Construction and Building Materials 185 (2018) 453–467

Contents lists available at ScienceDirect

Construction and Building Materials

journal homepage: www.elsevier.com/locate/conbuildmat

Monitoring interstory drift in buildings under seismic loading

using MEMS inclinometers
Shuang Hou a,⇑, Chuangshuo Zeng b, Haibin Zhang c, Jinping Ou c
a
State Key Laboratory of Subtropical Building Science, South China University of Technology, Guangzhou 510640, China
b
Department of Civil Engineering, The University of Mississippi, University, MS 38677, USA
c
School of Civil Engineering, Dalian University of Technology, Dalian 116024, China

h i g h l i g h t s

An approach for monitoring interstory drift using MEMS inclinometers was proposed.

Deformation mode of a column in different stages was studied.

An optimal sensor layout was designed.

The monitoring method was verified experimentally with acceptable error.

a r t i c l e i n f o a b s t r a c t

Article history: Accurate measurements of interstory drift are of great importance to the emerging performance-based
Received 17 February 2017 structural health monitoring (SHM). Previous studies have shown the potential of obtaining interstory
Received in revised form 5 July 2018 drift via rotation measurements. In the present study, an optimal layout of Micro Electro Mechanical
Accepted 11 July 2018
Systems (MEMS) inclinometers is proposed to improve the monitoring accuracy. First, detailed study
Available online 19 July 2018
on the deformation mode of a single column in both elastic and elastoplastic stages was carried out with
emphasis on the development of the plastic hinge in reinforced concrete (RC) columns. The optimal lay-
Keywords:
out for deploying inclinometers on a column was put forward with acceptable monitoring error antici-
Interstory drift
Inclinometer
pated using the deformation mode. Then, tests on three identical RC columns installed with
Plastic hinge inclinometers according to the optimal layout showed stable inclinometer performance and verified
Earthquake the feasibility of the proposed monitoring method by comparing with drift monitored by displacement
transducers. The test also verified the deformation mode in theoretical analysis and proved that the
development of plastic hinge length can roughly be monitored by the inclinometers. Finally, the influence
of various factors on the monitoring error was discussed in detail.
Ó 2018 Published by Elsevier Ltd.

1. Introduction updating. In contrast, the data-based approaches implements dam-

age diagnosis in SHM by establishing statistical models based upon
SHM is generally described as the process of implementing a experimentally acquired data. In addition, an emerging SHM
damage detection and characterization strategy for civil, aerospace approach, termed as performance-based SHM, utilizes
and mechanical engineering infrastructure, through which inher- performance-based earthquake engineering tools such as fragility
ent flaws or defects could be observed before they reach critical functions to correlate structural damage states and engineering
states [1–3]. Classical SHM algorithms are conducted using demand parameters (EDPs) [3]. By far the most commonly used
vibration-based methods which can be categorized in two groups: EDP in available fragility functions is the parameter called inter-
the model-based approaches and the data-based approaches [4,5]. story drift ratio (IDR) [5]. Interstory drift in a multistory building
The model-based approaches aim to detect damage by tracking is the relative translational displacement between two consecutive
variations in model and/or modal parameters (i.e., natural frequen- floors. The extended parameter, IDR, defined as the drift normal-
cies and mode shapes) via system identification and/or FE model ized by the corresponding story height, is also a universal indicator
for evaluating structural performance against seismic actions [3,6].
In seismic designs, limits on IDRs are used to guarantee structural
⇑ Corresponding author.
performance at acceptable deformation levels [7]. Studies have
E-mail address: [email protected] (S. Hou).

https://doi.org/10.1016/j.conbuildmat.2018.07.087
0950-0618/Ó 2018 Published by Elsevier Ltd.
454 S. Hou et al. / Construction and Building Materials 185 (2018) 453–467

