Chapter 3

Formulation and Validation

2.1 Formulation of Basic Force-Displacement Relations of Negative Stiffness Device

In this section basic force-displacement relation for the negative stiffness device are derived.
These equations of the device are valid for certain conditions or assumptions so that the
resulting algebraic equations are simple enough for use in analysis and design. The following
assumptions are made:

1. Frame members of the negative stiffness device are rigid

2. All elements are mass-less
3. Hysteresis in the joints of the device is negligible
4. Friction in joints is neglected.

Analysis of Negative Stiffness Device

Consider that the device is installed in a structure and its top is subject to a known lateral
displacement u as shown in Fig.. The following is a derivation of the forces produced by the
device in the direction of the imposed displacement.

Consider the free body diagram of the pivot plate shown in Figure. The forces acting on the
pivot plate (Fb, Fc, Fs) are shown. The figure also shows the GSA force Fg which does not
act on the pivot plate. the free body diagram of the bottom chevron is given in Figure.
Additional information on the NSD deformed shape is provided in Figure.
In addition to the main assumption listed above, the following assumptions are made:

The lever is assumed to have zero rotation and, therefore, the lever force acts in the horizontal
direction. This assumption also leads to point B and A having the same displacement. In
reality, the lever rotates; the rotation being very small when the lever long.
The height loss of the device due to its inverted pendulum motion is ignored. This height loss,
however small, results in small vertical displacement of point E, additional small lever
rotation and some small loss in the spring pre-load.

When a displacement u is imposed on the top of the device, the lever causes the pivot
plate to rotate and point B moves horizontally by the same amount as the imposed
displacement.
Moreover, point E, which is rigidly connected to the top channel through the top chevron,
Undergoes, the imposed displacement u. Point D moves horizontally in the direction
opposing u.
These three conditions are written:

uB  u
uE  u
l1
u D  u (4-1)
l2

Where, l2 is the distance from point C to point B and l1 is the distance from
point C to point D
Points B and D move down and up, respectively, by:

vB  l2  l22  u 2

2
 l 
vD  l1  l   u 1 
1
2
(4-2)
 l2 
The vertical distance between points E and D

2
 l 
vDE  l p  vD  l p  l1  l   u 1 
1
2
(4-3)
 l2 

Where lp is the length of the pre-compressed spring in the un-deformed configuration. The
spring length in the deformed configuration is obtained by using Equations (4-1), (4-2) and
(4-3)
 2
  l1  
2 2
 2 l1 
ls  l p  vD  l p  l1  l1   u 
2
 u 1   (4-4)
  l2    l2 
 

Taking moments about point C of the pivot plate in Figure 4-1, one can calculate the force at
B (this is the axial force in the lever):

Fs   l   u 
FB   cos  s  u 1   Fs sin  s  1   l1  (4-5)
l22  u 2   l2   l2  

Where θs is the inclination angle of the spring and Fs is the force of the pre-compressed
spring.
This force varies with displacement and is given by:
Fs  pin  ks (ls  l p ) (4-6)

Where pin is the pre-compression force of the spring (a positive value) and Ks is the stiffness
of the pre-compressed spring. Note that the force in the spring maximizes when the lateral
displacement is zero.
The sine and cosine of the spring inclination angle are given by:
u  l1 
sin  s  1  
ls  l2 

  l1  
2
1
cos  s  l p  l1  l1   u 
2
(4-7)
ls   l2  

The total force produced by the device, exclusive of the GSA force, is the force acting on
point C plus the horizontal component of forces in the double hinged columns. From vertical
equilibrium in Figure 4-3, the vertical component of the axial load in the double hinged
columns is equal to the vertical component of the spring force.

