STRESS FIELD EVALUATION IN CONCRETE GRAVITY DAMS USING THE PSEUDO-STATIC AND PSEUDO-DYNAMIC APPROACHES Paulo Marcelo VIEIRA

RIBEIRO Doctorate Student, Universidade de Braslia UnB Lineu Jos PEDROSO Professor, Universidade de Braslia UnB Silvio CALDAS Civil Engineer, Centrais Eltricas do Norte do Brasil - ELETRONORTE Brazil

Traditionally, concrete gravity dams have been designed and analyzed by a simplified procedure, which does not consider the structural elasticity and the fluid compressibility(Pseudo-Static Method). Some authors have already found out that the consideration of these aspects can produce significant stresses in the dam, indicating that the simple procedure might not be appropriate for the design of this type of structure [1,2]. A study conducted by [1] indicates an alternative procedure for evaluation of the seismic actions, incorporating the effects of the elasticity of the structure and the compressibility of the water. It is a simplified analytical procedure, based on the spectral response of the structure. Its application consists in the calculation of the seismic loading, which should be applied to the dam in a equivalent static analysis. In design conditions where the dam can be treated as a rigid body (period of vibration is approximately equal to zero), the proposed model by the Pseudo-Static method is appropriate, and a constant distribution of seismic accelerations along the dams height can be adopted, neglecting relative motion contribution. This method considers that the dam is provided with the same acceleration of an infinitely rigid foundation, and the peak ground acceleration produces the largest seismic effects on the structure.

However, there are conditions where the effects of the flexibility of the dam should be considered, and analysis of the dynamic response of the structure is needed. There are reports of accidents with dams that were designed by the traditional procedure, which were submitted to seismic actions much higher than those laid down by the usual rigid body procedure [1]. The Koyna dam, for example, was designed with a seismic coefficient of 0.05g. However, the 1967 earthquake, occurred in India, was able to produce actions that exceeded the design considerations. As a result, many cracks were formed along the dam. The structural flexibility is of fundamental importance for understanding the actions produced by the earthquake. A large amplification of the ground acceleration can occur with the consideration of this effect. Fig. 1 and 2 illustrate the total acceleration responses produced in a system of 1 degree of freedom (obtained by numerical integration, using a fourth order Runge-Kutta procedure), when subjected to a short range of the north-south component of the El Centro earthquake - 1940 [3]. The first system has a period of vibration equal to 0.02s, while the second system presents a period of vibration equal to 1s. The damping for the two situations is equal to 2%. It can be observed that in the first system (Fig. 1) the response approaches the ground acceleration. Thus, this indicates a rigid body motion. The second system presents an amplification of the acceleration response, with respect to the earthquake produced component. At a certain instant the total acceleration achieved by the system is twice the value of the ground acceleration. The behavior shown in Fig. 1 reveals an important characteristic of the seismic design spectra. On this diagram the spectral acceleration tends to move closer to the peak ground acceleration as the period of vibration of the structure decreases. This indicates that the system acquires properties similar to those of a rigid body motion, with the total acceleration response approximately equal to the ground acceleration. In these cases the use of Pseudo-Static method is acceptable. According to [4] the use of this procedure is valid for dams with fundamental periods of vibration less than 0.03s. Thus, the dam can be treated as a rigid body, with a constant coefficient equal to the ground produced acceleration, distributed throughout its height (sometimes called seismic coefficient method). Structural relative motion is neglected on this type of analysis. The behavior shown in Fig. 2 illustrates a dynamic amplification of the ground motion. In some cases the gain reaches 75% of the ground acceleration. A dam with the specific characteristics of this dynamic system, designed by the traditional procedure, would have a peak ground acceleration of approximately 0.04g, while the flexible structure reaches a maximum acceleration of 0.07g.

