Several years ago I designed a floor heating system for a new addition to a church. The original building was heated by a steel fire-tube boiler connected to a one-pipe steam system. The boiler had a tankless coil water heater and maintained a minimum standby water temperature year round. The new system would use an indirectly-fired storage water heater in place of the tankless coil. The existing tankless coil was replaced with a new (clean) coil that would serve as the heat source for the floor heating system. This allowed water in the steam system to remain isolated from that in the floor heating system. Not surprisingly, the existing boiler was well oversized, and had plenty of extra heating capacity for the addition. A schematic for the system is shown in Figure 1.
20/20 Hindsight: When the system was turned on we quickly discovered a problem. Although the floor heating system worked fine, we couldn’t generate hot water in the indirect tank. I assumed the DHW circulator was either defective, stuck or plugged with debris. We opened up the circulator and found everything fine inside. It ran smoothly when power was applied through the DHW tank aquastat. OK, so the circulator is running, but there’s no flow through the tank’s heat exchanger circuit. When does a circulator that’s running in a fully purged and unobstructed piping circuit not produce flow? As we discovered, it’s when the maximum pressure differential the circulator creates is smaller than the reverse pressure differential the system applies across it.
In this case the larger circulator, which was wired for constant circulation, was forcing enough flow through the tankless coil that its pressure drop was greater than the maximum pressure differential the small circulator could develop. The check valve in the DHW circuit was actually back-seated whenever the larger circulator was running. The small circulator just didn’t have the “humph” to get flow started in the intended direction. If only water would follow those little arrows on the schematic!
The lesson to be learned is this: Be careful if you’re installing different capacity circulators in parallel through a common heat source. If the header piping and heat source the circuits have in common have low head losses you may be fine. But watch out for heat sources, and/or header piping with high head loss characteristics, (like the tankless coil). They provide the necessary “bottleneck” to flow that can set up the situation I experienced.
Foresight Vs. Hindsight: It’s possible to predict when parallel pumps will interfere with each other. The answer lies in the curves. You’ll recall from previous columns that the intersection of a circulator’s pump curve and the piping’s system curve determines the flow rate and head differential the circulator operates at. By plotting a system curve for the common piping, heat source and distribution circuit supplied by the larger circulator, along with its pump curve, you can determine the flow rate through the common piping and heat source when the circuit is operating. To be conservative assume the three-way valve shown in Figure 1 is in the fully open position, and thus has no bypass flow of return water.
The next step is to use this flow rate to determine the head loss across the common piping and heat source, (blue piping between points A and B in Figure 1). This head loss is what the smaller circulator will have to overcome to establish flow in its own branch. Bottom line: If the head loss across the heat source and common piping is greater than the maximum head the small circulator can develop, there isn’t going to be any flow in the branch supplied by the small circulator.
Figure 2 shows the system curve and pump curve involved. Notice that the head loss of the common piping and heat source at the flow rate created by the large circulator is greater than the maximum (stagnation) head of the smaller circulator. For acceptable performance using this piping layout, (assuming both circulators are operating during domestic water heating), the maximum head of the DHW circulator should be judiciously higher than the head loss across the common piping and heat source. How much higher depends on what flow rate through the DHW heat exchanger is necessary for proper performance.
Multiple Solutions: The pump interference problem I’ve described could have been prevented by using primary/secondary piping to “uncouple” each circuit from a common boiler piping loop. A redesigned schematic using this approach is shown in Figure 3. In a true primary/secondary system each circuit operates as if it’s an isolated series circuit. It’s a great way to avert pumping problems when you’re planning complex multiload systems.
There are several details worth noting in Figure 3. The DHW tank is the first secondary load supplied by the primary circuit. This ensures the hottest water will be available to minimize DHW recovery time. Flow-check valves are shown on both up-going risers of the floor heating circuit to prevent undesired heat migration when the primary circuit operates for DHW only. Purging valves are shown on the return pipe of each secondary circuit to assist in filling and flushing the piping at start-up. A balancing valve is installed in the primary circuit between the tees leading to the tankless coil. This is adjusted to determine what portion of the primary loop flow will be pushed through the tankless coil. This keeps the primary loop flow up and, (when properly adjusted), prevents possible erosion damage to the coil from excessive flow velocity. The primary circulator pumps away from the expansion tank. And finally, the piping to the DHW tank is thermally trapped to prevent overheating the tank when the primary circuit is operating for floor heating, as well as stop reverse thermosiphoning flow that would otherwise be created by the vertical coil in a tank full of hot water when the primary circuit is off. A flow-check valve in the DHW secondary circuit could also accomplish this.
Another possible piping solution would be to couple the high flow resistance heat source to the header loop piping as its own secondary circuit, (with its own circulator). This essentially removes the head loss of the tankless coil from the flow path of the circulators. Keep in mind, however, that the head loss of the common header piping remains in the flow path of both branch circuits. It’s prudent to generously size both the supply and return header piping to minimize this head loss.
An electrical (as opposed to piping) solution for the original system was to install a relay that converted domestic water heating into a “priority” load. Whenever there’s a call for DHW heating the large circulator is temporarily turned off allowing the small DHW circulator to do its duty without interference. The large thermal mass of the floor heating system could easily withstand a temporary heat interruption during a call for domestic water heating. Because of its simplicity this is the method we used to remedy the pump interference problem in this system.
All circulators are not created equal. The bigger ones will definitely “bully” the smaller ones if given the opportunity. So use your knowledge of piping design to predict and eliminate a “circulation altercation” before installing the goods and seeing what happens when you throw the switch. This approach may not add to your collection of hydronics war stories, but it probably will add to your bottom line.