13th World Conference on Earthquake Engineering

Vancouver, B.C., Canada

August 1-6, 2004
Paper No. 404

MODELING EARTHQUAKE RESPONSE OF STRUCTURES WITH

HYSTERETIC DAMPING

Fred SEGAL1 and Dimitri V. VAL2

SUMMARY

The present paper examines differences between responses of single-degree-of-freedom (SDOF) systems
with viscous and hysteretic damping to earthquake ground motions. Only responses within yield limits are
considered. Structural behavior of SDOF systems with hysteretic damping is modeled by a generalized
elasto-slip model of the Masing type. The model leads to a force-displacement relationship similar to that
proposed by Ramberg and Osgood. Responses of the two systems are compared for a number of
earthquake records. To achieve consistent comparison parameters of the linear system with viscous
damping are determined by the secant stiffness method. In many cases significant differences in the peak
displacement responses and energy values have been observed between the systems. This indicates that
the prediction of the response of structures to earthquake ground motions even within elastic range by
using linear models with viscous damping may be far from accurate and further research on this issue is
needed.

INTRODUCTION

Performance-based seismic design implies the consideration of several performance levels, which are
associated with damage suffered by structures as a result of earthquake ground motions (e.g., SEAOC [1]).
At higher performance levels (such as fully operational and operational) very limited damage is allowed
that means that values of structural response parameters (e.g., deflections, internal forces) should mainly
stay within their elastic range. Usually, it is presumed that in such cases structural response can be
predicted accurately by using well-known techniques of elastic analysis based on a linear structural model
with viscous damping. However, this presumption is not necessarily correct since for traditional structural
materials (e.g., steel, concrete, masonry) dissipated energy depends mainly on the amplitude of vibrations
and not their frequency, i.e., damping is not viscous but hysteretic (e.g., Newmark [2]).

The present paper examines differences between responses of single-degree-of-freedom (SDOF) systems
with viscous and hysteretic damping to earthquake ground motions. Only responses within elastic range
are considered (i.e., the ductility ratio does not exceed unity). Behavior of SDOF systems with hysteretic

1
Israel Electric Corporation, Haifa, Israel
2
Faculty of Civil and Environmental Engineering, Technion – Israel Institute of Technology, Technion
City, Haifa 32000, Israel
damping is modeled by a generalized elasto-slip model of the Masing type (e.g., Lazan [3]). The model
leads to a force-displacement relationship which can be described in terms of the Ramberg-Osgood
expression (Ramberg [4]). Based on the model formulas, which allow to compute recoverable strain
energy and energy dissipated by hysteretic damping at any point in time, have been derived.

Responses of SDOF systems with viscous and hysteretic damping and natural periods between 0.05 and 3
s are compared for a number of recorded earthquake ground motions. To achieve consistent comparison
parameters of a linear system with viscous damping are determined by the secant stiffness method (e.g.,
Chopra [5]), i.e., its stiffness is set equal to the secant stiffness of the corresponding system with hysteretic
damping and the damping ratio is obtained by equating the energy dissipated per cycle by the two systems.
Methods employed in earthquake resistant design can be divided into (i) strength-based methods, (ii)
displacement-based methods, and (iii) energy-based methods. Performance-based design puts emphasis on
damage in structures, which is mainly associated with displacements and not strength. Thus, in this study
responses of SDOF systems with viscous and hysteretic damping are compared in terms of peak
displacement and energy values.

HYSTERETIC MODEL

Generalized elasto-slip model

Hysteretic behavior of a SDOF system is modeled using a generalized elasto-slip model (e.g., Lazan [3]).
The model can be represented as two thin elastic bars having the same modulus of elasticity, E, and cross-
sectional areas of A1 and A2 (see Figure 1). At their ends the bars can be loaded by axial forces, fS. The
interaction between the bars along the contact zone of length Lc is described by Coulomb friction; there
are also non-contact zones of length Ln. The maximum friction force between the bars determines the
ultimate axial force which can be applied to the system, i.e., it can be treated as the yield strength of the
system and it is denoted herein as fy.

