Anda di halaman 1dari 8

Proceedings of the 17th International Conference on Nuclear Engineering ICONE17 July 12-16, 2009, Brussels, Belgium

ICONE17-75474
NONLINEAR ANALYSES FOR THERMAL CRACKING IN THE DESIGN OF CONCRETE STRUCTURES
Randy J. James ANATECH Corp. San Diego, California, USA Ai-Shen Liu GE Hitachi Nuclear Energy San Jose, California, USA

ABSTRACT A Design by Analysis procedure has been developed to incorporate the results of thermal cracking analyses into the linear based design methods for reinforced concrete containment structures. Current practice employs linear based analyses and accounts for stress reduction due to thermal induced cracking on a section by section basis. Under thermal loading, in addition to a reduction of the section forces and moments for cracked sections, concrete cracking also reduces the structural stiffness and thus the constraint against thermal expansion or contraction for the whole structure, which in turn reduces the thermal induced stresses as compared to a linear (un-cracked) analysis. The design by analysis approach employs nonlinear thermal cracking analyses using detailed modeling of the complete structure. By comparing these stress distributions with those from a linear analysis of the same model, stress reduction factors can be developed for critical sections that also include global stress redistribution. These stress reduction factors can then be used to correct linear based thermal stresses in the design basis calculation for all load combinations that include thermal loads. INTRODUCTION A critical structural component in the design of new generation power plants using nuclear energy is the primary containment system, which must resist elevated internal pressures and temperatures in case of Loss of Coolant Accidents (LOCA). These structures are designed to resist deformations due to internal pressure, temperature, and other loads including impact and seismic loading. Because such structures with thick concrete sections and heavy reinforcement are very stiff, large internal stresses due to thermal differentials can also develop. Examples of testing and methods for accounting for this thermal induced cracking in the design of reinforced concrete containment structures is provided in References 1 and 2.

The design of the reinforced concrete containment structure is based on the membrane forces, shear forces, and bending moments that develop in the structural sections for the various load combinations that must be considered. A standard design practice is to obtain these section forces and moments using a global, linear finite element analysis with the proposed section sizes but ignoring any reinforcement and concrete cracking. The effect of thermal-induced concrete cracking is subsequently addressed for each member of the structure at the section level to determine the reduced thermal force and moment of the cracked section. The section sizing is then verified and the reinforcement is designed by assuming that the concrete cannot carry any tensile stress in the section. The resulting compressive stress in the concrete and the tensile (or compressive) stress in the reinforcement needed to carry the calculated section forces must then meet the allowable values defined in the applicable design codes. Under thermal loading, in addition to a reduction of the section forces and moments for cracked sections, concrete cracking also reduces the structural stiffness and thus the constraint against thermal expansion or contraction for the whole structure, which in turn reduces the thermal induced stresses as compared to a linear (un-cracked) analysis. This paper presents a Design by Analysis methodology that has been developed and used in the design of the primary containment system for the ESBWR. Detailed, nonlinear analyses are performed for thermal loading to establish the stress distributions allowing for concrete cracking and also concrete property degradation with elevated temperature. This captures the structure specific variation in stress reduction and stress redistribution due to thermal induced cracking. Because of the large number of load combinations that must be considered in the design, the use of such 3D nonlinear analyses is prohibitive for all load combinations. Thus, a procedure was developed to incorporate the results of the thermal cracking analyses into the linear based design methods.

