ML051360126
ML051360126 | |
Person / Time | |
---|---|
Site: | Quad Cities ![]() |
Issue date: | 04/30/2005 |
From: | Malone B General Electric Co |
To: | Document Control Desk, Office of Nuclear Reactor Regulation |
References | |
DRF 0000-0038-3018, Class I, NEDO-33191, Rev 1 | |
Download: ML051360126 (40) | |
Text
ENCLOSURE 2 ATTACHMENT 5 "Computational Fluid Dynamics Flow Visualization of Quad Cities Sub-scale Original Dryer Model As a Function of Reynolds Number," NEDO-33191, Revision 1, GE Proprietary, dated April 2005
GE Energy, Nuclear 3901 Castle Hayne Road NEDO-33 191 Wilmington, NAC 28401 Revision I DRF 0000-0038-3018 Class I April 2005 Computational Fluid Dynamics Flow Visualization of Quad Cities Sub-scale Original Dryer Model As a Function of Reynolds Number Prepared by: Bobby G. Malone
NEDO-33191, Revision I Table of Contents page
- 1. Summary ........................ 1-1
- 2. Introduction ......................... 2-1
- 3. Model Geometry ....................... 3-1
- 4. Numerical Modeling ....................... 4-1
- 5. Boundary Conditions ....................... 5-1
- 6. Fluid Properties ........................ 6-1
- 7. Runs Performed ........................ 7-1
- 8. Results of Cases Performed ....................... 8-1 8.1 100% Reynolds Scale, Base Case .8-2 8.2 Two Times the Sub-scale Model Reynolds Number .8-3 8.3 Five Times the Sub-scale Model Reynolds Number. 8-4 8.4 Ten Times the Sub-scale Model Reynolds Number .8-4 8.5 Fifty Times the Sub-scale Model Reynolds Number .8-4 8.6 One-Hundred Times the Sub-scale Model Reynolds Number .8-5 8.7 Five-Hundred Times the Sub-scale Model Reynolds Number .8-5 8.8 Full Size Model with Steam and at Pressure .8-5 ii
NEDO-33191, Revision I Table of Figures page Figure 1, Geometry of the CFD Dryer Model ................................................... 3-2 Figure 2, Nodal Configuration ................................................... 4-1 Figure 3, Close Up of Nozzle Mesh ................................................... 4-2 Figure 4, Nodal Structure on Hood ................................................... 4-4 Figure 5, Boundary Locations ................................................... 5-1 Figure 6, 100% Scale, y+ on the Front Hood, Seen From Rear ................................... 7-3 Figure 7, 50000% Scale, y+ on the Front Hood, Seen From Rear ............................... 7-4 Figure 8, Sub-Scale Velocity Streamlines, from Front, Re Ratio = 1............................ 8-7 Figure 9, Sub-Scale Velocity Streamlines, from Front, Re Ratio = 2............................ 8-7 Figure 10, Sub-Scale Velocity Streamlines, from Front, Re Ratio = 5.......................... 8-8 Figure 11, Sub-Scale Velocity Streamlines, from Front, Re Ratiio = 10....................... 8-8 Figure 12, Sub-Scale Velocity Streamlines, from Front, Re Ratio = 50 ........................ 8-9 Figure 13, Sub-Scale Velocity Streamlines, from Front, Re Ratio = 100...................... 8-9 Figure 14, Sub-Scale Velocity Streamlines, from Front, Re Ratio = 500 .................... 8-10 Figure 15, Full Size, from Front................................................... 8-10 Figure 16, Sub-Scale Velocity Streamlines, from Side, Re Ratio = 1 ......................... 8-11 Figure 17, Sub-Scale Velocity Streamlines, from Side, Re Ratio = 2......................... 8-11 Figure 18, Sub-Scale Velocity Streamlines, from Side, Re Ratio = 5......................... 8-12 Figure 19, Sub-Scale Velocity Streamlines, from Side, Re Ratio = 10 ....................... 8-12 Figure 20, Sub-Scale Velocity Streamlines, from Side, Re Ratio = 50 ....................... 8-13 Figure 21, Sub-Scale Velocity Streamlines, from Side, Re Ratio = 100 ..................... 8-13 Figure 22, Sub-Scale Velocity Streamlines, from Side, Re Ratio = 500 ..................... 8-14 Figure 23, Full Size Model, from Side ................................................... 8-14 Figure 24, Sub-Scale Velocity Streamlines, from Back, Re Ratio = 1........................ 8-15 Figure 25, Sub-Scale Velocity Streamlines, from Side, Re Ratio =2......................... 8-15 Figure 26, Sub-Scale Velocity Streamlines, from Back, Re Ratio = 5........................ 8-16 Figure 27, Sub-Scale Velocity Streamlines, from Side, Re Ratio = 10 ....................... 8-16 Figure 28, Sub-Scale Velocity Streamlines, from Back, Re Ratio = 50 ...................... 8-17 Figure 29, Sub-Scale Velocity Streamlines, from Side, Re Ratio 100 ..................... 8-17 Figure 30, Sub-Scale Velocity Streamlines, from Back, Re Ratio = 500 .................... 8-18 iii
