Journal of Nuclear Science and Technology ISSN: 0022-3131 (Print) 1881-1248 (Online) Journal homepage: http://www.tandfonline.com/loi/tnst20 Identification of the Relationship between Local Velocity Components and Local Wall Thinning inside Carbon Steel Piping Kyeong Mo HWANG, Tae Eun JIN & Kyung Hoon KIM To cite this article: Kyeong Mo HWANG, Tae Eun JIN & Kyung Hoon KIM (2009) Identification of the Relationship between Local Velocity Components and Local Wall Thinning inside Carbon Steel Piping, Journal of Nuclear Science and Technology, 46:5, 469-478, DOI: 10.1080/18811248.2007.9711554 To link to this article: https://doi.org/10.1080/18811248.2007.9711554 Published online: 19 Mar 2012. Submit your article to this journal Article views: 176 View related articles Citing articles: 15 View citing articles Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalinformation?journalcode=tnst20
Journal of NUCLEAR SCIENCE and TECHNOLOGY, Vol. 46, No. 5, p. 469 478 (2009) ARTICLE Identification of the Relationship between Local Velocity Components and Local Wall Thinning inside Carbon Steel Piping Kyeong Mo HWANG 1;, Tae Eun JIN 1 and Kyung Hoon KIM 2 1 Korea Power Engineering Co., 360-9 Mabuk-dong, Gihung-gu, Yongin-si, Gyeonggi-do 446-713, Korea 2 Kyunghee University, 1 Seocheon-dong, Gihung-gu, Yongin-Shi, Gyeonggi-do 446-701, Korea (Received October 10, 2008 and accepted in revised form February 12, 2009) A study for identifying the relationship between turbulent parameters and local wall thinning inside a carbon steel piping was performed. Experiments and numerical analyses for several types of downscaled piping components were conducted, and the obtained results were compared. Based on the results indicating that flow behaviors inside piping components can be sufficiently simulated by numerical analysis, numerical analyses for the models magnified to the actual sizes of plants were carried out. To determine the relationship between turbulent parameters and local wear rates, numerical analyses were performed for 17 piping components of 7 types installed in a feedwater system. Turbulent parameters resulting from numerical analyses were compared with the local wear rates based on the measured thickness data. From the comparison of the results, the vertical flow velocity component (V r ) flowing again to the wall after separation due to geometrical configurations or direct collision against the wall at an angle of some degrees was found to be analogous to the configuration of local wall thinning. From the least-squares fitting result, it was derived that the average relationship between V r and the local wear rate is proportional to 0.55-fold V r adding 0.1 to the standard deviation of 0.65. KEYWORDS: flow accelerated corrosion (FAC), local wall thinning, Tee, Orifice, FLUENT, numerical analysis, least-squares fitting I. Introduction Flow-accelerated corrosion (FAC) may lead to wall thinning and eventual rupture of the pressure boundary of carbon steel components exposed to flowing water or wet steam. The nuclear regulatory bodies in many countries have strengthened the wall thinning management related to FAC since the 1986 Surry 1,2) and 2004 Mihama 3) pipe rupture events. Based on the fact indicating that wall thinning typically occurs in localized regions, a study for deducing the relationship between the local wear rates and the turbulence parameters related to local wall thinning was performed. In this study, experiments using a downscaled experimental facility on the basis of the similitude law were carried out, and the feasibility of the numerical analyses was verified to be appropriate based on the comparison between the experimental and numerical analysis results. The numerical analyses after magnifying the experimental verification models to the actual sizes of plants were reperformed, and the turbulence parameters were compared with the measured wall thicknesses. A correlation between the local wear rates and the turbulence parameters related to local wall thinning was observed based on the numerical analysis results for 17 piping components of 7 types. Corresponding author, E-mail: hkm@kopec.co.kr ÓAtomic Energy Society of Japan II. Experiment and Numerical Analysis 1. Experiment The downscaled experimental facility consisting of various