The horizontal component of the axial load in the double hinged columns is given by:
1 1 u
FLX  Fs cos  s tan   Fs cos  s (4-8)
2 2 h
Where θ is the inclination angle of the double hinged columns.
The total force exerted by the NSD, inclusive of the GSA force, is given by:
FNSD   Fc  2 FLx  Fg (4-9)
where Fg is the GSA force and FC is the horizontal force at point C as calculated from
horizontal equilibrium of the pivot plate and given by:
Fc  FB  Fs sin s (4-10)
Substitutions of Equations (4-5) to (4-8) and (4-10) into (4-9) and after some algebra yields
the total NSD force as:
 
h  l p  l1   1  l22  u 2
l
 Pin  K s l p  l1   l p  l1 l2  l2  
FNSD    K s    2    u  Fg (4-11)
 ls   l2   l22  u 2 l1 h 
 
The force-displacement of the gap spring assembly (GSA) is given by:
 ks1u, 0  u  u 'y 
 
Fg   ' 
ks1u y  k  k  u  u y 
ks 2 ks1 (4-13)
'
u  uy 
'

 s2 s1 
Validation

The experimental work done by A.A. Sarlis et al. is considered to validate the analytical
model of the negative stiffness device. In their experiment, the 3-storey steel frame was taken
to study the changes in the different parameters of the structure by attaching the negative
stiffness device as an isolator. For this, the given structure with and without NSD was
subjected to different ground motion histories and then the parameters such as acceleration,
displacement, storey drift and base shear were observed. In the present validation work, the
same structure has been modelled in the FEM software package (SAP 2000) and attempt has
been made to validate the analytical model of the NSD by comparing the different parameters
of the structure obtained analytically with the same parameters observed in the experiment.

Description of Experimental Model

The 3-storey model is supported on four low damping elastomeric bearings. The
superstructure is a moment resisting frame in the longitudinal and braced frame in the
transverse direction. Five concrete blocks, each weighing 8.9 kN, were installed at each floor
and two more at the base. The total weight of the model (frame, base and added weight) on
top of the isolators was 196 kN (distributed as 53.2 kN at the base and 47.6 kN at each floor)
in the tests without the NSD and it was 201 kN in the tests with the NSD (the added 5 kN was
due to weight of the NSD connection components). All beams and columns are S3 x 5.7 (SI
designation S75 x 8.5) and all braces are L 1½ x 1½ x 1½ (SI designation L38 x 38 x 6.4). the
beam to column connections are fully welded and stiffened so that they are rigid. Horizontal
bracing of all floors at all bays achieves, together with the concrete blocks, rigid diaphragm
behaviour. The 3-story structure seats on a base-mat that consists of a grid of two longitudinal
W14x90 beams (SI designation W360 x 134) and four transverse W12 x 35 beams (SI
designation W310 x52), which are located at the superstructure’s column locations. Also, the
model features two HSS16x8x5/16 (SI designation HSS406.4x203.2x7.9) beams in the
transverse direction that are connected on the top of the W14 x 90 beams. These beams were
used to connect two NSD to the superstructure at the base-mat and the shake table.

Four isolators were placed below the W14x90 beams on a 122cmx244cm footprint as shown
in Figure 5-3 and Figure 5-4. The yellow plates seen at the bottom of the isolator-load cell
assembly in Figure 5-1 and Figure 5-4 were used to level the bearings and to raise them so
that the gravity loads on each isolator were approximately equal.

Two linear viscous dampers were installed in the isolation system between the shake table
and the base of the structure inclined at 36o with respect to the horizontal plane and 28o with
respect to a vertical plane as shown in Figure 5-4. The dampers could be easily connected and
disconnected during testing.

In the test, three configurations were tested. Configuration E is for the structure isolated at
base with elastomeric bearings only. Configuration ENB-LA is for the structure isolated by
elastomeric bearings and the negative stiffness device without GSA. Configuration EDNB-
LA is for the structure isolated by elastomeric bearings and negative stiffness device
accompanied with viscous dampers.