NEDO-33 191, Revision I Figure 31, Full Scale Velocity Streamlines, from Side ...................................... 8-18 List of Tables page Table 1, Viscosity Used and Resulting y+ 7-1 iv
NEDO-33191, Revision I IMPORTANT NOTICE REGARDING CONTENTS OF THIS REPORT Please Read Carefully GE proprietary information has been removed from this report as indicated by blank space in the document between double square brackets. (( ))
The only undertakings of GE respecting information in this document are contained in the Contract between Exelon and GE, and nothing contained in this document shall be construed as changing the contract. The use of this information by anyone other than Exelon, or for any purpose other than that for which it is intended, is not authorized; and, with respect to any unauthorized use, GE makes no representation or warranty, express or implied, and assumes no liability as to the completeness, accuracy, or usefulness of the information contained in this document, or that its use may not infringe privately owned rights.
Note: As part of the scale model test development the assumption was made that the scale model would duplicate the flow patterns in the plant; therefore, preservation of the Reynolds number in the scale model was not considered necessary. This report documents the verification of that assumption V
NEDO-33 191, Revision I
- 1. Summary This report describes the results of a computational fluid dynamics (CFD) analysis of a key assumption used in the development of the dryer sub-scale acoustic model. The assumption is that the flow in the full sized plant dryer and upper vessel is dynamically similar to a 1 /17 th scale model operating with air at near atmospheric conditions.
A commercial CFD code referred to as CFX version 5.7.1 has been used to perform this analysis. It is not a level 2 code as defined by GE procedures. Collectively, it is a suite of graphically driven computer codes that allow the creation of geometric models, nodalization of the models, solution of the numerical CFD problem, and graphical post processing of the results.
A CFD model was built from CAD drawings of the original Quad Cities dryer scale model. The effective Reynolds number of the model was varied by reducing the dynamic viscosity of the working fluid (air).
The flow patterns produced by the code were very consistent over the range of Reynolds numbers examined. Subject to the qualitative nature of the method of assessment, these results demonstrate that the flow does exhibit dynamically similar behavior over a wide range of Reynolds numbers. It follows that the principles of dynamic similitude analysis can be appropriately applied to the sub-scale model and extrapolations made to the full-scale prototype.
A CFX model of the prototype was also constructed using steam at pressure. Due to computer limitations, the fidelity of this model is less than that obtained for the scale model. However it also demonstrated similar flow characteristics to that of the scale model.
1-1
NEDO-33191, Revision I
- 2. Introduction In July 2002, the Quad Cities Unit 2 (QC2) was shut down as a result of steam dryer structural failure. GE is now in a research effort to understand the nature of mechanisms causing the failure. Part of this effort has been the development of a sub-scale model of the dryer and surrounding NSSS.
The QC2 sub-scale dryer has been built to 1/17.3th scale and uses air at near atmospheric conditions as the operating fluid. Mach number is matched between the scale model and prototype but Reynolds number is low compared to the prototype by approximately (a factor of) 700. Through dynamic similitude analysis, it has been concluded that there is little change in scaling factors due to Reynolds number [1].
Implying therefore that results from the sub-scale model tests can be scaled through derived scaling laws to infer conditions in the prototype.
An implicit assumption in dynamic similitude analysis is that there is dynamic similarity between the flow fields of the full-scale apparatus (the plant) and the modeled apparatus (the sub-scale model). A necessary consequence of this assumption is that the flow field follows similar paths, shows similar features and characteristics, and operates as a turbulent or laminar flow as conditions dictate.