components, such as tee and orifice, was constructed on the basis of the similitude law since it is very hard to replicate the actual operating conditions for the piping components of nuclear plants. From an engineering point of view, three types of similarities, i.e., geometrical, kinematic and dynamic similarities, need to be achieved between the experimental facility and actual piping components. The geometrical similarity was provided through 4.4:1 for the tee and 3.5:1 for the orifice scaled-down in all directions. The kinematic similarity was satisfied by single-path flow in the case of the orifice and by controlling 1:1 flow rates for upstream and branch compared with the actual piping components in the case of the tee. For the dynamic similarity, the approximate similarity was applied due to a very high Reynolds number inside the piping components of actual plants. It is known that the flow condition of very high Reynolds number is not sensitive to Reynolds number. 4,5) Based on the approximate similarity theory, the flow rate during experiment was controlled to acquire a high Reynolds number above 3 10 5. Table 1 shows the Reynolds numbers for the actual and experimental conditions. Based on the results indicating that flow behaviors inside piping components could be sufficiently simulated by numerical analysis, numerical analyses 469
470 K. M. HWANG et al. Table 1 Comparison of Reynolds numbers between actual and experimental conditions Item Unit Actual Condition Experimental Condition Components Tee Orifice Tee Orifice Inlet velocity m/s 2.80 7.36 2.75 2.82 Pressure Pa 9:73 10 6 9:57 10 6 1:01 10 5 1:01 10 5 Temp. C 232.22 137.39 15.0 15.0 Density kg/m 3 830.91 933.18 1,000 1,000 Viscosity Pa-s 1:16 10 4 2:03 10 4 1:14 10 3 1:14 10 3 ID m 0.66 0.43 0.15 0.13 Re number 1:33 10 7 1:46 10 7 3:62 10 5 3:09 10 5 Fig. 2 Configuration of tee Fig. 1 Layout of experimental facility for the models magnified to the actual sizes of plants were carried out. Figure 1 shows a layout of the experimental facility. Figures 2 and 3 show the configurations of the tee and orifice equipped with rubber tubes for measuring local pressure. These piping components were made of transparent acrylic material for inside visibility, while the other piping components were made of polyvinyl chloride (PVC). The experiments were carried out under the conditions of 15 C and 1:013 10 5 Pa. The inflow velocities at the tee and orifice were 2.75 and 2.82 m/s, respectively. Pressure differences between the base point of the inlet and the target points of the outlet, as shown in Figs. 4 and 5, were measured to identify flow behaviors inside the piping components since such behaviors depend on pressure distribution. Water and mercury manometers were used for pressure measurement. The spacings between the target points were 1 cm for the tee and 2 cm for the orifice. 2. Numerical Analysis To verify the experimental results, numerical analysis models were developed using the FLUENT code and Gambit program for mesh generation. Figures 6 and 7 illustrate three-dimensional (3D) cell divisions based on a Tet/Hybrid Fig. 3 Configuration of orifice grid. The grids for the tee and orifice are subdivided into 208,234 and 71,096 computational cells, respectively, based on susceptibility analyses. Boundary conditions for the analyses, enumerated in Table 2, are the same as those for the experiments. The downstream lengths of the tee and orifice were reflected in the models wherein the analysis results were not influenced by the length. Turbulence was incorporated into the code calculations through the use of the ReNormalized Group (RNG) k-" model on the basis of the instantaneous Navier- Stokes equation. 6) It is known that the separated and recombined flow behaviors, and the jet flow behavior can accurately be predicted by the RNG k-" model 7) compared with the standard k-" models. The standard wall function for predicting the flow behavior of an area adjacent to the wall was used. The numerical analysis was performed for the segre- JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY
Identification of the Relationship between Local Velocity Components and Local Wall Thinning 471 Fig. 4 Pressure-measuring points of tee Fig. 6 Cell divisions of tee Fig. 5 Pressure-measuring points of orifice downstream Fig. 7 Cell divisions of orifice Table 2 Boundary conditions gated condition as the solver and the steady state as the time. The convergence criteria were introduced into the models with 0.001 for continuity, k, ", and x-, y-, and z-velocities according to the normalized and unscaled residuals described in the FLUENT guideline. 3. Review of the Experimental and Numerical Analysis Results (1) Review of Tee The experimental result for the tee was compared with the result of numerical analysis based on the pressure difference between the base point of the inlet and the target points of the outlet. The flow entering the branch was allowed to collide against the side wall opposite the run pipe and then separated into two directions. According to this flow behavior, low pressure was formed at the junction portion between the run pipe and the branch. Figure 8 shows the comparison of the experimental and numerical analysis results for the pressure difference. The data comparison points are shown in Fig. 4. As delineated in Fig. 8, the analysis results agreed well with the experimental results even though the pressure gap gradually increased with the distance from the branch. It can be considered that the low pressure is gradually restored in both the numerical analysis and experiment. Item Unit Value Remarks Operating fluid Water Density kg/m 3 1,000 Viscosity kg/m-s 0.001139 Temperature C 15 Tee m/s 2.75 Magnitude and Inlet velocities Orifice m/s 2.82 direction Magnitude and direction (2) Review of Orifice The experimental result for the orifice was compared with the numerical analysis result based on the pressure difference between the base point of the inlet and the target points of the outlet. While the pressure in front of the orifice was very high, the pressure adjacent to the downstream of the orifice was relatively low and was gradually restored as the distance from the orifice increased. Figure 9 shows the comparison of the experimental and numerical analysis results for the pressure difference. The data comparison points are shown in Fig. 5. As delineated in Fig. 9, even though the pressure difference from the analysis was lower than that from the experiment, the analysis results agreed well with the experimental results qualitatively. VOL. 46, NO. 5, MAY 2009
472 K. M. HWANG et al. Operating conditions of components for numerical anal- Table 3 ysis Fig. 8 Comparison of pressure differences for tee Component Plant Pipeline ph DO Temp. ppb C OR1 HP HTRs bypass 9.58 0.15 137 Orifice OR2 HP HTRs bypass 9.71 0.20 136 OR3 HP HTRs bypass 9.46 0.17 137 Reducer BFP B suction from OR2 9.71 0.20 135 BFP B suction from OR4 9.45 0.22 135 Expander OR3 FWPs discharge 9.46 0.17 136 OR4 FWPs discharge 9.45 0.22 136 Lateral OR1 FWPs discharge 9.61 0.15 136 OR2 FWPs discharge 9.71 0.20 135 45 Elbow OR2 BFP A to FWP A 9.45 0.22 135 OR1 HTR 5B inlet 9.61 0.15 136 90 Elbow OR3 HTR 5B inlet 9.46 0.17 136 OR3 BFP A to FWP A 9.37 0.17 135 OR4 BFP A to FWP A 9.45 0.22 135 OR1 HTR 7s to SGs 9.61 0.15 232 Tee OR2 HTR 7s to SGs 9.70 0.20 232 OR2 BFPs suction 9.71 0.20 135 Notes: OR1, 2, 3, and 4: OPR-1000 Units 1, 2, 3, and 4 DO: Dissolved Oxygen HP: High Pressure HTR: Heater BFP: Booster Feedwater Pump FWP: Feedwater Pump Fig. 9 Comparison of pressure differences for orifice downstream III. Comparison of Numerical Analysis Results and Measured Thickness Data The comparison of the numerical analysis results and measured thickness data for the tee and orifice was only described in this paper, even though numerical analyses were performed for 17 components of 7 types, as shown in Table 3. 1. Analysis of Boundary Layer Characteristics A preliminary analysis was conducted for the selected type, namely, tetrahedron cell, to determine whether it can predict the boundary layer characteristics. The analysis of the boundary layer characteristics was performed for two 45 elbow models, which were the boundary layer model composed of dense cells near the pipe wall and the tetrahedron model. The numbers of cells are 179,349 for the boundary layer model and 314,546 for the tetrahedron model. For the two models, static pressure and radial- and axial-direction local velocity components at the location of 0.05r (radius) from the pipe wall were compared. Figures 10 and 11 show the cell divisions for the boundary layer and tetrahedron models, respectively. Figure 12 shows the calculation points. Figures 13 15 show the comparison of static pres- Fig. 10 Cell divisions of boundary layer model Fig. 11 Cell divisions of tetrahedron model JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY
Identification of the Relationship between Local Velocity Components and Local Wall Thinning 473 Fig. 12 Calculation points Fig. 15 Comparison of V t s between boundary layer and tetrahedron models Fig. 13 Comparison of static pressures between boundary layer and tetrahedron models Fig. 16 Layout of tee Fig. 14 Comparison of V r s between boundary layer and tetrahedron models Fig. 17 Cell divisions of tee sure and radial- and axial-direction local velocity components between the two models. As shown in the figures, it was determined that static pressure and radial- and axial-direction local velocity components are almost the same and, even if the tetrahedron-type cells of analysis models are constructed, the model constructed with a sufficient number of cells can reflect the boundary layer characteristics in the analysis results. 2. Comparison of Results for Tee The actual tee connected to two reducers on both sides is shown in Fig. 16. Figure 17 shows the 3D cell divisions based on the Tet/Hybrid grid subdivided into 429,261 com- VOL. 46, NO. 5, MAY 2009
474 K. M. HWANG et al. Fig. 18 Calculation points of tee Table 4 Analysis conditions for tee Fig. 20 Comparison between V r and WR of tee Unit Value Direction Inner diameter m 0.662 Operating condition Operating pressure MPa 9.63 Operating temperature C 232 Gravity m/s 2 9.81 y Boundary condition Inlet velocity (branch) m/s 5.60 +y Turbulence intensity % 10 Outlet turbulence intensity % 10 Material Density kg/m 3 831.2 Viscosity kg/m-s 1:16 10 4 Fig. 21 Comparison between V t and WR of tee model Fig. 19 Velocity magnitude profile of tee putational cells. Figure 18 shows the measurement and calculation points for data comparison. Actual ultrasonic test (UT) measurement was periodically performed for these points. The interval between the points is 11.18 cm, and the distance from the centerline of the main pipe is 95% of the pipe radius. The analysis conditions for the tee model are enumerated in Table 4, which are the actual operating conditions. Figure 19 shows the velocity magnitude profile of the tee based on the numerical analysis results, wherein two eddies in the junction of the branch and run pipe formed due to low pressure and flow stagnated on the side opposite the branch. Figure 20 depicts the comparison between the wear rate based on the UT measurement and the local velocity component in the radius direction (V r ) on the side opposite the branch. Figures 21 23 show the comparisons between the wear rate and the axial-direction local velocity component (V t ), turbulence intensity (TI), and turbulent kinetic energy (TKE). As shown in the figures, V r is most analogous to the wear rate configuration. V r at the location of 0.05r was calculated using Eq. (1)-expressed scalar and vector products. V r ¼ ~n ~V ¼ðn x ; n y ; n z ÞðV x ; V y ; V z Þ ð1þ ¼ðn x V x Þþðn y V y Þþðn z V z Þ; where ~n and ~V mean the unit and velocity vectors, respectively. Calculation locations are between the buffer zone beyond the viscous sublayer and the turbulent core beyond the buffer zone. The wear rates were calculated based on the data measured in the periods of preservice and first outage. The actual operating time of the tee was 11,256 h. JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY
Identification of the Relationship between Local Velocity Components and Local Wall Thinning 475 Fig. 25 Cell divisions of orifice Fig. 22 Comparison between TI and WR of tee model Fig. 23 Comparison between TKE and WR of tee model Fig. 26 Calculation points of orifice Table 5 Analysis conditions for orifice Unit Value Direction Fig. 24 Layout of orifice 3. Comparison of Results for Orifice The actual orifice connected to an elbow in front of it is shown in Fig. 24. Figure 25 shows the 3D cell divisions including the elbow and two pipes of upstream and downstream based on the Tet/Hybrid grid subdivided into 346,485 computational cells. Figure 26 shows the measurement and calculation points