Turbulence increases efficiency?
Last Post 06 Feb 2015 09:07 PM by newbostonconst. 16 Replies.
Printer Friendly
Sort:
PrevPrev NextNext
You are not authorized to post a reply.
Author Messages
riggerjackUser is Offline
New Member
New Member
Send Private Message
Posts:18

--
16 Nov 2014 01:16 PM
I didn't want to derail a previous thread, so I started this one.

I said:

"What I did (and don't recommend) is long loops.
The reason pros push parallel loops, is splitting your flow down parallel loops reduces fluid velocity in each loop. Reduced velocity results in less pumping resistance."

and joe.ami said:

"Understand there is more than one way to skin this cat. Tubulent flow helps extract btu's with less loop in the ground, but if you have enough loop you can reduce velocity and collect adequate btu's with laminar flow. That would offset much of the pump expense created by extra long loops."

Now, I'm no expert, and I've seen this come up before, so I wanted to make sure I understand this. For simplicity, I will only use the heating cycle for description.  The theory is that if the coolant moves fast enough, it will be more turbulent in flowing down the pipe. That this will cause the loop to be more efficient, as the temperature of the coolant will be more even in cross section, thus gaining more BTU's per inch of pipe, than a slower speed run, where friction on pipe walls would cause the fluid in the center to move faster than the fluid on the edge. This would result in a temperature gradient within the pipe cross section. Have I got this right?

The worst conductor in the ground loop is the plastic pipe, so anything that increases the temperature differential on each side of that insulating barrier is a good thing for our purposes. Let's throw some numbers out there.

Head loss is from: http://www.engineeringtoolbox.com/pressure-loss-plastic-pipes-d_404.html

If our GSHP calls for 8 GPM (number chosen for simplicity of working the chart)

If this all goes down one 1600' 3/4" loop, that is 9 GPM. Head loss is 201.6 feet.
If it splits between 2-800' 3/4" loops, at 4 GPM, head loss is 56 feet
If it splits between 4-400' 3/4" loops at 2 GPM head loss is 16 feet.

Increasing pipe size changes the numbers entirely, use the chart.

This is just the head loss of the loops, not including any header, bends, antifreeze type or other loss that links to the total pumping power needed.

A quick Googling shows:
http://www.geo-flo.com/downloads/Magna_benefits_flyer.pdf for an optimized comparison of pumping cost. Optimized for their purposes, of course.
http://ca.grundfos.com/content/dam/GCA/Data%20Sheets/Small%20UP/UP_26-99_F_BFC_0311.pdf a flow/head chart for a UP 26-99 pump.
http://ca.grundfos.com/content/dam/GCA/Data%20Sheets/Small%20UP/UP_26-116_F_BF_0311.pdf

An ideal loop field would only use 1 tiny pump to achieve all the flow necessary. A 26-99 pump is the smallest I'm aware of, in regular use. So our example above would use 1 26-99 pump for 4-400' loops, 2 26-116 pumps for 2-800' loops, and a pumper truck from the local fire dept for the single loop (slight exaggeration). Again, this is just the loops, and disregards other head loss.

That's my pie in the sky ideal scenario. In the real world, ground loops are pricey, land is not always available, and we all have to make tradeoffs. Raising fluid speeds makes the loops more efficient per foot. How much more efficient? Is there an online calculator?

I'm new here, and not trying to call anyone out. I'm just a geek trying to make sure I understand this. I also want to clarify concepts and the basis for calculations for the folks who are doing their due diligence before choosing a system/installer. The more we know, the better our decisions.



jonrUser is Offline
Senior Member
Senior Member
Send Private Message
Posts:5341

--
16 Nov 2014 03:08 PM
"Ground Loop Design" software calculates many of these things and has a option for turbulent, transition and laminar flows. In one quick test, laminar required ~8% more length that turbulent. As Joe suggests, something that could easily be accounted for (in various ways). For example, 3/4" pipe/turbulent and 1"/laminar perform the same at the same length and gpm (in another example I tried).
riggerjackUser is Offline
New Member
New Member
Send Private Message
Posts:18

--
17 Nov 2014 09:26 AM
Thanks! That was exactly the kind of thing I was hoping for. So, using that calcutor, how many GPM are required in a 3/4 loop for the turbulent flow?
jonrUser is Offline
Senior Member
Senior Member
Send Private Message
Posts:5341

--
17 Nov 2014 11:27 AM
It depends on the liquid and the temperature. You might also look at this (it will calculate loop flow parameters). Also be sure that the pipe layout is capable of reaching flows that allow it to be flushed.
joe.amiUser is Offline
Veteran Member
Veteran Member
Send Private Message
Posts:4377
Avatar

--
18 Nov 2014 09:05 AM
"It depends on the liquid and the temperature..."