Two NSD were connected to the shake table and to the HSS16x8x5/16 beams of the base
mat. The connection to the shake table was through an angle on the side of the table as shown
in Figure 5-6 and Figure 5-7. Stiffeners were welded to the angle in order to prevent bending
deformations. Four load cells between the NSD and the angle were used to measure the force
of the NSD.

S N

E
Description of Analytical Model

The superstructure, fixed at its base, was modelled in program SAP2000 using linear elastic
frame elements for all beams, columns and braces. The diaphragm bracing of the
superstructure was explicitly modelled and therefore no diaphragm constraints have been
assigned. The concrete blocks were modelled as lumped masses without mass moment of
inertia. The self-weight of the frame was explicitly captured using the steel density value for
the material in SAP2000. Additional small masses were added at the base-mat to capture the
difference in the total weight calculated by the program and the one obtained from
measurement by the load cells. This additional weight was contributed by elements not
accounted for in the model, such as steel connecting plates, stiffeners, bolts and connection
angles. Due to the large dimensions of the base-mat beams compared to the superstructure
elements, rigid beam elements have been used to connect the bottom of the columns to the
centreline of the W14x90 beams of the base-mat. Rigid offsets have not been used for any
beam-to-column connection in the rest of the structure. Table 7-1 presents results for the
modal properties of the model, fixed at the base, as obtained by program SAP2000 for the
first three modes. The damping ratio is the value assigned for each mode in SAP2000 for the
construction of the inherent damping matrix. Note that the assigned damping ratio values are
between the values identified in the experiments and presented in Table 6-4 and 6-5. There is
reasonably good agreement between the mode shapes and period values obtained in the
experimental identification (Table 6-4 and 6-5) and the results of the modal analysis in
SAP2000.

Modelling of elastomeric bearing

The elastomeric bearing is modelled by connecting five different link elements in parallel to
represent the exact non-linear behaviour. The behaviour of each link is represented by its
force-displacement relation in Fig. xx. There are 4 bearings in the experimental model. Table
XX presents values of the model parameters for each of the four bearings that were identified
from tests conducted on individual bearings by A.A Sarlis et al.
 Table xx : Properties of five elements representing each elastomeric bearing.

Property Bearing location on the shake table

NE NW SE SW
G1 (Multi-Linear Elastic Element)
Engagement displacement (cm) 2.36 1.60 1.37 1.83

Stiffness after engagement (kN/cm) -0.51 -0.53 -0.65 -0.56

G2 (Multi-linear elastic element)

Engagement displacement (cm) 5.56 5.84 6.10 5.72

Stiffness after engagement (kN/cm) 0.58 0.39 1.24 0.77

H1 (Wen Element)
Elastic stiffness (kN/cm) 175 175 175 175
Yield force 0.089 0.089 0.089 0.089
Yielding exponent 1 1 1 1
H2 (Wen Element)
Elastic stiffness (kN/cm) 1.52 1.52 1.52 1.52
Yield force 1.34 1.34 2.23 1.20
Yielding exponent 0.27 0.27 0.27 0.27
S1 (Linear elastic spring)
Horizontal Stiffness 2.87 3.00 3.08 3.15
1. The effective stiffness of all elements is zero except for S1 which should have a very small value
2. The vertical stiffness of all elements except for H1 is zero. For H1, the vertical stiffness is the vertical
stiffness of the bearing, equal to 964kN/cm
3. Element G1 has negative stiffness value and therefore negative force for positive displacement
Modelling of viscous damper

In a program SAP2000, the linear viscous dampers were modelled as link damper elements
together with their extenders as inclined elements in space exactly as installed in
experimental model. The damping coefficient of viscous damper was taken as 0.63 kN
sec/cm. This coefficient was obtained in the displacement-controlled test conducted by A.A
Sarlis.