fS A1
fS

A2
Ln Lc Ln

Figure 1. Generalized elasto-slip model

Load-deformation curves

Skeleton curve
When the system is first loaded by forces fS<fy it causes slip between the bars at the ends of the contact
zone. Diagrams of the axial forces forming in the bars as a result of this loading are shown in Figure 2,
where a1=A1/(A1+A2), a2=A2/(A1+A2), Lsl1=a2LcfS/fy and Lsl2=a1LcfS/fy are the lengths of the slip zones, and
Lst=Lc(1-fS/fy) is the length of the stick zone.
 fS
a 1 fS
x

fS
a 2 fS
x
Lsl1 Lst Lsl2
Ln Lc Ln

Figure 2. Diagrams of axial forces in the bars - loading

Generally, the deformation between the opposite ends of the bars, u, can be calculated as
N ( x)
u=
L
∫
EA
dx (1)

where N(x) is the axial force along the length of the bars. Substituting values of the axial force from the
diagrams in Figure 2 with the corresponding lengths and cross-sectional areas into Eq. (1) and integrating
gives (for both tension and compression)
1 + l + l  L  1 + 1 −3 
 a a  c  
u = fS 
1 2 a a f
1 +
1 2 S
 (2)
E ( A1 + A2 )  21 + l + l  f y 
  a1 a2  
where l=Ln/Lc. Introducing
E ( A1 + A2 )
k0 = (3)
1 + l + l  L
 a1 a 2  c

which can be considered as the initial stiffness of the system, and
1 + 1 −3
a1 a2
α= (4)

21 + l + l 
 a1 a 2 
reduces Eq. (2) to
f  f 
u = S 1 + α S  (5)
k 0  f y 
This equation of the skeleton curve represents a particular case of the relationship proposed by Ramberg
and Osgood (Ramberg [4]) for modeling nonlinear structural behavior.

Unloading/reloading curves
It is known that systems which represent different combinations of Coulomb friction and linear elastic
elements exhibit the Masing type of behavior (Newmark [2]). Thus, for the system shown in Figure 1
unloading/reloading curves should be described by the following equation
 u − u * f S − f S*  * 
= 1 + α f S − f S  (6)
2 2k 0  2 f 
 y 
where (u*,fS*) is the last point at which the loading process was reversed. This can be shown using the
approach that has been employed to derive Eq. (5) for the skeleton curve. As the applied load drops from
its peak value fS* to fS the length of the stick zone increases from Lst*=Lc(1-fS*/fy) at fS* to Lst=
Lst*+Lst1+Lst2, where Lst1=a2Lc(fS*+fS)/(2fy) and Lst2=a1Lc(fS*+fS)/(2fy). Correspondingly, the lengths of the
slip zones at the ends of the bars decrease from Lsl1*=a2LcfS*/fy and Lsl2*=a1LcfS*/fy to Lsl1= a2Lc(fS*-
fS)/(2fy) and Lsl2= a1Lc(fS*-fS)/(2fy) so that Lst1+ Lsl1= Lsl1* and Lst2+ Lsl2= Lsl2*. Diagrams of the axial forces
forming in the bars after unloading are shown in Figure 3. Note that all inclined parts of the diagrams have
the same absolute value of the slope equal to fy/Lc, i.e., the maximum value of the friction force between
the bars per unit length. Eq. (6) can be derived by substituting values of the axial force from the diagrams
in Figure 3 with the corresponding lengths and cross-sectional areas into Eq. (1).

fS+ a2(fS*-fS)/2

a1fS - a1(fS*-fS)/2
fS x

fS+ a1(fS*-fS)/2
- a2(fS*-fS)/2 a2fS
fS
x

Ln Lsl1 Lst1 Lst* Lst2 Lsl2 Ln

Figure 3. Diagrams of axial forces in the bars – unloading/reloading

Energy evaluation
Relationships of the type defined by Eqs. (5) and (6) are employed quite often to describe hysteretic
behavior of nonlinear structural systems. Usually, they are used to formulate empirical models since with
the proper selection of the parameters they provide a good approximation to experimentally obtained
force-displacement (or stress-strain) curves. However, such an approach creates difficulties in evaluation
of energy of a structural system. Although energy absorbed by the system (i.e., work performed by force fS
on displacements u) can be easily calculated by integration of Eqs. (5) and (6) it cannot be decomposed
into two essential parts - recoverable elastic strain energy, ES, and irrecoverable hysteretic energy, EH,
without additional assumptions (of course, except of extreme points at the end of loading cycles when EH
equals the area of completed hysteretic loops). A different approach is chosen in the present paper where
formulas for the evaluation of ES are EH are directly derived using the generalized elasto-slip model
(Figure 1) that allows calculating these energies at any point of cyclic loading.

