Validation: ASHRAE Elbow Loss Coeffs

LKF · April 13, 2009, 19:37

I'm curious, has anyone studied / compared CFX results to tabulated ASHRAE elbow loss coeffecients (or something similar)?

I've been studying a 60 inch diameter 90 degree elbow (5 Gore) for air between 2,000 fpm and 6,000 fpm.

Currently I'm seeing a CFX estimated Co that is about 0.65 times the ASHRAE value. Comparison against other reference values suggests similar under-prediction.

ghorrocks · April 13, 2009, 21:19

Hi,

Did you model the full experimental test setup, such as inlet and exhaust pipes? Have you done sensitivity analysis of all important variables?

Glenn Horrocks

LKF · April 13, 2009, 21:58

Hi Glenn,

I've modeled 3 diameters of duct length on either end - mass flow inlet w. zero gradient turbulence (to approximate fully developed flow), opening BC at 0 Prel on outlet. I understand there are possible improvements to the BC's, however in terms of sensitivity, I've considered different inflow velocity profiles such as 1/7 power law and uniform inlet velocity, with negligable change to the results. I could perhaps benefit from extending the outlet (as I've been advised from CFX tech support) and switching to an outflow condition. I've considered wall roughness, but have since reverted to a no slip smooth wall condition since there was little response to a rough wall.

I demonstrated grid indepence at about 300,000 nodes (using symmetry along the center, results are consistent for the whole domain). Y+ has only negligable impact at an average value of Y+ < 80 (that is Y+ in POST, not the solver y+ which is 1/4 of that value). Mesh spacing is about 2" and I've got 12 nodes of inflation that is about 2.5" thick.

I've compared both the standard k-e model and the SST model (with the appropriate mesh correction to get Y+ ~3). In all cases, I've converged to a max residual (not RMS) of 1E-4 (on some cases less than 5E-5). Mostly I've run double precision, but again haven't seen appreciable differences between single and double precision.

I believe I've done a fairly comprehensive analysis . . . and I still observe total pressure loss through the elbow that is around 0.65 times references of elbow loss data that I have access to (I would have expected +/-15% and would even be happy to get +/-20%). I should mention I'm using area based averaging instead of mass flow averages. I know mass flow averages are theoretically correct, but a lot of industry measurements I come across are area averages. Also, the mass flow based losses have even less agreement.

Do you have any experience with comparing this sort of scenario to reference / experimental results?

April 13, 2009, 19:37	Validation: ASHRAE Elbow Loss Coeffs	#1
LKF New Member Join Date: Mar 2009 Posts: 16 Rep Power: 17	I'm curious, has anyone studied / compared CFX results to tabulated ASHRAE elbow loss coeffecients (or something similar)? I've been studying a 60 inch diameter 90 degree elbow (5 Gore) for air between 2,000 fpm and 6,000 fpm. Currently I'm seeing a CFX estimated Co that is about 0.65 times the ASHRAE value. Comparison against other reference values suggests similar under-prediction. Last edited by LKF; April 13, 2009 at 22:00.

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Need help for loss coefficient of elbow diffuser	Pete	Main CFD Forum	0	November 9, 2004 02:54

April 13, 2009, 21:19		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,700 Rep Power: 143	Hi, Did you model the full experimental test setup, such as inlet and exhaust pipes? Have you done sensitivity analysis of all important variables? Glenn Horrocks

April 13, 2009, 21:58		#3
LKF New Member Join Date: Mar 2009 Posts: 16 Rep Power: 17	Hi Glenn, I've modeled 3 diameters of duct length on either end - mass flow inlet w. zero gradient turbulence (to approximate fully developed flow), opening BC at 0 Prel on outlet. I understand there are possible improvements to the BC's, however in terms of sensitivity, I've considered different inflow velocity profiles such as 1/7 power law and uniform inlet velocity, with negligable change to the results. I could perhaps benefit from extending the outlet (as I've been advised from CFX tech support) and switching to an outflow condition. I've considered wall roughness, but have since reverted to a no slip smooth wall condition since there was little response to a rough wall. I demonstrated grid indepence at about 300,000 nodes (using symmetry along the center, results are consistent for the whole domain). Y+ has only negligable impact at an average value of Y+ < 80 (that is Y+ in POST, not the solver y+ which is 1/4 of that value). Mesh spacing is about 2" and I've got 12 nodes of inflation that is about 2.5" thick. I've compared both the standard k-e model and the SST model (with the appropriate mesh correction to get Y+ ~3). In all cases, I've converged to a max residual (not RMS) of 1E-4 (on some cases less than 5E-5). Mostly I've run double precision, but again haven't seen appreciable differences between single and double precision. I believe I've done a fairly comprehensive analysis . . . and I still observe total pressure loss through the elbow that is around 0.65 times references of elbow loss data that I have access to (I would have expected +/-15% and would even be happy to get +/-20%). I should mention I'm using area based averaging instead of mass flow averages. I know mass flow averages are theoretically correct, but a lot of industry measurements I come across are area averages. Also, the mass flow based losses have even less agreement. Do you have any experience with comparing this sort of scenario to reference / experimental results?