Aalborg Universitet

The IEA Annex 20 Two-Dimensional Benchmark Test for CFD Predictions

Nielsen, Peter Vilhelm; Rong, Li; Cortes, Ines Olmedo

Published in:
Clima 2010 : 10th Rehva World Congress

Publication date:
2010

Document Version
Publisher's PDF, also known as Version of record

Link to publication from Aalborg University

Citation for published version (APA):

Nielsen, P. V., Rong, L., & Cortes, I. O. (2010). The IEA Annex 20 Two-Dimensional Benchmark Test for CFD
Predictions. In Clima 2010 : 10th Rehva World Congress: Sustainable Energy Use in Buildings. Antalya: Clima
2010 : 10th Rehva World Congress.

General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

? Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
? You may not further distribute the material or use it for any profit-making activity or commercial gain
? You may freely distribute the URL identifying the publication in the public portal ?
Take down policy
If you believe that this document breaches copyright please contact us at [email protected] providing details, and we will remove access to
the work immediately and investigate your claim.

Downloaded from vbn.aau.dk on: April 16, 2016

The IEA Annex 20 Two-Dimensional Benchmark Test for CFD Predictions

Peter V. Nielsen1, Li Rong1 and Ins Olmedo2

1
Aalborg University, Denmark
2
Crdoba University, Spain

Corresponding email: [email protected]

SUMMARY

This paper describes a benchmark test which can be used for tests of CFD predictions of
room air distribution. The benchmark describes a box-like room with a supply slot along the
side wall. Laser-Doppler measurements and hot-wire measurements are given for comparison
with the obtained CFD predictions both for isothermal flow and for nonisothermal flow. The
benchmark is defined on a web page, which also shows about 50 different benchmark tests
with studies of e.g. grid dependence, numerical schemes, different source codes, different
turbulence models, RANS or LES, different turbulence levels in a supply opening, study of
local emission and study of airborne chemical reactions. Therefore the web page is also a
collection of information which describes the importance of the different elements of a CFD
procedure.

The benchmark is originally developed for test of two-dimensional flow, but the paper also
shows results for three-dimensional steady flow.

INTRODUCTION

The IEA Annex 20 two-dimensional test case was developed by the participants of the Annex
20 work in 1990 as a benchmark [1]. Figure 1 shows the dimensions of the geometry.

Figure 1. The two-dimensional benchmark test.

The air distribution in the benchmark corresponds to the situation in a room with a slot inlet.
The dimensions and flow conditions are the following: L/H = 3.0, h/H = 0.056 and Re = 5000,
where L, H, h, Re are length, height, slot height and Reynolds number, respectively.
Velocities are measured in a model with W/H = 1.0 [2], and streaklines are photographed in a
model with W/H = 4.7, see figure 2, [3].
Figure 2. Streaklines obtained by photographing metaldehyde particle flow. The particles are
illuminated in the symmetry plane of the room.

The velocity profiles for the inlet slot are also given in [1], and the following conditions
should be used in most benchmark tests

ko = 1.5 (0.04 uo)2, o = ko1.5/lo and lo = h/10 (1)

where ko is turbulent kinetic energy in the supply opening, o is the dissipation, and lo is the
length scale.

The benchmark test is divided into an isothermal case, 2D1, as described above, and a
nonisothermal case, 2D2, where the temperature distribution is measured along a horizontal
plane in the room with heat released along the whole floor, figure 3 [3]. A small Archimedes
number and a high velocity ensure an almost isothermal flow, and the temperature distribution
corresponds to a concentration distribution from an even distributed source along the floor. A
nonisothermal test with high Archimedes number and restricted penetration of the plane wall
jet into the room is also introduced [1]. Those measurements are based on Scwenkes
experiments [4].

Figure 3. Position of measurement points for temperature distribution in the benchmark test.

APPLICATION OF THE BENCHMARK

The benchmark is located together with three other benchmarks on the web page:

www.cfd-benchmarks.com
The other benchmarks are about computer simulated manikins. They deal with mixing
ventilation and displacement ventilation, thermal comfort and personalized ventilation.

About 50 different applications of the IEA Annex 20 two-dimensional benchmark (made

between 1974 and now) can be found on the web page. This gives the possibility to compare a
prediction with other authors results (other turbulence models, other numerical schemes,
different inlet conditions, etc.). The basic knowledge behind the benchmark test has thus been
expanded during the years and made the benchmark test more useful. A few examples are
discussed in the following sections.

