ISTP-6, 25, PRAGUE 6 TH INTERNATIONAL SYMPOSIUM ON TRANSPORT PHENOMENA WIND AFFECTED ISOTHERMAL AIRFLOW PATTERNS IN A SCALE MODEL OF A SWINE BARN MEASURED WITH PIV SYSTEM Pavel Bárta Czech Technical University, Faculty of Mechanical Engineering, Department of Fluid Dynamics and Power Engineering, Division of Fluid Dynamics and Thermodynamics Address: Technická 4, 66 7 Praha, Czech Republic pavel_barta@centrum.cz, +42 724 52 533, fax: +42 224 3 292 Keywords: Airflow patterns, Ventilation, Wind, PIV Measurement Abstract This project investigates a new ventilation system, which improves living conditions for livestock. The main aim is to determine the wind effect on the airflow patterns inside the barn and to find the flue inlet position which ensures steady and wind unaffected airflow patterns inside the building. To reach this goals both experiments in wind tunnel, including PIV (Particle Image Velocimetry) measurements, and numerical simulation by CFD program COSMOS/Flow was used. Introduction In society today people are as much concerned with the living conditions for animals as for humans, especially the animals used in animal production as they are here to serve us. In this case pigs in the swine barn. Outlet Breathing zone Investigation of this ventilation system, that has never been used, could be helpful not only for swine barn but for other places like for example cow stables, horse stables and other facilities. The biggest advantage of the investigated ventilation system is that the manure fumes does not pass the breathing zone of the animals. This is possible because the ventilation outlets are placed below the barn and the inlets in the roof. Thus fresh air enters the barn and goes out from under, normally it s the other way around. The manure fumes will disappear under the feet of the animals and never reach their breathing zone. It s worth considering the fact that this system is very energy consuming. The research about the ventilation system was made in Denmark (Research Centre Bygholm), a country with high living standards. In Denmark the expenses of the system are not as big an issue as the welfare of the animals. Inlet Inlet Manure fumes Outlet Figure : Scheme of classic (left) and investigated (right) ventilation system.
Pavel Bárta. Goals The ventilation system is equipped with two flues, one is on the leeward side and the other one is on the windward, so the flow-around can cause different flue inlet conditions. Subsequently unsteady (on wind velocity depending) and asymmetrical airflow patterns inside the barn appear, which are not desired. Thus the main object of this project was to determine optimal height of flues, when the airflow patterns within the barn will not be influenced by wind conditions on the outside. This was done by using both experimental measurements and numerical simulations. sidewalls and 25-degree roof slope (see Figure 2). The model is equipped with two flues in the middle of each side of the roof, baffles and exhaust ventilation system. Air was exhausted from the model through a tube (placed in the bottom of a wall), which was connected to an exhauster that could provide different airflow rates of ventilated air. 2. Description of the Scale Model Experiment The experiments were carried out with a scale model in a wind tunnel at Bygholm Research Centre. This was done under isothermal conditions and with the wind direction perpendicular to the sidewalls. 2. Wind tunnel The open circulation wind tunnel was.2 m long with cross section (2.78 x 2.78) m and equipped with four axial fans at the beginning of the wind tunnel which could provide different velocity of air. There was a working table at distance of 5.55 m from the fans. The dimensions of working table were.5 m high, 5 m long and 2.2 m wide. A surface on the table was made of carpet for simulation of roughness. 2.2 Barn A :2 scale model of a typical growing/finishing barn was used for the experiments. The scale model was made of plexiglas, which is important for observation of the airflow patterns and measuring. The dimensions were scaled down from naturally ventilated barn with section being m long and m deep with 2.4 m high Figure 2: 3D wire frame of the barn. 2.3 Instrumentation The velocity profiles inside flues were measured in the middle section of the model. A Pitot tubes were used for measuring these velocity profiles in both flues (in leeward and windward side). The setup of the Pitot tubes is shown in Figure 3, where is: H flue height, r. radius, where the Pitot tubes were placed - positive values are in the wind direction. A traversing system was used for scrolling of the laser sensor (Laser Doppler Anemometer). The laser sensor measured the velocity in the middle section of the model in two different heights above the floor h = 9 mm and h 2 = 95 mm. 2
