# Results for case (not set)

Code: Code_Saturne

Version: 2.0-beta2

Authors: E. Moreau

## Method and Numerical Options

The previous version 1.3 has tested the same cases of the head losses, but only with a constant time step (idtvar = 0). Here we have tested two cases: idtvar = 0 and idtvar = -1 (stationary).

• In the first configuration, three simulations (laminar) have been performed:
1 idtvar = 0 and iphydr = 1
2 idtvar = 0 and iphydr = 0
3 idtvar = -1 and iphydr = 1
This case allows to compare the results with and without an improved modeling of the head losses in the pressure gradient.

• In the second configuration, we use a %$k-\varepsilon$% model with tilted head losses with respect to the channel axis. In this case the head losses tensor is diagonal. Both the influence of the gradient calculation (IMRGRA) and of iphydr are studied. The angle %$\alpha$% used is of %$\pi/3$%: as a consequence, the fluid must undergoes a deviation of %$\pi/6$%.

• The third configuration allows also an analytical study of the head losses. The mesh undergoes two rotations of angle %$\pi /6$% with respect to Oy then Oz.

## Mesh

You can download both meshes (parallel with Ox and Oy axis and not) here: meshes.

## Description of the results files

Text files called cas1_P.dat and cas1_V.dat are used to plot graphs below.

Data are stored in two columns; the first is the distance on Ox axis (m) the second is:

And results given by the cases models are classed in vertical by 'index' (separate by two new lines) as below :

• index 0: "Axial velocity (m/s) : iphydr=1"
• index 1: "Axial velocity (m/s) : iphydr=1"
• index 2: "Axial velocity (m/s) : iphydr=1"
• index 3: "Axial velocity (m/s) : iphydr=1"

## Reference Publications

### Additional information (links, pictures, etc.)

The simulations show that we recover results of the previous version 1.3.1 from a qualitative and analytical point of view.

First and third configurations

• The results are presented on two graphs below. We can see that:
o The option iphydr = 1 allows to recover the accurate value of the head losses: %$\Delta P = 10 000~\rm{Pa}$%.
o The cases idtvar = 0 and idtvar = -1 yield the same results.
o The same jumps as for the version 1.3 are generated on the velocity field for iphydr = 0.
o The error on the pressure jump, for iphydr = 0, is of 0.18% for the version 1.3 against 0.64% for the 2.0
But:
o Contrary to the version 1.3, the version 2.0 does not present any divergence when the option IMRGRA = 3 is activated (probably due to bug fixes in the extended neighborhood handling). Moreover we recover similar results for IMRGRA = 0 and IMRGRA = 3.
o Il should be noted that the third configuration, with extra diagonal term in the head losses tensor, leads to an error in the stationary case (idtvar = -1). Hence the results presented here concerning this case have been done with a constant time step (idtvar = 0).

Gnuplot Error:

"/tmp/cas1P.gnu", line 12: warning: Warning - difficulty fitting plot titles into key


The results are given figures (1) to (3). Only a qualitative analysis is done. The results are conform to those expected, and we recover on figure (3) the global shape obtained with the version 1.3.1.

Figure 1 - Pressure field (second configuration).

Figure 2 - Velocity field (second configuration).

Figure 3 - Velocity vectors (second configuration).

It should be noted that a simulation without using the xml file have been performed (we have used only the fortrans files) for the first configuration (laminar). We recover exactly the same results.

Current Tags:
create new tag
, view all tags
Topic attachments
I Attachment Action Size Date Who Comment
dat cas1_P.dat manage 4.5 K 2010-09-23 - 08:47 JacquesFontaine cas1Pdata
dat cas1_V.dat manage 3.7 K 2010-09-23 - 08:47 JacquesFontaine cas1Vdata
gz mesh.tar.gz manage 192.0 K 2010-09-23 - 14:48 JacquesFontaine meshes
jpg ptot.jpg manage 17.8 K 2010-09-23 - 15:26 JacquesFontaine pdcconf2
jpg speed.jpg manage 21.4 K 2010-09-23 - 15:26 JacquesFontaine pdcconf2speed
jpg speedvect1.jpg manage 55.2 K 2010-09-23 - 15:49 JacquesFontaine pdcconf2speedvect
Topic revision: r37 - 2012-12-12 - 17:29:50 - JacquesFontaine
CfdTm Web
09 Dec 2019

##### CfdTm Pages:
Home
People
Research
Vacancies
Events
Calendar
Links

Computational Fluid Dynamics and Turbulence Mechanics
@ the University of Manchester
Copyright &© by the contributing authors. Unless noted otherwise, all material on this web site is the property of the contributing authors.