A computational fluid dynamics (CFD) analysis has been performed on the scale model of the original dryer. The aim of the analysis was to investigate the effect of changing Reynolds number on the flow patterns observed in the region outside the dryer and into the steam lines. By performing an assessment of the "similarity" between the various cases, the assumption of dynamic similitude can be examined.
Reynolds number is a dimensionless quantity given by the following equation.
Rn= p.U-L
'I 2-1
NEDO-33 191, Revision I To vary the Reynolds Number over a wide range, the most practical parameter to change is the viscosity. Increasing the velocity would void the matching of the Mach number and introduce compressible flow conditions, which would violate the assumption of similar flows (shock waves). Changing density would consequently change the sonic velocity of the fluid, again voiding the Mach number similarity. Changing the size is possible but requires the time consuming process of developing a unique model for each case.
A full sized model with steam at pressure was developed by enlarging the geometry of the sub-scale model through features of the CFX code. Due to computer limitations this full sized model does not provide equivalent fidelity to that of the sub-scale models.
The assessment of these results by necessity is of a qualitative nature. There is no practical way to quantitatively assess the similarity of the flow streamlines produced by the code. The approach is to visually compare the flow patterns produced by the CFD code and make a judgment on their similarity.
Due to computer limitations, an acoustic analysis of these models is not possible.
Ideally, power spectral density (PSD) plots could be made of the numerical results using fast Fourier transforms (FFT). These would then compared to the PSD results from the scale model giving greater confidence in the fidelity of the CFD model. However, the time step size required to accurately track the sonic wave propagation is significantly less that that used in this study. In addition, thousands of time steps are required to produce satisfactory PSD plots rather than the hundred or so performed for these cases.
2-2
NEDO-33 191, Revision I
- 3. Model Geometry The model geometry used to assess the effect of Reynolds number is based on a simplification of CAD drawings used to create the scale model of the old QC2 dryer.
Difficulties were found in the use of the CAD drawing files as direct input to the CFX program. Therefore, the CFD model was essentially rebuilt with CFX drafting software using the CAD drawings as a guide. This bypassed problem areas in the CAD model that produced difficulties in the nodalization of the geometry. Difficulties with the importation of "dirty" CAD drawings are unfortunately a common problem in CFD development.
Basic geometry was retained and results are believed to be unaffected by these simplifications. Figure 1, shows a portion of the geometry of the model used in this analysis. The full extent of the model is shown in Figure 5. The steam line continues past the SRVs on the line shown.
Quarter symmetry has been assumed in order to make this a tractable problem with the available computer resources. It is a common practice to utilize symmetry in CFD analysis. However, this can introduce some limitations to be aware of. A potential consequence is that flow features involving a scale of similar or larger size to that of the quarter of the model assumed here will not be developed accurately by the solution.
For example, a large irregular vortex that might at times occupy both sides of the physical model would be forced by the assumed boundary conditions to configure itself to meet the symmetric boundary condition. A second consequence is that the flow of fluid will not cross the line of symmetry. For example, this could occur during a swing or oscillation of flow from one side of the dryer to another. Such behavior, if it exists, will therefore not be seen in these results.
In order to reduce the size of domain, the interior of the dryer was not modeled.
Instead, the flow was allowed to enter the upper head directly through the inlet slots of the dryer as a boundary condition. This is believed to cause little reduction in the fidelity of the CFD model for this application.
3-1
NEDO-33 191, Revision I NO I
Figure 1, Geometry of the CFD Dryer Model In addition, the entire length of the scale model steam line was not modeled. The line length was truncated just downstream of the SRVs. The fluid was then modeled as exhausting to the ambient air through a loss coefficient of one. Again, little reduction in the fidelity of the CFD model will be experienced for the cases under consideration.
3 -2 Co[
I-
NEDO-33191, Revision I
- 4. Numerical Modeling Numerical modeling involves the nodalization of the geometric model, the selection of appropriate time step, turbulence model, and working fluid characteristics.