WIND AFFECTED ISOTHERMAL AIRFLOW PATTERNS IN A SCALE MODEL OF A SWINE BARN MEASURED WITH PIV SYSTEM H Measuring plane r r Wind Leeward flue Pitot tubes Windward flue Figure 3: The set up of the 3. Pitot Numerical tubes inside Model flues. 2.3. 2D-PIV (Particle Image Velocimetry) For PIV measurement a DANTEC apparatus in following configuration has been used: a Nd-YAG pulse LASER New Wave Gemini 5 Hz PIV, 2 mj, 2 cavities CCD camera Dantec HiSense PIV-PLIF, 28 x 24 pixel 2 bit, CCD chip of type progressive scan interline, max. 9 Hz single frame mode, 4,5 Hz double frame mode PIV processor Dantec FlowMap 5, GB buffer PC 2 x XEON 3 GHz, HDD 3 Gb, Windows XP Software FlowManager v. 4.3 + FlowManager Advanced measurement synchronization (8S32) + FlowManager MatLab Links (8S36) + other modules 2.4 Wind Profiles Two required wind profiles (3.2 m/s and 5m/s) were achieved with nets of different fineness used to cover the fans. The wind profile H equation U = U ref was used as a H ref model of the wind (for details see [2]). a All CFD simulations were carried out in the COSMOS/Flow program. 3. Geometry The geometry of the problem as well the numerical model can be divided into three parts: the first part is the model of the inside of the barn with its dimensions, the next part includes the surrounding space of the barn with its outside dimensions, the last part creates sufficient wind profile in the model. 3.. Barn The dimensions of the model are scaled with the factor :2 ([4]). For the simulation of the outside, as well as the inside, dimensions are important. These dimensions were taken from the physical model. 3..2 Surroundings The surroundings represents atmospheric boundary layer. In the table below are the dimensions of the surroundings of the barn, which are chosen in ratio to the size of the barn. It is important to keep this ratio big enough. Otherwise the 3
Pavel Bárta boundaries of the surroundings can influence the simulation. part of the flow. That means the measured data and CFD simulation data is comparable. Barn model (mm) Model of surr. (mm) Ratio (surr./barn) Width 5 2 4 Height 3 9 3 Length 55 28 5. Table : Ratio between surroundings and physical model dimensions. 3..3 Fans Fans create two sufficient wind profiles (boundary conditions) 3.2 m/s and 5 m/s, therefore the fans consists of ten blocks (each 9 mm high). Each block has its own defined velocity input. This velocity was calculated from the wind profile equation (see 2.4). 3.2 Meshing High-order tetrahedral elements, standard mesher type and a few mesh controls were used for all the models. The mesh was refined in and around inlets (flues) and inside the barn model. The numerical model with cm high flues has 42 22 elements and 63 226 nodes. 3.3 Flow and Fluid Properties It is used steady state internal turbulent incompressible flow with k-ε turbulence model for problem description. Air with stagnation temperature 293 K, stagnation pressure 325 Pa and density.247 kg/m 3 was chosen as a fluid, according to the measurement. 4. Results The data of the numerical simulation were extracted of the model according to the physical model instrumentation (see 2.3). Velocity (m/s) Velocity (m/s) Comparison of the wind profiles for 3.2 m/s 4 3.5 3 2.5 2.5.5 2 4 6 8 Height (cm) Results from COSMOS Empirical model Wind tunnel data Comparison of the wind profiles for 5. m/s 6 5 4 3 2 2 4 6 8 Height (cm) Results from COSMOS Empirical model Wind tunnel data 4. Comparison of the Wind Profiles The data is in very good agreement for the height above 3 cm and is close in the lower Figure 4: Comparison of wind profiles for 3.2 m/s and 5 m/s. 4
WIND AFFECTED ISOTHERMAL AIRFLOW PATTERNS IN A SCALE MODEL OF A SWINE BARN MEASURED WITH PIV SYSTEM 4.2 Inlet Data Evaluation It is assumed that the influence of wind at a given wind velocity is dependent on the inlet velocity at no wind (with suction only). One of a practical ways of presenting data for different combinations of the wind velocity W, and inlet velocity at no wind V, is to use the inlet velocity ratio and the wind ratio [8]. The resulted graphs shows how inlet velocity ratio depends on wind ratio. The inlet velocity ratio is the inlet velocity with wind divided by the inlet velocity at no wind - Vw Vo. The wind ratio is the wind velocity divided by the inlet velocity at no wind - W Vo. Inlet velocity ratio Wind ratio Data Figure 5: Scheme of resulted charts. 4.2. Comparison of Normalized Velocity Ratios,4 Comparison - H = cm Noramlised inlet velocity ratio Vw/Vo Comparison - H = 5 cm,4,2,8,6,4,2 2 3 4 5 Comparison - H = cm,4,2,8,6,4,2 2 3 4 5 Leeward measuring Windward measuring Leeward COSMOS Windward COSMOS Comparison - H = 5 cm Noramlised inlet velocity ratio Vw/Vo,2,8,6,4,2 2 3 4 5 Leeward Measuring Windward measuring Leeward COSMOS Windward COSMOS,4,2,8,6,4,2 2 3 4 5 Figure 6: Comparison of normalised velocity ratios. 5
