Effect of Viscous Damping on Inelastic Design Spectra

E. Bazán-Zurita, S. Jarernprasert
DiGioia, Gray and Associates, LLC, 570 Beatty Road, Monroeville, PA, USA;
PH: 412-372-4500; FAX: 412-372-1972; [email protected];
[email protected]

J. Bielak
Department of Civil and Environmental Engineering, Carnegie Mellon University,
Pittsburgh, PA, USA; PH: 412-268-2958; FAX: 412-268-7813;
[email protected]

ABSTRACT:
Seismic regulations define design seismic ground motion as elastic spectra associated to 5-percent viscous
damping. Reduction factors are applied to the 5-percent spectra to consider energy dissipation in inelastic
structural systems. In some instances, however, viscous damping differs from 5 percent. For instance, ASCE
Manual 113 requires that the structures of electrical substations be designed for 2-percent damping. Correction
factors are prescribed for other than 5-percent damped systems. These factors are based on studies considering
elastic behavior. However, the effects of viscous damping on inelastic spectra could differ significantly from
those on elastic systems. This paper presents a study over the impact of viscous damping on spectra of inelastic
structures with various types of hysteric behavior. The seismic input consists of earthquake records from
Californian and Mexico City sites. We propose simple equations for modifying inelastic spectra from 5 to other
percents of critical damping, which depend on the ductility demand.

Keywords: spectra, inelastic, damping, hysteretic, degrading

1. INTRODUCTION

Most current codes stipulate elastic design spectra corresponding to 5% of critical damping. However,
different damping ratios may be either prescribed or judged to be adequate for some structures or
components. Thus, damping correction factors (DCFs) are applied to the 5%-damped. Early DCF
equations were proposed by Arias and Husid (1962) and Newmark and Rosenblueth (1971). Recently,
Cameron and Green (2007) have reviewed the literature on the topic and developed factors in terms of
structural period, frequency content of the ground motion, and tectonic setting. All these studies deal
with elastic spectra. However, current seismic design regulations recognize that, under severe events,
the structures will undergo significant nonlinear deformations. On account of this beneficial inelastic
behavior, results of elastic analyses are reduced by a factor, R. Typically, R is specified in terms of
the type of system adopted to resists lateral forces and on the natural period of the structure. As an
example, ASCE Manual 113 (2008) and IEEE Standard 693 (2006) for electrical substations, have
adopted specific damping correction factors (DCF) independent of inelastic behavior. At the same
time, Manual 113 stipulates R-factors ranging from 1.3 to 3. When the viscous damping differs from
5%, the DCFs specified for elastic spectra are applied to the R-reduced spectra. This study shows that
this practice could be inaccurate and that DCFs also depend significantly on the types of inelastic
behavior.

2. EFFECT OF VISCOUS DAMPING ON ELASTIC SYSTEMS

DCFs are developed by comparing elastic spectra for various damping ratios from ensembles of
earthquake records. In this study, to evaluate the current IEEE DCFs (adopted by ASCE 113) and to
develop factors applicable to nonlinear structures, we use two sets of earthquake records: one
consisting in 87 records from Californian stiff to medium stiff sites, with dominant periods between
0.2s and 0.3s, and the second comprising 66 record from the soft lakebed of Mexico City with
dominant periods of nearly 2s. The records of each set were separately normalized to have the same
Arias Intensity and to yield average peak ground acceleration equal to that of gravity. Average 5%-
damped elastic spectra of the two ensembles are presented in Figs. 1 and 2.

3.5
exact 2%
Normalized Acceleration 3.0
exact 5%
2.5 exact 10%
2.0 IEEE 2%
IEEE 10%
1.5

1.0

0.5

0.0
0.0 1.0 2.0 3.0
Period (s)

Figure 1. Elastic Calculated and Approximate (IEEE approach) Spectra, Californian Records

exact 2%
5.0
Normalized Acceleration

exact 5%
exact 10%
4.0
IEEE 2%
3.0 IEEE 10%

2.0

1.0

0.0
0.0 1.0 2.0 3.0 4.0
Period (s)

Figure 2. Elastic Calculated and Approximate (IEEE approach) Spectra, Mexico City Records

Fig. 1 compares the calculated average elastic spectra for the Californian records used in this study
with approximate values resulting from using of the above IEEE CDFs for 2 and 10%-damped spectra.
The excellent agreement between calculated and approximate spectra indicates that the IEEE CDFs are
adequate for Californian stiff to medium stiff sites. Fig. 2 displays a similar comparison for the
ensemble of Mexican records. The agreement is still very good except for periods in the vicinity of
the site dominant period (2s): the IEEE factors underestimate for 2%-damping spectral peak and
overestimate the peak for 10% damping. The reason is that records from the soft lakebed of Mexico
City exhibit a much narrower frequency content than Californian records.
Standard IEEE 693 stipulates the following damping correction factors:

DCF        for 0 ≤ f ≤ 8 Hz (2.1)

DCF  1  0.04  133  f    for 8 ≤ f ≤ 33 Hz (2.2)

DCF  1 , for f > 33 Hz (2.3)

  1.5173  0.3213 lnd  (2.4)

d = percent of damping ratio

f = frequency in Hz = inverse of period, T, in seconds.