Copyright 2009 by ASME

DESIGN BY ANALYSIS PROCEDURE The analysis procedure consists of the following steps for each LOCA condition considered; 1. A steady state thermal analysis is performed to initialize the temperature distribution for normal operating conditions for ambient conditions. Cases for both winter and summer extreme conditions are considered. Transient thermal analyses are performed to establish the temperature distributions during the 72 hours after the LOCA event based on the specified temperature transients for the different locations within the containment and with the external ambient conditions. A linear stress analysis is performed, first for the steady state operating temperature distribution, and then using the temperature distributions at 5 seconds, 6 minutes, 10 hours, and 72 hours after the LOCA. This stress analysis does not allow the concrete to crack but considers the effects of degrading modulus with increasing temperature consistent with the analysis procedure used in the design basis. Using the same finite element model, a nonlinear stress analysis is performed at the same time points for the same temperature distributions, now considering both concrete cracking and modulus degradation with temperature. For these analyses, several load increments are performed to transition between the different temperature distributions in time. For each load increment, iterations for the deformations and stresses are needed to maintain equilibrium as concrete cracking develops. At predefined, critical section cuts throughout the structure, section forces and moments are calculated from the continuum element based stress distributions on the sections. These section forces and moments are calculated for both the linear and the cracking analyses. At pre-defined points on the internal steel components attached to and affected by the response of the Reinforced Concrete Containment Vessel (RCCV), the Von Mises or Effective stress is calculated from the stress components. Again, this is done for both the linear analysis and the cracking analysis. The thermal knockdown ratios are calculated by taking the results from the cracking analysis and dividing by the results from the linear analysis. These ratios are calculated for each section force component (or stress point) for each section cut for each of the 4 time points considered for each of the 2 ambient conditions.

concrete sections are calculated by a post-processing procedure. The procedure for calculating the section forces and moments from the stress based finite element models is as follows; 1. The element stresses, including the stresses in the embedded rebar, are integrated over the element volumes to obtain nodal point forces. This is a standard procedure in finite element analysis needed at each loading increment or iteration to develop the internal forces for achieving static equilibrium with the externally applied loads. This step is performed by the host FE program, and the nodal point forces are stored for the elements identified for the various section cuts of interest. For each section cut, the geometry, nodal forces, and cracking information are extracted from the postprocessing file for all the elements lying along one side of the section cut. The nodal forces are summed at common nodes from the different element surfaces making up the section cut. This results in a plane of nodes with 3 components of force for each node in the global coordinate system. A local coordinate system is constructed for the surface of the cut from the coordinate information of the nodes, and the global force components are transformed into the local coordinate system. The section force components are then calculated by summing up the appropriate contributions from each node. For example, the membrane force is calculated by summing all the nodal force components that are normal to the cut surface. The bending moments are calculated by summing the appropriate force component times its distance from the neutral axis. For the linear analyses, the neutral axis is the centroid of the section. For the cracking analysis, the neutral axis is determined from the cracking depth across the section. For the cracking analysis, the resulting section forces are transformed back to the centroid of the section so that the ratios of the cracking results to the linear results have a common basis.

2.

3.

2.

3.

4.

4.

5.

5.

6.

6.

7.

7.

The continuum-based finite element models provide stress distributions for the elements across the wall or slab sections. The design procedure is configured to use section forces and moments as the basis for section sizing and rebar design. The section forces and moments for a cut across the reinforced

REINFORCED CONCRETE MODELING The modeling of the concrete for these temperature dependent cracking analyses is provided through the ANACAP-U constitutive model [3]. The behavior of concrete is highly nonlinear having small tensile strength, shear stiffness and strength that depend on crack widths, and possibly compressive plasticity. The main components of the concrete model for these analyses are tensile cracking, post-cracking shear performance, compressive yielding or crushing if the compressive strength is reached, and property degradation with temperature.

Copyright 2009 by ASME

Tensile cracking in the concrete is governed by the magnitude of the load in the directions of principal strain. Cracks are assumed to form perpendicular to the directions of largest tensile strains. Multiple cracks are allowed to form at each material point, but they are constrained to be mutually orthogonal. If cracking occurs, the normal stress across the crack is reduced to zero and the distribution of stresses around the crack is recalculated through equilibrium iterations. This allows stress redistribution and load transfer to reinforcement or other load paths in the structure. Once a crack forms, the direction of the crack remains fixed and can never heal. However, a crack can close, resist compression and shear, and re-open under load reversals. The cracking criterion is based on an interaction of both stress and strain as illustrated in Figure 1. The model predicts cracking when the generalized stress and strain state exceeds the limit state shown. Thus, biaxial and triaxial tension stress states are treated consistently with uniaxial conditions, but the associated cracking will now occur at a slightly higher stress and slightly lower strain. This model can also capture split cracking, for example cracking near a free surface under high compressive stress, where the cracks form under near zero stress and a tensile strain approximately twice that of uniaxial tensile cracking.