Pavel Bárta All the compared values are in good agreement, especially results for flue height H = cm and H = 5 cm. Values from the compared methods have the same tendencies: The leeward flue sucks more than windward flue for all the heights except one (H = cm). Data from both sources has almost the same amplificatory trends. Both sources of data show that flue inlet for H = 5 cm is not influenced with the wind ratio. That means the flue inlets at such height could ensure steady (wind unaffected) and symmetrical airflow patterns inside the barn, which is desirable. Values for low flues (H = cm and H = 5 cm) are slightly different. It is probably because of the inlet velocity profile evaluation or different behaviour of Cosmos (recirculating flow see below 4.2.2 or [4]). 4.3 Comparison of the Recirculating Flow The main difference between COSMOS simulation and physical model is in the recirculating flow that occurs behind the ridge of the roof. Otakar Horejš [6] made few pictures with a digital camera of the recirculation flow on the physical model. He used smoke and laser sheet for visualization. There is clear recirculating flow behind the building. He concludes that cm and 5 cm high leeward flue is within the zone of the recirculating flow. He also found that cm high flue is on the border. The recirculating flow has not been modeled properly in the COSMOS simulation. There is only a small flow for the flue height H = cm. This recirculating flow is just around flue (approximately above 5% of roof area). Measured and CFD simulation data support the result that the leeward flue with height 5 cm is out of the recirculating flow which should ensure similar inlet conditions for both of the flues (both flue inlets reach the height where barn doesn t affect wind flow). 4.5 PIV Measurement Results These results clearly shows the recirculation flow and therefore supports measured data. Detail inside investigation was impossible due to reflection of the laser sheet by the plexiglass model. Figure 7: Resultant velocities in the animal zone (plane 9 mm above the floor), flow rate 55 m 3 /h, H = 5 cm, W = 3.2 m/s. 6
WIND AFFECTED ISOTHERMAL AIRFLOW PATTERNS IN A SCALE MODEL OF A SWINE BARN MEASURED WITH PIV SYSTEM 4.4 Airflow Pattern inside the Barn One of the other important topics was to clarify the impact of the heights of the flues on the overall airflow patterns inside the building. Velocities inside the barn (obtained from two horizontal planes) are in best agreement for flue height H = 5 cm, in the lower plane. The data are slightly different for the lower flue heights. The reason is probably that the airflow is steadier with increasing height of flue. Moreover we can observe that the distribution inside barn is influenced by position of the inlet flue height, as was supposed. Airflow pattern on the surveyed planes is relatively symmetric for flue height H =5 cm. The symmetry of airflow pattern deteriorates with decreasing flue height. Figure 7 shows resultant velocities plots with asymmetrical airflow patterns for flue height 5 cm. 5. Conclusion The resultant data shows that distribution inside the barn is influenced by position of the inlet. The airflow pattern is relatively symmetric for flue height H = 5 cm. The symmetry of the airflow pattern deteriorates with decreasing flue height. The comparison of measuring and numerical simulation data on the model with flue height H = 5 cm clearly shows that the inlets are unaffected at this height. This inlet is.6 cm above the ridge of the roof (model scale is :2). different inlet angles of wind or to achieve similar results by CFD simulation. Finally I also recommend applying more suitable model of turbulence (for example k-ω) for following CFD simulations, because I realized few disagreements between numerical and physical model (recirculating flow). References [] ASHRAE. 996. Handbook HVAC Systems and Equipment. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. 996 [2] ASHRAE. 997. Handbook of Fundamentals. American Society of Heating, Refrigerating and Air- Conditioning Engineers, Inc. 997 [3] Bárta P., Horejš O.: Isothermal Airflow Patterns in a Scale Model of a Swine Barn with a Flue Effect. Vitus Bering, Centre for Higher Education, Horsens, 22 [4] Bárta, P.: Wind Affected Airflow Patterns in a Scale Model of a Swine Barn. ČVUT, Praha, 23 [5] Etherige, D., Sandberg,M.: Building Ventilation Theory and Measurements. John Wiley & Sons Ltd., 996 [6] Horejš, O.: Wind Affected Airflow Patterns in a Scale Model of a Swine Barn. ČVUT, Praha, 23 [7] Morsing, S., Ikeguchi, A., Bennetsen; J.C., Strom, J.S., Ravn, P., Okushima, L..: Wind Induced Isothermal Airflow Patterns in a Scale Model of Naturally Ventilated Swine Barn with Cathedral Ceiling. Structures & Environment Division of ASAE, 2 [8] Morsing, S.,Strøm J.S.: Wind Protection of Sidewall Inlets a New Windbreak Design. Livestock Enviroment V. Proceedings of the fifth International Symposium, Blooomington, Volume II, Minnesota, 997 I conclude that the flue inlet should be at least 2.2 m above the roof ridge of this barn. In real life the wind will come from different often fluctuating directions and the ventilation rate will change with changing temperatures. Because of these reasons it is a good idea to carry out full scale measurements in commercial barn in order to valid the scale model data. Other possibilities are to execute more laboratory experiments with a scale model for 7