shown that IDR correlates best with observed damage compared to adopted for bridge deformation monitoring. A team led by Hoult
other response quantities such as peak floor acceleration [8]. installed seven sensor nodes on a bridge to evaluate the potential
Therefore, accurate measurements of IDRs are of great importance of wireless sensor networks for use in bridge management [15].
to the success of the novel performance-based SHM. Among those nodes, three MEMS inclinometers were mounted
Interstory drift is most commonly measured by double integra- on two of the bridge’s bearing pads to measure any change in their
tion of measured accelerations. This can yield acceptable accuracy inclination with temperature. The lateral deformation of bridge
for linear responses, but its unstable performance during signifi- piles has also been monitored using MEMS inclinometers [16].
cant inelastic deformation makes it unsuitable in many applica- Six MEMS inclinometers were placed along the depth of each pile
tions [3]. A more direct, displacement-based approach employs a to measure their rotation under lateral loading applied at the pile
displacement sensor, typically a linear variable differential trans- cap. The deformation profile provided by the MEMS inclinometers
ducer (LVDT) with a wire diagonally strung across a bay of the agreed well with that provided by an automated laser survey. A
structure to be measured. But that method is impractical for build- group led by Ha has developed a wireless MEMS inclinometer
ings with numerous partition walls because of the obstructive monitoring system by assembling existing inclinometers (SCA
setup. And it is impossible to completely eliminate the influence 103T) for monitoring the structural health of civil engineering
of wire slack and thus dynamic interaction, making the small- and building structures. The reliability and the accuracy of the
amplitude response of an LVDT unreliable [3,9]. There are also method were verified in a tilt measurement experiment [17].
some non-contact methods which have been applied in laboratory Applications of MEMS inclinometers to underground structures
experiments. One involves a laser crosshair projection technique have so far been studied only in the laboratory. Using a wireless
first proposed by Bennett and Batroney [10] and later improved sensing network, the relative deformation of a long tunnel can be
by Chen and his colleagues [11]. That method detects any change measured with MEMS inclinometers to yield an accuracy of
of the positions of two consecutive floors from the output of pho- 0.05°, acceptable for tunnel structures [18]. The application for a
todetectors which receive light from a crosshair beam generator. It buried pipeline had also been tested, and the influence of sensor
performs well in monitoring interstory drift and chord drift, but spacing and load distribution on monitoring errors has been dis-
only within a limited range. A semi-automated method has been cussed [19].
proposed to rapidly measure post-disaster structural damage by Generally inclinometers are pendulum-based and can be classi-
applying close-range photogrammetry surveying techniques and fied into three categories: solid-mass pendulum, liquid pendulum
augmented reality visualization. The interstory drift can be and gas pendulum [20–22]. All the pendulum-type inclinometers
obtained by comparing the shifted position of each floor after an are designed based on the influence of the gravitational force to
earthquake with the outer geometry of the building before the the pendulum [20,22,23]. In this paper, a MEMS solid-mass pendu-
earthquake if it is available from the building’s owners or a govern- lum inclinometer that uses capacitance as transduction mecha-
ment database. Indoor experiments have shown that the measure- nism was adopted for experiment. Its internal structure is
ment accuracy is highly related to optical-sensing, positioning illustrated as a simplified analytical model in Fig. 1. A tiny mass
techniques and the resolution of the camera. Deformation during block inserted between two electrodes is connected with the base
an earthquake is not obtainable. This approach is suitable for dam- of the inclinometer through an elastic beam. The mass block is
age diagnosis of civil structures following seismic events but not regarded as a rigid body that no deformation occurs during rota-
applicable for monitoring drift during an earthquake since an tion. As the inclinometer rotates, the gravity of the block bends
absolute reference is absent [12]. the elastic beam and changes the distances between the block
In addition to the approaches discussed above, previous studies and the electrodes. As a result, the capacitance between the two
carried out by Cheung and Kiremidjian indicate the possibility of electrodes and the output voltage changes accordingly. Therefore,
estimating residual drift in single-column structures using rotation the rotation of the inclinometer can be detected by the change of
measurements [13]. This idea derives from the common view that output voltage.
the deflection of a structural member can be calculated by inte- Using MEMS inclinometers for interstory drift measurements
grating the rotation along its longitudinal direction. To extend this offers several advantages over the existing approaches. First, per-
idea to accurate measurements of interstory drift in more complex meant installations of sensors are finished during construction.
structures, however, an appropriate sensor layout should be This avoids difficult and/or repetitive setup which is common in
designed based on the deformation mode of the corresponding some contact (e.g., LVDT measurements) and noncontact (e.g.,
investigated structure. Furthermore, successful implementation vision-based measurements) approaches. Second, environmental
of the rotation-based approach for interstory drift measurement effects such as temperature change and noise can be mitigated
also relies on accurate measurements of rotations. Although gyro- since the sensors are embedded within structural members.
scopes are ideal for tilt measurement due to its stable performance Finally, the MEMS inclinometers measure rotation by taking the
in the presence of high shock and vibration, they are too expensive
to be extensively used in civil engineering structures. In this paper,
a pendulum-type MEMS inclinometer is suggested to be used for
rotation measurements on a single column within building
structures.
Inclinometers have long been widely applied to monitor
deformation of bridges, embankments, landslides and so on. As
micro-electronic mechanical system technology has matured,
MEMS-based inclinometer has drawn attention in civil engineering
because of its small size, high sensitivity, low power consumption
and unobtrusive installation. Today, however, it is still new and its
application is still relatively rare. Soil stability monitoring using
MEMS inclinometers for highway embankments and landfill slopes
has been reported in which MEMS inclinometers with working
range of ±15°, a resolution of 0.009°, and long-term stability of less
than 0.004° were used [14]. MEMS inclinometers have also been Fig. 1. Internal structure of a solid-mass pendulum inclinometer.
 S. Hou et al. / Construction and Building Materials 185 (2018) 453–467 455

direction of gravity as the reference which is unchangeable under In elastic stage, a single column is assumed to be subjected to a
any circumstances. However, noncontact methods using remote pair of shear forces and a pair of bending moments at its two ends
sensing technologies generally require reference frames which to simulate the real loading conditions. The free body diagram of
might be unobtainable for some building structures. this column and its internal forces is shown in Fig. 4. Through inte-
To summarize, obtaining accurate interstory drift could help gration the rotation and displacement distribution along the col-
advance performance-based SHM since IDR is by far the best indi- umn height can be easily obtained.
cator to correlate with observed damage levels of structures. Exist- Furthermore, the rotation angle of the column along the vertical
ing measurement methods are not qualified for engineering direction is also determined once the interstory drift ratio, say a1,
applications. To accurately obtaining interstory drifts via rotation is known.
measurements, this paper presents a new approach using MEMS
6a1  
inclinometers and proposes an optimal sensor layout. Firstly, the hðxÞ ¼ 2
lx  x2 ð3Þ
deformation modes of a single column in elastic and elastoplastic l
stages were analyzed theoretically. A design method for the opti- The lateral displacement along the vertical direction is
mal deployment of sensors, therefore, can be obtained with accept-
a1  
able measurement error anticipated using the deformation mode. yðxÞ ¼ 2
3lx2  2x3 ð4Þ
Three identical RC columns were then tested to verify the feasibil- l
ity of the method, and the monitoring results from the MEMS incli- When the column enters the elastoplastic stage, plastic hinges
nometers were compared with direct LVDT measurements. Finally, emerge at both ends. The analysis of plastic hinge length (lp) is
the monitoring accuracy was quantified, and the sources of widely accepted as a key step in estimating or measuring column
monitoring error will be discussed in terms of the evolution of deformation [24–26]. The plastic hinge of a member can be consid-
the nonlinear behavior of the structure. ered as the region that experiences large inelastic curvature and
over which severe damage is observed. The plastic hinge of a RC
2. Monitoring principles column consists of three zones: a bar yielding zone, a concrete
crushing zone and a curvature concentration zone where the cur-
For a shear-type structure as shown in Fig. 2, its interstory drift vature increases rapidly along the height [27]. This study empha-
can be calculated by integrating the rotation of a column over its sized the curvature concentration zone because it is the most
length [13]. closely related to rotation-based interstory monitoring. For ease
Z of calculation, the elastic deformation is assumed to be constant
y¼ hðxÞdx ð1Þ throughout the elastoplastic stage, that is, ye,u = a1,ul. So the
newly-developed deformation has two components: yp caused by
the rotation within the plastic hinge region, and yr caused by rigid
where h is the angle at any point along the column in the vertical
body rotation of the elastic segment, as shown in Fig. 3.
direction and x is the distance from the point to the base of the col-
In fact, determining the rotation distribution within the plastic
umn. The column’s rotation can be captured by a set of inclinome-
hinge region is not straightforward due to the great scatter in the
ters and the interstory drift can be expressed as
curvature distribution along the length caused by the large disper-
X
n1
ðhi þ hiþ1 Þ sion of concrete material [28]. It should be noted that the rotation
y¼ D xi ð2Þ of the plastic hinge accounts for most of the total deformation
i¼1
2
when the column enters the elastoplastic stage. So correctly mon-
where hi is the rotation angle measured by the ith inclinometer and itoring the development of rotation within the plastic hinge region
Dxi is the distance between the ith and the (i + 1)th inclinometer. is vital when measuring interstory drift. That makes it necessary to
Deploying the inclinometers involves a prior understanding of investigate the distribution of rotation within the plastic hinge
the potential rotation gradient of the column and its nonlinear region and to design a reasonable sensor deployment pattern.
behavior. This requires study of the column’s deformations in the The well-known plastic hinge concept proposed by Park and
elastic and elastoplastic stages. Deformation modes at the elastic Paulay expresses the ultimate inelastic deformation area as is
and elastoplastic limits are shown in Fig. 3. The IDRs of a column shown in Fig. 5(a) [29]. Wu’s group has reported typical curvature
which is fixed at both ends at the elastic and elastoplastic limits distributions for cantilever RC columns under different lateral
are a1,u and a2,u respectively. The final plastic hinge length lp of forces and defined the plastic hinge length as the distance from
the column is a0l, where l is the length of the column and a0 is the column’s bottom to the fixed point dividing the linear part
the ratio of lp to l. and nonlinear part of the curvature distribution [28]. As shown
in Fig. 5(a), increasing lateral force results in a curvature distribu-
tion with a larger amplitude in plastic hinge region, while the
curvature distribution outside of plastic hinge region is linear
and keeps to a straight line.
Based on the characteristics of plastic hinges and the curvature
distribution model presented, a simplified plastic hinge deforma-
tion model of a half-scale column shown in Fig. 5(b) is proposed.
The half-scale column was adopted for analysis since the deforma-
tion of an integral column is supposed to be symmetrical about its
middle section when its two ends are regarded as fixed. The curva-
ture distribution along the height can be simplified to a trilinear
curve consisting of an elastic part and a plastic part. lpa represents
the height below which plastic curvature has developed. As indi-
cated by the simplified model, the curvature of the plastic hinge
tapers linearly from the column’s bottom along its height until
lpp is reached, while the elastic part of the curvature distribution
Fig. 2. Deformation diagram for interstory drift in a shear-type structure. remains the same.
456 S. Hou et al. / Construction and Building Materials 185 (2018) 453–467

