27 International Nuclear Atlantic Conference - INAC 27 Santos, SP, Brazil, September 3 to October 5, 27 ASSOCIAÇÃO BRASILEIRA DE ENERGIA NUCLEAR - ABEN ISBN: 978-85-99141-2-1 PRESSURE DROP OF FLOW THROUGH PERFORATED PLATES José A. Barros Filho 1, Moysés A. Navarro 1, Felipe A. P. Magalhães 2, André A. Campagnole dos Santos 2 1 Centro de Desenvolvimento da Tecnologia Nuclear (CDTN / CNEN - MG) Rua. Professor Mário Werneck, s/n Cidade Universitária 3123-97 Belo Horizonte, MG jabf@cdtn.br navarro@cdtn.br 2 Departamento de Engenharia Mecânica DEMEC Universidade Federal de Minas Gerais - UFMG Av. Antônio Carlos, 6627 31271-91 Belo Horizonte, MG fapmbr@yahoo.com.br acampagnole@yahoo.com.br ABSTRACT This paper presents the results of the first part of a research program devised to develop and validate a CFD methodology to simulate the flow of water through perforated plates with geometry similar to the ones of the bottom end piece of a Pressurized Water Reactor. This component drives the coolant into the core, promotes an adequate mixing of dissolved substances and holds the fuel assembly mechanically. It accounts for a great part of the pressure loss through the core. The experimental pressure losses of water flow through 5 different perforated plates are presented here. The profile of the recovering of the pressure right downstream the plates was also measured. The experimental results were compared with numerical modeling performed with the commercial code CFX 1.. Three different turbulence models of the two equation type were used in the simulations. The analysis of the results shows that the agreement between calculations and experiments decrease for flows with higher disturbance of the current. 1. INTRODUCTION There is an increasing interest of the nuclear industry in the application of three dimensional computational fluid dynamics (CFD) codes as a supplement to or in combination with system codes. With further progress in safety analysis techniques, an increasing use of CFD codes for nuclear applications is expected. At present, the main objective with respect to CFD codes is generally to improve confidence in the available analysis tools and to achieve a more reliable approach to safety relevant issues [1, 2, 3]. Aiming at these goals, the Thermohydraulics Laboratory of the Centro de Desenvolvimento da Tecnologia Nuclear - CDTN started a research program to develop and validate a CFD methodology to simulate the flow of water through components of a PWR s core. The first component chosen for the program is the bottom end piece of the fuel element. This component, consisting basically of a nozzle and a perforated plate, drives the coolant into the core, promotes an adequate mixing of dissolved substances and holds the fuel assembly mechanically. It accounts for a considerable amount of the core s pressure loss. Perforated plates have also been studied as flow conditioners for measuring stations [4, 5] as
distillation trays to increase the heat and mass transfer between fluids [6] and as flow controller emerging from diffuser [7]. The agreement between the pressure loss coefficients obtained from some experiments [8] and the ones defined by Idelchik [9] are not always acceptable. Even though there is a considerable number of studies on numerical simulation of flow through perforated plates [1, 11, 12] some issues related to turbulence models, near wall treatment and refinement of the mesh have not been properly explained yet. This work presents the results of the pressure loss of water flow through perforated plates with geometry similar to the ones of the bottom end piece of a PWR fuel assembly. The profile of the pressure recovering right downstream the plates was measured as well. It also presents the numerical modeling of the tests, performed with the commercial CFD code CFX 1. [13]. 2. EXPERIMENTAL SET UP AND MEASUREMENTS A schematic of the facility used in the experiments is shown in Fig 1. Water, at room temperature, is pumped from a storage tank through an orifice plate flow meter and into the test section. Assembled in the flange of the inlet of the test section there is a perforated plate flow conditioner with the purpose to decrease flow perturbation caused by pipe bends upstream the section. The test section is a square duct made of clear acrylic with internal side of 72.14 mm and a total length of 3435.9 mm. The section is divided in two halves with flanges on both ends. The perforated plates, also made of acrylic, were positioned in the middle of the section by means of a holding flange. Sixteen pressure taps were drilled along one wall of the test section. The upper and lower ones were placed in the undisturbed region of the flow (see Fig 1), to measure the pressure loss coefficient. The other fourteen were placed, very close to each other, right downstream the plate, to determine the profile of the pressure in the recovering region. The absolute pressure was taken on the opposite wall near the plate. The test temperature was measured at the inlet of the section. The tests were performed at room temperature. The perforated plates are 2 mm thick and have 25 identical holes in a square arrangement with a pitch of 14.3 mm. Two distinct hole diameters were chosen so as to cover the range of free flow area ratios most commonly found in bottom pieces of PWRs. Plates with straight and chamfered hole inlets were tested. Table 1 lists the geometrical features of the plates tested. In Fig. 1 there is a photo of a plate already assembled in the holding flange. The signals from the transducers are sent to a data processing station which performs an error analysis, converts them in engineering data and stores the results. 3. NUMERICAL METHODOLOGY The simulations were performed with the commercial CFD code CFX 1. [13]. After some simulations that showed that the flow at the inlet of the test section could be considered symmetric, it was decided to perform the simulations with the 1/8 symmetry, since it
drastically reduces computing time. The level of turbulence is also requested as inlet boundary condition. As long as there is a conditioner at the inlet of the section, the level of turbulence was set to the maximum allowed by the code. Here again simulations with different degrees of turbulence at the inlet were performed to check its influence on the results, showing no relevant differences. Figure 1. Schematic of experimental set up. The mesh used was structured in three different regions. A finer mesh was used in the region from 2 mm upstream to 3 mm downstream the plate. The characteristic dimension of this part of the mesh was set to.5 mm. From 3 mm to 23 mm downstream the plate, a mesh with a characteristic dimension of 1.5 mm was used. This is supposed to cover the region of disturbed flow, where velocity and turbulence fields change abruptly. The rest of the test section was covered by a coarser mesh with a characteristic dimension of 3 mm. Those values were chosen after a mesh refinement study which showed that further refinement would increase computational efforts and give no correspondent increase in the quality of the results. Near the walls, 1 layers of structured mesh with increasing thickness away from the wall were used to capture the effects of the boundary layer in all simulations. An effort was made to keep the value of the thickness of the first layer as small as possible (y + less than 5) to take best advantage of the near wall treatments of the code.