Skeleton curve
The sum of the recoverable strain energy and the dissipated hysteretic energy equals work, W, performed
by the force fS as the displacement increases from 0 to u along the skeleton curve, i.e.,
W = ES + EH (7)
The work can be calculated as
fS

W = f S u − u ( f S )df S
∫ (8)
0
Substituting Eq. (5) into Eq. (8) and integrating results in
f 2  1 2α f S 
W= S  + (9)
k 0  2 3 f y 
The recoverable strain energy of the system shown in Figure 1 can be evaluated as
N 2 ( x)
ES = ∫
L
2 EA
dx (10)

Substituting values of the axial force from the diagrams in Figure 2 with the corresponding lengths and
cross-sectional areas into Eq. (10) and integrating leads to
f 2  1 α f S 
ES = S  + (11)
k 0  2 3 f y 
Then using Eqs. (7), (9) and (11) the following formula for the dissipated hysteretic energy is obtained
αf S2 f S
EH = (12)
3k 0 f y

Unloading/reloading curves
The recoverable strain energy of the system after unloading (the applied force decreases from its peak
value fS* to fS, see Figure 4) is calculated using Eq. (10) and the axial force diagrams presented in Figure 3
that gives
1
(
E S = E S* − (u * −u ) f S* + f S
2
) (13)
where ES* is the recoverable strain energy at (u*,fS*).

fS
u'
(u*,fS*)
Skeleton W'
curve
Unloading
curve
(u,fS) fS '

Figure 4. Force-displacement curves

In order to evaluate hysteretic energy dissipated during unloading introduce a new coordinate system u'-fS'
with the origin at (u*,fS*). The transformation of the coordinates between the two coordinate systems is
u ' = u * −u
(14)
f S ' = f S* − f S
According to Eq. (13) in the new coordinate system the change in the recoverable strain energy during
unloading will be equal to
1
(
E S − E S* = − u ' 2 f S* − f S '
2
) (15)
At the same time in the new coordinate system (ES-ES*) should also be equal to the sum of the work done
by the force fS* on the displacement increasing from 0 to u' and the recoverable strain energy, ES', supplied
to the system due to the work, W', done by fS' on the same displacement. The work done by fS* is negative
(the direction of fS* is opposite to the direction of u') and its absolute value equals fS* times u' (fS* remains
constant while the displacement changes) so that
E S − E S* = − f S*u '+ E S ' (16)
The recoverable strain energy ES' can be calculated as
E S ' = W '− E H ' (17)
where EH' is the hysteretic energy dissipated during unloading. From Eqs. (15), (16) and (17) follows that
1
E H ' = W '− f S ' u ' (18)
2
According to Eq. (6)
f ' f ' 
u ' = S 1 + α S  (19)
k 0  2 f y 
while W' can be calculated using Eq. (8) with fS and u replaced by fS' and u', respectively (see Figure 4)
f ' 2  1 α f S ' 
W'= S  + (20)
k 0  2 3 f y 
Finally, substituting Eqs. (19) and (20) into Eq. (18) gives

EH ' =
αf S ' 2 f S '
=
(
α f S − f S* )
2
f S − f S*
(21)
12k 0 fy 12k 0 fy

EQUIVALENT LINEAR SYSTEM

The equation of motion of a nonlinear SDOF system with hysteretic damping subjected to earthquake
ground motion is
mu&& + f S (u ) = −mu&&g (t ) (22)
where m is the mass of the system, u the relative displacement, fS(u) the restoring force, and üg (t) the
ground motion acceleration. The equation does not include a term representing viscous damping since for
a structural system built from traditional construction materials (e.g., steel, concrete) and without specially
installed viscous dampers the contribution of viscous damping to energy dissipation is negligible (e.g.,
Newark [2]). For the model considered above the relationship between the restoring force and the
displacement can be derived for the skeleton curve from Eq. (5)
fy  
f S (u ) = − 1 − 1 + 4αk 0 u sgn u  sgn u (23)
2α  fy 

and for an unloading/reloading curve from Eq. (6)
fy  2αk 0 
f S (u ) = f S* − 1 − 1 + (u − u *) sgn (u − u *) sgn(u − u *) (24)
α  fy 

