AC 6-07 CFD Simulations
Draft tube
Application Challenge 6-07 © copyright ERCOFTAC 2004
CFD Simulations
Overview of CFD Simulations
Two international workshops, focused on this test case, have been organised: Turbine-99 - Workshop on Draft Tube Flow, Porjus, Sweden, 20-23 June 1999 (for Proceedings see Gebart et al., 2000a, hereafter in the text called Proc.T99W1), and Turbine-99 - Workshop 2 on Draft Tube Flows, Älvkarleby, Sweden, June 18-20, 2001 (for Proceedings see Engström et al. 2003, http://epubl.luth.se/1402-1536/2000/11/index-en.html, hereafter in the text called Proc.T99W2)). Brief summaries of the workshops have also been presented at the 20th and 21st IAHR Symposia on Hydraulic Machinery and Systems in Charlotte, N.C. 2000 (Gebart et al., 2000b), and in Lausanne, respectively, (Engström et al., 2002).
The first workshop was organised as a blind test and attracted 18 participants who contributed with comparison data and papers to the proceedings. The geometry and the axial and tangential velocity components at the inlet section were specified, while most of the remaining options were left to the participants to decide upon, e.g. grid and radial velocity component at the inlet. The focus was from an engineering point of view, to get an understanding of the difficulties of CFD-simulations and possible variations between different CFD-simulations of the test case. The first workshop showed that the case was very sensitive to different settings and the resulting Cp varied with up to ± 50 %. The reasons were as stated above: the choice of inlet boundary condition for the radial velocity had a large effect on the pressure recovery, many solutions were obtained on very coarse grid, various turbulence models were used, etc. To summarize: most of the computations were not performed according to the ERCOFTAC Best Practice Guidelines for Quality and Trust in Industrial CFD, which formally came out in the beginning of 2000 (Casey and Wintergerste, editors, 2000). Detailed results from the first workshop are available from Luleå University of Technology, Div. of Fluid Mechanics, see Proc.T99W1, Gebart et al (2000a). Since these results to a large extent are superseded by the results from the second workshop, we will not further discuss the results from the first workshop.
The second workshop focused more on quality and it was recommended to use the ERCOFTAC Best Practice Guidelines (Casey and Wintergerste, 2000) as a guide when performing computations. For this workshop the grid, the radial velocity component, turbulence model etc were specified for one test case, (case T), to get a reference solution from every participant in order to reduce the spread between the calculations. And for the 12 participants the variation in Cpr was reduced to ± 20 %. Engström et al. 2001, Engström et al., 2003. The difference was still very big based on Cpr wall, considering that all affecting variables should be the same. If one instead looks at Cprmean the difference is only ± 7 %.
In this document a brief comparison between the CFD-simulations will be given to demonstrate the agreement and differences between CFD results and experiments, with focus on the results from the second workshop. Results regarding engineering quantities (of which many are exactly the design or assessment parameters, DOAPs, defined for the test case) will be given for computations by all participants. A “typical” simulation of the standard test case T is then chosen to enable more detailed comparisons with the experiments.
The participating groups and their acronyms are given in Table 3.1
Table 3.1. List of acronyms
|
Acronym |
Authors |
Affiliation |
CFD code |
|
AEAT |
Holger Grotjans |
AEA Technology GmbH |
CFX-TASCflow |
|
Germany |
|||
|
CKD |
Ales Skotak |
CKD Blansko Engineering, a.s. |
FLUENT |
|
Czech Republic |
|||
|
HQ |
Maryse Page |
IREQ – Institute de recherche |
CFX-TASCflow |
|
Anne-Marie Giroux |
d’Hydro-Quebec |
FIDAP | |
|
Canada |
FINE/Turbo | ||
|
HTC |
Katsumasa Shimmei |
Hitachi, Ltd. |
|
|
Takanori Ishii |
|||
|
Kazuo Niikura |
|||
|
NUT |
Sadao Kurosawa |
Toshiba Corporation, Japan |
|
|
Tomotatsu Nagafuji |
Nagoya University, Japan |
||
|
Debasish Biswas |
Toshiba Corporation, Japan |
||
|
TEV |
Per Egil Skåre |
Sintef Energy Research, Norway |
FLUENT |
|
Ole Gunnar Dahlhaug |
Norwegian University of |
||
|
Science and Technology, Norway |
|||
|
VUAB |
Staffan Jonzén |
Vattenfall Utveckling AB |
FLUENT |
|
Bengt Hemström |
|||
|
Urban Andersson |
|||
|
Iowa |
Yong. G. Lai |
Iowa Institute of Hydraulic |
U2RANS |
|
V.C. Patel |
Research, U.S.A. |
||
|
LTU |
Michel Cervantes |
Luleå University of Technology |
CFX-4.3 |
|
Fredrik Engström |
Sweden |
||
|
EXA |
Alain Bélanger |
Exa Corporation, U.S.A. |
PowerFlow |
|
Experiment |
Urban Andersson |
Vattenfall Utveckling AB |
|
Table 3.2 contains a summary description of all test cases.