3. EFFECT OF VISCOUS DAMPING ON INELASTIC NONDEGRADING SYSTEMS

In this study we consider the two types of inelastic force-displacement curves depicted in Fig. 3. The
first curve represents nondegrading bilinear behavior with second slope equal to 2 percent of the initial
one, yield strength Fy, and initial stiffness k, and with stable hysteretic cycles. The second curve
shows total stiffness degradation associated to structures where steel X-bracing is the primary
component to resist lateral forces (Bazán and Rosenblueth, 1974).

Figure 3. Bilinear Non-Degrading (a) and Degrading Hysteretic Curves

Our first finding is that appreciable better agreement between 5%-damped calculated spectra and the
IEEE approximation for Mexican records is achieved when the spectrum yielding a ductility of unity
is considered instead of the entirely elastic spectra, as illustrated in Fig. 4. To define a ductility-one
spectrum we conducted nonlinear analysis of non-degrading systems to determine the yield resistance
that results on an average ductility of 1.0. The maximum shear force induced by some records is
smaller that the yield resistance producing a ductility smaller than unity, while other records generate a
shear force larger than the yield resistance with an ensuing ductility bigger than 1.0. The difference
between the two concepts is apparent when comparing the 4.4 calculated peak of the 5%-damped of
the strictly elastic spectrum in Fig. 2 with the 5.3 calculated peak of the 5% ductility-one spectrum.
Clearly, even a limited amount of inelastic behavior smoothes the peaks of narrow banded spectra.
We conducted a similar exercise with the Californian records finding no appreciable difference
between the 5%-damped strictly elastic spectrum and the ductility-one spectrum. Thus, we conclude
that the IEEE approach is adequate for ductility-one spectra in both Californian and Mexico City sites.
 5.0 exact 2%

Normalized Acceleration
exact 5%
4.0 exact 10%
IEEE 2%
3.0 IEEE 10%

2.0

1.0

0.0
0.0 1.0 2.0 3.0 4.0
Period (s)

Figure 4. Spectra for  = 1, Non-Degrading Systems, Mexican Records

1.6
exact 2%
1.4
Normalized Acceleration

exact 5%
1.2
exact 10%
1.0
approx 2%
0.8
approx 10%
0.6
0.4
0.2
0.0
0.0 1.0 2.0 3.0
Period (s)
Figure 5. Inelastic Spectra for  = 2, Non-Degrading Systems, Californian Records

To quantify the impact of viscous damping ratios on inelastic nondegrading structures, we have
calculated inelastic spectra for average ductility demands of 2 and 4, using the ensembles of
Californian and Mexico City records. Fig. 4 presents the calculated inelastic spectra of Californian
records for an average ductility of 2, and viscous damping ratios of 2, 5 and 10%. Also presented in
Fig. 4 are approximate spectra for 2 and 10% damping obtained by multiplying the 5%-damped
spectrum by CDFs defined as follows:

 1 
 
 n   
DCF    DCF 1 (3.5)

n   1  1.2 ln   (3.6)

DCF(1) is the factor given by Eqs. (2.1) to (2.4) and includes the correction for damping ratio other
than 5%.

Fig. 5 shows that Eqs. (3.5) and (3.6) lead to accurate estimates of the calculated inelastic spectra.
Additional spectral comparisons are presented in Fig. 6 for  = 4 and Californian records, Fig. 7 for 
= 2 and Mexican records, and Fig. 8 and for  = 2 and Mexican records. In all cases, the
approximations provided by Eqs. (3.5) and (3.6) agree very well with the calculated inelastic spectra
over the entire ranges of studied periods.

1.2
exact 2%
Normalized Acceleration

1.0
exact 5%
0.8 exact 10%
approx 2%
0.6
approx 10%
0.4

0.2

0.0
0.0 1.0 2.0 3.0
Period (s)

Figure 6. Inelastic Spectra for  = 4, Non-Degrading Systems, Californian Records

1.6
exact 2%
1.4
Normalized Acceleration

exact 5%
1.2 exact 10%
approx 2%
1.0
approx 10%
0.8
0.6
0.4
0.2
0.0
0.0 1.0 2.0 3.0 4.0
Period (s)

Figure 7. Inelastic Spectra for  = 2; Non-degrading Systems; Mexico City Records

 1.2
exact 2%

Normalized Acceleration
1.0 exact 5%
exact 10%
0.8
approx 2%
0.6 approx 10%

0.4

0.2

0.0
0.0 1.0 2.0 3.0 4.0
Period (s)

Figure 8. Inelastic Spectra for  = 4, Non-degrading Systems, Mexico City Records

4. EFFECT OF VISCOUS DAMPING ON INELASTIC DEGRADING SYSTEMS

The hysteretic loops of most inelastic structures degrade appreciably under repeated load cycles. To
assess the impact of viscous damping on degrading systems, we have considered the total stiffness
behavior represented in Figure 3b. Degrading in this type of system is rather severe since for cycles of
the same amplitude, no energy is dissipated by hysteresis after the first cycle. Using this degrading
load displacement curve, we have repeated the calculations described in the previous section for to
calculated inelastic spectra for average ductility demands of 2 and 4, again using the ensembles of
Californian and Mexico City records.

