Would it be bad to use a T connector on it's side?
 

I'm going to replace my res with a fill and bleed and I was wondering if it would be bad to use the T connector sideways, as in:

http://www.gregstiffler.com/fill.JPG

I know that 90 degree elbows hamper flow, so this would basically be doing the same thing. I am wondering how much it would hurt my performance.

I just did a web search and the closest answer I can find is in this plumbers handbook.
If you look on page 35
HEAD LOSS (H) = A x (Q/N)(squared)
A = fitting loss factor (see table)
Q = flow rate in liters/sec
N =calculation factor for DN100B

Now, guessing a bit about the ID/OD relationship of "nominal" metric copper tubing sizes - I would guess that nominal DN15 is actually about 1/2" ID. (Speaking of guessing - I'm also guessing you're using 1/2" ID)

Look on page 36 for 'N' values for DN15. There are several types of copper tubing - you're not using copper tubing, so the best we're going to get is a guesstimate. I'd use 0.1

Now look at page 37, "Fitting loss factors - A"
Under DN15 / 'T's , there's
Branch (turn the corner) 0.122
Flow (straight through) 0.046

You didn't mention what pump you were using. pHaestus did a pump comparison which might have your pump listed.

I'm going to guess 1GPM (or about 0.063 L/sec)

So, you've got
headloss = A x (Q/N)(squared)
Branch 0.122 x (0.063 / 0.1)(sq) or .122 x 39.69 or 4.8
Flow 0.046 x (0.063 / 0.1)(sq) or .046 x 39.69 or 1.8

So using the branch offers about 2.7 times the head loss as straight through (pretty much the same as the relationship in 'A' values) but your flow will go down as resistance goes up, which means your head loss goes down.

I'm pretty sure head loss is in meters, so, refer back to pHaestus' table see what 4.8M head loss will do to your flow.

Erm... pretty bad, isn't it. Maybe it's really head loss in cm. Maybe I did something else wrong. Maybe it really is pretty bad.
Anyone else?
Feel free to call me an idiot/moron/sloppy thinker/missed something important :)

Bob
PS: Apropos plumbing, I just got a call from our commonhouse (like a clubhouse - I live in a cohousing community) that there's the sound of running water and a big puddle in the furnace room. I'm off to go deal with US sized pipes... :(

No time to check your reference right now, Bob, but (.063/.1)^2 = 0.63^2 = 0.3969. Reduces your H by factor of 100.

Yep, you're right (duh!)
so:
Branch 0.122 x (0.063 / 0.1)(sq) or .122 x .3969 or .048
Flow 0.046 x (0.063 / 0.1)(sq) or .046 x .3969 or .018

or more like a half-meter head loss for the branch - looks a lot more reasonable (might still be wrong 'cause I screwed up somewhere else :))
Thanks!

PS: If you look at the 'A' factor for other things that might be in the loop instead of the 'T' fitting
elbow 0.112
bend 90degree (long radius) 0.061

Thinking about that 'elbow' - I'm pretty sure it's not a milled elbow. but just a short radius turn. The 'T's listed probably aren't milled either. The 'T' Stiffler's thinking about using probably is milled, though, which will make it (AFAIK) worse - and which means I'm wasting my time with that plumbers handbook... :(

And a caveat - I'm pretty sure none of these plumber's fittings are milled fittings - and I have read that milled fittings are lots worse for flow than their formed

ROFL, I really didn't expect someone to take the time to work all that out, but thank you. I'm using a Dolphin Poind 385GPH pump, here are the specs: Max Flow Rate: 385gph, Max Pressure: 9ft, Max Power: 20W.

I'm thinking I should find a way to route my tubing straight through the T. It'll take some inginuity, but nothing I can't handle.

Thanks for the replies.

How does .048m become .5m? Looks like you're having a rough day!

You're right - I was. Trying to do too many things at once, including trying to do this while semi-involved in a phone conference, doing online product research and debugging some C++. ...and slipping decimals is exactly what I do when multitasking to the point of distraction.
So... almost five centimeters head loss. Doesn't sound like much at all, does it?
Especially if there's going to be a bend in the curcuit there anyway, so
0.061 x (0.063 / 0.1)(sq) or .061 x .3969 or .024M, so the additional resistance of the 'T' fitting would be .048 - .024 or just 2.4cm head loss (assuming I didn't screw up that decimal point as well :))

Wow, 2.4cm isn't much at all. If that's true, then my routing will look MUCH better than having the traditional way...