Table 1. Dimensions of the plates Plate Pitch [mm] Width [mm] Hole diameter [mm] Free flow area ratio Characteristics of the chamfer Dimension l [mm] Angle θ [degree] 1 8.66.283 2 8.64.283 1.792 6 3 14.3 2 1.68.43 4 1.69.431 1.445 6 5 1.69.431 1.17 3 Among the various turbulence models available in the code, the standard k-ε, k-ϕ and the (Shear Stress Transport) were chosen for this work. These two equation turbulence models fall into the category defined as eddy viscosity models. They were adopted in this work mainly because of their numerical robustness and fast convergence. 4. RESULTS The experimental results are summarized in Tab. 2. The evaluated uncertainties for the measured parameters are: mass flow (q m ):.23 kg/s; absolute pressure (P):.2 bar; test temperature (T): 1 o C ; Reynolds number (Re): 1533 and differential pressure (DP):.1 mbar A comparison between experimental and calculated results is shown in Table 3. The values of DP were taken from pressure taps and 15, in the flow regions undisturbed by the plate. As can be easily noticed from Table 3, there is a good agreement between simulations with the different turbulence models. The agreement between simulations and experiments, however, indicates that, the more disturbed the flow, the worse is the result. The simulations of the plates with sharp inlet holes gave discrepancies two times greater than those with inlet chamfered, for the same hole diameter. Also, comparing only the plates with sharp inlet holes, plate number 1, with a smaller free flow area ratio, has a greater discrepancy than plate number 3. Plate number 5 has a chamfer angle that offers less resistance to the flow. It therefore presents the best agreement between simulation and experiment. Table 4 shows a comparison of the pressure loss coefficient K obtained in this work with the values calculated with the Idelchik handbook [9]. This coefficient is given by the formula: DP = (K ρ v 2 )/2 (1) Where ρ is the density and v the velocity in the flow channel. DP is the pressure through the plate. The comparison refers only to the plates with the sharp inlet because the handbook does not account properly for chamfered inlets. Figure 2 shows plots comparing the calculated and experimental recovering of the pressure downstream the plates. The experimental points refer to the pressure taps 1 to 14. In the figure, two vertical lines show the position of the plate
Table 2. erimental results Plate 1 2 3 4 5 Test conditions Tap pressure drops [mbar] q m [ kg/s] 3.318 3.324 3.322 3.317 3.321 P [bar].88.695.69.686.691 T [ o C] 31.9 33.4 28.7 28.8 3.9 Re 6376 62365 56411 56493 59166-1.62.47.56.48.48-2 3.95 29.66 12.64 11.17 1.88-3 26.88 25.57 8.8 7.52 7.9-4 23.79 22.3 7.93 6.48 6.6-5 21.97 2.3 7.5 6.2 5.62-6 2.76 18.91 7.23 6.2 5.51-7 2.17 18.46 7.8 5.91 5.41-8 2.7 18.2 7.1 5.8 5.37-9 19.96 18.32 7. 5.88 5.32-1 19.94 18.46 6.89 5.87 5.48-11 19.91 18.12 7. 6.3 5.31-12 19.96 18.13 6.97 6.2 5.45-13 2.3 18.14 7.9 5.96 5.4-14 19.96 18.13 7.9 5.98 5.41-15 2.3 18.64 7.39 6.35 5.74 Table 3. Comparison between experimental and calculated data Plate DP experimental DP calculated [mbar] [mbar] (taps and 15) k-ε Diff % k-ω Diff % Diff % 1 2.3 26.12 28.66 25.89 27.54 26.21 29.11 2 18.64 2.93 12.29 2.73 11.21 2.52 1.9 3 7.39 8.98 21.52 8.97 21.38 9. 21.79 4 6.35 6.99 1.8 7.7 11.34 6.98 9.92 5 5.74 5.77.52 5.9 2.79 5.65-1.6 Table 4. Comparison between experimental and Idelchik pressure loss coefficient. Plate K experimental K calculated k-ε Diff % Diff % k-ω Diff % Idelchik Diff % 1 9.36 12.22 3.56 12.26 3.98 25.89 27.54 1.44 11.53 3 3. 3.78 26. 3.78 26. 8.97 21.38 3.16 5.33
