Figure 1
If you attended the RPA RadFest East this past October, you may remember the piping schematic in Figure 1 as an example of what not to do.

The problem here is the lack of a thermal trap. True, you do see a trap in the plumbing sense. The injection riser piping does drop at least 18 inches below the primary loop before making the U-turn. But this piping geometry doesn't constitute a thermal trap. The reason is that the closely spaced tees that connect the injection risers to the primary loop are not at least 18 inches higher than the tees that connect to the distribution circuit.

Figure 2
Figure 2 shows the situation inside the trap when the injection pump shuts off after having shuttled heat to the distribution system.

The “U-tube” shape of the injection risers below the dashed line in Figure 2 is “buoyancy-balanced.” They do not create any unbalanced forces that would lead to buoyancy-induced heat migration. This is true for both the hot and cool U-tubes, even though the weight of hot water is less than that of cool water. It remains true regardless of how deep the U-shaped piping assembly is built.

Unfortunately, the injection risers above the dashed line are not buoyancy-balanced. Although the hot injection riser contains the same volume of water as the cool injection riser, the density of hot and cool water are not the same. This means the weight of water in the hot riser is less than that of the cool water in the cool riser.

Just like adding a slight extra weight to one side of an otherwise balanced seesaw causes it to rotate, the unbalanced buoyancy forces tend to keep the injection flow moving after the injection pump turns off. As long as hot water continues to flow through the primary loop (i.e., to supply other secondary circuits), this situation will persist. The slow thermosyphon effect continually “refuels” itself with hot water from the primary loop. The result will be heat migration into the distribution system, especially if it operates with continuous circulation. You will get a callback on this.

Figure 3
One relatively simple fix for this situation is to bring a portion of the distribution system down below the primary loop so that the closely spaced tees connecting the injection risers to the latter are at least 18 inches below the tees connecting the risers to the primary loop. This is shown in Figure 3.

Other important details for injection-mixing assemblies include:

  • Closely spaced tees at both ends of injection riser piping;
  • Providing purging provisions for the distribution system;
  • Providing a minimum of six diameters of straight pipe upstream and four pipe diameters downstream of the closely spaced tees;
  • Keeping electrical components away from potential water drips; and
  • Providing straight piping at all circulator inlets.


Figure 4

Harvey's Solution

My long-time associate Harvey Youker of Hytech Heating has developed a relatively simple detail for providing these injection mixing details in a good-looking piping assembly. I call it the Harvey Horseshoe. It's shown in Figure 4.

Harvey has used this detail on a number of multiload hydronic systems where two or more simultaneous water temperatures are needed. The approach is really pretty simple: Create a common primary loop and “populate it” with a separate mixing assembly for each water temperature required. If necessary, each distribution circuit can then be divided into two or more independently controlled zones using circulators, zone valves or valve actuators on manifolds. A call for heat from any of attached loads powers up the associated distribution circulator, injection controller, and allows the boiler and primary loop circulator to operate.

Harvey installs mixing assemblies that supply high temperature loads near the beginning of the primary loop and the low temperature subsystems near the end. In some systems, this allows the design temperature drop around the primary loop to be 20 or even 30 degrees F. A small 55-60 watt circulator can almost always serve as the primary loop circulator.

Figure 5
Figures 5 and 6 show how nicely this all looks in the mechanical room for a system supplying three mixed water temperatures and DHW. Be sure you check out the details such as:
  • A definite 18-inch thermal trap;
  • Closely spaced tees at top and bottom of injection risers;
  • The “drip loop” on the MC cable to prevent water from ever following the cable into the circulator junction boxes;
  • All circulator junction box openings pointing down for the same reason;
  • Approximately 12 diameters of straight pipe ahead of all circulators;
  • At least 6 diameters of straight pipe ahead of, and at least 4 diameters downstream of, closely spaced tees;
  • Electrical controls are not located where water can drip on them;
  • Easy accessibility to all serviceable hardware;
  • Well-supported piping; and
  • Neatly organized and well-protected wiring.


Figure 6
This approach does have some redundant hardware relative to designs that consolidate two or more mixing functions into a single controller. However, redundancy isn't all bad. If a given controller or its transformer fails, only that portion of the overall system is off-line until help arrives.

Harvey and I occasionally toss around ideas for further improving the horseshoe detail. Perhaps each mixing assembly could be made more compact with new primary/secondary fittings, or specialized purging valves. Maybe portions of the assembly could use bent tubing rather than 45-degree ells for the offsets. All of us should be looking for such potential improvements every time we design or fabricate a system. It's not only part of being a hydronic heating professional, it's fun! We welcome any suggestions that you might have.

Wishing you and your loved ones a blessed Christmas.