Fig. 3. Deformation modes of a single column.

Fig. 4. Analytical model of a column: (a) Free body diagram; (b) Shear force diagram; (c) Moment diagram.

The curvature distribution can thus be expressed as and the total displacement is
 
/ðxÞ ¼ /y ðxÞ þ /p ðxÞ 2 up;u 3 up;u 2  
(   /    y ¼ a1;u l þ up;u lpa  lpa þ up;u lpa  lpa l  2lpa ¼ a2 l
/y;u 1  2xl  p;u x  lp 0 6 x 6 lpa ð5Þ 3lp 2lp
lp
¼    
/y;u 1  2xl lpa < x 6 0:5l ð8Þ

6a where a2 is the interstroy drift ratio in the elastoplastic stage.

where /y;u ¼ l1;u is the ultimate yield curvature.
When the plastic hinge region fully develops, that is lpa=lp,
Integrating the curvature along the height yields the rotation
distribution 1 1
a2;u l ¼ a1;u l þ /p;u lp l  /p;u l2p ð9Þ
8 6a1;u   u   2 3
< 2 lx  x  2lp;up x2 þ up;u 0 6 x 6 lpa
2

h ð xÞ ¼ 6 a l  ð6Þ So
: 1;u lx  x2   up;u l2 þ u lpa lpa < x 6 0:5l
l2 2lp pa p;u
a2;u l  a1;u l
/p;u ¼ 2
ð10Þ
The displacement distribution is 1
l l
2 p
 13 lp
8 a1;u   u u  
To estimate the value of lp, many researchers have developed
>
> 3lx2  2x3 þ 2p;u x2  6lp;up x3 0 6 x 6 lpa
>
< l2
a1;u   u 2 u 3
empirical models for the plastic hinge length based on numerous
y ð xÞ ¼ 2 3lx2  2x3 þ 2p;u lpa  6lp;up lpa ð7Þ experiments. Bae and Bayrak have conducted a detailed study of
>
>  l

: þ u l  up;u l2 x  l 
> 
lpa < x 6 0:5l
 the influence of axial load, the shear span-to-depth ratio, and the
p;u pa 2lp pa pa
amount of longitudinal reinforcement on lp through a serious of
 S. Hou et al. / Construction and Building Materials 185 (2018) 453–467 457

Fig. 5. Typical and simplified plastic hinge modes.

RC column tests [24]. They also concluded several typical models of normally about 1/12 to 1/8 of l. Therefore, lp can be taken as
plastic hinge length shown in Table 1, and compared the measured 3/32 l (i.e. a0 = 3/32). According to China’s Code for Seismic Design
results from their tests with the calculated results using the mod- of Buildings (GB 50011-2010) [30], IDR in the elastic stage should
els to examine the accuracy of the predicted lp using a new expres- be limited to a1,u = 1/550 and in the elastoplastic stage to a2,u =
sion proposed by them. Both the experimental results and the 1/50. Here, a2,u = 1/25 was taken for calculation considering extre-
models gave lps mostly ranging from 0.5 h to 1.0 h under a variety mely large column deformation patterns.
of conditions. Therefore, 0.75 h was adopted in this study for pre- The proposed plastic hinge development model yields the rota-
dicting the plastic hinge length. The height of the cross-section is tion profiles at different drift levels along the height shown in
Fig. 6. The maximum rotation angles at the elastic limit and the
elastoplastic limit are 0.16° and 2.4° respectively. It can be seen
Table 1 that at the elastic limit the rotation varies smoothly along the
Empirical models for plastic hinge length. vertical direction, while in the elastoplastic stage there is drastic
variation in the rotation angle within the plastic hinge region.
Author Plastic hinge length (lp)
Baker  z 0:25
lp ¼ k1 k2 k3 d ð1956Þ [31]
d
 0:25 3. Designing an optimal layout
lp ¼ k1 k2 k3 dz d ð1956Þ [32]
  0 qffiffiffiffiffiffiffi
Mattock  pffiffi 
lp ¼ 2d 1 þ 1:14 dz  1 1  qq q
d
16:2 [33] Given the features of this deformation mode, inclinometers
b

lp ¼ þ 0:05z (simplified) [34]

d should be densely placed in the plastic hinge region and more spar-
2
Corley lp ¼ þ 0:2 pdffiffiz [35]
d sely placed in the elastic region. Within each region the inclinome-
2
Park et al. lp ¼ 0:08l þ 6db [36] ters should be distributed uniformly. The deployment layout is
Sheikh and Khoury lp ¼ 1:0h (under high axial load) [37] presented in Fig. 7, where m inclinometers are equally spaced
h     i
Bae lp ¼ 0:3 PP0 þ 3 AAgs  0:1 l þ 0:25h [24] along the column’s elastic region and n inclinometers are equally
spaced in each of the pre-assumed plastic hinge regions.
458 S. Hou et al. / Construction and Building Materials 185 (2018) 453–467