The Pseudo-Static method seismic actions are derived from the hypothesis of a rigid body moving towards an incompressible fluid. Thus, the structural mass is treated with a uniform distribution of acceleration, equal to the infinitely rigid foundation acceleration. The seismic loading (distributed per unit of height) to be applied in an equivalent static analysis is composed of two parts: the inertia force and the hydrodynamic pressures. The first one is given by the product of the corresponding mass distribution in the analyzed section with the design established rigid body acceleration. The hydrodynamic pressure distribution is based on studies developed by [5], which were recently reviewed by [6], and represents the inertial effects of the reservoir along the fluid-structure interface.

0.05 0.04 0.03

total acceleration (Runge

ground acceleration (El C

acceleration (g)

0.02

0.08 0.01 0.06 0.00 0.00 0.04

total acceleration (Runge

ground acceleration (El C

0.3

Fig. 2 Single degree of freedom total acceleration response (1s period)

celeration (g)

0.02 -0.02

0.00 -0.03 0.00 -0.02

0.10

0.20

0.3

This procedure is also known as the Seismic Coefficient Method and was widely used for the seismic resistant design of dams. The simplified character and the routine procedure is one of the major advantages of this method, despite its drawbacks, which include not considering: the structural elasticity, damping forces, variation of the foundation acceleration over time, and the alternation and short duration characteristics of the seismic loading [7]. The Pseudo-Static Method seismic actions are given by Eq.[1]:
C.S .( y ) = vg wS ( y) + g p( y) g

[1 ]

where vg stands for the horizontal ground acceleration, g indicates the gravity acceleration, wS represents the distributed dam weight per unit of height and g p is related to the hydrodynamic pressure distribution along the fluid-structure interface (relative to a rigid body motion towards an incompressible fluid [2]). Thus, the resulting seismic forces and moments in an analyzed section are given by application of this seismic loading in an equivalent static analysis.

2.1.

EQUIVALENT STATIC ANALYSIS

Stress analysis procedures can be applied to the structure once the seismic loading (distributed per unit of height) has been established. The procedure involved in the Pseudo-Static method is simple, and consists in applying this loading to an equivalent static analysis. The stress analysis calculation may be driven by analytical or numerical methods. One the highlights of the analytical tools is the Gravity Method [8] with a formulation that provides the internal stress distribution in the dam geometry. These are divided into three in plane components: two normal stresses (to the vertical and horizontal planes, respectively) and a shear stress. These are given, respectively, by Polynomial Equations [2], [3] and [4]. A great attraction of this method is the ease of use, in addition to the excellent results obtained when compared to more refined numerical solutions [9]. Fig. 3 and 4 illustrate, respectively, results from the stress field distribution obtained with this methodology and with the application of the Finite Element Method. A clear similarity between these two stress fields can be observed. Generally, this comparison tends to lose quality as the selected section approaches the foundation. In sections where the hypothesis of trapezoidal distribution of vertical normal stresses is not verified, the method collapses, since this a basic assumption of this procedure. However this method remains as one of the most widely used approaches for preliminary dam design.
Z ( y) = a + b y
ZY ( y ) = a1 + b1 y + c1 y 2

[ 2] [

 Y ( y ) = a2 + b2 y + c2 y 2 + d 2 y 3

3] [ 4]

where a, b, a1, b1, c1, a2, b2, c2 and d2 are constants related to the analyzed section, which depend on parameters such as geometry, forces and moments on the design elevation. A more detailed explanation can be found on [2].

Fig. 3 Gravity Method results (units: kPa)

Fig. 4 Finite Element Method results (units: kPa)

Another alternative for the analytical stress evaluation is the application of the procedure developed by [10]. This author proposed the stress calculation only at the dams upstream and downstream slopes, which usually provide the critical design regions of this type of structure when facing seismic actions. This reduces the design equations to a procedure simpler than the Gravity Method, based on the infinitesimal prisms equilibrium at upstream and downstream slopes. On these formulations the involved terms are reduced to: normal stresses on horizontal planes, hydrostatic pressures, hydrodynamic pressures and slope angles. All these components are immediately provided, except for the normal stresses, which can be obtained by application of the classical beam theory. The latter is a function of the moment of inertia, and the resultant of forces and moments in the analyzed section. These actions can be obtained in practice with the interpretation of seismic loading by means of line segments [10], forming trapezoidal loading areas. Thus, the calculation of the seismic action is reduced to the solution of a loading formed by a combination of straight segments, largely simplifying the procedure for calculating the seismic resultant forces and moments and providing

excellent results if an appropriate number of design divisions is chosen.