is retained in the stress-strain matrix when a crack forms. The shear modulus for each direction in the plane of the crack is immediately reduced by 60% when a crack forms. The shear stiffness is further reduced using a hyperbolic variation with the opening strain normal to the crack. Perhaps the most important feature of concrete modeling is the ability to capture the shear capacity in cracked concrete. The concrete model is equipped with a shear-shedding feature to track the shear stress capacity across an open crack. The shear retention model reduces the incremental shear modulus across on open crack as discussed above. The shear stress capacity for an open crack is also a function of the crack opening strain, as illustrated in Figure 2. The shear-shedding feature reduces the shear stresses previously supported across an open crack if the crack continues to open. Again, equilibrium iterations are needed to redistribute the loads. Cracks form in the principal strain directions so that, in general, there is no shear across a crack when it first opens. However, continued loading resulting in shear deformations will be carried in shear across the crack if possible. This loss of shear capacity as cracks open is a major contributor to structural failures in reinforced concrete structures.

Figure 2. Example of Shear Stress on Open Cracks In the compression regime, the continuous stress-strain curve is defined from uniaxial test data, which is then generalized to multi-axial stress/strain states using the uniaxial equivalence of the multi-axial state, namely, the effective stress and the effective strain. The uniaxial behavior is generalized to multi-axial behavior, within the analytical framework of isotropic hardening plasticity formulation, using a DruckerPrager surface to represent the loading surface under multiaxial compression. In this formulation, the loading surface is a function of the hydrostatic pressure, the second invariant of the deviatoric stress tensor, and the yield strength. This type of formulation incorporates the effects of low to moderate confinement stress levels, which typifies the behavior of civil structures. These relations allow for linear behavior for compressive stresses below about 50% of the compressive strength, and then strain hardening behavior until the compressive strength is reached. Strain softening behavior develops for continued compressive loading past the strain

Figure 1. Crack Initiation Criteria Curve The surfaces of cracks that develop due to tensile stress in concrete are usually rough and irregular. When a shear force is applied along a crack, tangential shear sliding occurs, and this induces displacements normal to the crack surface as the crack surfaces ride up on each other. When this normal displacement is restrained by reinforcement crossing the crack, tensile stresses will develop in the steel bars, which will then induce compressive stresses across the crack in the concrete. The resistance to shear is provided by the frictional force generated by the compressive stress across the crack. The crack width is the primary variable affecting this mechanism of shear transfer. Smaller crack widths correspond to greater shear stiffness and strength. Aggregate size, reinforcement design, and concrete strength are other important factors. In order to account for the effect of cracking on shear stiffness, a reduced shear modulus

Copyright 2009 by ASME

capacity at the compressive strength. Another important characterization of concrete material is the response under cyclic load, particularly under high compressive loads. Under cyclic loading, the model exhibits a continually degrading stiffness and strength, as illustrated in Figure 3 for 5 ksi concrete, which is in agreement with experimental test data.

reference, but for these analyses, a small tensile strength is enforced for compatibility with design codes, which does not allow credit for tensile stress in concrete. The degradation due to time held at constant temperature can be ignored since the time span of interest is less than 100 hours. The effects of creep are also considered negligible for this time period of interest.

Figure 3. Behavior Under Cyclic Compressive Load Exposure to elevated temperatures is known to have a deteriorating effect on the physical and mechanical properties of concrete. At elevated temperatures, microcracking will develop between the aggregate and the cement paste due to differences in the thermal expansion characteristics between these materials. At temperatures above 93 C (200 F), it is also observed that concrete properties will continue to degrade even when the concrete is held at constant temperature. This degradation is associated with the movement of free water and the water of hydration within the hardened cement paste. This movement of water also results in shrinkage of the paste. These chemical and mechanical changes cause degradation of the modulus, compressive strength, and tensile strength. Furthermore, this degradation of properties is non-recoverable upon return to lower temperatures since the microcracking and chemical changes are irreversible. For these analyses, the degradation of the concrete properties is separated into 2 parts, degradation as a function only of temperature, and degradation as a function of time held at elevated temperature. For the degradation of concrete properties with temperature, the relations defined in Reference 4 are used. These characterizations for modulus and compressive strength as functions of temperature were based on a comprehensive review of test data that include a wide variety of test conditions, concrete mixes, and temperature conditions, and are recommended by the Department of Energy as a basis for the design of concrete structures at elevated temperatures. In these analyses the reduction factor for degradation with temperature is taken to be the average between the lower bound and upper bound values from the reference. This characterization for modulus and compressive strength degradation as a function of temperature is illustrated graphically in Figure 4 for nominal 5 ksi concrete. Note that relations for the tensile strength are also included in the