((
Figure 2, Nodal Configuration 4-1
NEDO-33191, Revision I Selection of the node size and the proper distribution of different sizes is the key to successful generation of high fidelity results. The opposing constraints are the speed, memory and run time of the available computer. The overall exterior surface node size and distribution arrived at are illustrated in Figure 2. In general, node sizes of 0.05 to Figure 3, Close Up of Nozzle Mesh 0.1 inches were specified for areas closer to regions of interest (e.g., front hood and nozzle entrance to the steam line). Larger nodes were specified for regions away from the regions of interest. The very fine nodal structures seen in Figure 2 were generated by the code to meet internal proximity requirements. The CFX code recognizes the 4-2
NEDO-33191, Revision I proximity of surfaces to one another, such as the surfaces of the dryer to the vessel wall. It reduces the size of the nodes in these regions to satisfy a required minimum number of nodes between surfaces.
On surfaces near the front hood and nozzle entrance, the mesh was further refined to allow resolution of the boundary layer. Ideal resolution was obtained for sub-scale Reynolds numbers ratios of less than 100 as discussed in the section on y+. This was not true of the full sized plant model again due to computer limitations.
Figure 3 shows a close up of the region near the entrance to the steam line. Such resolution is adequate to resolve flow patterns expected in this region.
Figure 4 shows the node structure on and near the front hood as seen from the interior of the dryer. The node size on the hood and adjacent areas has been made small to capture the turbulence that is expected to be generated there. Again, finer nodes are present, particularly near the dryer inlets due to close proximity of other structures.
While CFX has the ability to solve for a steady state solution, this capability was only used to initialize the problem. Afterward, the simulation was run in a fully transient mode. CFX utilizes an implicit solver allowing a relatively large time step. The time step is generally selected by picking a suitably small fraction of the flow transport time.
For example, the transport time across the region of interest (the front hood and nozzle entrance) is conservatively about 0.005 seconds. Taking one tenth of this gives a value of 0.0005 seconds. Transient information was retained at alternate steps, or at 0.001 second intervals.
In order to accurately track the sonic wave, the time step must be such that the distance traveled by the sonic wave is smaller that the size of the nodes. This would require a time step approximately fifty times smaller that that actually used. However the flow structures of interest are not significantly influenced by sonic wave transmission time, so the value of 0.0005 seconds is quite adequate.
4 -3
NEDO-33191, Revision I
((
11 Figure 4, Nodal Structure on Hood A high fidelity turbulence model was selected for this problem. The shear stress transport (SST) model was selected from the menu of possible models offered by the CFX 5.7.1 code. It provides highly accurate predictions of the onset and the amount of flow separation under adverse pressure gradients. It is considered one of the better turbulence models offered in CFX.
4 -4
NEDO-33 191, Revision I CFX's "high resolution" scheme for the advection was utilized. It is a near second order method. The method is accurate and bounded, preventing unrealistic over and undershoots of the solution.
A second order, implicit, backward Euler scheme was used to advance in time.
The working fluid used in the simulation is air treated as an ideal gas. Since the energy equation was not solved in the simulation, the flow is considered to be thermodynamically incompressible. This is a reasonable assumption due to the relatively low Mach numbers (<0.2) occurring in the scale model. However, density is allowed to vary with local pressure. The properties inherent to CFX for air were utilized.
The transient was run with a relative error (RMS) control of 10-5 . This value is well within the range recommended for the CFX code.
The initial cases were run using single precision. During the case for 50 times the Reynolds number, it was noted that the convergence became unsteady. Good convergence returned when double precision was activated. For this case and all higher values, double precision was used.
For a final case, a full-scale model was developed. Desirable nodal sizes were not achievable with the full-scale model due to computer limitations. The computer resources available allowed only approximately 1.5 million nodes. Results were obtained, but they are considered to not be of the same fidelity as the previous sub-scales cases. It is estimated that approximately double this number of nodes, well distributed would be required for fidelity similar to that of the sub-scale results.
4-5
NEDO-33 191, Revision I
- 5. Boundary Conditions Various types of boundary conditions are available for use with the CFX code.
However, boundary conditions in CFX are generally specified as flow (velocity) at the inlet and pressure (static) at the outlet.
CFXP y
21j Figure 5, Boundary Locations The small green area at the end of the steam line segment shown in Figure 5 is the outlet to the ambient air conditions of 14.7 psia through a loss coefficient of one. This is as far into the sub-scale model steam line as the simulation extended. The dryer inlet (in red) was set up to supply air at a velocity of 15.4 ft/sec. The velocity was determined given a specified volumetric flow rate of 167 cfm, and the area of the inlet.