Elastic limit Elastoplastic limit respectively, and the vertical distance between them is xAB. The
angle between line AB and the vertical direction is hAB. The vertical
1/550) ( 4%)
1.0
deformation between A and B could be ignored assuming small
deformation, so the lateral deformation between A and B can
Elastic region approximately be expressed as
0.8 Plastic region
yAB  xAB hAB ð11Þ
Height (l)

0.6
hAB is estimated by the mean value of hA and hB, and the result-
1% 2% 3%
0.4 ing error is
 B
hAB  hA þh
0.2 d¼ 2
ð12Þ
hAB
0.0 Using Eq. (3), the rotations at points A and B are
0 1 2
Drift ratio Rotation ( )  
hA ¼ 6a1 aA  a2A ð13Þ
Fig. 6. Rotation profile evolution. and
 
hB ¼ 6a1 aB  a2B ð14Þ
respectively. Thus, the mean value of hA and hB is
hA þ hB   
¼ 3a1 aA þ aB  a2B þ a2A ð15Þ
2
Using Eq. (4), the lateral displacements at point A and B are
 
yA ¼ a1 3a2A  2a3A ð16Þ
and
 
yB ¼ a1 3a2B  2a3B ð17Þ
respectively. Thus, the angle between the line AB and the vertical
direction is
yAB yB  yA   
hAB ¼ ¼ ¼ a1 3ðaA þ aB Þ  2 a2A þ a2B þ aA aB ð18Þ
xAB aB  aA
hAB is actually unobtainable and is approximated by the mean
value of hA and hB, hence the absolute error is
Fig. 7. Instrumentation layout of MEMS inclinometers.  
hA þ hB
hAB  ¼ a1 ðaB  aA Þ2 ¼ a1 Da2 ð19Þ
2
Moreover, an optimal layout for inclinometers requires the The absolute monitoring error of the lateral displacement
study on minimizing the monitoring error caused by the disparity between points A and B is
between the real and the measured rotation profile. The deforma-  
tion between any two adjacent inclinometers is presented hA þ hB 3
hAB  xAB ¼ a1 ðaB  aA Þ ¼ a1 Da3 ð20Þ
schematically in Fig. 8. The angles at points A and B are hA and hB 2

Fig. 8. Schematic deformation between two adjacent inclinometers.

 S. Hou et al. / Construction and Building Materials 185 (2018) 453–467 459

A single column installed with m inclinometers along the verti- the sensors’ signals can also be influential. So m = 8 is suggested
cal direction are divided into m + 1 segments with lengths denoted here. For a typical column with height of 3 m, the distance between
by Da1, Da2. . . Dam+1, so the absolute error of lateral deformation sensors is about 333 mm.
of the entire column is In the elastoplastic stage the plastic hinge propagates from the
ends of the column as the interstory drift increases. Since accurate
X
mþ1
a1 Da31 þ a1 Da32    þ a1 Da3mþ1 ¼ a1 Da3i ð21Þ prediction of the real length of the plastic hinge is difficult, the dis-
i¼1 parity between the real length and the pre-assumed length should
be considered in estimating the monitoring errors. The ratio of the
Using Jensen’s inequality, real value of lp to the pre-assumed value was assumed to be 0.5, 1.0
a1 Da31 þ a1 Da32    þ a1 Da3mþ1 or 1.5 for considering under-, proper- or over-estimation of the
 3 plastic hinge length, respectively. The number n was assumed to
X
mþ1
1 a1
¼ a1 Da3i P a1 ðm þ 1Þ ¼ ð22Þ be 1, 2 or 3. The deployments of inclinometers within the pre-
i¼1
mþ1 ðm þ 1Þ2 assumed plastic hinge region are presented in Fig. 10. By compar-
ing the drift ratio deduced from measured rotations using Eq. (2)
which means inclinometers should be equally spaced along the col- and the given real drift ratio, the monitoring error can be obtained.
umn height to minimize the monitoring error. So the minimum rel- First, the number of inclinometers deployed within the plastic
ative error of the interstory drift is hinge region was assumed to be 3 and the monitoring error varia-
a1 1 tions with different actual values of lp, were studied, see Fig. 11(a).
¼ ð23Þ The actual plastic hinge length was then supposed to be fixed at 1.0
ðm þ 1Þ2 a1 ð m þ 1Þ 2
lp and the monitoring error variations with different values of n
The relationship between monitoring error and number m in were compared, see Fig. 11(b). The two figures show that the mon-
the elastic stage is shown in Fig. 9. It can be seen that when m is itoring error increases rapidly as the column deformation enters
larger than 4, the monitoring error can be controlled down to 5% the elastoplastic stage, then deceases gradually as the drift ratio
that is acceptable from the perspective of engineering. More incli- increases further. Fig. 11(a) shows clearly that once enough sensors
nometers than predicted should probably be deployed because are installed within the plastic hinge region, the length of the ulti-
some other error sources such as installation error and noise in mate plastic hinge has little influence on the amplitude of any
monitoring error. However, it can be seen from Fig. 11(b) that
25 under the assumption of a correctly-estimated plastic hinge length,
too few sensors deployed in the plastic hinge region can greatly
20 reduce the monitoring accuracy (e.g. the maximum monitoring
error when n = 1 increases to 7.6%). So for the sake of an acceptable
monitoring accuracy, inclinometers should be densely distributed
15
Error (%)

in the plastic hinge region. For a general column with estimated

lp of 300 mm at both ends, the sensor distance is 100 mm, close
10 acceptable limit to the seismic requirement for stirrup spacing at column ends.
MEMS inclinometers thus can easily be fixed on the nearby stir-
5 rups. So n = 3 is suggested by these results.
It is worth nothing that the two IDR limits used here might be
different from real cases because they could be influenced by var-
0 ious parameters (e.g., steel ratio, concrete strength and column’s
0 2 4 6 8 10
length). It is possible that the actual IDR exceeds 4% if the struc-
Sensor number m
ture is subjected to extreme peak ground acceleration. However,
Fig. 9. Relationship between monitoring error and the number of inclinometers m the difference between assumed and actual IDR limits has little
for the elastic stage. effect on the accuracy. Firstly, Eq. (23) indicates that within the

Fig. 10. Inclinometer layout in the plastic hinge region.