4 32 24 16 5 4 3 2 8 Plate 2-2 2 4 6 8 1 12 14 16 2 16 12 8 1 Plate 1-2 2 4 6 8 1 12 14 16 15 12 9 6 4 Plate 3-2 2 4 6 8 1 12 14 16 15 12 9 6 3 Plate 4-2 2 4 6 8 1 12 14 16 [cm] 3 Plate 5-2 2 4 6 8 1 12 14 16 Figure 2. Calculated and experimental pressure recovering 5. CONCLUSIONS erimental tests were performed where the pressure loss of water flow through perforated plates was measured. The results were compared with simulations performed with the commercial code CFX. The simulations were done with three different turbulence models, the standard k-ε, k-ω and the. Although they presented very close results to each other, there seems to be a direct correspondence between the degree of disturbance of the flow and the discrepancy in the results. This fact seems to confirm other works reporting the inadequacy of the standard k-ε model for flows in complex geometries with sudden area change [14]. The k-ω and models, however, were expected to give better results, once they incorporate improvements over the k-ε to model the separation of the current after sudden expansions. It is important to stress here that those models were chosen among others more advanced available in the code, because of their numerical robustness, fast convergence and straightforward use. These features are expected in a calculation intended for licensing purposes in the nuclear industry.
The results of the tests with plates 3, 4 and 5 also show the strong influence of the orifice chamfer in the pressure loss. In the sequence of the work initiated here, other turbulence models will be tested, along with more structured meshes. Different types of perforated plates will also be tested. ACKNOWLEDGMENTS The authors express their thanks to the Fundação de Amparo à Pesquisa do Estado de Minas Gerais - FAPEMIG, for the financial support. REFERENCES 1 Yadigaroglu, G.. Andreani, M.. Dreier. J., Coddington. P., Trends and needs in experimentation and numerical simulation for LWR safety, Nuclear Engineering and Design, 221, pp 25 223 (23). 2 IAEA, Thermohydraulics Relationships for Advanced Water Cooled Reactors, TECDOC-123, (21). 3 IAEA, Use of computational fluid dynamics codes for safety analysis of nuclear reactor systems, Summary report of a technical meeting jointly organized by the International Atomic Energy Agency and the Nuclear Energy Agency of the Organization for Economic Co-operation and Development. Pisa. Italy. 11 14, TECDOC-1379, (22) 4 Schlüter, Th. and Merzkirch, W., PIV measurements of the time-averaged flow velocity downstream of flow conditioners in a pipeline, Flow Meas. Instrum., pp 7, 173-179 (1996). 5 Spearman, E. P., Sattary, J. A. and Reader-Harris, M. J., Comparison of velocity and turbulence profiles downstream of perforated plate flow conditioners, Flow Meas. Instrum., 7, 181-199 (1996). 6 Liu, R., Ting, D. S-K. e Rankin, G. W., On the generation of turbulence with a perforated plate, erimental Thermal and Fluid Science, 28, pp. 37-316 (24) 7 Sahin, B. and Ward-Smith, A. J., The use of perforated plates to control the flow emerging from a wide-angle diffuser with application to electrostatic precipitator design, J. Heat Fluid Flow, 8, pp 124-131 (1987). 8 Gan, G. and Riffat, S. B., Pressure loss characteristics of orifice and perforated plates, erimental Thermal and Fluid Science, 14, pp 16-165 (1997). 9 Idelchik, I. E., Handbook of hydraulic resistance, third edition, CRC Press Inc., Boca Raton, US (196) 1 Erdal, A. and Andersson, H. J., Numerical aspects of flow computation through orifices, Flow. Meas. Instrum., 5, pp 27-37 (1997). 11 Erdal, A., A numerical investigation of different parameters that affect the performance of a flow conditioner, Flow. Meas. Instrum., 8, pp 93-12 (1997). 12 Frattolillo, A. and Massarotti, N., Flow conditioners efficiency a comparison based on numerical approach, Flow Meas. Instrum., 13, pp 1-11.(22). 13 CFX-1., User manual, ANSYS-CFX, (26) 14 Tzanos, C. P., Performance of k-ε Turbulence Model in the Simulation of LWR Fuel Bundle Flow, Trans. Am. Nucl. Soc., 84, 197-23 (21).