Help with pump sizing and head loss calculation

I am trying to design a radiant floor heating system for my house. I think that I have enough knowledge to just go ahead with what I know so far but I really want to design this system with all aspects of efficiency in mind.

This brings me to the difficulty of pump sizing. After scouring the internet I was finally able to find a pdf document that did a fair job of explaining how to calculate feet of head loss but I still have questions.

formatting link
How do you calculate for parallel circuits between headers? I'll attempt some ascii art here to explain.

_______________________________________ | | | | | | | | | | | | | | | pump | | | | | | | | | | | | | | | | | | | ---------------------------------------

The first problem I have is which came first; the chicken or the egg? You have to know the flow to calculate head loss but you have to know the amount of head to see what the flow is for a given pump curve.

Sorry, back to the drawing. Assume that all pipe is 3/4" copper; the supply and return headers are 100' long each and each parallel circuit is 25' long. (This is not the actual design just a simple example).

The headers elbows and tees are easy enough to calculate but I can't figure out the circuits. According to that pdf I think they would say to just include the circuit with the most head loss. I don't think that's right though. Other things I have tried are to combine the pipe areas and calculate the equivalent diameter of pipe; divide the given flow by the number of circuits and add their head losses together; and I've thought about calculating for equivalent circumfrence of the combined pipes. None of the options really jumps out at me as the answer. Can somebody please help clarify?

Reply to
ElMalo
Loading thread data ...

Hey, I followed you from alt.hvac, lot of crap on that site.... I think (based on paper you showed) you need to calculate head from farthest run, and flow from total of all pipes. (Same size pipe is kinda' irrational.)

I did instrumentation for years, and could tell you how full a high tank was from pressure, then HVAC controls and we only considered height of building for close off pressure on valves.

For these engineers.... Doesn't the height of a system (building) affect the head that a pump needs to create? If you read paper (below) they show coils in examples at different heights, then no allowance for height. In a sealed pumped system, does height not matter? Does it not matter until certain limits? 33ft?

formatting link

Reply to
Tim or Marty Shephard

Dear ElMalo:

formatting link

Do this instead:

----+-+-+-+ | | | | +-+-+-+--+ |

------------+

A little goofy looking, but will provide the most equal paths possible without valving and meters.

Yes, but it settles down pretty quickly. Usually in 3 or 4 iterations you'll find it is between two reasonably close flow rates.

With these assumptions:

100 + 25/n + 100 would be my first pass estimate, with n being the number of parallel paths.

Copper pipe has limitations to the number of feet per second they recommend you flow through it (to prevent erosion). Find out what that is, and pick a recirc pump based on this flow, and maybe a little less.

David A. Smith

Reply to
dlzc1.cox

Dear Tim or Marty Shephard:

yes, but the fittings are cheaper. The math is more complex also, for the equivalent "circuit".

If you have to lift it up into the building, yes. Once it is within a story or two of the point of use, and pressurized as hot water needs to be, you don't much need to worry about cavitation.

Always need to pay attention to the "net positive suction head". As I said, of it is already pressurized because it is a hot water system, usually no need to worry.

If you are going to suck the fluid out of the ground, then 32 feet will be a stretch. 10 feet is even out of the capacity of cheap pumps. Most hot water recirculators are even worse than this.

David A. Smith

Reply to
dlzc1.cox

ElMalo,

You have taken on a project that is very difficult to figure out *exactly*. This brings on some questions. Do you understand all that applies? Going by the example you have, I'd say not, but that is very presumptuous. Do you want the technical challenge of an *exact* solution, or do you want a practical solution that will work good enough? I'll suggest some searches and references available on the www, add some comments, and let you decide. Good luck.

Not only do the flows and pressure drops have to be figured out, but how well the flows are distributed in the system to be installed is of utmost importance. That is if you want all the parallel circuits to have equal flows; you may not.

I suggest you use your favorite search engine, mine is Google, and search on some keywords and phrases. I ran two searches using "FLOW DISTRIBUTION" MOMENTUM EFFECT (where the parentheses are placed is critical) and "ANALYSIS OF FLOW DISTRIBUTION". Two excellent articles are:

formatting link
formatting link
Btw, you are using tubing, not pipe, and the headers should be larger in diameter than the parallel circuits. For good flow distribution using cylindrical pipe and tubing (no tapered headers), you want most of the pressure drop in the parallel circuits and very little in the headers.

Reply to
John

Height only matters if you are pumping out of a system such as tank filling. In a completely filled system, the pump is only moving fluid around the system so height is not a factor. This is the case whether the system is sealed or vented.