Yes prop glycol uses more pumping energy than methanol for instance.
Joe Hardin
www.amicontracting.com
We Dig Comfort!
www.doityourselfgeothermal.com
Dig Your Own Comfort!
byundtUser is Offline
New Member
New Member
Send Private Message
Posts:5

--
31 Jan 2015 03:12 PM
You determine the transition between laminar and turbulent flow using the Reynolds number (fluid density * velocity * pipe ID/viscosity times conversion factor to eliminate all engineering units). Osborne Reynolds did the original experiments in the 1870s using a filament of colored fluid in the middle of clear water. He found that at low velocities there was no mixing of the colored fluid with the clear fluid. But above a certain point, which depended on the entrance conditions in the pipe, the colored fluid would begin mixing. http://en.wikipedia.org/wiki/Laminar-turbulent_transition My fluid dynamics textbooks taught that the transition between laminar and turbulent was inevitable when the Reynolds number exceeded 2100.

In my part of Florida, the vertical wells are filled with water, which reaches its highest viscosity of 1.79 centipoise at 32 F (just before freezing). Assuming 3/4" SDR-17 pipe, you would reach a Reynolds number of 2100 at 1.1 GPM. If the viscosity were lower (such as through use of higher temperature water), then turbulent flow would occur at even lower well velocities. Conversely, if you use a glycol/water mixture, you would have a higher viscosity at the minimum leaving temperature, and would need a higher well velocity to remain turbulent.

I have a variable speed pump on my WaterFurnace Series 7, so I wanted to make sure that flow was consistently turbulent even during part-load conditions. My coldest leaving water temperature is expected to be 50 F, for a Reynolds number of 6500 in 3/4" SDR-17 pipe at 2.5 GPM. This gives me a 3:1 turndown while remaining turbulent during part-load conditions, which I believe is acceptable.
docjenserUser is Offline
Veteran Member
Veteran Member
Send Private Message
Posts:1400

--
01 Feb 2015 01:57 PM
I am not sure if SDR 17 pipe has a good enough pressure rating for vertical wells, it is only rated for 100 PSI static and 50 PSI recurrent surge pressure. 100 PSI is 260ft of water column. Now ad some PSI from the pump and possibly an pressurized system. There is a reason why SDR 11 is pretty much the standard. We tried SDR 13.5 in horizontal loop field, the pipe did kink too easy, we went back to SDR 11.

While you might loose 8% of in laminar versus turbulent (0.75' versus 1"), keep in mind that surface area is larger with the 1" pipe, allowing more heat transfer. Most importantly, the pumping energy is lower.

So I learned that using more surface area and more loop, and designing for a lower pressure drop, especially with the supply header, will benefit operational costs via less pumping power needed. Especially in a low conductivity pipe the mixing is less important than in a high conductivity environment, like a heat plate or co-ax exchanger. There you need turbulent flow, a lot of it!

If you are having a Reynolds number of 6500 in your loop field, you are wasting pumping power. During part load conditions, your loop field is way oversized, thus turbulent flow is of lesser importance.
www.buffalogeothermalheating.com
jonrUser is Offline
Senior Member
Senior Member
Send Private Message
Posts:5341

--
01 Feb 2015 03:12 PM
I agree, don't design for excessive flow to allow turndown during partial load - turbulent loop flow during partial load doesn't matter (much).

It would be interesting to see some example cost numbers on pumping cost vs upsizing the pipe.
byundtUser is Offline
New Member
New Member
Send Private Message
Posts:5

--
01 Feb 2015 04:05 PM
I'll bite. Why are you so concerned over pumping power? Most of the energy consumption is in the compressor motor.

For my six wells at Re=6500 and 250 feet deep, I have a flow of 15 GPM and a pressure drop of 2.15 psi. Assuming 100% efficient pump & motor, my hydraulic power is 2.06*2.32*15/3960 = 0.019 HP or 14 Watts. Since real pumps have closer to 40~50% efficiency and a motor efficiency around 75%, my real world full-load power consumption might be 47 Watts. Assuming 2000 hours/year operation and $0.12/kWh electricity, pump operating cost without turndown is $11.27 per year. Since I have a variable speed pump, it will actually be substantially less than that. In other words, pump power is not a big number.