Symmetric boundary conditions were assumed by dividing the full geometry half way between pairs of steam lines.
5-1 C O
NEDO-33 191, Revision I
- 6. Fluid Properties The properties internal to CFX for dry air as an ideal gas were used in this analysis.
The dynamic viscosity was 1.831 E-05 kg/m/s for the sub-scale model and then reduced proportionally for the higher values of Reynolds number. The molar mass was 28.96 kg/kmole. The ambient pressure at the steam line exit was specified to be 14.7 psia.
The dryer inlet temperature was specified to be 100 deg. F, which was typical of the test conditions within the sub-scale model. The conservation of energy equation was not solved for these cases, so the temperature within the simulation is held constant.
Density of the air however, is permitted to vary with local pressure.
6-1
NEDO-33 191, Revision I
- 7. Runs Performed The purpose of this analysis was to determine the sensitivity of the flow patterns produced by the sub-scale geometry to variations in Reynolds number. Of the factors determining Reynolds number, the viscosity is the most practical to change in this situation. Since the variation in Reynolds number ranged from a factor of 1 to a factor of 500, a change in velocity of this magnitude would introduce compressible sonic flow which is unphysical and would obviously not be dynamically similar. Changing density leads to changes in the speed of sound, which is also undesirable since it would change the Mach number. A change in the physical dimensions is possible, but time consuming requiring that a differently nodalized model be built for each case. Therefore, each of the runs was scaled by changing the default dynamic viscosity in CFX in inverse proportion to the desired change in Reynolds number.
Viscosity Used and Resulting y Scaled Dynamics Viscosity Approximate Reynolds Kg/(m-s) Average y Number On Hood 1 1.831E-05 2 2 9.155E-06 3 5 3.662E-06 6 10 1.831E-06 10 50 3.662E-07 50 100 1.831E-07 80 500 3.662E-08 400 700 1.92E-05 NA (Plant)
Table 1, Viscosity Used in Simulation and Resulting y+
7-1
NEDO-33191, Revision I The above table (Table 1) shows the values of viscosity used and the resulting average value of y+ obtained. The mesh was not refined between cases but held constant.
Satisfactory results can be expected for the SST turbulence model when y+ remains below 100. It should be noted that this degree of mesh refinement was only used on surfaces near the regions of interest: the front hood, and MSL nozzle.
Starting with the scaled Reynolds number of 50, it was observed that convergence became unsteady. A change to a double precision solver returned the solution to a reasonable convergence (10-5 RMS).
The above table also shows an approximate average value of y on the front hood of the scale dryer. y+ is a dimensionless number that is used to assess the distance of the first node away from the wall. It is necessary for y+ to be in a prescribed range to achieve the maximum fidelity from the selected turbulence model. As seen in Figure 6 and 7, the value of y+ varies over the surface, so an approximate average is reported.
These values are not typical of the rest of the model. To keep the solution size and run time tractable, only the hood region was covered (inflated) with very thin nodes that allow values of y4 this small. The physical height of the first node off the surface to achieve this value was 0.0005 inch. Accurate solutions are generally assured with values of y+ less than 100. Average values less than 100 were maintained for all cases of scale factor 100 or less. For cases with scale factors larger than 100, y+ grew such that at a scale factor of 500 an approximate average of y+ was near 400. A similar plot of y' for the 500 times sub-scale Reynolds number case is shown in Figure 7. The effect that this increasing value has on the fidelity of the results is a function of the flow pattern sensitivity to the boundary layer development in this region. The results obtained are believed to be sufficiently resolved for the purposes of this qualitative analysis. The flow patterns going from a Reynolds scale of 100 to 500 did not change to any significant degree indicating a small sensitivity to the change in y+.
The value of y+ is a function of viscosity. Therefore, it is to be expected that as Reynolds number increases, the value of y+ also increases.
7-2
NEDO-33191, Revision I Figure 6, 100% Scale, y+on the Front Hood, Seen From Rear 7 -3 Co2~
NEDO-33 191, Revision I 11 -
Figure 7, 50000% Scale, y on the Front Hood, Seen From Rear 7-4 CO .