Modelling of NSD

The negative stiffness device was modelled by multilinear elastic link element. For this
element, the input parameters are the force and the corresponding displacement and the
direction in which the given force-displacement relation is valid. The force displacement
relation for the NSD can be obtained using the analytical model developed in the section xx.
The force-displacement relation was developed for the specifications of the NSD mentioned
in the table xx. The NSD of these specifications was used in the test. The force-displacement
curve for these specifications is represented in Fig. xx. The link was then attached as a 1-joint
link at the mid-point in the longitudinal direction of the base mat made up of W360 x 134.
The analytical model for both configuration i.e with GSA and without GSA were modelled in
SAP 2000.

NSD properties used in the experimental study

Quantity Symbol Value

Length BC of pivot plate l1 25.4 cm
Length CD of pivot plate l2 12.7 cm
NSD spring stiffness Ks 1.4 kN/cm
NSD spring preload Pin 16.5 kN
Double hinged column height h 124.5 cm
Lever length llv 67.3 cm
NSD engagement u’y 1.65 cm
displacement
GSA spring S1 stiffness ks1 4.9 kN/cm
GSA spring S2 stiffness ks2 0.3 kN/cm
GSA spring S2 preload Pis2 8.1 kN
Loading and Analysis

The model was tested for the ground motion history of PS-10317 component of Denali
Alaska earthquake. The scaled peak ground acceleration was chosen as 0.32 g in the test. The
magnitude of this earthquake was 7.9. For the test, the original earthquake motions were
compressed in time by a factor of 4 in consistency with length scale factor of 2. The ground
motion was applied only in the North-South direction in the experiment and in the positive x
direction (U1) in the analytical model developed in SAP 2000. For the purpose of time
history analysis in SAP 2000, the ground motion data of PS-10317 was taken from PEER
(Pacific Earthquake Engineering Research Center). The accelerogram of PS-10317 from
PEER was already baseline corrected. In SAP 2000, the non-linear time history analysis was
performed with the time scale factor of 0.25 and length scale factor of 2. In the analysis
options, translational degrees of freedom in x and z direction and rotational degrees of
freedom about y axis were set as the active degrees of freedom.

0.2

0.15

0.1

0.05
Acceleration [g]

-0.05

-0.1

-0.15

-0.2

-0.25

-0.3
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
Time [sec]

Results and comparison

In this section, the results of the experimental test and the results obtained from analytical
model are compared. The fundamental time period obtained from the test is 0.299 sec and the
computed period in the analytical model is 0.291 sec for the fixed base model.

Displacement response
In the test conducted by A.A Srlis, it was observed that the displacement of the structure with
negative stiffness device and elastomeric bearings is much higher than the structure with only
elastomeric bearings. The displacements were reduced by the addition of viscous damper in
the isolation system. The displacements computed in SAP 2000 and the displacements
observed in the test are presented in table xx. The displacements computed in the analytical
model are in good agreement with the displacements observed in the test. Displacement
response against the time duration of PS 10317 for E (Structure + Elastomeric Bearings),
ENB-LA (Structure + Elastomeric Bearings + NSD), EDNAB-LA (Structure + Elastomeric
Bearings + NSD + Viscous Damper) configuration obtained from SAP 2000 is represented in
Fig.xx to Fig.xx.

Displacements of the storeys for configuration E

Analytical Experimental Error
Base Displacement (mm) 57.998 58.00 0.003%
1st Storey Displacement 61.438 61.200 0.385%
2nd Storey Displacement 65.281 65.320 0.059%
3rd Storey Displacement 67.506 68.614 1.615%

Displacements of the storeys for configuration ENB-LA

Analytical Experimental Error
Base Displacement (mm) 67.943 68.900 1.389%
1st Storey Displacement 70.730 70.547 0.259%
2nd Storey Displacement 72.750 73.383 0.863%
3rd Storey Displacement 74.408 75.213 1.071%

Displacements of the storeys for configuration EDNB-LA

Analytical Experimental Error
Base Displacement (mm) 43.126 43.100 0.060%
1st Storey Displacement 45.161 44.747 0.925%
2nd Storey Displacement 47.535 46.851 1.459%
3rd Storey Displacement 49.006 48.590 0.856%
Acceleration Response

The acceleration was reduced by the addition of NSD in to the isolation system. Further by
the addition of viscous damper in to the isolation system reduced the acceleration slightly.
The test results and the analytical results are represented in the table xx. The plots of
acceleration and the time duration of PS 10317 for the storeys are represented in Fig.xx to
Fig.xx for all three configurations.