460 S. Hou et al. / Construction and Building Materials 185 (2018) 453–467

2.5 8 n=1
n=2
0.5 lp 7 n=3
2.0
1.0 lp 6
the actual plastic hinge lenth
is assumed to be 1.0 lp
1.5 lp 5
1.5

Error (%)
Error (%)

Elastic limit 4
1.0 3

2
0.5 the number n is assumed to be 3
1
Elastic limit
0.0 0
0 1 2 3 4 0 1 2 3 4
Drif Ratio (%) Drift Ratio (%)
(a) (b)
Fig. 11. Monitoring error vs. drift ratio with: (a) different plastic hinge lengths; (b) different sensor numbers.

elastic stage, the monitoring error is dominated by the number of dominated and contributes more and more to the total deforma-
measured points on the column. Secondly, for a single column tion of the column.
undergoing large deformation in elastoplastic stage, the displace-
ment caused by rigid-body rotation of the elastic segment, 4. Experimental program
namely yr shown in Fig. 3, makes up most of the column’s defor-
mation. Thus, the monitoring error will not be significant as long 4.1. Specimen details
as the plastic hinge rotation is measured accurately, which is
guaranteed by densely-distributing sensors within pre-assumed In order to verify the feasibility of using MEMS inclinometers
plastic hinge region. Moreover, as IDR continues to increase after for interstory drift monitoring, a series of loading tests on RC col-
large inelastic deformation, decreasing monitoring error is umns were carried out. Three identical concrete columns denoted
expected because the portion of rigid-body displacement has C1, C2 and C3 were tested. The dimensions and reinforcement

Fig. 12. Details of the test specimens.

 S. Hou et al. / Construction and Building Materials 185 (2018) 453–467 461

layout of the columns are shown in Fig. 12. Each specimen con- on two parallel thin steel wires using plastic straps. The steel wires
sisted of a 1600 mm  400 mm  400 mm column cast integrally were then soldered to the adjacent stirrup. Inclinometers were
with a 1500 mm  650 mm  600 mm stub at the bottom, and a deployed along the central axis of the column where the influence
700 mm  650 mm  400 mm cap at the top for convenient load- of concrete damage and reinforcement yielding was assumed to be
ing. Each tested column represents the lower half part of a com- the least. Fig. 15 shows the test system.
plete column in a frame structure, and each stub represents a As recommended by the discussion of lp expression in the pre-
footing which could be regarded as a rigid body in an earthquake. vious section, lp was assumed to be 3/4 of the height of the
Eight HRB 400 steel bars (Hot-rolled Ribbed Bar, fyk = 400 cross-section (i.e. 300 mm) for each column, and four inclinome-
N/mm2) with a diameter of 22 mm were used as longitudinal rein- ters (Inc1–4) were placed in the pre-assumed plastic hinge region,
forcement, and HRB 335 steel bars 10 mm in diameter (fyk = 335 with Inc1 placed at the base of the column to measure the rigid
N/mm2) were used as stirrups with a spacing of 100 mm in the body rotation of the stub. For convenience of installation, the spac-
densely-distributed region. The concrete strength for all specimens ings between two adjacent inclinometers are 130 mm, 100 mm
was designed to be grade C30. Concrete cubes cured in the same and 100 mm from the bottom up. Based on the analysis presented
environment as the specimens were also taken to monitor the in Section 3, for a half-scale column, two MEMS inclinometers are
strength, and the average measured cube strength was 33.5 MPa needed for the rotation measurement in the elastic region. In fact,
after 28 days of curing. in this experiment four were used (Inc5–8) to be conservative. The
distances between inclinometers from Inc4 to Inc8 were 250 mm,
4.2. Measurement and test setup 300 mm, 250 mm and 300 mm. The top of the column housed Inc8.
Two LVDTs were mounted on a steel frame to monitor the lat-
To simulate real seismic loading, all of the columns were tested eral displacement of the column top relative to the column base
under constant axial load combined with reserved cyclic lateral to obtain the column deformation, which were assumed to give
force, as shown in Fig. 13. A constant axial force of 800 kN was benchmark results for checking the accuracy of monitoring results
applied first with an axial load ratio of 0.22 on the top surface of based on inclinometers.
the column cap using a hydraulic jack to simulate the dead load
in a real structure. The reversed cyclic lateral displacement load 5. Test results and discussions
was applied through a servo-hydraulic test machine (MTS High-
Force servo hydraulic test system, Model 311.41, USA). 5.1. Structure response
The MEMS inclinometer adopted was manufactured using a SCA
100T chip developed by Murata Electronics Oy (Vantaa, Finland) as The drift ratios of the three columns as measured by the LVDTs
the sensing element. The deployment of the sensors is shown in are presented in Fig. 16. Deformation control was used during the
Fig. 13. Each column was fitted with eight MEMS inclinometers cyclic loading with increasing amplitudes. The LVDT measure-
denoted by Inc1 to Inc8 for rotation measurement. The inclinome- ments were normal except for Column3 during the first 150 s,
ters were 25 mm  25 mm  10 mm, as shown in Fig. 14(a). Before when the core wire of the LVDT somehow became stuck in one
casting the concrete, the inclinometers were pre-positioned in the direction. The maximum drift ratios for Column1, Column2,
rebar case as shown in Fig. 14(b). Each inclinometer was first fixed and Column3 were 2.2%, 3.3%, 3.1%, respectively. The

Fig. 13. Details of the sensor deployment.

462 S. Hou et al. / Construction and Building Materials 185 (2018) 453–467

(a) Sensor dimensions (b) Installation

Fig. 14. Details of the MEMS inclinometer.

(a) Loading test set-up (b) Data acquistition

Fig. 15. Test system.

force-deformation hysteresis loops of the three columns are shown The plastic hinge regions of the three columns are shown in
in Fig. 17. The hysteresis loops were plump, indicating good ductil- Fig. 18. The lengths of the plastic hinges lie in the range of
ity of the three columns. The maximum horizontal thrust was 130 to 230 mm, within which only one or two MEMS inclinome-
about 220 kN for each column. A vertical load of 800 kN was main- ters were placed apart from the one at the bottom of the
tained throughout the loading. column.