NEDO-33 191, Revision I
- 8. Results of Cases Performed Seven cases were run varying the Reynolds number from 1 to 500. These cases were all run as transients rather than simply forcing a steady state solution. Most complicated flows are actually transients having no true steady state solution. The cases were run as a transient long enough for the convergence to reach a specified value and to show a regular pattern indicating a quasi-steady state had been obtained.
There are three key flow features observed in these cases that will be used to assess the similarity of the various Reynolds numbers.
The first is seen in the frontal view of the flow field as it enters the region between the front hood and the vessel and proceeds into the steam line. ((
1]
8-1
NEDO-33 191, Revision I 8.1 100% Reynolds Scale, Base Case The first group of figures is from the CFD run made for the sub-scale model as tested.
Figure 8 shows the flow paths as would be seen from outside the vessel looking in the plane of the steam line. The streamline flow paths are created by integrating along a velocity vector starting from the dryer outlet into the vessel and steam lines. Only a small fraction of the possible paths are shown.
((I 1))
Figure 16 is from the same run but from the side looking into the vortex adjacent to the wall. ((
Figure 24, also is from the same run but here the view is orientated to look from the rear (inside the dryer) outward directly into the axis of the steam line. ((
8-2
NEDO-33 191, Revision I 1]
Work is underway to simplify the problem to two dimensions in an attempt to resolve these features. ((
)) If they are found to be similar, the basic assumption inherent in the use of dynamic similitude analysis is upheld.
8.2 Two Times the Sub-scale Model Reynolds Number The viscosity of the air is reduced by a factor of two and the case rerun. Figures 9, 17, and 25 show the same set of streamlines from the same directions as previously discussed. A side-by-side visual comparison of Figure 9 with Figure 8 shows no significance difference.
Figures 17 and 16 are both from the side showing the development of a vortex adjacent to the front hood. ((
Figures 25 and 24 show the flow paths from the rear. ((
Given that these are transients, some variance is to be expected. Later results show that the cases do not continue to deviate. ((
8-3
NEDO-33191, Revision I
)) However, these cases are still very similar so no violation of the dynamic similitude assumptions would be made in doubling the Reynolds number. This assessment is of course qualitative since no quantitative assessment can be made.
8.3 Five Times the Sub-scale Model Reynolds Number Figures 10, 18, and 26 show the same velocity streamlines made with a case where the viscosity has been reduced by a factor of five, thereby increasing the Reynolds number by five. ((
)) Otherwise the figures are quite similar. It is therefore concluded that the flow patterns are quite similar up through a Reynolds number increase of a factor five.
8.4 Ten Times the Sub-scale Model Reynolds Number Figures 11, 19, and 27 show the same velocity streamlines made with a case where the viscosity has been reduced by a factor of ten, thereby increasing the effective Reynolds number by ten over the sub-scale model. These figures show very little change from the previous set. The figures are very similar. It is therefore concluded that the flow patterns are quite similar up through a Reynolds number increase of a factor ten.
8.5 Fifty Times the Sub-scale Model Reynolds Number Figures 12, 20, and 28 show the same velocity streamlines made with a case where the viscosity has been reduced by a factor of fifty, thereby increasing the effective Reynolds number by fifty over the sub-scale model.
During the initialization of this case, it was observed that the requested convergence (10-5) was no longer being maintained. Upon switching to a double precision solver, the required convergence was maintained for this and all future cases.
8-4
NEDO-33191, Revision I These figures show almost no change from the previous two sets. It is therefore concluded that the flow patterns continue to maintain similar flow patterns up through a Reynolds number increase of a factor fifty.
8.6 One-Hundred Times the Sub-scale Model Reynolds Number Figures 13, 21, and 29 show the same velocity streamlines made with a case where the viscosity has been reduced by a factor of one-hundred, thereby increasing the effective Reynolds number by one-hundred over the sub-scale model The streamlines for all views are very similar to previous cases. Therefore flow similarity is being maintained.
8.7 Five-Hundred Times the Sub-scale Model Reynolds Number Figures 14, 22, and 30 show the same velocity streamlines made with a case where the viscosity has been reduced by a factor of five-hundred, thereby increasing the effective Reynolds number by five hundred.
The flow patterns and features continue to show similarity through out the entire range of Reynolds numbers evaluated.
