problem is that most buffer tanks are not designed to efficiently flush dirt at the bottom of the tank out of the tank. Dirt that happens to settle near the drain valve may get entrained with flow out of the lower tank drain valve.

The low local flow velocities in other lower areas of the tank cannot effectively entrain dirt, and thus are unable to carry it to the drain connection. Therefore, most buffer tanks can eventually separate and accumulate dirt but are not good at flushing that dirt to a drain. This is where a dirt separator that can generate sufficient internal flow velocity during a flush to entrain accumulated dirt is the preferred choice.

The heat stored in a buffer tank can also be used for domestic water heating, or preheating, using either an internal coil heat exchanger suspended in the upper portion of the tank, or the “on-demand” assembly shown in Figure 3. This assembly has been discussed in previous articles (see HPAC May/June 2013, www.hpacmag.com). It is an excellent way to harvest heat from any thermal storage tank.


The piping shown in Figures 1, 2 and 3 all involve four principal piping connections to the buffer tank – two into the upper portion and two into the lower portion. Although these principal connections can function well, they are not the only way to connect a buffer tank into the system.

After looking over many schematics from European sources, especially those associated with biomass boilers, I have noticed a trend that places the buffer in a different arrangement relative to the heat source and load. This alternate arrangement is shown in Figure 4.

In this layout, flow from the heat source does not pass through the buffer tank on its way to the distribution system. Instead, the distribution system extracts flow from the piping between the buffer tank and heat source. Any hot water flow from the heat source that is not pulled off into the distribution system passes downstream into the buffer tank, as shown in Figure 5.

In this two-pipe buffer tank scenario, the flow velocity entering the buffer tank is lower than with the “four-pipe” arrangement shown in Figures 1 through 3. Lower entering flow velocities help preserve temperature stratification and thus maintain the warmest water at the top of the tank, ready for transfer to the load.

If the heat source is off, stored hot  water from the buffer tank flows backward from the tank and into the distribution system at point A. It is also possible for some flow to enter the distribution system from the heat source, while the remainder of the required flow comes from the buffer tank. This happens when the distribution system requires more flow than is currently passing through the heat source.

Another benefit of the two-pipe buffer tank configuration is that the distribution system has access to the hottest water in the system before that water passes through the upper portion of the buffer tank. This would be an advantage if the buffer tank has cooled over several hours before the next call for heat occurs. Under such a condition, the cooler water in the upper portion of the tank would mix with incoming hot water from the heat source. This would thermally dilute the water temperature supplied to the load until the upper p
ortion of the tank has warmed back to the normal operating temperature. This effect would be especially noticeable when the system is first started and the
water in the buffer tank is at room temperature.

Yet another benefit is that the piping shown in Figure 4 eliminates two of the principal sidewall connections on the buffer tank. This should also reduce the cost of the tank, with all other specifications being the same.

Still, in my opinion, it is better to work with buffer tanks that have more connections than is absolutely necessary for a given application. The extra connections can always be plugged off if not needed. They can also be used to attached devices such as sensor wells, thermometers, sight gauge connections, or piping connections that better match incoming flows to the likely temperature stratification of the tank. Extra connections might also be used to accommodate the on-demand DHW assembly shown in Figure 3.


The degree of hydraulic separation between the distribution circulator and heat source circulator provided by the system in Figure 4 depends on the length and size of piping between the tees where the distribution subsystem connects and the connections to the tank. This piping should be short and generously sized to minimize head loss; thus the callout for short/fat headers in Figure 4.

If the buffer tanks have the four principal piping connections shown in Figures 1 through 3, it is possible to join two or three of them together in a “close coupled” arrangement as shown in Figure 6.

This piping allows the two buffer tanks to maintain flow dynamics very similar to the single “two-pipe” buffer tank shown in Figures 4 and 5. To maintain good hydraulic separation across the tank array, the couplings between the tanks should create minimal head loss. A short flexible connector works well and compensates for any slight misalignments of the tank connections due to manufacturing tolerances, or a less than perfectly flat floor slab under the tanks.


Either the traditional buffer tank piping shown in Figures 1 to 3, or the alternate method shown in Figure 4 and 5 can work. Both have been used on many successful installations. It is likely to come down to what tanks are available, how the piping connections on the tank are sized and located, and how those tanks would be optimally located in relationship to the other components and subsystems. <>

John Siegenthaler, P.E., is a mechanical engineering graduate of Rensselaer Polytechnic Institute and a licensed professional engineer. He has over 34 years experience in designing modern hydronic heating systems. He is also an associate professor emeritus of engineering technology at Mohawk Valley Community College in Utica, NY.

See John at Modern Hydronics-Summit 2015 on September 10 at the International Centre in Mississauga, ON.

Visit www.modernhydronicssummit



Stories continue below