I'm not an installer, but rather a happy customer with a new geothermal system. Since I spent 40 years as an industrial heat transfer engineer, I did a lot of calculations behind the scene based on what my installer (20 years experience in geothermal) had quoted. I was initially uncomfortable with the rule of thumb that you size the wells for 20 F difference between ground temperature and water temperature leaving the wells. It was only when I realized that most of the system operation would be at part load that I began to feel comfortable with my installer's sizing practices.

Both the installer and the local utility performed Manual J calcs, and my HVAC was downsized from its original 7 tons of conventional to 5 tons in the geothermal. The Manual J calcs are for peak load at the design day, and they assume that all parts of the house experience those conditions at the same time. In actual fact, there will be a favorable diversity of the loads, because the east-facing part of the house will have a peak cooling load in the morning while the west-facing part will have it in the afternoon. Manual J calcs therefore overestimate the peak load for the house. This is coupled with the fact that most of the operating hours (on a bin basis) will be at more favorable outside temperatures.

I expect relatively few hours of operation at full geothermal capacity. Most of the time, my system will be operating at part load, so the water temperature leaving the well will have a closer approach to ground temperature. I'd like to enjoy the extra benefit from turbulent flow at that point because the closer the water temperature is to ground temperature, the less power the compressor uses.

joe.amiUser is Offline
Veteran Member
Veteran Member
Send Private Message
Posts:4377
Avatar

--
02 Feb 2015 09:20 AM
byundt,
your heat pump and your pumping requirements may be all you are concerned with, but some of us have larger jobs and larger requirements to consider. Multi unit, multi ton, and hydronic jobs are routinely tanked in efficiency by a lack of attention to pumping power. further the use of propylene glycol has a penalty as well.

as an engineer you will recognize the term, best practice. you also recognize that your compressor is more efficient by virtue of many extra stages.

your numbers look good at first glance, but you didn't do the other math half which is demonstrate how much you actually expect to save by raising the LWT a few degrees. and of course there is extra wear on the pump and extra erosive wear on the coax.
Joe Hardin
www.amicontracting.com
We Dig Comfort!
www.doityourselfgeothermal.com
Dig Your Own Comfort!
byundtUser is Offline
New Member
New Member
Send Private Message
Posts:5

--
03 Feb 2015 02:19 AM
WaterFurrnace performance data for a 5 ton Series 7 at 50% load, 6.0 GPM water flow, 700 CFM air flow shows 1.59 kW with 80 F Entering Water Temperature and 1.33 with 70 F EWT. That's about 26 Watt more energy for each degree F increase in EWT. If there is a 10 F approach between EWT and ground temperature and a 6 F temperature drop through the well, then the LMTD in the well is 12.8 F. If turbulent flow improves heat transfer by 8%, then I'd expect about 1 F improvement in LMTD by switching from laminar to turbulent flow. A 9 F approach would give an LMTD of 11.7 F. So that 8% improvement in heat transfer efficiency due to turbulent flow will reduce EWT by 1 F and thereby save 26 Watt of power.

The pressure drop during partload cooling at 6 GPM flow to 6 wells (1 GPM/well) & Re=4000 will be 0.16 psi. Assuming 40% pump efficiency, 75% motor efficiency, the expected pump power is 4 Watt.

So I am spending some fraction of 4 Watt in pump power to change from laminar to turbulent flow, and thereby gain the benefit of 26 Watt saving in energy used by the Series 7.

The issue of extra wear and tear on the pump and coax is moot because I am operating at WaterFurnace recommended flows. Those flows won't change if the wells are running laminar rather than turbulent. The laminar vs turbulent issue really has to do with how many wells you have in parallel to get the desired amount of heat transfer surface in the ground. In my situation, laminar flow during part load means 12 wells each 125 feet deep as opposed to 6 wells that are 250 feet deep.

After watching the well drilling operation at my house, the driller completed one well per day because there was a fair bit of time spent setting up for each well. The actual drilling and pipe stuffing goes fairly fast. I'd expect several thousand dollars more drilling cost if the installer had specified 12 shorter wells (rather than 6 deeper ones) to run turbulent at full load and laminar at part load.

