NoFossil gets surprised

[Nofossil]

Just when you think you've seen it all - I have a client with a zone where, as the supply temp drops, the return temp rises! Since he's using our controls, I had to take a look. Here's a graph. Supply is the magenta line at the top of the chart - top of storage. Return temp from the zone is the red line. As the top of storage temp drops, you can see the return temp go up from about 100 to almost 110.

So what's going on? The zone in question supplies heat to a greenhouse through a flat plate heat exchanger. We control the speed of the circulator on the hot water side to maintain a glycol temperature of 120 (yellow line) on the greenhouse side. As the supply temp drops, the circulator has to speed up from about 20% to about 25% (green line). The higher speed means that the water doesn't lose as much heat passing through the HX, so the return temp actually increases. Very cool and unexpected. Here's a plumbing diagram. We're varying the speed of P15:

 

[ewdudley]

The higher speed means that the water doesn't lose as much heat passing through the HX,

Perhaps you're confusing heat loss per unit time with temperature.

 

[Nofossil]

Perhaps you're confusing heat loss per unit time with temperature loss per unit mass.

I should have said "doesn't lose as much temperature". Clearly, the heat transfer in BTU/minute stays pretty much exactly the same. The glycol-side flow rate is unchanged and the glycol-side heat load is pretty much constant over the period in the chart.

I hadn't thought about it too carefully, and I would have expected that even though the circulator has to speed up a bit, a 30+ degree drop in supply temp would result in at least a small drop in return temp. As it turns out, with this particular heat exchanger in this particular situation, that's not what happens.

 

[maple1]

What happens to the Glycol return temp in that period? Stays constant? Or does it also rise with the rising supply return temp?

 

[Nofossil]

What happens to the Glycol return temp in that period? Stays constant? Or does it also rise with the rising supply return temp?

We don't have that instrumented. Since the flow rate, supply temp, and heat load on the glycol side are all constant the glycol return temp has to be constant as well.
Basically, the amount of heat transferred into the greenhouse did not change over the chart period, and from the greenhouse point of view everything was very stable.
Since the supply temp on the water side dropped, we clearly needed to increase the flow rate. Same BTU/hr with higher flow rate means lower delta T.

On the water side, the initial Delta T was about (188 - 102), or 86 degrees. As the supply temp dropped to around 155, the return temp increased to about 106 for a delta T of 49 degrees.

One unknown variable is the actual flow rate at 20% drive vs. 25% drive. It looks like it must be almost twice the flow. I've always suspected that circulators are pretty non-linear when driven this way (we're using a modified chopped waveform). The circulator is a Grundfos 15-58.

 

[jebatty]

Don't forget the relationship between flow rate and pump head. Double the flow, pump head increases by 3.4. How does this fit the pump curve for the 15-58? Are you really getting doubled flow?

 

[Bob Rohr]

Does P-17 have a check valve?

 

[Hansson]

NoFossil what drawing program are you using?

Your gaugepanel picture make mine look like someting a 6 year child have drawn
http://www.e-logger.se/pub?hansson&p=1

 

[Nofossil]

So - I'm inferring that the P15 flow rate has increased dramatically because it's transferring the same amount of energy per unit of time to the Glycol, but the delta T of the water going through the HX is *much* lower. The flow restriction on the water side is pretty low - almost all 2" pipe. The variable speed drive is pretty non-linear at low speeds.

All the circulators have check valves.

I use an ancient version of Visio, although there's a Google Apps plugin called LucidChart that's pretty dang good even in the free / limited version.

I actually keep a virtual Windows XP machine that I run on my Linux workstation just to run Visio, a circuit board design package, and TurboTax.