Early predictions

Figure 4 shows some early two-dimensional predictions based on the vorticity-stream

function equations, -, and on the RANS equations, [3] and [5] as well as predictions based
on RANS equations with a one equation turbulence model, [6].

Figure 4. Early predictions of the velocity profile at x/H = 2.0 with different sets of transport
equations, and different turbulence models.

Different numerical schemes

Figure 5 shows a set of predictions based on a low Reynolds number k- model (the LRN
model), [7]. The results are compared at two different vertical sections. The recirculating flow
is well predicted in the simulations, although the magnitude of the counter flow is slightly
underestimated. The maximum velocity in the occupied zone is accurate with respect to both
location and magnitude.

Figure 5. Velocity profiles and turbulent kinetic energy predicted by a LRN model.
It is possible to compare the measured RMS value u '2 with the square root of the turbulent
kinetic energy k because the flow can be considered as a flow with wall jet profiles ( k ~
1.1 u '2 ). The figure shows that the predicted level of turbulence is in good agreement with
the measurements.

Large Eddy Simulations, LES, can be used to obtain detailed information on turbulence,
which can not be achieved by the traditional time-averaged turbulence models. The turbulence
is predicted in details, and it is possible to make a direct prediction of u '2 . Figure 6 shows
some early predictions by Davidson and Nielsen [8]. It was difficult to get sufficient
integration time for the calculation of u and u '2 .

Figure 6. Time-averaged velocity and turbulence intensity in the symmetry plane predicted by
LES.

A B

Figure 7. Probability density function of u uo close to the ceiling at x/H = 2.0, and close to the
floor at x/H = 1.0. See [8].

The probability density function of u uo close to the ceiling at x/H = 2.0 has a well-defined
mean velocity, and the velocity fluctuates around the mean velocity in a symmetrical
distribution, figure 7A, while the flow close to the floor at x/H = 1.0 is asymmetric (has a
skewness) with a probability of both a negative and a positive velocity, figure 7B.

Different turbulence models

Two-dimensional steady state predictions with different turbulence models indicate some
differences, especially in two of the corners as shown in figure 8, [9]. The grid is identical in
all four cases, which means that the difference in the flow can be ascribed to the turbulence
models used to predict the two-dimensional flow. The predictions with the SST model show
especially a large recirculating flow in the occupied zone below the supply slot 0 < x/H < 1.5.
Similar results have been obtained by Voigt [10]. It is not possible to fully exclude this flow
pattern from the observation of the streaklines shown by photos as in figure 2, but it does not
corresponds to the Laser-Doppler measurements in the benchmark.

Figure 8. Two-dimensional isothermal and steady state simulations of the Annex 20 2D

benchmark test. Four different turbulence models are used together with the RANS equations.

THREE-DIMENSIONAL STEADY STATE PREDICTIONS

The benchmark was originally selected for the simulation of steady two-dimensional flows for
the IEA Annex 20 work back in 1990. The flow in the benchmark was considered to be two-
dimensional and steady, but three-dimensional unsteady flows could, at least in principle,
occur. LES predictions indicate an asymmetric probability density function at the floor at x/H
= 1.0 with a probability of both negative and positive velocities, figure 7B, and measurements
in rooms with L/H 4.0 and W/H = 4.7 do also show unsteady flow at the floor for 0 < x/H <
1.0, [3].

The predictions have been repeated as steady three-dimensional flow in the benchmark with
W/H = 1.0 to examine if this geometry gives results which are more identical to each other
when different turbulence models are used. Figure 9 shows the flow in the symmetry plane of
the room for the k- model and the SST model. The three-dimensional prediction with the k-
model looks similar to the two-dimensional predictions in figure 8, while the three-
dimensional prediction with the SST model is slightly different from the two-dimensional
version in figure 8. The large recirculating flow in the occupied zone below the supply slot is
reduced compared to the flow in the two-dimensional case. This change in the flow towards
the flow predicted by the other turbulence models could be a result of having three-
dimensional flow, but it could not be fully excluded that the use of a different grid might have
some influence. (The two-dimensional predictions, figure 8, are made in a structured grid,
while the three-dimensional predictions in figure 9 and 10 are made in an unstructured grid).

k- model k- model

SST model

Figure 9. Three-dimensional isothermal and steady state simulations of the Annex 20 2D

benchmark test. Three different turbulence models are used together with the RANS
equations.