for data comparison. The intervals between the points are 13.3 cm, and the distance from the centerline of the main pipe is 95% of the pipe radius. Inner diameter m 0.432 Operating condition Pressure MPa 9.47 Temperature C 137 Gravity m/s 2 9.81 z Boundary condition Inlet velocity (branch) m/s 7.35 +x Turbulence intensity % 10 Outlet turbulence intensity % 10 Material Density kg/m 3 933.2 Viscosity kg/m-s 2:03 10 4 The analysis conditions for the orifice model are enumerated in Table 5. Figure 27 shows the velocity magnitude profile of the orifice based on the numerical analysis results. While the pressure in front of the orifice was very high and the flow stagnated, the pressure of the adjacent downstream of the orifice was very low, and the flow through the orifice hole curved to VOL. 46, NO. 5, MAY 2009
476 K. M. HWANG et al. Fig. 27 Velocity magnitude profile of orifice Fig. 30 Comparison between V t and WR of orifice model (extrados Fig. 28 Comparison between V r and WR of orifice (extrados Fig. 31 Comparison between V t and WR of orifice model (intrados Fig. 29 Comparison between V r and WR of orifice (intrados the extrados side due to the effect of the upstream elbow. Figures 28 and 29 depict the comparisons between the wear rates based on the UT measurement results and V r at the pipe walls of the extrados and intrados, respectively. Figures 30 and 31 show the comparisons between the wear rate and V t for the extrados and intrados, respectively. Figures 32 and 33 show the comparisons between the wear rate and the TI for the extrados and intrados, respectively. Figures 34 and 35 show the comparisons between the wear rate and the TKE for the extrados and intrados, respectively. The wear rates were calculated based on the data measured in the periods of preservice and third outage. The actual operating time of the orifice was 27,096 h. In Fig. 29, the negative values of V r indicate that the fluid flows to a direction opposite the normal flow direction. As shown in Figs. 28 35, even though there are UT measurement errors, the radial direction velocity components in terms of local wall thinning are more strongly related to the wear rates in general than the other parameters. This means that, even though the turbulence intensity and turbulent kinetic energy may be applied to identifying the susceptible components, such as the elbow, tee, and reducer, to wall thinning, it is hard to detect the small location susceptible to wall thinning inside piping components. JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY
Identification of the Relationship between Local Velocity Components and Local Wall Thinning 477 Fig. 32 Comparison between TI and WR of orifice model (extrados Fig. 34 Comparison between TKE and WR of orifice model (extrados Fig. 33 Comparison between TI and WR of orifice model (intrados Fig. 35 Comparison between TKE and WR of orifice model (intrados IV. Relationship between V r and Local Wall Thinning To review the influences of turbulence parameters based on the numerical analysis results on local wall thinning, numerical analyses were performed for 17 components of 7 types, and the parameters were compared with the wear rates based on the measured thickness data. It is identified that V r has an effect on the local wall thinning, and the location colliding vertically against the pipe wall is changed by the flow velocity and geometrical configuration installed upstream. The obtained results correspond to those of a study for a feedwater heater 8) and a main feedwater isolation valve. 9) To determine the relationship between V r and local wall thinning, the wear rates based on the measured thickness data and V r s based on the numerical analysis results were extracted point by point. Then, the wear rates calculated as negative values and the V r s flowing on the side opposite the pipe wall were eliminated. Additionally, the data of which the error between the V r s and wear rates was outside 68:3% based on the average values were eliminated. When the wear rates by radial-direction velocity components based on the curve fitting method were outside 68:3%, the data were eliminated, considering UT measurement errors. 