Figure 4. Property Degradation Versus Temperature EXAMPLE An illustrative example of this design by analysis method is provided for a beam in bending due to an applied thermal gradient. The example considers a reinforced concrete beam subjected to a linear temperature distribution across the depth of the beam. The ends of the beam are restrained such that a moment develops to resist the thermal demand for curvature. The thermal expansion in the axial direction is not restrained so that only a thermal moment develops. Figure 5 illustrates the modeling used for this study. The entire beam is initialized to a reference temperature of 60 F, and then the temperatures are increased incrementally with the bottom surface kept at the reference temperature and the top surface reaching 340 F with a linear temperature distribution between the surfaces. A linear analysis without concrete cracking is first performed as a baseline. A nonlinear analysis is performed allowing concrete cracking and modulus degradation with temperature to develop. The study considered a 42 deep by 12 wide beam with 1 in2 tensile rebar (~.2% tensile rebar ratio to concrete area) using a coefficient of thermal expansion of 4.735E-6 in/in-F. Figure 6 illustrates the knockdown factor on concrete compressive stress that is calculated for this configuration. This figure plots

Copyright 2009 by ASME

comparisons of the concrete compressive stress versus temperature of the hot surface for 2 different values of modulus represented by the different concrete strengths shown. That is, one should compare the red curve to the black curve for 4 ksi concrete and the green curve to the blue one for 5 ksi concrete. This problem also considers a nominal value of tensile strength for the concrete. Note that the stress in the nonlinear analyses follow that for the linear analyses until the tensile strength is exceeded and cracking develops. Also note that for the higher strength concrete, less cracking develops for higher temperatures because of the slightly higher tensile strength, and the knockdown factor at 340 F is thus slightly higher for the higher strength concrete. This analysis also allowed further verification of the software used to convert section stresses into section forces and moments. This post-processing software was used to calculate the moments for the linear and cracking analyses. The linear moment is verified against the theoretical value, defined by M = EIT/t, where E is the elastic modulus, I is the bending moment of inertia, is the coefficient of thermal expansion, T is the temperature change from the reference temperature, and t is the depth of the beam. The moment with concrete cracking is verified for consistency with the calculated concrete compressive stress and the cracked section properties and with the moment that develops at the fixed reaction point for force equilibrium. In addition, the analytical results are consistent with experimental testing described in References 1 and 2.

It is noted that the thermal reduction factor depends on the stiffness (section size, modulus, and reinforcement ratio) and the temperature range and distribution. This factor can also be different for the various section force components in a 3dimensional section. Thus, the design by analysis methodology computes a reduction factor for each force component for each structural section and temperature distribution of interest. APPLICATION This methodology has been used in the structural design of the reinforced concrete primary containment boundary and parts of the reactor building for the ESBWR. Half symmetric, detailed 3D brick element models are constructed, as illustrated in Figure 7 for the thermal analysis model and Figure 8 for the stress analysis model. The stress analysis model uses higher order, 20-node brick elements with reduced integration as providing the best performance for concrete cracking analyses for a given level of mesh refinement. The thermal analysis model uses bi-linear, 8-node brick elements to avoid known convergence issues associated with bi-quadratic elements and the implicit formulation used in the thermal solution. The thermal models use eight 8-node brick elements for each 20node brick in the stress analysis model for ease of transferring the nodal temperature distributions from the thermal model to the stress model. The major internal steel components and all liners are included using plate and membrane elements, as illustrated in Figures 9 and 10, respectively. The reinforcement is included in the stress model for concrete cracking as illustrated in Figure 11, which shows some of the bars in the RCCV wall, the Pedestal Wall, and Suppression Pool floor slab.