John

Reply to
John Manders

Dear John:

The problem occurs right at 3/4" too. Commonly 3/4" *is* pipe, with a 3/4" ID. But it can also be obtained as tube size 3/4", namely 3/4" OD. Big surrpises if you don't trial fit the parts ahead of time.

David A. Smith

Reply to
dlzc1.cox

formatting link

Reply to
MLD

Take a look at this:

formatting link
EPANET is a Windows 95/98/NT program that performs extended period simulation of hydraulic and water-quality behavior within pressurized pipe networks. A network can consist of pipes, nodes (pipe junctions), pumps, valves and storage tanks or reservoirs. EPANET tracks the flow of water in each pipe, the pressure at each node, the height of water in each tank, and the concentration of a chemical species throughout the network during a simulation period comprised of multiple time steps. In addition to chemical species, water age and source tracing can also be simulated.

EPANET can also be used to plan and improve a system's hydraulic performance. Pipe, pump and valve placement and sizing, energy minimization, fire flow analysis, vulnerability studies, and operator training are just some of the activities that EPANET can assist with.

Reply to
John

Thank you and many apologies. First, it doesn't look like my ascii survived past the original format. Second, cross posting seems to have created two threads; maybe that's why it's a no no. Third, I'm posting via google so I have to wait a few hours before this appears so I'm slow to respond.

Anyway,

mmm, that's a loaded statement. I'll need to absorb that for awhile.

I'm not an engineer but sometimes I impersonate one. I acquired a coffee mug that says engineer on it and it fools 'em every time (until they start asking questions then it's time for evasive maneuvers).

I think technically (and remember I'm not really an engineer; just acting) that head is determined by the resistance in the system. The pump will only produce so much flow at a certain head.

I don't think they intended fo rthe coils to look like they're at different heights.

I think you're right here. Height doesn't usually matter in a closed system but you might experience some cavitation at extreem heights or low pressures. That would just be bad design though and could be remedied by pressurization and better pump placement and sizing.

Is this from experience? I thought that the flow would be higher in the middle two runs regardless of which way you ran the water. Oh wait, my drawing didn't show it but the headers would extend 2.5 diameters past the last run to avoid having velocity pressure in the last run and static pressure in the rest. Maybe you're right though. Putting elbows on the first and last runs might really even out the flows. I'm not that concerned about it though.

I think that I've just discovered that somehow the water heater or boiler that I use is what will dictate the flow rate. I'm not sure why this is yet or how. I'm looking at a Takagi Jr. boiler right now. Don't know enough about boilers yet.

no

somewhere in between or both.

I don't think even flows is that important in this application. The temperature rise should be pretty gradual and the circuits should have a tendency to stabilize toward each other.

Thanks for the links and search tips. I have yet to check them out but I definately will.

Actually, I'm using two 45' long 3/4" headers and eight circuits of PEX tubing each being 233' long. The PEX diameter in this case is being dictated by the necessary BTUs. I would have liked to go with

1/2" PEX but at only 20 BTUs per foot for a crawl space installation I won't be able to meet my requirement of 55,000 BTUs. 1" copper is out of the question because I'm on a budget and I can't justify the huge jump in price.

Do you think flow distribution will really be a problem?

MLD-Something else that my brain can't absorb just yet (It's late and I've spent way too long responding to this thread). The technique sounds promising though.

There are simple calculations for determining volatge drops across parallel circuits. I guess I was hoping there would be something similar in the fluids world.

Thanks everyone. Sorry for the long post that runs off the page; I hate those. And TIA for future answers and comments, I really appreciate them (sometimes sooner than later depending on brain cell activity or how far into my "engineering" facade (coffee) I've fallen into).

Reply to
ElMalo

hello all,

Elmalo, first produce a schematic of your model. it's a close network, ok specify each flow rates of your heaters. this is your start point. once you know all the flow rates then draw your network on a scaled plan. which fluid are you using? ( for ex water) get a nomogram for water at X degrees ( different temp of water would have different viscosity) the nomogram would give you head losses for the pipe material ( you would get this chart from the pipe supplier / dealer ) for ex: nomogram for copper pipe ( grade XX) for water at XX degrees.

david posted this pipe arrangement, it's called the two pipe reverse connection.