68.3% represents the 2 in the normal distribution, where means the standard deviation. Even though the radial-direction local velocity components flowed to the side opposite the pipe wall, the thicknesses were measured by the full grid measuring method of the UT test. The thickness data were eliminated in the statistic treatment because of their unsusceptibility to wall thinning. Figure 36 shows a least squares fitting result between V r and local wear rate based on the least-squares fitting method including the upper and lower 95% prediction limits. The average relationship between V r and local wear rate is developed and shown as WR ¼ 0:55V r þ 0:1 ðfor 0 < V r < 1:2Þ; ð2þ where the standard deviation is 0.65, the correlation coefficient r is 0.97, the associated probability p is 0.001, and the degree of freedom is 86. VOL. 46, NO. 5, MAY 2009
478 K. M. HWANG et al. Fig. 36 Least-squares fitting result between V r and wear rate To determine the relationship between turbulent parameters and local wear rate, numerical analyses were performed for 17 components of 7 types included in the main feedwater system, which were similar in terms of reactor type and water chemistry. The turbulent parameters from the numerical analysis results were compared with the local wear rates from the measured thickness data. From the comparison of results, V r flowing again to the wall after separation due to geometrical configuration or direct collision against the wall at an angle of some degrees was found to be analogous to the configuration of local wall thinning. From the least-squares fitting result, it was derived that the average relationship between V r and local wear rate is proportional to 0.55-fold V r adding 0.1 to the standard deviation of 0.65. This result may be used for selecting the locations and scopes of UT measurement, verifying wall thinning causes, and modifying pipeline designs, and contribute to upgrading the techniques for wall thinning prediction and improving the plant safety related to wall thinning. From the equation, it may be conjectured that the local wall thinning progresses more quickly due to the mass transfer under the influence of the magnitude of V r, which is related to the difference in ferrous ion concentration between the surface of the pipe and bulk fluid. Because the piping included in feedwater systems was analyzed in this study, Eq. (2) can be applied to only V r, ph, dissolved oxygen, and temperature spans as follows: 0 < V r < 1:2 m/s ph: 9 10 Dissolved Oxygen: 0 2 ppb Temperature: 130 240 C. V. Conclusions The experiments and numerical analyses for several types of piping components, which are used in plants in general, were performed, and the obtained results were compared. Based on the results indicating that the flow behaviors inside piping components can be sufficiently simulated by numerical analysis, numerical analyses for the models magnified to the actual sizes of plants were carried out. A preliminary analysis was conducted for the selected type of cell to determine whether it can predict the boundary layer characteristics. References 1) NRC, Thinning of Pipe Walls in Nuclear Power Plants, Bulletin 87-01 (1987). 2) NRC, Feedwater Line Break, Supplement 3, Notice 86-106 (1988). 3) Maeda, Amano, Automatic Shut-down of Mihama Unit 3, 4 th Report, NISA/METI Press Release (2004). 4) M. C. Potter, E. P. Scott, Thermal Science: An Introduction to Thermodynamics, Fluid Mechanics, and Heat Transfer, Thomson, Pacific Grove, California, USA (2005). 5) Y. A. Cengel, J. M. Cimbala, Fluid Mechanics: Fundamental and Applications, McGraw-Hill, Seoul, Korea (2006). 6) D. Choudhury, Introduction to the Renormalization Group Method and Turbulence Modeling, Technical Memorandum TM-107, Fluent, Inc. (1993). 7) CHAM, An Introduction into the Method for Implementing Multi-Block Grids and/or Grids with Refinements in PHOENICS, Ver. 2.1 CHAM TR/401 (1994). 8) K. M. Hwang, W. Lee, T. E. Jin, K. H. Kim, A study on the shell wall thinning causes identified through experiment, numerical analysis and ultrasonic test of high pressure feedwater heater, Nucl. Eng. Des., 238[1], 25 32 (2008). 9) K. M. Hwang, T. E. Jin, K. H. Kim, A study on wall thinning causes identified through experiment, numerical analysis and ultrasonic test of main feedwater isolation valve, J. Nucl. Sci. Technol., 45[1], 1 7 (2008). JOURNAL OF NUCLEAR SCIENCE AND TECHNOLOGY