Double Precision Solver

anj2012 · July 10, 2012, 17:31

Its a convective heat transfer problem. I have water flowing through a channel of dimensions 0.2x0.1x12.5mm, with constant heat flux applied on the walls of channel. The inlet and outlet faces are 0.2x0.1mm. Should i use double precision solver ?

flotus1 · July 11, 2012, 00:59

It depends on other details of the setup than you gave.

Is there a large difference between the extent of the smallest and the biggest cell in your grid, for example because you are resolving the boundary layer with Y+ < 1?

Or is there any need to account for very small values of a variable?

If in doubt, go for double precision. It is just the amount of memory usage that will be doubled, the computational time will increase only slightly.

anj2012 · July 11, 2012, 11:22

thanks flotus1
The cells are of uniform size.
I was worried because channel walls are of silicon with thermal conductivity of 148W/m.K, while that of water thermal conductivity is 0.6W/m.K (ratio of thermal conductivities is high). Also, the cross section of channel is much smaller compared to its length.

July 10, 2012, 17:31	Double Precision Solver	#1
anj2012 New Member Anj Join Date: Jul 2012 Posts: 2 Rep Power: 0	Its a convective heat transfer problem. I have water flowing through a channel of dimensions 0.2x0.1x12.5mm, with constant heat flux applied on the walls of channel. The inlet and outlet faces are 0.2x0.1mm. Should i use double precision solver ?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Steady state chemistry	linnemann	OpenFOAM Running, Solving & CFD	14	April 7, 2015 13:10
return a normal number to solver in double precision mode in udfs	payam_IUST	FLUENT	1	February 13, 2012 07:06
Double precision & User Fortran	Martijn	CFX	3	April 4, 2009 05:43
Missing math.h header	Travis	FLUENT	4	January 15, 2009 11:48
Double precision CFX	Vanessa	CFX	8	August 13, 2006 04:34

July 11, 2012, 00:59		#2
flotus1 Super Moderator Alex Join Date: Jun 2012 Location: Germany Posts: 3,399 Rep Power: 46	It depends on other details of the setup than you gave. Is there a large difference between the extent of the smallest and the biggest cell in your grid, for example because you are resolving the boundary layer with Y+ < 1? Or is there any need to account for very small values of a variable? If in doubt, go for double precision. It is just the amount of memory usage that will be doubled, the computational time will increase only slightly.

July 11, 2012, 11:22		#3
anj2012 New Member Anj Join Date: Jul 2012 Posts: 2 Rep Power: 0	thanks flotus1 The cells are of uniform size. I was worried because channel walls are of silicon with thermal conductivity of 148W/m.K, while that of water thermal conductivity is 0.6W/m.K (ratio of thermal conductivities is high). Also, the cross section of channel is much smaller compared to its length.