Figure 5. Beam Model for Thermal Stress Example

Figure 6. Stress Reduction due to Cracking in Beam

Figure 7. Illustration of Model for Thermal Analysis

Copyright 2009 by ASME

Figure 8. Illustration of Model for Cracking Analysis Figure 11. Illustration of Reinforcement Included in the Modeling for Cracking Analysis The temperature distributions are assumed to be independent of the stress solution, that is, the temperatures do not depend on displacements. The analyses first use transient thermal solutions to establish the temperature distributions for 72 hours after the LOCA event. Transient thermal solutions are employed to capture the correct distribution of the temperature gradient through the thick concrete walls with time. These analyses use temperature and convection boundary conditions for exposed or wetted walls and apply the time histories of temperatures for the drywell and wetwell air spaces based on the conditions and performance of the containment that develop during a LOCA. The temperature distributions at the nodes for the specified time points are read into the stress analysis from the thermal models. The stress analysis is based on static equilibrium without any inertial effects. For the cracking analyses, several load increments are used to transition between the different temperature distributions. At each of the load increments, equilibrium iterations are used to redistribute the loads and section stresses as cracking develops. Yielding and strain hardening of steel components is also considered in the nonlinear cracking analyses. A linear stress analysis without concrete cracking, but including concrete property degradation with temperature, is performed as the reference for establishing the stress reduction and redistribution due to concrete cracking. Figure 12 illustrates the temperature distribution that develops in the RCCV at 72 hours for a LOCA occurring during extreme winter conditions. This case is generally more severe because higher thermal gradients develop across the thick concrete walls. The dark blue represents 10 C, and the

Figure 9. Illustration of Modeling for Internal Steel Components

Figure 10. Illustration of Modeling for Liners

Copyright 2009 by ASME

dark red represents 150 C. The temperatures on the top slab of the RCCV are dependent on the water pools above the slab. The yellow color represents temperatures for water boiling in the various pools.

Figure 14. Concrete Compressive Stresses in RCCV at 72 Hours, Winter LOCA, Cracking Analysis Section forces and moments at critical sections through the concrete walls are calculated for both the linear and the cracking analyses. Figure 15 illustrates locations where section cuts are defined. Several locations in theta around the RCCV at these locations are also considered. Thermal reduction factors are then established by dividing the section forces and moments calculated from the nonlinear cracking analysis by the corresponding section forces and moments from the linear analysis. These thermal ratios can then be used to factor the section forces and moments calculated from the linear design based analyses of the thermal load cases to include the effect of thermal induced cracking into the design.

Figure 12. Thermal Contours for RCCV at 72 Hours, Winter LOCA Figures 13 and 14 illustrate the associated distribution of concrete compressive stress in the RCCV at 72 hours for the linear analysis and cracking analysis, respectively. The linear analysis shows peak concrete compressive stress of 35.6 MPa, slightly above the design strength of 5,000 psi. The peak compressive stress in the non-linear cracking analysis is reduced to 26.2 MPa (3800 psi). The red color in the cracking analysis indicates areas of essentially zero compressive stress.

Figure 16. Illustration of Section Cut Locations for Defining Thermal Ratios on Section Forces SUMMARY AND DISCUSSION A design by analysis methodology is developed to account for thermal stress reduction due to concrete cracking for design of thick reinforced concrete structures subjected to elevated temperatures and thermal gradients. The methodology

Figure 13. Concrete Compressive Stresses in RCCV at 72 Hours, Winter LOCA, Linear Analysis