Accelerations of the storeys for configuration E

Analytical Experimental Error
Base Acceleration (mm) 3.273 3.041 7.625%
1st Storey Acceleration 3.381 3.335 1.367%
2nd Storey Acceleration 3.536 3.532 0.124%
3rd Storey Acceleration 3.725 3.728 0.075%

Accelerations of the storeys for configuration ENB-LA (NSD)

Analytical Experimental Error
Base Acceleration (mm) 2.397 2.452 2.263%
1st Storey Acceleration 2.313 2.354 1.758%
2nd Storey Acceleration 2.603 2.649 1.725%
3rd Storey Acceleration 2.767 2.747 0.735%

Accelerations of the storeys for configuration EDNB-LA (ANSS)

Analytical Experimental Error
Base Acceleration (mm) 2.158 2.243 3.929%
1st Storey Acceleration 1.962 2.100 7.033%
2nd Storey Acceleration 2.256 2.300 1.936%
3rd Storey Acceleration 2.551 2.596 1.780%

Base Shear

According to the test, the base shear was reduced in the case of the structure with NSD and
elastomeric bearing when compared to the structure with only elastomeric bearings. Further
the base shear was reduced by the addition of viscous damper in to the isolation system. The
results of test and the analytical model in SAP 2000 are represented in table xx. Also the plots
of base shear against the time duration of PS 10317 are represented in the fig.xx to fig.xx for
all three configurations.

Base Shear (kN)

Configuration Analytical Experimental Error
E 66.263 65.6 1.011%
ENB-LA 43.300 44.626 3.062%
EDNB-LA 34.600 35.306 2.040%

Displacement
Peak Centers of Displacement
50 E-PS10317
Displacement (mm)

-50
-57.99835

0 5 10 15 20 25 30
Time (sec)

Base Displacement
 Displacement
Peak Centers of Displacement

50
E-PS10317

Displacement (mm)

-50
-61.43875

0 5 10 15 20 25 30
Time (sec)

1st Storey Displacement

Displacement
Peak Centers of Displacement
50 E- PS 10317
Displacement (mm)

-50
-65.2814

0 5 10 15 20 25 30
Time (sec)

2nd Storey Displacement

 Displacement
Peak Centers of Displacement
50 E-PS 10317
Displacement (mm)

-50
-67.50605

0 5 10 15 20 25 30
Time (sec)

3rd Storey Displacement

66.26308

Base Shear
Peak Centers of Base Shear
50 E-PS 10317
Base Shear (kN)

-50

0 5 10 15 20 25 30
Time (sec)

Base Shear
 4
Acceleration

3.27339
Peak Centers of Acceleration
E-PS 10317
Acceleration (m/sec2) 2

-2

-4
0 5 10 15 20 25 30
Time (sec)

Base Acceleration

4
Acceleration
3.38176

Peak Centers of Acceleration

Acceleration (m/sec2)

-2

-4
0 5 10 15 20 25 30
Time (sec)

1st Storey Acceleration

 Acceleration
4 Peak Centers of Acceleration

3.5366
Acceleration (m/sec2) 2

-2

-4
0 5 10 15 20 25 30
Time (sec)

2nd Storey Acceleration

Acceleration
3.7255

4 Peak Centers of Acceleration

Acceleration (m/sec2)

-2

-4
0 5 10 15 20 25 30
Time (sec)

