Numerical and experimental investigation of flow in an opentype subsonic wind tunnel
 137 Downloads
Abstract
A numerical and experimental investigation of flow, inside the test section of an opentype subsonic wind tunnel, has been incorporated in the present paper. Experimental data are collected at different selected locations along the wind tunnel length and inside the test section. For detail assessment of the spatial variation of flow variables, numerical analysis is carried out. Boundary conditions have a significant influence on the validation of numerical simulation. A novel approach of system curve generation by experimental analysis of wind tunnel is adopted, instead of using the conventional approach of fantype boundary condition. Mass flow rate and pressure jump obtained by system curve are utilized for inlet and outlet boundary conditions, respectively. Comparison of numerical and experimental flow fields at the test section suggests maximum error of 9.84% in case of areaweighted average wind velocity along the length of the test section, whereas along the height of the test section a maximum error of 7.75% is observed.
Keywords
Wind tunnel testing Numerical simulation Boundary conditions Honeycomb and screens System curve Flow assessmentList of symbols
 L
Length of the honeycomb cell
 d_{h}
Hydraulic diameter of the honeycomb cell
 CFD
Computational fluid dynamics
 VFD
Variable frequency drive
1 Introduction
Wind tunnels are one of the essential tools for aerodynamic studies. It has a very diverse application in many engineering fields [1]. Designing a wind tunnel is somewhere between an art and science, with occasional excursions into propitiatory works [2]. Due to a wide variety of requirements and especially the working section configuration, it is unwise to lay down firm design rules [3]. However, critical areas of concern in wind tunnel design are addressed, and an attempt to establish proper design criteria has been made [4]. The complete design process for opentype wind tunnel has also been reported [3, 4, 5]. With the advent of the modern computational facility, CFD seems to be an affordable, valuable tool for not only to design and optimization of the critical components of the wind tunnel but also being complementary to the experimental work. Generally, while designing a wind tunnel, the capabilities of the CFD can be explored to examine ways of improving the design of individual components. Many researchers have demonstrated the feasibility of using CFD for improving the design of some conventional wind tunnel component [6, 7, 8, 9, 10, 11].
The flow conditions in the test section with numerical simulation of the entire wind tunnel, on the basis of the relationship between the total pressure loss and flow rate with fan performance curve, have been presented by Moonen et al. [12]. A different approach of replacing the fantype boundary condition with the use of the calculated pressure based on the section loss coefficient is found as well [13]. Numerical analysis of the fullscale wind tunnel can produce valuable information for detailed interpretations of flow physics. It may provide a thorough understanding of flow interaction between the individual components and its effect on the flow condition at the test section. The CFD simulation could be used to assess the spatial variations that can often be insufficiently covered by experiments. Thus, the derived variables such as shear stress and turbulent kinetic energy (that are difficult to determine experimentally) can be evaluated by numerical simulation [14]. One can integrate experimental measurements with a wind tunnel to corresponding numerical simulations and thus substantially improving the accuracy and efficiency of the flow analysis as compared to ordinary simulation [15].
The present paper incorporates a survey of wind velocity variations inside the test section of a renovated opentype subsonic wind tunnel with experimental and numerical analysis. The experimentation is carried out at five different operating fan speeds from 10 to 30 Hz in the interval of 5 Hz. The numerical methodology includes the entire configuration of the wind tunnel as a CFD model, to account the impact of various design features on the flow in the test section. The precise correspondence of the numerical simulation with the experiment requires calibration and validation of the numerical results, and this can be achieved by providing stable boundary conditions. A novel approach of system curve generation is utilized to overcome the problem associated with the use of the fantype boundary condition for numerical simulation. In this approach, the system curve based on measured velocity inside the test section and wall static pressure at the downstream end of divergent section, at different fan speeds, is determined. This will take care of resistance throughout the tunnel length and its effect on flow parameters at the test section and correlate the numerical simulation with the experiment to a reasonable accuracy.
2 Experimental analysis
3 Numerical analysis
A number of studies on free stream turbulence control using honeycomb and screens of various configurations, individually and also their combinations had been carried out [17, 18, 19, 20, 21, 22, 23, 24]. A honeycomb with square cells of L/d_{h} of 4–12 and at a downstream of 4d_{h} has turbulence intensity in between 0.15 and 0.2 [19]. The present wind tunnel has a square cell of L/d_{h} of 4, and the first screen is placed at a distance of 4d_{h} from the downstream end of the honeycomb. So it can safely be assumed that at the entrance of the first screen, the turbulence intensity will be in between 0.15 and 0.2. The axial turbulence reduction factor for a single screen of different sizes, when experimentally determined, was found to be in between 0.5 and 0.7; however, for a series of screens, the total turbulence reduction would be equal to the product of the individual reduction factor [25]. Thus the total reduction factor for a series of two screens is in between 0.25 and 0.49. So a typical combination of the honeycomb and two screens will bring the turbulence intensity in between 0.0375 and 0.098, even at worst inlet condition to the settling chamber. Thus for the numerical simulation, the turbulence intensity at the inlet is kept at 10%, considering all possible scenario.