2
(a) Column 1
0

-2
0 20 40 60 80 100 120 140 160 180 200
Drift ratio (%)

2 (b) Column 2
0

-2

0 50 100 150 200 250

2 (c) Column3
0

-2 LVDT stuck

0 50 100 150 200 250

Time (s)

Fig. 16. Drift ratios monitored by LVDT.

 S. Hou et al. / Construction and Building Materials 185 (2018) 453–467 463

Fig. 17. Lateral force-displacement hysteresis loops of: (a) Column1; (b) Column2; (c) Column3.

(a) Column 1 (b) Column 2 (c) Column 3

Fig. 18. The plastic hinge regions observed after tests.

All of the inclinometers worked well except Inc2 of Column 2 The relative rotations within the pre-assumed plastic hinge
which failed during installation. The rotation angles measured by region (i.e. Inc1–4) and the relative rotation between Inc4 and
Inc2 to Inc8 for the three columns are shown in Fig. 19. The max- Inc8 are presented in Fig. 21. The figure shows that the plastic
imum value was about 2°. hinge developed mostly in the area between Inc1 and Inc2 and
Column rotation profiles at different peak deformation levels to a much smaller extent in the area between Inc2 and Inc3 if
are shown Fig. 20. It shows that after the elastic limit a plastic the drift ratio was larger than 2%. But it rarely developed in the
hinge developed in a pattern consistent with the results of the ana- area between Inc3 and Inc4. As has been discussed, the plastic
lytical model shown in Fig. 6. After the column deformation hinge can be thought of as a region where large inelastic curvature
entered the elastoplastic stage, the additional rotation concen- occurs. Fig. 21(a) and (c) show that the relative rotation between
trated in the plastic hinge region. Inc2 and Inc3 increased rapidly as the drift ratio exceeded 2%,

Inc2~Inc8
1
(a) Column1
0

-1

0 50 100 150 200

2 Inc3~Inc8
1
(b) Column 2
Rotation ( )

0
-1
-2
0 50 100 150 200 250

2 Inc2~Inc8
1 (c) Column 3
0
-1
-2
0 50 100 150 200 250 300
Time (s)

Fig. 19. Rotation along the vertical direction.

464 S. Hou et al. / Construction and Building Materials 185 (2018) 453–467

drift ratio (%) drift ratio (%) drift ratio (%)

1500 0.36 0.81 1.36 1.95 0.23 0.76 1.27 1.89 2.46 3.01 0.16 0.62 1.14 1.70 2.25 2.79

2.21 3.28 3.06

1000
Height (mm)

Inc2~Inc8

Inc3~Inc8

Inc2~Inc8
500

0
0 0.5 1.0 1.5 0 0.5 1.0 1.5 2.0 0 0.5 1.0 1.5 2.0
Rotation ( ) Rotation ( ) Rotation ( )
(a) Column 1 (a) Column 2 (a) Column 3
Fig. 20. Column rotation profiles at different deformation levels.

1.0 1.5
Inc2-Inc1 Inc3-Inc1 Inc2-Inc1
Relative rotation ( )

Inc3-Inc2 Inc4-Inc3 1.0 Inc3-Inc2

Inc4-Inc3 1.0 Inc8-Inc4 Inc4-Inc3
Inc8-Inc4 Inc8-Inc4
0.5
0.5
0.5

0.0 0.0 0.0

0 1 2 0 1 2 3 0 1 2 3
Drift ratio (%) Drift ratio (%) Drift ratio (%)
(a) Column 1 (b) Column 2 (c) Column 3
Fig. 21. Relative variation of the angle with deformation level.

LVDT Inclinometer
2
1 (a) Column 1
0
-1
-2

0 20 40 60 80 100 120 140 160 180 200

3.0
Drift ratio(%)

1.5 (b) Column 2

0.0
-1.5
-3.0
0 20 40 60 80 100 120 140 160 180 200 220 240 260
3.0
1.5 (c) Column 3
0.0
-1.5
-3.0
0 20 40 60 80 100 120 140 160 180 200 220 240 260
Time (s)

Fig. 22. Drift ratios measured by the MEMS inclinometers and LVDTs.

which means the large curvature had propagated to the region nometers’ output. Based on the development of relative rotation
between those two inclinometers. The development of plastic between inclinometers, the approximate ultimate lengths of
hinge length can therefore be roughly monitored using the incli- the plastic hinges for columns 1 and 3 were 0.35 h and 0.5 h
 S. Hou et al. / Construction and Building Materials 185 (2018) 453–467 465

respectively, which is smaller than predicted. The monitoring of concrete and yielding of the steel bars, the increased deforma-
results conform well to the observation to columns’ exterior dam- tion concentrated in the plastic hinge region. In the theoretical
age as shown in Fig. 18. analysis the transition stage was not considered, and that is where
The growth rate of the relative rotation between Inc4 and Inc8 the difference comes from. In addition, the P-D effect caused by the
dropped dramatically as the drift ratio came to exceed 1.5% to 2%. axial force and the increase in the drift ratio could also have led to
The plastic hinge region had apparently developed to the extent the large deformation in the upper part of column at the beginning
that increased deformation of the plastic hinge made up almost of the elastoplastic stage when the plastic hinge region was not
all of the newly-developed column deformation. This is different seriously damaged.
from the theoretical assumption that the deformation outside the
plastic hinge region remains the same in the elastoplastic stage.
The discrepancy can be explained by the fact that the bottom part 5.2. Error analysis
of the column was far from severely damaged when the deforma-
tion first entered the elastoplastic stage, so the effect of the plastic The accuracy of the drift ratios obtained from rotation monitor-
hinge was not very significant. However, with the further crushing ing was evaluated by comparing them with those from the LVDT
measurements. Fig. 22 shows the monitored drift ratio given by
the MEMS inclinometers and the LVDTs. The waveforms from the
10 two measurements appear consistent except for the cases where
the LVDT failed. The monitoring errors in the drift ratios from the
Column 1
inclinometers for the three columns are presented in Fig. 23. It
8 Column 2
shows that the monitoring error increased significantly as soon
Column 3
as the columns entered the elastoplastic stage and then decreased
6 gradually with further increases in the drift ratio. The maximum
Error (%)

monitoring errors for Columns 1, 2 and 3 are 5.9%, 8.7%, and 8.1%
respectively, with a mean value of 7.6%. The monitoring error
4
development shows a pattern similar to that indicated in the the-
oretical analysis but with greater amplitudes, though still in an
2 acceptable range.
The larger amplitude of monitoring error in experiment might
0 have two causes. First, the plane cross-section assumption adopted
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 in formulating the theoretical model might be invalid due to yield-
Drift Ratio (%) ing of the steel bars and the crushing of concrete in the plastic
hinge region. As a result, the rotation monitored by the inclinome-
Fig. 23. Variation of the monitoring error with column drift ratio. ters deployed in the plastic hinge region would be inaccurate, since

7 10 10
6
8 8
5
Error (%)

4 6 6
3 4 4
2
All used All used All used
2 2
1 Inc5 & Inc7 not used Inc5 & Inc7 not used Inc5 & Inc7 not used
0 0 0
0.0 0.5 1.0 1.5 2.0 0 1 2 3 0 1 2 3
Drift Ratio (%) Drift ratio (%) Drift ratio (%)
(a) Column 1 (b) Column 2 (c) Column 3
Fig. 24. The influence of the number of inclinometers deployed in the elastic region on monitoring error.