The difference between a two-dimensional k- prediction and a three-dimensional k-

prediction of the flow can be explained by looking at the flow in the occupied zone. Figure 10
show the three-dimensional steady flow predicted by the k- model. The flow in the middle
plane is very different from the flow close to the side wall. A horizontal section in the
occupied zone shows how the recirculating flow in the middle plane has a higher velocity
than the flow close to the side walls. The flow in the middle of the room therefore creates two
horizontal circulations in the region below the supply opening, which ensures a fully vertical
recirculation in the middle plane. The flow is unsymmetric in the room, although the room is
symmetric.

The measured flow in the benchmark, [1], does not show this large difference between the
velocity distribution in the middle plane and the velocity distribution at a distance of 0.1W
from the side wall.

Flow in middle plane Flow close to side wall (0.1W from

side wall)

Flow in horizontal plane in the

occupied zone (0.5h above floor)
Figure 10. Three-dimensional isothermal and steady state simulations of the Annex 20 2D
benchmark test with a k- turbulence model solved together with the RANS equations.
A study by Susin et al. [11] using three-dimensional flow concludes that the k- model
performed best compared to the RNG k- model and the k- model, and that the most
significant differences between calculated and experimental velocity profiles were also found
along the floor.

CONCLUSIONS

The two-dimensional benchmark test, called the IEA 2D test case was defined in 1990 to be
used in the IEA Annex 20 work.

The benchmark is defined on a web page, so it can also be used in other CFD work. There
have been a very large number of papers that use this geometry, and about 50 works are
reported on the web page. This gives the possibility to compare a prediction with other
authors results (other turbulence models, other numerical schemes, different inlet conditions,
etc.). The basic knowledge behind the benchmark test has thus been expanded during the
years and made the benchmark test more useful.

Predictions by different turbulence models show some variation in the obtained streamline
pattern. Especially the k- SST model gives a flow different from the flow obtained by the
other models.

The benchmark was originally selected for the simulation of steady two-dimensional flow,
and the flow in the benchmark was considered to be two-dimensional and steady, but three-
dimensional unsteady flow could, at least in principle, occur.

Predictions by a steady state three-dimensional flow show streamline pattern similar to the
two-dimensional predictions, but the SST model introduces a change in the flow which is
slightly towards the flow predicted by the other turbulence models.

An analysis of the three-dimensional flow predicted by the k- model shows a large

difference of the flow in the middle plane compared to the flow in a plane close to the side
wall. This difference could not be confirmed by the measurements in the benchmark.

REFERENCES

1. Nielsen, P V. 1990. Specification of a Two-Dimensional Test Case, Aalborg University, IEA

Annex 20: Air Flow Patterns within Buildings.

2. Restivo, A M. 1979. Turbulent Flow in Ventilated Rooms, PhD thesis, Imperial College,
London.

3. Nielsen, P V. 1974. Strmningsforhold i luftkonditionerede lokaler, PhD thesis, Technical

University of Denmark. (English translation: Flow in Air Conditioned Rooms, 1976).

4. Schwenke, H. 1975. ber das Verhalten ebener horizontaler Zuluftstrahlen im begrenzten

Raum, Luft- und Kltetechnik, Nr. 5, Dresden.

5. Nielsen, P V, Restivo, A, and Whitelaw, J H. 1978. The Velocity Characteristic of Ventilated

Rooms, Journal of Fluid Engineering, Vol. 100, pp. 291-298.
6. Davidson, L and Olsson, E. 1987. Calculation of Some Parabolic and Elliptic Flows Using a
New One-Equation Turbulence Model, Proc. 5th. International Conference on Numerical
Methods in Laminar and Turbulent Flow, Montreal.

7. Skovgaard, M and Nielsen, P V. 1991. Simulation of Simple Test Case, Case 2D1, Internal
report for the International Energy Agency, Annex 20, Aalborg University, ISSN 0902-7513
R9131.

8. Davidson, L and Nielsen, P V. 1996. Large Eddy Simulation of the Flow in a Three-
Dimensional Ventilated Room, Proc. of Roomvent96, Yokohama.

9. Rong, L and Nielsen, P V. 2008. Simulation with Different Turbulence Models in an Annex 20
Room Benchmark Test Using Ansys CFX 11.0, DCE Technical Report No. 46, Aalborg
University.

10. Voigt, L K. 2000. Comparison of Turbulence Models for Numerical Calculation of Airflow in
an Annex 20 Room, Department of Energy Engineering, Technical University of Denmark, ET-
AFM 2000-01, ISBN 87-7475-225-1.

11. Susin, R M, Lindner, G A, Mariani, V C, and Mendonca, K C. 2009. Evaluating the Influence of
the Width of Inlet Slot on the Prediction of Indoor Airflow: Comparison with Experimental
Data, Building and Environment, 44, pp. 971-986.