3rd Storey Acceleration

 3
Acceleration

2.31347
Peak Centers of Acceleration
2 ENB-LA PS 10317
Acceleration (m/sec2)

-1

-2
0 5 10 15 20 25 30
Time (sec)

Base Acceleration

3
2.39674

Acceleration
Peak Centers of Acceleration
2 ENB-LA PS 10317
Acceleration (m/sec2)

-1

-2
0 5 10 15 20 25 30
Time (sec)

1st Storey Acceleration

 2.60308
Acceleration
Peak Centers of Acceleration
2
ENB-LA PS 10317
Acceleration (m/sec2)

-2
0 5 10 15 20 25 30
Time (sec)

2nd Storey Acceleration

4
Acceleration
2.7669

Peak Centers of Acceleration

ENB-LA PS 10317
Acceleration (m/sec2)

-2

0 5 10 15 20 25 30
Time (sec)

3rd Storey Acceleration

 50

44.62627
45 Base Shear
40 Peak Centers of Base Shear
35
ENB-LA PS 10317
30
Base Shear (kN) 25
20
15
10
5
0
-5
-10
-15
-20
-25
-30
0 5 10 15 20 25 30
Time (sec)

Base Shear

Displacement
Peak Centers of Displacement
ENB-LA PS 10317
50
Displacement (mm)

-50
-67.94341

0 5 10 15 20 25 30
Time (sec)

Base Displacement
 Displacement
Peak Centers of Displacement
50
ENB-LA PS 10317
Displacement (mm)

-50
-70.73043

0 5 10 15 20 25 30
Time (sec)

1st Floor Displacement

Displacement
Peak Centers of Displacement
50 ENB-LA PS 10317
Displacement (mm)

-50
-72.7499

0 5 10 15 20 25 30
Time (sec)

2nd storey displacement

 Displacement
Peak Centers of Displacement
50 ENB-LA PS 10317

Displacement (mm)

-50 -74.40709

0 5 10 15 20 25 30
Time (sec)

3rd Storey Displacement

50
Displacement
Peak Centers of Displacement
EDNB-LA PS 10317
25
Displacement (mm)

-25
-43.12614

-50
0 5 10 15 20 25 30
Time (sec)

Base Displacement
 Displacement
Peak Centers of Displacement
25
EDNB-LA PS 10317
Displacement (mm)

-25
-45.16152

-50
0 5 10 15 20 25 30
Time (sec)

1st Storey Displacement

Displacement
40
Peak Centers of Displacement
Displacement (mm)

20

-20
-47.53523

-40

0 5 10 15 20 25 30
Time (sec)

2nd Storey Displacement

 40
Displacement
Peak Centers of Displacement
EDNB-LA PS 10317
Displacement (mm)

20

-20

-40
-49.00578

-60
0 5 10 15 20 25 30
Time (sec)

3rd Floor Displacement

Acceleration
2.2429

Peak Centers of Acceleration

2 EDNB-LA PS 10317
Acceleration (m/sec2)

-2

0 5 10 15 20 25 30
Time (sec)

Base Acceleration
 Acceleration

2.10061
2 Peak Centers of Acceleration
EDNB-LA PS 10317
Acceleration (m/sec2)

-1

-2
0 5 10 15 20 25 30
Time (sec)

1st Floor Acceleration

2.30045

Acceleration
Peak Centers of Acceleration
2 EDNB-LA PS 10317
Acceleration (m/sec2)

-2

0 5 10 15 20 25 30
Time (sec)

2nd Floor Acceleration

 2.59567
Acceleration
Peak Centers of Acceleration
2 EDNB-LA PS 10317
Acceleration (m/sec2)

-2

0 5 10 15 20 25 30
Time (sec)

3rd Floor Acceleration

40
Base Shear
35.30573

Peak Centers of Base Shear

EDNB-LA PS 10317
20
Base Shear (kN)

-20

-40
0 5 10 15 20 25 30
Time (sec)

Base Shear