10 10 12

8 8 10
Error (%)

8
6 6
6
4 4
All used 4 All used
Inc2 not used Inc2 not used Inc2 not used
2 2 2
Inc2 & Inc3 not used Inc2 & Inc3 not used Inc2 & Inc3 not used

0 0 0
0.0 0.5 1.0 1.5 2.0 0 1 2 3 0 1 2 3
Drift ratio (%) Drift ratio (%) Drift ratio (%)
(a) Column 1 (b) Column 2 (c) Column 3
Fig. 25. The influence of the number of inclinometers in the plastic hinge region on monitoring error.
466 S. Hou et al. / Construction and Building Materials 185 (2018) 453–467

they were installed along the central axis of the column. The theo- crushing were ignored in the theoretical analysis. The sub-
retical analysis suggests that the accuracy of the plastic hinge rota- stantial scatter in the curvature distribution was also inade-
tion measurements is vital to the monitoring error. Then, as quately represented in the simplified analytical deformation
discussed in Section 2, the curvature distribution in the plastic model.
hinge region has a lot of scatter while in the theoretical model it (4) Too few inclinometers deployed in the elastic region
has been simplified. That would be an important difference adversely affected the monitoring quality within the small
between the theoretical rotation model and the actual situation drift ratio range, more so than at larger deformations. On
leading to differences between the deduced monitoring error and the other hand, sparse deployment of inclinometers in the
that experimentally observed. Second, the experimental results plastic hinge region leads to greater errors when the drift
show clearly that the curvature distribution is much more concen- ratio is small than when the deformation is large.
trated in the region below Inc2 than predicted by the theoretical
analysis. Thus the rotation development below Inc2 was not accu- The present study has proved the feasibility of using MEMS
rately monitored. Too few sensors were deployed in that region inclinometers to monitor interstory drift in a single-column exper-
based on the theoretical model. iment with the optimal sensor layout. It’s noteworthy that imple-
The influence of the number of inclinometers deployed in both menting the method to single-story structures or more complex
the elastic region and the plastic hinge region on the monitoring structures could further test its applicability and accuracy. Future
error was also studied. For the elastic region, Inc5 and Inc7 are studies should focus on clarifying the relationship between mea-
assumed to be missing for drift monitoring, and the monitoring sured interstory drifts and damage states, which is crucial to the
errors of the three columns in that case are shown in Fig. 24. If emerging performance-based SHM approaches. In addition, the
Inc5 and Inc7 were not used, the monitoring error would be signif- increasingly mature wireless sensors network could greatly reduce
icantly greater, especially in the small drift ratio range. The mean the cost and thus potentially actualize wide application of the pro-
value of the maximum errors of three columns increases by 20.8%. posed method.
For the plastic hinge region, two cases were considered. One
assumed that Inc2 was not used and the other that both Inc2 and
Conflict of interest
Inc3 were eliminated. The variations of monitoring errors for the
three columns are shown in Fig. 25. Significant increases were found
The authors declared that there is no conflict of interest.
in both cases, especially in the large drift ratio range. In the two
cases, the mean maximum error of the three columns would
increase by 19.6% and 45.0% respectively. So skimping on incli- Acknowledgements
nometers in the plastic hinge region leads to lower monitoring accu-
racy in the elastoplastic stage, though it has less influence in the Joint financial supports from the National Science Foundation of
elastic stage, as predicted in the analysis of Section 3, see Fig. 11(b). China (Grant No. 51261120376) and the National Basic Research
Program of China (973 Program, Granted No. 2015CB057704) are
gratefully acknowledged.
6. Conclusions
References
This paper proposed an optimal layout of MEMS inclinometers
to improve the accuracy of monitoring interstory drift in building [1] C.R. Farrar, K. Worden, Structural Health Monitoring: A Machine Learning
Perspective, first ed., Wiley, 2012.
structures via rotation measurements. The interstory drift can be [2] M. Radzieński, Ł. Doliński, M. Krawczuk, M. Palacz, Damage localisation in a
obtained through the integration of measured rotations. A column stiffened plate structure using a propagating wave, Mech. Syst. Signal Process.
in a typical shear-type structure was represented in a simplified 39 (2013) 388–395.
[3] D.A. Skolnik, J.W. Wallace, Critical assessment of interstory drift
plastic hinge development model, and the monitoring error of this
measurements, J. Struct. Eng. 136 (2010) 1574–1584.
method throughout the whole nonlinear deformation process was [4] R. Barthorpe, On Model- and Data-based Approaches to Structural Health
analyzed using different sensor deployment layouts and different Monitroing (Ph.D. thesis), The University of Sheffield, 2015.
[5] H.Y. Noh, D.G. Lignos, K.K. Nair, A.S. Kiremidjian, Development of fragility
pre-assumed plastic hinge lengths. The optimal sensor layout
functions as a damage classification/prediction method for steel moment-
was developed from theory and validated experimentally. Three resisting frames using a wavelet-based damage sensitive feature, Earthq. Eng.
identical column specimens under combined axial load and Struct. Dyn. 41 (2011) 681–696.
reserved cyclic lateral force were tested to simulate real earth- [6] J. Shan, H.T.Y. Yang, W. Shi, D. Bridges, P.K. Hansma, Structural damage
diagnosis using interstory drift-based acceleration feedback with test
quake loading conditions, with MEMS inclinometers and LVDTs validation, J. Eng. Mech. 139 (2013) 1185–1196.
deployed. The following conclusions can be drawn from the [7] Building Seismic Safety Council (BSSC), Prestandard and commentary for the
experiment. seismic rehabilitation of buildings, FEMA-356, Washington, D.C., 2000.
[8] B.B. Algan, Drift and Damage Consideration in Earthquakeresistant Design of
Reinforced Concrete Buildings (Ph.D. thesis), University of Illinois at Urbana-
(1) Fixing the inclinometers on two parallel thin steel wires sol- Champaign, 1982.
dered to the nearby stirrup was proved to provide reliable [9] D.A. Skolnik, W.J. Kaiser, J.W. Wallace, Instrumentation for structural health
monitoring: measuring interstory drift, in: 14th World Conference on
readings throughout the whole deformation process. The Earthquake Engineering, Beijing, China, 2008.
performance of the MEMS inclinometers was found to be [10] K.D. Bennett, C.B. Batroney, Laser crosshair projection technique for interstory
immune to the influence of factors such as concrete cracking drift measurement, in: SPIE Fiber Opt. Laser Sensors XIV, Denver, CO, 1996, pp.
302–310.
and crushing or rebar bucking.
[11] W. Chen, K.D. Bennett, J. Feng, Y.-P. Wang, S. Huang, Laser technique for
(2) Deploying four inclinometers in the pre-assumed plastic measuring three-dimensional interstory drift, in: SPIE 3555, Opt. Fiber Opt.
hinge region and four in the elastic region yielded acceptable Sens. Syst., 1998, pp. 305–310.
[12] F. Dai, S. Dong, V.R. Kamat, M. Lu, Photogrammetry assisted measurement of
errors in the measured drift ratio. The errors increased sig-
interstory drift for rapid post-disaster building damage reconnaissance, J.
nificantly as the column deformation entered the elastoplas- Nondestruct. Eval. 30 (2011) 201–212.
tic stage, but decreased gradually with the further increases [13] A. Cheung, A.S. Kiremidjian, Development of a rotation algorithm for
in the drift ratio. The average maximum error was 7.6%. earthquake damage diagnosis, Earthq. Spectra. 30 (2013) 1381–1401.
[14] S. Swaminathan, A.J. Muck, P.W. Chou, Real-time slope and wall monitoring
(3) The monitoring error differed from that predicted by the and reporting using 3-D MEMS-based, in-place instrumentation system,
theoretical analysis because steel bar yielding and concrete GeoFlorida 2010 Adv. Anal. Model. Des., ASCE, Orlando, FL, 2018.
 S. Hou et al. / Construction and Building Materials 185 (2018) 453–467 467