The equation of motion of a linear SDOF system with viscous damping can be written as
 mu&& + cu& + ku = − mu&&g (t ) (25)
where c is the viscous damping coefficient and k the elastic stiffness of the system. To be able to compare
responses of the two systems to an earthquake ground motion "equivalence" between the systems needs to
be established, i.e., it should be defined how to select parameters of one of the systems when parameters
of the other system are given. A number of methods have been proposed for the replacement of a
nonlinear system with an "equivalent" linear system. The methods differ by the determination of two
fundamental parameters of the equivalent linear system – the equivalent damping ratio, ζe=ce/(2meωe), and
the natural frequency of vibrations, ωe=√ke/me (or the natural period of vibrations, Te=2π/ωe), where ce, me
and ke denote the viscous damping coefficient, mass and elastic stiffness of the equivalent linear system,
respectively. The methods can be divided into two groups – analytical methods based on harmonic loading
(e.g., Iwan [6]) and empirical methods (e.g., Gulkan [7], Iwan [8]).

In this study responses of the two systems are compared within elastic range, i.e., when the ductility ratio
µ=um/uy does not exceed unity (um denotes the maximum displacement of the system and uy a yield
displacement, i.e., the displacement corresponding to fy). The empirical methods have been developed for
yielding systems and equations of these methods to estimate Te and ζe are only applicable when µ>1. The
analytical methods can be used for equivalent linearization within the whole range of deformations. There
have been a number of studies on the performance of the different analytical methods for the linearization
of nonlinear systems subjected to earthquake loading. However, these comparative studies were limited to
the case of nonlinearity associated with yielding so that a nonlinear system was usually presented by a
simple bilinear model (e.g., Jennings [9], Iwan [6], Hadjian [10]). Thus, no data are currently available
which could give a clear indication what of the analytical linearization methods is the most suitable for the
purpose of the present study.

The secant stiffness method is the most commonly used among the analytical linearization methods (e.g.,
Miranda [11], Kwan [12]) and will be used herein. In this method the mass of the equivalent linear system
is taken as the mass of the original nonlinear system. The stiffness of the linearized system is determined
as the secant stiffness of the nonlinear system at the maximum displacement. Since in this study the
systems are compared only within elastic range, the maximum displacement is equal to the yield
displacement (i.e., um=uy). The secant stiffness at uy can be found from Eq. (5)
k
ke = 0 (26)
1+α
The natural frequency of vibration of the equivalent linear system is then
k ω0
ωe = e = (27)
m 1+α
where ω0=√k0/m can be considered as the initial instant natural frequency of the nonlinear system. The
equivalent damping ratio is found by equating the energy dissipated per cycle of harmonic vibrations with
amplitude of uy by the equivalent linear system with viscous damping and by the nonlinear system with
hysteretic damping. The energy dissipated per cycle by the equivalent linear system, ∆ED, is (e.g., Chopra
[5])
ω
∆E D = 2πζ e k e u 2y (28)
ωe
where ω is the exciting frequency. The energy dissipated per cycle by the nonlinear system (i.e., the area
enclosed by a hysteretic loop), ∆EH , can be obtained using Eq. (21)
4αf y2
∆E H = (29)
3k 0
Substituting Eqs. (5), (26) and (27) into Eq. (28) and equating it to Eq. (29) gives
2α ω0
ζe = (30)
3π (1 + α ) ω
32

In Eq. (30) damping of the nonlinear system is represented by the parameter α. However, damping is
usually measured by the loss factor, ξ, which is defined as fractional part of the strain energy, ESm (i.e.,
strain energy at the maximum displacement which herein equals uy), dissipated during one cycle of motion
and divided by 2π
1 ∆E H
ξ= (31)
2π E Sm
Finding ESm from Eq. (11) (when fS=fy), ∆EH from Eq. (29) and substituting that into Eq. (31) allows to
obtain the following relationship between α and ξ
3πξ
α= (32)
2(2 − πξ )
Substituting Eq. (32) into Eq. (30) results in
ξ 2 − πξ ω 0
ζe = (33)
(2 + πξ 2 )3 2 ω
As can be seen the equivalent damping ratio depends on the exciting frequency. Thus, in order to evaluate
the equivalent damping ratio for a nonlinear system subjected to an earthquake ground motion the
frequency content of this motion should be characterize with a single parameter. Several frequency
content parameters have been proposed (see Rathje [13]). In this study the frequency content will be
represented by the average period, Tav, which is defined using a fast Fourier transformation (FFT) of the
ground motion accelerogram as