----+-+-+-+ | | | | +-+-+-+--+ |

------------+

this will allow equal flow of the fluid in the heating units.

now for sizing the lines. start from the furthest heating unit, take it's flow rate. comme to the 2nd least furthest unit, add its flow rate with the

1st one, continue like that until you come to your pump . use the nomogram set a fixed pressure drop say 4cm/m of column of water ( 1m column of water = 9.8 Kpa) this would mean for every 1 m of piping, you're restricting the drop to 4 cm of column of water. using the nomogram you will get different pipe dimensions.

now coming to the pumping capacity. add all the different flow rates to have an indication of the pump flow required. then choose the heating unit with longest path from the pump having the highest head loss ( adding losses due to pipe length and fittings) . if the pump is able circulate water to this point, then it will be able to supply to the rest of the network ( less head losses)

you have an indication of the pump flow rate and the drop to overcome. you can select the required pump from catalogues from this information. now to be sure of your choice, you have plot the system curve (different flows along the network) , where the network flow rate and the pump flow rate coincide, you will have the optimum pump capacity.

it's not very evident to understand all that i've said, it's a bit confusing i would say. but hope this helps..

-- Hasta Luego

Irshaad (Faster than Bruce Lee)

remove the X to reply snipped-for-privacy@softhome.net

Reply to
Irshaad

There are several things to watch out for during your design.

  1. The required flow rate in each zone is a function of many variables. Tube size, thermal resistance, zone heat load, liquid temperature, pressure drop. You should engage a radiant heat manufacturer such as Wirsbo to provide you with the basic parameters of the tubing layout. They use software that will calculate the necessary length, size, flow rate and , temperature required for each zone. Without this, you are taking a big risk with the effectivness, and operation of your system. There should only be enough tubing in each zone to *match the heat* load. Any less and you will have some zones that can't keep up. If there is too much tubing, the flywheel effect of temperature control will increase. I believe that Wirsbo will do a grid design for you for free if the product you spec is their product. Also other manufacturers may do the same.
  2. Once you know the design parameters for each loop. (flow rate, temp. drop, friction loss) ,then the pump should be sized to supply enough pressure to the zone with the highest pressure drop. IE when all of the zones are calling for heat, the total flow will be passing through the main header. Add the pressure drop of the total flow through the header and return plus the pressure drop of the parallel circuit that has the *highest* pressure loss.
  3. Also consider what happens to the flow rate through the boiler when all the zones are not calling for heat. IE does the flow through the boiler stop? If so, a pressure bypass is required or a primary secondary loop system should be considered to always have fluid circulating through the boiler.
  4. The approx. maximum temperature in the supply to the tubes should be not much more than 100F. You need a way to achieve this temperature. This is done by several methods with more information available here.

Download the following literature: from

formatting link
E 006 Control of Radiant Floor Heating Zones

formatting link

Only the circuit with the highest resistence is considered and added to the supply and header losses. Its that simple.

Kevin

Reply to
Kevin

Dear ElMalo:

There are, which is why I wrote down 100+25/n+100. However, you are using a smaller diameter for the cross tubes (making 25/n wrong), which makes good sense, but makes the math more difficult. And electrons don't have dominant momentum, as does fluid, and tee connections and elbow connections are different with fluid than with electrons.

Mesh and node equations will solve it all, but the resistance-to-flow of the pipes will need to be correctly modelled. And the resistance value is

*not* linear with flow rate unless your flow is laminar. This will most likely be true, unless you put "more power" in (grunt, grunt).

You can estimate the relative resistance of the various pipe/tube IDs, using the Hazen-Williams formula:

formatting link
David A. Smith

Reply to
dlzc1.cox

Thanks again everybody. Lot's of great information which is exactly what I was searching for. Sometimes I can Google all day to no avail just because I don't know what words to search. I also got straightened out on a few things. That bit about the two middle circuits having more flow than the others was some remnant of information concerning passive solar collectors; a completely different dynamic there.

I knew that what I was asking for was some pretty detailed stuff. I don't think that'll I go through and try to get an exact answer but I can get a much better one with the info I have now.

I actually have some Wirsbo software. They sent it to me a long time ago but it looked mostly like a job price calculator for contractors.

Anyway, thanks again for all the info and thanks for not nailing me to the wall. I'm always interested in all the info I can get so if someone has more please post it and I'll be checking back. Please don't send me any email unless it's spam. I don't have time to check The Bad Priests email only his spam.

The Bad Priest

Reply to
ElMalo

Dear ElMalo:

Yes. When transferring heat out of water, the kinematic viscosity changes. This will allow more hot water to enter that space more easily. The net result on almost any "reasonable" design from transferring heat from water to a space is that you achieve a net "constant" temperature surface. Not so on transferring heat *to* water.

Good luck.

David A. Smith

Reply to
dlzc1.cox

Dear ElMalo:

formatting link
kinematic viscosity of water is in the denominator of the Reynolds number. As the water cools:

- the kinematic viscosity of the water increases,

- the Reynolds number for a particular pipe decreases,

- the head loss for a given flow decreases (being positively associated with head loss),

- the flow increases bringing in more hot fluid.

Nothing quite like a more complete answer... for a searcher.

David A. Smith

Reply to
dlzc1.cox

PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.