Table 3.3 is a description of all available data files which the computors were asked to provide from their computations. The data from VUAB:s computation are chosen to illustrate typical computational results for the case T(r ).
|
NAME |
|
|
| |||||
|
|
|
|
|
|
|
|
| |
|
CFD 1a (T(r)-case) |
|
|
|
|
|
|
|
|
|
CFD 2a (R(r)-case) |
|
|
|
|
|
|
|
|
Table 3.2. Summary description of all test cases
|
|
|
|
| |||
|
|
|
|
|
|
| |
|
CFD 1a |
 TrIa_VelCp_CFD.dat |
TrIa_keps_CFD.dat |
TrIb_CFD.dat |
TrII_CFD.dat |
TrIII_CFD.dat |
TrIII_keps_CFD.dat |
|
CFD 2a |
RrIa_VelCp_CFD.dat |
RrIa_keps_CFD.dat |
RrIb_CFD.dat |
RrII_CFD.dat |
RrIII_CFD.dat |
RrIII_keps_CFD.dat |
|
|
|
|
|
|
| |
|
|
|
|
|
|
| |
|
CFD 1a |
TrIaW_CFD.dat |
TrIVa_CFD.dat |
TrIVb_CFD.dat |
TrUcl_Cp_CFD.dat |
TrLcl_Cp_CFD.dat |
TrDOAP_CFD.dat |
|
CFD 2a |
RrIaW_CFD.dat |
RrIVa_CFD.dat |
RrIVb_CFD.dat |
RrUcl_Cp_CFD.dat |
RrLcl_Cp_CFD.dat |
RrDOAP_CFD.dat |
Table 3.3. Description of all available data files, and simulated parameters.
Simulation Case
Solution strategy
The different workshop participants used different codes, both commercial and in-house codes, based on either FVM and FEM, as shown in Table 3.1 where some basic information is available. It was also planned that each group should fill in a Table like Table 3.4. This was however not possible. As an example, however, we show Table 3.4 filled in by ourselves (VUAB), since VUAB:s results for test case T has been chosen as representing typical results.
All relevant settings for the group can be seen in Table 3.4.
Table 3.4. Solution strategy chosen by VUAB, test case T.
|
Solution acronym |
VUAB |
|
Test case |
T |
|
Operating condition |
|
|
CFD code (version) |
Fluent 6.0.20 |
|
Authors |
Jonzén S., Hemström B. & Andersson U. |
|
Affiliations |
Vattenfall Utveckling AB. |
|
Report date |
June 2001, later revised |
|
References |
Jonzén S., Hemström B. & Andersson U: “Turbine 99 – Accuracy in CFD Simulations on Draft Tube Flow”,paper in Proc.T99W2. |
|
Computational topics |
Description |
Parameter |
|
Discretization method |
||
|
Governing equations |
Continuity, momentum, turbulent kinetic energy, turbulent dissipation. |
|
|
Turbulence modelling |
Standard k-epsilon |
|
|
Near-wall treatment |
Standard wall functions |
K=0.42, E=9.793 |
|
Discretization scheme |
QUICK. Second order upwind for pressure |
|
|
Mesh |
Modification of first grid by Lai, 707760 cells |
|
|
Fluid properties |
ρ=1000 kg/m3, μ=1006*10-3 |
|
|
Inlet boundary conditions |
Velocities, k and ε are specified |
|
|
Outlet boundary conditions |
Static pressure=0 |
|
|
Convergence criteria |
Stable residuals and monitor value of velocity. |
Although not provided in the same format here or in ProcT99W2, the corresponding information for the other computations according to the table of acronyms can be extracted from the separate papers included in Proc.T99W2.