Tav =
∑
C i (1 f i )
(34)
∑ Ci
where Ci's are the Fourier amplitudes of the accelerogram, and fi's the discrete Fourier transform
frequencies. Eq. (33) then becomes
ξ 2 − πξ Tav
ζe = (35)
(2 + πξ 2)3 2 T0
where T0=2π/ω0.

COMPARISON OF THE SYSTEMS

Responses of the two systems are compared for three earthquake records: (1) Pacoima, USA, 196o,
02/09/1971, Tav=0.3420 s; (2) El Centro, USA, 270o, 05/18/1940, Tav=0.4623 s; and (3) Mexico City –
Station 1, Mexico, 270o, 09/19/1985, Tav=1.6312 s (source of the records: NISEE, U.C. Berkeley, CA,
USA; the average periods were computed in this study). Parameters of the system response being
compared are: the maximum displacement, the maximum input energy, the dissipated energy, and the
maximum sum of the elastic strain energy and the kinetic energy. The maximum displacement obtained
for the nonlinear system is always equal to uy. This is achieved by adjusting the yield strength of the
system. For both of the systems the input energy, EI, is evaluated as
E I = − ∫ mu&& g u&dt (36)
and the kinetic energy, EK, as
 mu& 2
EK = (37)
2
For the nonlinear system the energy dissipated in hysteretic damping, EH, is estimated using Eqs. (12) and
(21), and the elastic strain energy, ES, by Eqs. (11) and (13). For the linear system the energy dissipated in
viscous damping, ED, is defined as
E D = ∫ c e u& 2 dt (38)
where ce=2ζemωe, and the elastic strain energy as
ku 2
ES = (39)
2
When the same symbol is used to denote a parameter, for example, like EI for the input energy, in order to
distinguish between the two systems in the following subscripts h and v will be used for the symbols
associated with the nonlinear system with hysteretic damping and the linear system with viscous damping,
respectively. For each of the earthquake records responses of the systems are compared for the initial
period of vibration of the nonlinear system, T0, ranging from 0.05 s to 3.0 s with period increments of 0.05
s.

In the first series of analyses the loss factor, ξ, representing damping of the nonlinear system, is set equal
to 0.05 which is a typical value for steel elements subjected to cyclic loading with the amplitude of fy (e.g.,
Lazan [3]). The equivalent damping ratio of the linear system is defined by Eq. (35), i.e., ζe changes
depending on T0. Results of the analyses are shown in Figure 5.

2.2 2.2
Pacoima Pacoima
1.8 El Centro 1.8 El Centro
Mexico City Mexico City
1.4 1.4
m,h

I,h
E /E
/u
m,v

I,v

1 1
u

0.6 0.6

0.2 0.2
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
Initial period of hysteretic system, T (s) Initial period of hysteretic system, T (s)
0 0

2.2 2.2
Pacoima Pacoima
1.8 El Centro 1.8 El Centro
Mexico City Mexico City
K+S,h

1.4 1.4
H
E /E

/E
D

K+S,v

1 1
E

0.6 0.6

0.2 0.2
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
Initial period of hysteretic system, T (s) Initial period of hysteretic system, T (s)
0 0

Figure 5. Ratios between response parameters of linear and hysteretic systems (ζe variable)

In the second series of analyses the damping ratio of the linear system, ζe, is assumed to be constant and
equal to 0.02. The corresponding value of the loss factor for the nonlinear system, ξ, found from Eq. (35)
for Tav/T0=1 is 0.0436 (α=0.1102). The aim of these analyses is to check if it is possible using a single
value of the damping factor (i.e., independent of the system period of vibration) to obtain a good
approximation of the response of a nonlinear hysteretic system to an earthquake ground motion. Results of
the analyses are presented in Figure 6.