I'll grant you that more complex systems may have a different optimum design. But for the specific case of my house, I believe my installer made a fairly reasonable selection of the number and depth of the wells.


The issue of SDR-11 vs SDR-17 pipe is interesting. I asked what type of pipe would be used and wrote down SDR-17, but I may have asked the wrong person. Even if it truly is SDR-17, however, that represents the standard practice of an installer with a lot of experience installing geothermal systems in this section of Florida. The water table in my neighborhood is somewhere between 10 and 25 feet down, depending on season. My section of Florida has low elevations and very porous soil, so high water tables occur throughout my installer's territory. My lot is 25 feet above sea level and 2 blocks from a bayou feeding an intercoastal bay, so I believe the well driller's assertion on water table depth. Given that, the hydrostatic pressure difference the pipe wall sees is less than 15 psi. That is well within the pressure rating for SDR-17 pipe.

I agree, however, that the water table may be much lower in other locations. So your experience that SDR-17 is irresponsibly flimsy is quite valid for those locations.
jonrUser is Offline
Senior Member
Senior Member
Send Private Message
Posts:5341

--
04 Feb 2015 07:13 AM
If water is 25' down, then for a 125' well, 20% of it has noticeably less thermal performance. Only 10% of a 250' well. This factor may exceed the others.
byundtUser is Offline
New Member
New Member
Send Private Message
Posts:5

--
04 Feb 2015 09:32 AM
jonr,
I thought you were arguing the other side of this issue. Are you now favoring fewer and deeper wells with turbulent flow even in part-load conditions? :-)
byundtUser is Offline
New Member
New Member
Send Private Message
Posts:5

--
04 Feb 2015 07:12 PM
I stand corrected on the wall thickness of well piping.

I doublechecked the wall thickness of the well piping with the field supervisor on my install. He said it was definitely SDR-11, and laughed when I told him the office contact had said it was SDR-17.

Relative to riggerjack's original question, flow will be turbulent at 1.0 GPM of water at 32 F flows through the 3/4" SDR-11 pipe. It will remain turbulent at higher temperatures, and will have a Reynolds number of 4800 at that flow at 90 F.

If the fluid in riggerjack's well is a glycol mix, the viscosity will be higher and you will need more flow to remain turbulent. A 20% mix of propylene glycol and water has a viscosity of 4.23 centipoise at 32 F--and you would need a flow of 2.4 GPM to become turbulent.
docjenserUser is Offline
Veteran Member
Veteran Member
Send Private Message
Posts:1400

--
05 Feb 2015 03:44 PM
Watts Total gpm
16 5.1
23 5.7
26 6.2
33 6.7
53 7.9
85 9.5
122 10.8

See above the electrical consumption in watts needed for a Grundfos Magna 140 pump at different flow rates. Pumping power is different from electrical consumption of a pump. Higher efficiency does not mainly come from turbulent flow, but from a higher delta T between the fluid and the refrigerant dues to the fluid having lesser time to get colder, since it remains shorter period of time in the heat exchanger.
www.buffalogeothermalheating.com
jonrUser is Offline
Senior Member
Senior Member
Send Private Message
Posts:5341

--
06 Feb 2015 07:31 PM
My point is "run the numbers" and don't get stuck on something like "it's always best to have turbulent loop flow".
newbostonconstUser is Offline
Advanced Member
Advanced Member
Send Private Message
Posts:735

--
06 Feb 2015 09:07 PM
I took a killawatt meter and plugged one pump at a time into it. I changed speeds on the pump each day to compare the watt usage to the previous day (provided the temp, wind, and sun were similar between the days). I found the most efficient setting. It was easy and the difference between settings was more then I thought. In one room I increased the pump speed from optimal to help keep the temp more constant across the floor surface.

I have five zones with thermostat's one each zone and three speed grundfos pumps.
"Never argue with an idiot. They will only bring you down to their level and beat you with experience." George Carlins
You are not authorized to post a reply.

Active Forums 4.1
Membership Membership: Latest New User Latest: janvin New Today New Today: 0 New Yesterday New Yesterday: 0 User Count Overall: 34705
People Online People Online: Visitors Visitors: 143 Members Members: 0 Total Total: 143
Copyright 2011 by BuildCentral, Inc.   Terms Of Use  Privacy Statement