Computational Domain
To reduce the variation in results from the first workshop, a common grid was supplied to the workshop participants. The grid is described in http://epubl.luth.se/1402-1536/2000/11/index-en.html.
Due to geometrical problems at the first part of the grid and a rather bad grid quality at the elbow, one conclusion from the second workshop was that an improved grid was needed. Some authors modified in various ways the given grid or used their own grid.
Boundary Conditions
A full summary of experimental data needed to set up a CFD-problem can be found in http://epubl.luth.se/1402-1536/2000/11/index-en.html . Here will be found information on the two main velocity components at the inlet, information on the wall roughness, wall pressures at the outlet and suggested assumptions for the third (radial) velocity component at the inlet that can be used to verify CFD-calculations against other simulations.
Both workshops used experimental data from (r) data set and for verification of the quality of a code it is recommended that these data is used with the specifications of Case 1 (T) of the second workshop since this offers the best opportunity for comparisons with other computations. However, for the future validation of the codes the calculations should be focused on the (n) data set since this offers the most consistent data set with a ‘complete’ set of velocity measurements from section I to section III. In the next chapter some additional requirements to obtain simulations with a good agreement with the experiments will be discussed.
Application of Physical Models
Most groups used wall functions.
For the second workshop a constant turbulent length scale at the inlet was suggested. Several workshop participants pointed out that this was not the best possible estimate. A better estimate would be a smaller length than recommended by the organisers.
Numerical Accuracy
Bergström (2000) made a thorough study of the numerical accuracy. The grid error was < 7 % and the iterative error was less than 0.8 %, however, a new grid caused an increase in e.g. Cp of 20%. This shows that the grid topology is very important and that ordinary refinement strategies might not be enough to reveal the total grid error.
The variation in the results between the participants serves as a measurement of the total accuracy comparing different CFD-codes and discretisation schemes. For Case T(r) most participants used the recommended grid, which facilitates comparisons between different computed results.
Comparing different computations revealed that the evaluation of the DOAPs was extremely sensitive to the evaluation algorithm. Therefore, these parameters were recomputed by the organisers to reduce the error and to give a constant numerical error. Details are given in Engström et al (2003).
CFD Results
Grid:
Workshop 2-grid
New grid
Derived data for comparison:
ASCII-files with DOAPs for the different cases submitted to the workshop for comparisons.
DOAPs calculated at available cross sections for case T(r). Both workshop 1 and 2.
DOAPs calculated at available cross sections for case R(r). Workshop 2.
Case T
Velocity and pressure values:
Section Ia
Computed velocity components and normalised pressure (Cp)
at cross section Ia, for Case T(r). (See Figure 1.5).
Computed turbulent kinetic energy and turbulent dissipation rate
at cross section Ia, for Case T(r). (See Figure 1.5).
Computed normalised wall pressure (Cp) at cross section Ia, for Case T(r).
(See Figure 1.5).
Section Ib
Computed velocity components and pressure normalised pressure (Cp) at cross section Ib,
for Case T(r). (See Figure 1.5).
Section II
Computed velocity components and normalised pressure (Cp) at cross section II
for Case T(r). (See Figure 1.4).
Section III
Computed velocity components and normalised pressure (Cp) at cross section III,
for Case T(r). (See Figure 1.4).
Computed turbulent kinetic energy and turbulent dissipation rate at cross section III,
for Case T(r). (See Figure 1.5).
Section IVa
Computed velocity components and normalised pressure (Cp) at cross section IVa,
for Case T(r). (See Figure 1.4).
Section IVb
Computed velocity components and normalised pressure (Cp) at cross section IVb,
for Case T(r). (See Figure 1.4).
Computed normalised wall pressure (Cp) at cross section IVb, for Case T(r).