This analytical procedure was developed by [1] as an alternative to more general procedures, which required the use of a computer in order to evaluate the structural seismic response. It consists on a simplified analysis of the spectral response, which determines the structures response in the fundamental vibration mode due to a horizontal ground motion [11]. This author observed that the response of structures with short periods of vibration, such as concrete gravity dams, when subjected to seismic actions, was largely influenced by the fundamental vibration mode. It was also concluded that the vertical components of the ground acceleration exerted little influence on the structural response. Based on these conclusions this author suggested a simplified methodology for preliminary analysis of concrete gravity dams. The dam, which in the Pseudo-Static Method was supposed rigid, is now treated as flexible (see Fig. 5), and the water in the reservoir considered as a compressible fluid. The seismic loading now depends on the fundamental mode seismic response of the structure, associated with the corresponding horizontal ground motion.

Fig. 5 Pseudo-Dynamic Method seismic response [1]

The seismic coefficient of the Pseudo-Dynamic Method takes into account the particular characteristics of each ground motion. The spectral acceleration is obtained by a acceleration response spectrum corresponding to the design

earthquake, and depends on the fundamental vibration period and the structural damping. The seismic forces calculated using the spectral acceleration are used in an equivalent static analysis. The main disadvantages of this procedure involve the alternation and short duration characteristics of the seismic loading, which are not considered [7]. It is, in fact, an estimate of the maximum response produced by the fundamental mode. A more refined analysis would consist on a study of the dynamic response of the structure, where the entire history of displacements and other response values would be studied over time. The seismic loading of this level of analysis, including the effects of the reservoir, is given by Eq.[5]. A more detailed explanation of the origin and application of this loading can be obtained in [1,11].
C.S .( y ) = 4 Sa wS ( y) ( y) + g p ( y) 1 g

[5]

where S a indicates the spectral acceleration corresponding to the structures fundamental vibration period (considering the reservoir effects), g is the gravity acceleration, wS represents the distributed dam weight per unit of height, stands for the fundamental mode shape function and g p1 is related to the hydrodynamic pressure distribution along the fluid-structure interface (relative to a flexible boundary motion towards a compressible fluid [1]).

3.1.

PROPOSED
PROCEDURE

MODIFICATIONS

TO

THE

EQUIVALENT

STATIC

ANALYSIS

The Gravity Method [8] is an excellent analytical tool for evaluation of the stress field distribution. However, its original formulation was conceived for the Pseudo-Static actions, and does include those provided by the Pseudo-Dynamic Method. Through a more detailed interpretation of this procedure, [2] introduced in a simplified manner, the seismic actions produced by the procedure developed by [1]. The modifications proposed by [2] include the adoption of a second degree polynomial fundamental mode shape function (simplifying largely the analytical procedure for evaluation of the equivalent forces and moments produced by the inertia force) and the use of hydrodynamic pressure functions similar to those proposed by [5], with the inclusion of correction coefficients. Fig. 6 and 7 illustrate graphics with comparisons of these simplifications with the considerations originally proposed by [1].