2.2 2.2
Pacoima Pacoima
1.8 El Centro 1.8 El Centro
Mexico City Mexico City
1.4 1.4
m,h

I,h
E /E
/u
m,v

I,v
1 1
u

0.6 0.6

0.2 0.2
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
Initial period of hysteretic system, T (s) Initial period of hysteretic system, T (s)
0 0

2.2 2.2
Pacoima Pacoima
1.8 El Centro 1.8 El Centro
Mexico City Mexico City
K+S,h
1.4 1.4
H
E /E

/E
D

K+S,v

1 1
E

0.6 0.6

0.2 0.2
0 0.5 1 1.5 2 2.5 3 0 0.5 1 1.5 2 2.5 3
Initial period of hysteretic system, T (s) Initial period of hysteretic system, T (s)
0 0

Figure 6. Ratios between response parameters of linear and hysteretic systems (ζe constant)

According to the results presented in Figures 5 and 6 better agreement between responses of the systems is
observed when the damping ratio of the linear system was taken as a constant value, independent of the
period of vibration. Especially, this concerns the prediction of the maximum displacement. However, even
in this case at certain values of T0 the difference between the maximum displacements obtained for the
two systems exceeds 20%. Moreover, as can be seen in Figure 6 the ratios between the responses of the
systems fluctuate without any pattern. This shows that no matter what value of the damping ratio is
selected if it is constant (i.e., ζe, does not depend on the period of a nonlinear system) there are always
periods at which differences between the responses of the systems are significant. It can be also noted that
the largest differences are observed between the responses of the systems to the Mexico City earthquake
record, which has a very long average period.

CONCLUSIONS

Responses within yield limits (the ductility ratio did not exceed unity) of SDOF systems with viscous and
hysteretic damping to earthquake ground motions were compared. In many cases significant differences
between the peak displacements and energy values obtained for the hysteretic system and the equivalent
linear system were observed. This indicates that the prediction of the response of structures to earthquake
ground motions even within elastic range by using linear models with viscous damping may be far from
accurate and further research on this issue is needed.
 REFERENCES
1. SEAOC Vision 2000 Committee. "Performance-based seismic engineering." Report prepared by
Structural Engineers Association of California, Sacramento, CA, 1995.
2. Newmark NM, Rosenblueth E. "Fundamentals of earthquake engineering." Englewood Cliffs, NJ:
Prentice-Hall, 1992.
3. Lazan BJ. "Damping of materials and members in structural mechanics." New York: Pergamon
Press, 1968.
4. Ramberg W, Osgood WT. "Description of stress-strain curves by three parameters." Technical Note
902. NACA, 1943.
5. Chopra AK. "Dynamics of structures: theory and applications to earthquake engineering." Upper
Saddle River, NJ: Prentice-Hall, 2001.
6. Iwan WD, Gates NC. "Estimating earthquake response of simple hysteretic structures." Journal of
Engineering Mechanics Division ASCE 1979; 105(EM3): 391-405.
7. Gulkan P, Sozen M. "Inelastic response of reinforced concrete structures to earthquake motions."
ACI Journal 1974; 71(12): 604-610.
8. Iwan WD. "Estimating inelastic response spectra from elastic spectra." Earthquake Engineering and
Structural Dynamics 1980; 8(4): 375-388.
9. Jennings PC. "Equivalent viscous damping for yielding structures." Journal of Engineering
Mechanics Division ASCE 1968; 94(EM1): 103-116.
10. Hadjian AH. "A re-evaluation of equivalent linear models for simple yielding systems." Earthquake
Engineering and Structural Dynamics 1982; 10: 759-767.
11. Mirando E, Ruiz-Garcia J. "Evaluation of approximate methods to estimate maximum inelastic
displacement demands." Earthquake Engineering and Structural Dynamics 2002; 31: 539-560.
12. Kwan WP, Billington SL. "Influence of hysteretic behavior on equivalent period and damping of
structural systems." Journal of Structural Engineering ASCE 2003; 129(5): 576-585.
13. Rathje EM, Abrahamson NA, Bray JD. "Simplified frequency content estimates of earthquake
ground motions." Journal of Geotechnical and Geoenvironmental Engineering ASCE 1998; 124(2):
150-159.