(See Figure 1.4).
Centrelines
Computed normalised wall pressure along the upper centre line, for Case T(r).
(See Figure 1.7).
Computed normalised wall pressure along the lower centre line, for Case T(r).
(See Figure 1.7).
Computed normalised wall shear stress (Cf) along the upper centre line,
for Case T(r). (See Figure 1.7).
Computed normalised wall shear stress (Cf) along the lower centre line,
for Case T(r). (See Figure 1.7).
Case R
Velocity and pressure values:
Section Ia
Computed velocity components and normalised pressure (Cp)
at cross section Ia for Case R(r). (See Figure 1.5).
Computed turbulent kinetic energy and turbulent dissipation rate
at cross section Ia, for Case R(r). (See Figure 1.5).
Computed normalised wall pressure (Cp) at cross section Ia, for Case R(r).
(See Figure 1.5).
Section Ib
Computed velocity components and pressure normalised pressure (Cp)
at cross section Ib, for Case R(r). (See Figure 1.5).
Section II
Computed velocity components and normalised pressure (Cp)
at cross section II for Case R(r). (See Figure 1.4).
Section III
Computed velocity components and normalised pressure (Cp)
at cross section III, for Case R(r). (See Figure 1.4).
Computed turbulent kinetic energy and turbulent dissipation rate
at cross section III, for Case R(r). (See Figure 1.5).
Section IVa
Computed velocity components and normalised pressure (Cp)
at cross section IVa, for Case R(r). (See Figure 1.4).
Section IVb
Computed velocity components and normalised pressure (Cp)
at cross section IVb, for Case R(r). (See Figure 1.4).
Computed normalised wall pressure (Cp) at cross section IVb, for Case R(r).
(See Figure 1.4).
Centrelines
Computed normalised wall pressure along the upper centre line, for Case R(r).
(See Figure 1.7).
Computed normalised wall pressure along the lower centre line, for Case R(r).
(See Figure 1.7).
Computed normalised wall shear stress (Cf) along the upper centre line, for Case R(r).
(See Figure 1.7).
Computed normalised wall shear stress (Cf) along the lower centre line,
for Case R(r). (See Figure 1.7).
References
Bergström, J. (2000). “Modeling and Numerical Simulation of Hydro Power Flows”. Doctoral Thesis 2000-06, Luleå University of Technology, Department of Mechanical Engineering, Division of Fluid Mechanics.
Casey M. & Wintergerste T. (editors). (2000) ERCOFTAC Best Practice Guidelines. ERCOFTAC Special Interest Group on “Quality and Trust in Industrial CFD”.
Gebart B.R., Gustavsson L.H. & Karlsson R.I. (editors) (2000a) "Turbine 99 – Workshop on Draft Tube Flow”. Technical Report 2000-11, Luleå University of Technology, Department of Mechanical Engineering, Division of Fluid Mechanics.
Gebart B.R., Gustavsson L.H. & Karlsson R.I. (2000b) "Report from Turbine 99 – Workshop on Draft Tube Flow in Porjus, Sweden, 20-23 June 1999”. Paper presented at the 20th IAHR Symposium Hydraulic Machinery and Systems. Aug. 6-9, 2000, Charlotte, North Carolina, U.S.A.
Engström T.F., Gustavsson L.H. & Karlsson R.I. (2002) "Report from Turbine 99 – Workshop 2 on Draft Tube Flow”. The second ERCOFTAC Workshop on Draft Tube Flow, held in Älvkarleby, Sweden, June 18-20, 2001. Paper presented at the 21st IAHR Symposium Hydraulic Machinery and Systems, Lausanne, Switzerland, Sept. 2002
Engström, T.F., Gustavsson, L.H., & Karlsson, R.I. (2003), Proceedings of Turbine-99 - Workshop 2. The second ERCOFTAC Workshop on Draft Tube Flow. Älvkarleby, Sweden, June 18-20 2001. http://epubl.luth.se/1402-1536/2000/11/index-en.html
In text called Proc.W2.
© copyright ERCOFTAC 2004
Contributors: Rolf Karlsson - Vattenfall Utveckling AB