Copyright 2009 by ASME

accounts for both the stress reduction on each section due to concrete cracking and also that due to stress redistribution from global reduction in stiffness due to the thermal induced cracking. The methodology can be incorporated into design procedures that require linear analyses of load cases due to the large number of load combinations that need to be considered by developing structure specific thermal ratios for each component of section forces at critical section cuts. In general, the thermal ratios are less than 1 at locations where the thermal stresses from the linear analysis are high because of the relief and redistribution of stress as the concrete cracks. In some cases, the thermal ratio may be greater than 1 because of the redistribution of the section forces and moments due to concrete cracking. This typically occurs at sections where the thermal stresses from the linear analysis are low, and a small increase in stress develops from redistribution in the non-linear analysis. One question regarding this methodology is the level of conservatism when results from nonlinear analyses are combined with results from linear analyses. For example, the sections forces due to pressure or other mechanical loading is based on linear analysis results. These are combined with section forces for thermal loads based on nonlinear analyses, whether the section forces for thermal loads are used directly or when the thermal ratios are applied to linear based section forces for thermal loads. A demonstration analysis study was undertaken to address this issue. The first case considers a linear analysis for pressure loads combined with the nonlinear cracking analysis for thermal loads. For this case, the nonlinear cracking analysis is performed for the DBA thermal conditions at 72 hours. The DBA thermal conditions at 72 hours are considered for the demonstration because this condition maximizes the effect of thermal induced cracking with the deepest penetration of temperatures into the concrete walls. A separate linear analysis is then performed for application of the DBA pressure loads (no concrete cracking), and the stress results, both concrete compression and rebar tension, from these 2 analyses are added together. Thus, Case 1 provides the linear based stresses due to pressure combined with the nonlinear calculated stresses due to thermal. Case 2 then considers a true nonlinear analysis where both the pressure and temperature loads are applied simultaneously and incrementally with concrete cracking and stress redistribution. The maximum concrete compressive stresses and the maximum rebar tensile stresses are then compared between Case 1 and Case 2 for the section cuts used in calculating the thermal ratios from the previous analyses. The results show that the design by analysis procedure is conservative for concrete compressive stresses. The concrete stresses from the linear based pressure and nonlinear based thermal are consistently higher than those due to nonlinear based analyses of both pressure and thermal. The results for rebar tensile stresses were not as clear cut, some sections showed higher rebar stresses for the nonlinear analysis of both pressure and temperature. It should be noted that the

demonstration analysis using linear based stresses for concrete can only approximate the true situation, because the procedure used in the design contains an additional step that is not considered here. In the design procedure, the sections forces and moments are calculated for the load combinations and then the concrete section calculations are performed that eliminate any tensile stress in the concrete due to the linear analyses that is used to determine the section forces and moments. Thus, the linear analysis for pressure in the demonstration analysis can contain concrete tensile stresses that will not be present in the final design calculations, and this reduces the tensile stress in the rebar. This final step in the design procedure may increase the concrete compressive stress slightly and will certainly increase the rebar tensile stress as the concrete tensile stress is eliminated and redistributed to the rebar. The design calculations force the rebars to take the tensile load that develops in the concrete from the linear analyses. ACKNOWLEDGMENTS This work was performed for GE Hitachi Nuclear Energy, with partial funding provided by the Department of Energy Under Award Numbers DE-FC07-05ID14635 and DE-FC0705ID14636. Disclaimer: "This report was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government nor any agency thereof, nor any of their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately owned rights. Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise does not necessarily constitute or imply its endorsement, recommendation, or favoring by the United States Government or any agency thereof. The views and opinions of the authors expressed herein do not necessarily state or reflect those of the United States Government or any agency thereof. REFERENCES 1. Shibasaki, N., et al, An Experimental Study on Thermal Stress of Reinforced Concrete Members Under Short-Term Loading, 7th International Conference on Structural Mechanics in Reactor Technology, August 1983, J4/3. 2. Irino, K, et al, Studies on Thermal Stress Design Method for Reinforced Concrete Members of Nuclear Power Plant, 7th International Conference on Structural Mechanics in Reactor Technology, August 1983, J4/5. 3. ANACAP-U, Version 2.5, Theoretical Basis, ANATECH Corp. San Diego, California. 4. Burns and Roe, State of the Art Report on High Temperature Concrete Design, prepared for the U. S. Department of Energy, Document DOE/CH/94000-1, November 1985.

Copyright 2009 by ASME

Anda mungkin juga menyukai