The fantype boundary condition is a lumped parameter model that allows to input an empirical fan curve that governs the relationship between pressure rise and flow rate. In the present numerical analysis, the fantype boundary condition is replaced by flow rate (mass flow rate) and a constant pressure jump as boundary conditions for inlet and outlet, respectively. The required mass flow rate and pressure jump for the analysis are obtained from the system curve generated by experimental analysis.
The mass flow rate and static pressure for inlet and outlet boundary conditions
VFD frequency (Hz)  Average velocity (m/s)  Temperature (K)  Density (kg/m^{3})  Mass flow rate (kg/s)  Average static wall pressure at diffuser end (Pa) 

10  5.68  303.9  1.1615  0.614  − 9.81 
15  8.64  303.96  1.1613  0.933  − 24.5 
20  11.56  303.98  1.1612  1.249  − 39.2 
25  14.45  303.95  1.1613  1.561  − 58.9 
30  17.51  304  1.1611  1.891  − 88.3 
4 Results and discussion
Thus the qualitative agreement of the pressure and velocity variation as obtained by the numerical simulation and experimental measurements indicates that the boundary conditions based on the system curve can generate simulation which resembles the actual operating condition with reasonable accuracy. A similar approach of replacing the fantype boundary condition with other types of boundary condition can also be found in the work of Calautit et al. [13] and suggests that the CFD model can reproduce the wind tunnel measurements with an error of below 10% for mean velocity, pressure coefficient and turbulent intensity measurement.
The velocity contours (refer Figs. 13, 14, and 15) also reveal the acceleration of flow from upstream to downstream in the test section for all three horizontal planes at all speed. On observing the velocity contours of experimental and numerical flow fields at zl plane, at low speed, i.e., at 10 Hz, the extent of the velocity variation for the experimental and numerical flow field is 3.25% and 0.65%, respectively. As the speed of the fan is increased, the extent of the velocity variation for experimental and numerical flow field increases and reaches a peak of 8.14% and 1.16%, respectively. Similarly, on observing the velocity contours at the zm plane, the flow field is more or less the same as the flow field on the zl plane at all speed range. The extent of the velocity variation for experimental and numerical flow field, at low speed, is 3.96% and 0.65%, respectively. As the wind tunnel operates at a higher velocity range, the extent for the experimental and numerical flow field increases and reaches a peak of 8.23% and 1.16%, respectively. However, on observing the flow field at the zu plane, at a lowspeed range, the extent for the experimental and numerical flow field is 1.95% and 0.651%, respectively. However, as the higher speed range proceeds, the extent of the velocity variation for experimental and numerical flow field increase and reaches a peak of 5.81% and 1.16%, respectively.
The extent of the velocity range of the flow fields also depicts the magnitude of nonuniformity in the flow. On observing the extent of the velocity range of the three horizontal planes, the experimental flow field at zu planes appears to be relatively more uniform than the flow field at zm and zl planes. Although the zu and zl planes are symmetric about zm plane, the flow fields on the respective planes are quite different. The flow fields at zu plane appear to be more uniform (maximum extent of velocity variation is 5.81%) than the zl plane (maximum extent of velocity variation is 8.14%). This is on account of the dynamic oscillations experienced by the sensing probe mounted at the end of the telescopic arm of the measuring instrument. The models supported by a cantilever arm (sting) inside the test section often experiences dynamic oscillations due to unsteady flow and induces an error in the measurements [32].
It is to be noticed that the velocity range of the numerical flow fields mostly covers the peaks of the contour legend. The high gauge pressure, in case of numerical results (refer Fig. 9), is also a reflection of the highvelocity head inside the test section. This implies that numerical flow fields overpredict. This overprediction of the velocity field could also be because of inadequate mass flow rate provided at the inlet boundary conditions besides the inherent shortcoming of the turbulent models and lack of exact boundary or initial conditions. The LogTchebycheff method, which accounts for the velocity profile, can provide accurate flow rate [33, 34]. However, due to certain constraints in the accessibility of the test section, the mass flow rates are calculated using average (timeaveraged) velocity measured at the centerline of the test section. Generally, the centreline wind velocity in the test section is higher than the average velocity and using the centreline velocity for mass flow rate calculation will produce a higher mass flow rate. For a better prediction from numerical analysis, authors suggest LogTchebycheff method for obtaining the flow rates.