[15] N.A. Hoult, P.R.A. Fidler, P.G. Hill, C.R. Middleton, Long-term wireless structural [26] O. Youssfa, M.A. ElGawady, J.E. Mills, Displacement and plastic hinge length of
health monitoring of the ferriby road bridge, J. Bridg. Eng. 15 (2010) 153–159. FRP-confined circular reinforced concrete columns, Eng. Struct. 101 (2015)
[16] M. Abu-Farsakh, S. Yoon, D. Ha, W.A. Marr, Z. Zhang, X. Yu, Development of a 465–476.
substructure instrumentation system at the new I-10 twin span bridge and its [27] X.M. Zhao, Y.F. Wu, A.Y.T. Leung, Analyses of plastic hinge regions in
use to investigate the lateral behavior of batter piles, Geotech. Test. J. 34 reinforced concrete beams under monotonic loading, Eng. Struct. 34 (2012)
(2011) 332–343. 466–482.
[17] D.W. Ha, H.S. Park, S.W. Choi, Y. Kim, A wireless MEMS-based inclinometer [28] C. Jiang, Y.F. Wu, G. Wu, Plastic hinge length of FRP-confined square RC
sensor node for structural health monitoring, Sensors 13 (2013). columns, J. Compos. Constr. 18 (2014).
[18] B. He, Y. Ji, R. Shen, Wireless inclinometer for monitoring deformation of [29] R. Park, T. Paulay, Reinforced Concrete Structures, Wiley, 1975.
underground tunnel, Opt. Precis. Eng. 21 (2013) 1464–1471 (in Chinese). [30] Ministry of Housing and Urban-Rural Development, GB 50011-2010: Code for
[19] X. Peng, S. Li, Safety monitoring of buried pipeline subjected to external Seismic Design of Buildings, 2010 (in Chinese).
interference using wireless inclinometers, in: Int. Conf. Pipelines Trenchless [31] A.L.L. Baker, The Ultimate-Load Theory Applied to the Design of Reinforced &
Technol., ASCE, Xiamen, China, 2014, pp. 13–22. Prestressed Concrete Frames, 1st ed.,., Concrete Publications Ltd, 1956.
[20] Y.G. Lee, K.K. Jang, D.H. Kim, C.G. Kim, Development of a mirror mounted fiber [32] A.L.L. Baker, A.M.N. Amarakone, Inelastic hyperstatic frames analysis, in: SP-
optic inclinometer, Sens. Actuators A Phys. 184 (2012) 46–52. 012 Flexural Mech. Reinf. Concr., Farmington Hills, MI, 1964.
[21] L. Tang, K. Zhang, S. Chen, G. Zhang, G. Liu, MEMS inclinometer based on a [33] A.H. Mattock, Rotational capacity of hinging regions in reinforced concrete
novel piezoresistor structure, Microelectronics J. 40 (2009) 78–82. beams, in: SP-012 Flexural Mech. Reinf. Concr., Farmington Hills, MI,
[22] D. Welch, G. Julius, J.B. Christen, Fully differential current-mode MEMS dual- 1964.
axis optical inclination sensor, Sens. Actuators A Phys. 192 (2013) 133–139. [34] A.H. Mattock, Discussion of ‘‘Rotational capacity of hinging regions in
[23] F.S. Alves, R.A. Dias, J.M. Cabral, J. Gaspar, L.A. Rocha, High-resolution MEMS reinforced concrete beams”, J. Struct. Div. 93 (1967) 519–522.
inclinometer based on pull-in voltage, J. Microelectromechanical Syst. 24 [35] W.G. Corley, Rotational capacity of reinforced concrete beams, J. Struct. Div. 92
(2015) 931–939. (1966) 121–146.
[24] S. Bae, O. Bayrak, Plastic hinge length of reinforced concrete columns, ACI [36] M.J.N. Priestley, R. Park, Strength and ductility of concrete bridge columns
Struct. J. 105 (2008) 290–300. under seismic loading, ACI Struct. J. 84 (1987) 61–76.
[25] O. Bayrak, S.A. Sheikh, Plastic hinge analysis, J. Struct. Eng. 127 (2001) 1092– [37] S.A. Sheikh, S.S. Khoury, Confined concrete columns with stubs, ACI Struct. J. 90
1100. (1993) 414–431.