1.00 0.80 0.60 0.40

CHOPRA PARABOLIC

Fig. 6 Proposed fundamental mode shape function [2]

y/HS 1.00
0.20
Fig. 7 Proposed hydrodynamic pressure functions [2]

= 0.8 (

Mode S
0.40

0.60

It can be observed that the proposed fundamental mode shape function leads to conservative results in evaluation of the seismic loading, producing values always higher than those obtained with Chopras proposed fundamental mode shape. It should be also noticed that the proposed solution to the hydrodynamic pressure is non-conservative, producing results lower than those achieved by [1]. In general, it is expected a compensation between the overestimated inertia force and the underestimated hydrodynamic pressures. Results of tests conducted by [2] confirm the latter.

y/H

0.00 0.00

0.20

0.40 0.20

0.00 0.00

0.20

0.40

The Pine Flat dam non-overflow section (see Fig. 8) was analyzed by both the Pseudo-Static and the Pseudo-Dynamic Methods. This study will present the main differences obtained on the stress field distribution for this profile. In the equivalent static analysis the Gravity Method was applied, with the modifications proposed by [2], for the introduction of the pseudo-dynamic seismic actions in this procedure. In the pseudo-static analysis the dam will be treated as rigid, accelerated with the peak ground acceleration, which will be adopted as equal to 0.2g. The pseudo-dynamic seismic loadings are evaluated using the simplifications proposed by [2]. Fig. 9 illustrates the seismic response spectrum applied on these analyses. Due to the difficulties of obtaining a typical seismic response spectrum for the Brazilian territory, a curve was adopted for a specific earthquake in the North American region. This spectrum is suitable for seismic design - in regions of firm ground in California to ground motions with a similar intensity of the earthquakes recorded in Taft, during the Kern Country earthquake in July 1952 [1]. This is a characteristic response spectrum of an earthquake with PGA (peak ground acceleration ground) equal to 0.2g.

Fig. 8 Pine Flat Dam non-overflow section

Fig. 9 Seismic design spectrum [1]

4.1.

STATIC ANALYSIS (NORMAL OPERATION ACTIONS)

Fig. 10 and 11 illustrate, respectively, the analyses of maximum and minimum principal stresses (were positive sign indicates compression), obtained from the normal operation actions, including the concrete self-weight and the hydrostatic pressures. Uplift pressures were neglected on both analyses.

4.2.

EQUIVALENT STATIC ANALYSIS (PSEUDO-STATIC METHOD)

In this type of analysis it is assumed that the structure is rigid moving toward the incompressible fluid reservoir. For the seismic coefficient, values ranging between 0.05g and 0.10g are usually adopted (or a fraction of the peak ground acceleration). The response spectrum illustrated on Fig. 9, for a vibration period equal to zero, results in a seismic coefficient equal to 0.2g, which corresponds to a value much higher than usually adopted. Still, this is the value that should be used, because it corresponds to the peak ground acceleration (PGA) in this example. In this analysis the following loadings are considered: concrete self-weight, hydrostatic pressures, inertia forces and hydrodynamic pressures. The dam will be examined with horizontal seismic acceleration towards the upstream direction. This means that seismic forces will act in the downstream direction. Fig. 12 and 13 illustrate, respectively, the analyses of maximum and minimum principal stresses, obtained with the application this procedure.

Fig. 10 Maximum principal stresses results (units: kPa)

Fig. 11 Minimum principal stresses results (units: kPa)

Fig. 12 Fig. 13 Maximum principal stresses results Minimum principal stresses results (units: kPa) (units: kPa) 4.3. EQUIVALENT STATIC ANALYSIS (MODIFIED PSEUDO-DYNAMIC METHOD)

In this analysis the applied seismic loading will be similar to the one proposed by [1], with the simplifications made by [2]. This is not a pseudo-dynamic analysis itself, because the seismic loading is not implemented exactly as defined by [1]. However, the results of previous studies performed by [2] demonstrate an excellent agreement of the simplified procedure (defined as Modified PseudoDynamic Method) with the results obtained by [1]. Fig. 14 and 15 illustrate the maximum and minimum principal stress distribution, obtained with the application of the Modified Pseudo-Dynamic Method, for a horizontal seismic acceleration oriented towards the upstream direction. It can be observed an increase in the magnitude of the seismic actions when compared to the previous item (resulting in smaller stresses at the upstream slope and in an increase of these values at the downstream slope).