5 Conclusion

A comparison of the pressure variations obtained by the numerical simulation and experiment indicates good qualitative agreement. However, small discrepancies in flow properties like static pressure and velocity inside the test section are observed due to ideal and constant initial conditions as well as the inherent shortcoming of the turbulence model. The mass flow rate based on the centreline velocity approach is also a culprit in overprediction of flow properties in numerical simulation.

A comparison for the similarity of average velocity along the length of the test section shows a maximum error of 9.84% and along the height of the test section shows a maximum error of 7.75%. Thus, novel approach for boundary conditions is able to produce simulation results with reasonable accuracy.
Notes
Compliance with ethical standards
Conflict of interest
On behalf of all authors, the corresponding author states that there is no conflict of interest.
References
 1.Gonzalez Hernandez MA, Moreno Lopez AI, Jarzabek AA et al (2013) Design methodology for a quick and lowcost wind tunnel. Wind Tunn Des Their Divers Eng Appl. https://doi.org/10.5772/54169 CrossRefGoogle Scholar
 2.Bradshaw P, Pankhurst RC (1964) The design of lowspeed wind tunnels. Prog Aerosp Sci 5:1–69. https://doi.org/10.1016/03760421(64)90003X CrossRefGoogle Scholar
 3.Mehta RD, Bradshaw P (1979) Technical notes design rules for small low speed wind tunnels. Aeronaut J R Aeronaut Soc 83:443–449. https://doi.org/10.1017/S0001924000031985 CrossRefGoogle Scholar
 4.Eckert WT, Mort KW, Jope J (1976) aerodynamic design guidelines and computer program for estimation of subsonic. NASA Tech note 197Google Scholar
 5.William H (1999) Lowspeed wind tunnel testing. Wiley, New YorkGoogle Scholar
 6.Gordon R, Imbabi MS (1998) CFD simulation and experimental validation of a new closed circuit wind/water tunnel design. J Fluids Eng 120:311–318CrossRefGoogle Scholar
 7.Leifsson L, Koziel S (2015) Simulationdriven design of lowspeed wind tunnel contraction. J Comput Sci 7:1–12. https://doi.org/10.1016/j.jocs.2014.12.004 CrossRefzbMATHGoogle Scholar
 8.Leifsson L, Koziel S, Andrason F et al (2012) Numerical optimization and experimental validation of a lowspeed wind tunnel contraction. Procedia Comput Sci 9:822–831. https://doi.org/10.1016/j.procs.2012.04.088 CrossRefGoogle Scholar
 9.Lummer M (2019) Aircraft noise generation and assessment. CEAS Aeronaut J. https://doi.org/10.1007/s13272019003825 CrossRefGoogle Scholar
 10.Rodríguez Lastra M, Fernández Oro JM, Galdo Vega M et al (2013) Novel design and experimental validation of a contraction nozzle for aerodynamic measurements in a subsonic wind tunnel. J Wind Eng Ind Aerodyn 118:35–43. https://doi.org/10.1016/j.jweia.2013.04.008 CrossRefGoogle Scholar
 11.Zanoun ES (2017) Flow characteristics in lowspeed wind tunnel contractions: simulation and testing. Alexandria Eng J. https://doi.org/10.1016/j.aej.2017.08.024 CrossRefGoogle Scholar
 12.Moonen P, Blocken B, Roels S, Carmeliet J (2006) Numerical modeling of the flow conditions in a closedcircuit lowspeed wind tunnel. J Wind Eng Ind Aerodyn 94:699–723. https://doi.org/10.1016/j.jweia.2006.02.001 CrossRefGoogle Scholar
 13.Calautit JK, Chaudhry HN, Hughes BR, Sim LF (2014) A validated design methodology for a closedloop subsonic wind tunnel. J Wind Eng Ind Aerodyn 125:180–194. https://doi.org/10.1016/j.jweia.2013.12.010 CrossRefGoogle Scholar
 14.Gartmann A, Fister W, Schwanghart W, Müller MD (2011) CFD modelling and validation of measured wind field data in a portable wind tunnel. Aeolian Res 3:315–325. https://doi.org/10.1016/j.aeolia.2011.07.002 CrossRefGoogle Scholar