As show in the previous table and in Figures 6 and 7, the value of y+ continues to grow with the decreasing viscosity. With this case, it has reached an approximate average value of 400. This is outside the ideal range of the SST turbulence mode. But as can be seen, there is still no significant change in the flow pattern obtained from the model. It is therefore concluded for this qualitative assessment, there is good similarity shown by the CFD modeling of the sub-scale geometry. It can be inferred that principles of dynamic similitude can be successfully applied to this situation.
8.8 Full Size Model with Steam and at Pressure Figures 15, 23, and 31 show the same velocity streamlines made with a case where the full sized plant has been modeled. Because of computer limitations, there are 8-5
NEDO-33 191, Revision I significant shortcomings in the modeling of this case. The scale model was constructed with approximately 600,000 nodes. The full sized model (scaled by a factor of 17.3) has approximately 1.5 million nodes, which is at the limits of the available computer resources. Because of the relatively larger nodalization, the model did not converge well, exhibiting undesirable characteristics. A transient was not made as for the other cases, but an approximate steady state solution was obtained. These results are shown with the understanding that due caution be made in the use of them.
Figure 15 shows a frontal view. The overall flow patterns show similar characteristics to the subscale. ((
)) The velocity scale has been expanded by the ratio of the Mach numbers to allow a more accurate comparison with the color images.
Figure 23 shows a weaker vortex from the side, than observed in the scale model. For this case the, the boundary layer on the front hood was not resolved by fine inflation layers as done for the scale model. ((
))
Figure 31 shows the full size model results as seen from the rear. [f 1))
It can tentatively be concluded that given proper nodalization, the full sized model would show a greater similarity to the scale results.
8-6
NEDO-33191, Revision I Figure 8, Sub-Scale Velocity Streamlines, from Front, Figure 9, Sub-Scale Velocity Streamlines, from Front, Re Re Ratio = Ratio =2 11 11 8-7
NEDO-33191, Revision I Figure 10, Sub-Scale Velocity Streamlines, from Front, Figure 11, Sub-Scale Velocity Streamlines, from Front, Re Ratio = 5 Re Ratiio = 10
[I 8-8
NEDO-33191, Revision I Figure 12, Sub-Scale Velocity Streamlines, from Front, Figure 13, Sub-Scale Velocity Streamlines, from Front, Re Ratio = 50 Re Ratio = 100 1] 1]
I[ I 8-9
NEDO-33191, Revision I Figure 14, Sub-Scale Velocity Streamlines, from Front, Figure 15, Full Size, from Front Re Ratio =500 I II 8-10
NEDO-33 191, Revision I 8-11
NEDO-33e191, Revision I Figure 18, Sub-Scale Velocity Streamlines, from Side, Figure 19, Sub-Scale Velocity Streamlines, from Side, Re Ratio = 5 Re Ratio = 10 1] 1]
8-12
NEDO-33 191, Revision I Figure 20, Sub-Scale Velocity Streamlines, from Side, Figure 21, Sub-Scale Velocity Streamlines, from Side, Re Ratio = 50 Re Ratio = 100 11 1]
8-13
NEDO-33 191, Revision I Figure 22, Sub-Scale Velocity Streamlines, from Side, Figure 23, Full Size Model, from Side Re Ratio = 500 ((
1,. ..
8-14
NEDO-33 191, Revision I Figure 24, Sub-Scale Velocity Streamlines, from Back, Figure 25, Sub-Scale Velocity Streamlines, from Side, Re Ratio = 1 Re Ratio = 2 l 1]3 8-15
NEDO-33 191, Revision I Figure 26, Sub-Scale Velocity Streamlines, from Back, Re Figure 27, Sub-Scale Velocity Streamlines, from Side, Re Ratio Ratio = 5 = 10
] 1))
8-16
NEDO-33191, Revision I Figure 28, Sub-Scale Velocity Streamlines, from Back, Figure 29, Sub-Scale Velocity Streamlines, from Side, Re Ratio = 50 Re Ratio = 100 1][
8-17
NEDO-33191, Revision I Figure 30, Sub-Scale Velocity Streamlines, from Back, Figure 31, Full Scale Velocity Streamlines, from Side Re Ratio = 500 ((
((
1] 1]
8-18