1.4 exact 2%
Normalized Acceleration

1.2 exact 5%
1.0 exact 10%
approx 2%
0.8
approx 10%
0.6
0.4
0.2
0.0
0.0 1.0 2.0 3.0
Frequency (Hz)

Figure 9. Inelastic Spectra for  = 4, Stiffness-Degrading Systems, Californian Records

Fig. 9 shows the calculated and approximated inelastic spectra for an average ductility demand of 4.
Examining this figure and the results for ductility demand of 2 (omitted for brevity), we find that CDF
factors defined by Eq. (3.5) applies to approximate the spectra for 2 and 10% damping if Eq. (3.6) is
replaced by:
 n   1  0.3 ln  (4.7)

Similar results corresponding to Mexico City records are presented in Fig. 10. Examination of this
figure and the results for ductility demand of 2 (omitted for brevity), we find that CDF factors are
practically insensitive to the ductility demand. The approximation shown in Fig. 10 was obtained by
straightforward use of the IEEE CDFs (Eqs. (2.1) to (2.4)). This observation, as well as the low
coefficient of 0.3 affecting ln() term in Eq. (4.7), reflect the limited hysteretic energy dissipation
capability of systems with severe degradation. Thus, viscous damping plays a more significant role in
the seismic response than in non-degrading systems.

2.5
exact 2%
Normalized Acceleration

2.0 exact 5%
exact 10%
1.5 approx 2%
approx 10%
1.0

0.5

0.0
0.0 1.0 2.0 3.0 4.0
Period (s)

Figure 10. Inelastic Spectra for  = 4, Stiffness-Degrading Systems, Mexico City Records

5. CONCLUSION REMARKS

Step-by-step analyses of nonlinear single-degree-of-freedom systems reveal that the effects of viscous
damping on bilinear systems depend on their ability to dissipate hysteretic energy. Thus, the
modification of a design 5%-damped spectrum to represent a different damping ratio depends on the
type of hysteretic behavior and on ductility demand.

The procedure of Standard IEEE 693 is very accurate for elastic spectra of Californian records.
However, errors up to 19% occur for elastic spectra of events recorded in the soft lakebed of Mexico
City. The IEEE approach becomes significantly more accurate when applied to the spectrum yielding
a ductility of unity instead of the entirely elastic spectra. In our opinion, code design spectra to be
reduced due to inelastic behavior are more correctly interpreted as being ductility-one.

Our study of non-degrading bilinear systems indicates that viscous damping correction factors (DCFs)
developed to adjust elastic spectra must be appreciably modified in terms of the ductility demand.
Eqs. (3.5) and (3.6) can be used for this purpose. As an example, Fig. 11 shows how inaccurate are
the IEEE equations when directly applied to spectra for ductility 4 of Mexico City. In particular,
spectral values can be underestimated by more than 20% for damping ratios of 10%.
 1.4
exact 2%
1.2

Normalized Acceleration
exact 5%
1.0 exact 10%

0.8
IEEE 2%
IEEE 10%
0.6

0.4

0.2

0.0
0.0 1.0 2.0 3.0 4.0
Period (s)

Figure 11. Inelastic Spectra for  = 4; Non-Degrading Systems; Mexico City Records

Our analyses of systems with total stiffness degrading behavior indicates that changes in viscous
damping have a similar effect as in entirely elastic systems. This result was expected since in the
absence of significant hysteretic loops, viscous damping becomes the dominant source of energy
dissipation.

REFERENCES

ASCE (2008). “Manual 113: Substation Design Guide.” Edited by L. Kempner Jr. Reston, VA.
Bazán-Zurita E., Bielak, J., DiGioia, A.M. Jr. and Jarenprasert, S. (2009). “Seismic Design of Substation
Structures,” ASCE - Electrical Transmission and Substation Structures Conference, Fort Worth,
Texus, USA
Bazán, E. and E. Rosenblueth, 1974, “Seismic Response of One Story X-Braced Frames,” Techincal Note,
Journal of the Structural Divison, ASCE 100, ST2, EUA.
Cameron, W.I. and R.A. Green, 2007, “Damping Correction Factors for Horizontal Ground-Motion Response
Spectra,” Bulletin of the Seismological Society of America, Vol. 97, No. 3, pp 934-960.
Arias, A. and R. Husid, 1962, “Influencia del Amortiguamiento sobre la Respuesta de Estructuras Sometidas a
Temblor,” Rev. IDIEM, 1(3), 219-228.
IEEE (2006). “IEEE Standard 693: Recommended Practice for Seismic Design of Substations,” American
National Standards Institute.
Newmark, N.M. and E. Rosenblueth, 1971, “Fundamentals of Earthquake Engineering,” Prentice-Hall,
Englewood Cliffs, N. J.