 15.Nisugi K, Hayase T, Shirai A (2005) Fundamental study of hybrid wind tunnel integrating numerical simulation and experiment in analysis of flow field. JSME Int J, Ser B 47:593–604. https://doi.org/10.1299/jsmeb.47.593 CrossRefzbMATHGoogle Scholar
 16.Hnidka J, Rozehnal D (2017) Pressure field in measurement section of wind tunnel. MATEC Web Conf. https://doi.org/10.1051/matecconf/201710700075 CrossRefGoogle Scholar
 17.Dryden HL, Schubauer GB (1947) The use of damping screens for the reduction of wind tunnel turbulence. J Aeronaut Sci 14:221–228. https://doi.org/10.2514/8.1324 CrossRefGoogle Scholar
 18.Ghorbanian K, Soltani MR, Manshadi MD (2011) Experimental investigation on turbulence intensity reduction in subsonic wind tunnels. Aerosp Sci Technol 15:137–147. https://doi.org/10.1016/j.ast.2010.06.009 CrossRefGoogle Scholar
 19.Kulkarni V, Sahoo N, Chavan SD (2011) Simulation of honeycombscreen combinations for turbulence management in a subsonic wind tunnel. J Wind Eng Ind Aerodyn 99:37–45. https://doi.org/10.1016/j.jweia.2010.10.006 CrossRefGoogle Scholar
 20.Laws EM, Livesey JL (1978) Flow through screens. Annu Rev Fluid Mech 10:247–266. https://doi.org/10.1146/annurev.fl.10.010178.001335 CrossRefzbMATHGoogle Scholar
 21.Loehrke RI, Nagrib HM (1976) Control of freestream turbulence by means of honeycombs: a balance between suppression and generation. J Fluid Eng 98:342–351. https://doi.org/10.1115/1.3448313 CrossRefGoogle Scholar
 22.Roach PE (1987) The generation of nearly isotropic turbulence by means of grids. Int J Heat Fluid Flow 8:82–92. https://doi.org/10.1016/0142727X(87)900014 CrossRefGoogle Scholar
 23.Santos AM, Souza DB, Costa FO et al (2016) Effects of screens set characteristics on the flow field in a wind tunnel. J Phys Conf Ser. https://doi.org/10.1088/17426596/733/1/012001 CrossRefGoogle Scholar
 24.Stathopoulos T (1984) Design and fabrication of a wind tunnel for building aerodynamics. J Wind Eng Ind Aerodyn 16:361–376. https://doi.org/10.1016/01676105(84)900187 CrossRefGoogle Scholar
 25.Scheiman J, Brooks JD (1981) Comparison of experimental and theoretical turbulence reduction characteristics for screens, honeycomb, and honeycombscreen combinations. J Aircr 18:638–643. https://doi.org/10.2514/3.57538 CrossRefGoogle Scholar
 26.ANSYS (2011) Fundamentals of ANSYS CFX, vol 15317. ANSYS, Canonsburg, pp 724–746Google Scholar
 27.Azzawi IDJ, Mao X, Jaworski AJ (2016) Design, fabrication and characterization of low speed openjet wind tunnel. Proc World Congr Eng II:1–6Google Scholar
 28.Calautit JK, Hughes BR (2016) CFD and experimental data of closedloop wind tunnel flow. Data Br 7:216–220. https://doi.org/10.1016/j.dib.2016.02.033 CrossRefGoogle Scholar
 29.Moonen P, Blocken B, Carmeliet J (2007) Indicators for the evaluation of wind tunnel test section flow quality and application to a numerical closedcircuit wind tunnel. J Wind Eng Ind Aerodyn 95:1289–1314. https://doi.org/10.1016/j.jweia.2007.02.027 CrossRefGoogle Scholar
 30.Canonsburg TD (2013) ANSYS CFX reference guide, vol 15317. Ansys Inc, Canonsburg, pp 724–746. https://doi.org/10.1186/1756038131 CrossRefGoogle Scholar
 31.Kato H, Obayashi S (2011) Integration of CFD and wind tunnel by data assimilation. J Fluid Sci Technol 6:717–728. https://doi.org/10.1299/jfst.6.717 CrossRefGoogle Scholar
 32.Young PC, Buehrle RD, Balakrishna S, Kilgore WA (1994) Effects model of vibration on inertial attitude measurement devices. NasaTm109083Google Scholar
 33.Ben Mosbah A, Flores Salinas M, Botez R, Dao T (2013) New methodology for wind tunnel calibration using neural networks–EGD approach. SAE Int J Aerosp 6:761–766. https://doi.org/10.4271/2013012285 CrossRefGoogle Scholar
 34.Caré I, Bonthoux F, Fontaine JR (2014) Measurement of air flow in duct by velocity measurements. EPJ Web Conf 77:1–5. https://doi.org/10.1051/epjconf/20147700010 CrossRefGoogle Scholar