Fig. 14 Maximum principal stresses results (units: kPa)

Fig. 15 Minimum principal stresses results (units: kPa)

Two application examples of procedures for seismic analysis of typical profiles of concrete gravity dams, subjected to a design spectrum, were presented. The pseudo-static procedure treats the dam as a rigid body, ignoring the amplification effects of the dynamic response. However, there are some case were this effect cannot be neglected. Only structures with very small periods of vibration present dynamic responses that approach the ground motion (see Fig. 1), featuring a rigid body movement. In other situations, the dynamic characteristics of the dam should be taken into account, in order to include the structural response amplification effects (see Fig. 2). The pseudo-static procedure, which was traditionally used in the design of many dams (also known as the Seismic Coefficient Method), produces results that underestimate the seismic actions when the structure cannot be treated as a rigid body. Application of this procedure is recommended only to structures with a fundamental vibration period less than 0.03s [4]. The achieved results provide that the actions produced by the Pseudo-Static Method are far lower than those produced by the Pseudo-Dynamic procedure. The stress distribution in the pseudo-static analysis exerts little influence in

the design of a dam, because the tensions are almost nonexistent (see Fig. 13, with maximum tension equal to 50 kPa), and the added compression will hardly exceed the concrete strength usually employed in this type of structure. However, in the modified pseudo-dynamic procedure, tensions can be significant (see Fig. 15, with values up to 1500 kPa), directly influencing the choice of the concrete design resistance, due to the low tensile strength resisted by this material. In some cases, special mixtures are needed to ensure the necessary strength to regions were very high tensions are expected.

ACKNOWLEDGEMENTS

The authors are grateful for the financial support provided by CAPES and CNPq agencies.

REFERENCES

[1] [2]

[3] [4]

[5] [6]

[7] [8] [9]

[10]

CHOPRA A .K. "Earthquake resistant design of concrete gravity dams" ASCE Journal of Structural Division, vol. 104, n. ST6, pp. 953-971, Jun. 1978. RIBEIRO P. M. V., "Uma metodologia analtica para a avaliao do campo de tenses em barragens de concreto durante terremotos". Masters thesis. Universidade de Braslia, 2006. CHOPRA A. K. Dynamics of Structures Theory and Application to Earthquake Engineering. Prentice Hall, 2001. GHRIB, F., LGER P., TINAWI R., LUPIEN R., VEILLEUX M. Seismic safety evaluation of gravity dams. In: International Journal on Hydropower & Dams, v. 4, n. 2, p. 126-138, 1997 WESTERGAARD H. M., "Water pressure on dams during earthquakes" Transactions ASCE, v. 98, n. 1835, pp. 418-433, 1933. SILVA S. F., PEDROSO L. J. "Estudo analtico-numrico do campo de presses e da massa adicional em barragens durante terremotos" Relatrio Tcnico de Pesquisa. RTP-SF2-05-2005, 2005. PRISCU R. Earthquake Engineering for Large Dams. Bucaresti, 1985. USBR, Design of Gravity Dams. Denver: United States Department of the Interior Bureau of Reclamation, 1976. RIBEIRO P. M. V., PEDROSO L. J, "Analytical procedure for stress field solution in concrete gravity dams," in International Symposium on Solid Mechanics. So Paulo: Universidade de So Paulo, 2007. RIBEIRO P. M. V., PEDROSO L. J. "Uma aplicao de referncia do mtodo pseudo-dinmico para a anlise ssmica de barragens de concreto gravidade" in XXVII Seminrio Nacional de Grandes Barragens. Belm, 2007.

[11]

FERC (Federal Energy Regulatory Commission). Chapter III Gravity Dams. In: Federal Energy Regulatory Commission, Office of Hydropower Licensing. Engineering guidelines for evaluation of hydropower projects. Washington, 2002