Content uploaded by Gaston A. Fermandois
Author content
All content in this area was uploaded by Gaston A. Fermandois on Sep 29, 2021
Content may be subject to copyright.
17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 27th to October 2nd, 2021
CLIENT-SERVER APPLICATION FOR MULTI-PLATFORM
COORDINATION IN REAL-TIME HYBRID SIMULATION TESTING
G. Fermandois(1), D. Mera(2)
(1) Assistant Profesor, Departamento de Obras Civiles – Universidad Técnica Federico Santa María, Santiago, Chile,
gaston.fermandois@usm.cl
(2) M.S. Student, Departamento de Obras Civiles – Universidad Técnica Federico Santa María, Santiago, Chile,
diego.mera@sansano.usm.cl
Abstract
Real-Time Hybrid Simulation (RTHS) is an experimental technique with a physical and computational interface to
evaluate complex structural systems. This technique is based on the substructuring methodology in which difficult-to-
model components are tested in an experimental setup. Meanwhile, structural components and materials whose behavior
can be predicted are modeled numerically. The main feature of RTHS is that all calculations, measurements, and signal
processing must be carried out in brief time intervals, typically less than one millisecond. The literature reveals that simple
and linear numerical models are used most of the time and are implemented directly in the microcontroller unit for real-
time execution. The reason being most research prioritized the study of compensation methods. However, these methods
have already reached sufficient maturity, and working with more complex numerical substructures is imperative to
achieve realistic results. Thus, it is essential to integrate the capabilities of finite element analysis with the control software
used in the experimental setup of RTHS tests. Unfortunately, the lack of detailed information on existing
integration/coordination methods for RTHS testing makes the development and implementation of these coupled systems
difficult, especially for researchers who just start working on RTHS.
This study’s motivation is to provide a reasonable and detailed procedure to achieve the coupling of general FE software
with the control software in RTHS testing. Computational tasks are interconnected using a client-server communication
approach. First, the client is chosen as a Matlab/Simulink program to implement the control algorithms for RTHS.
Meanwhile, the servers are multiple OpenSees programs that simulate either numerical or experimental substructures.
Examples of a virtual RTHS benchmark are provided to validate the integration method and illustrate the various uses of
the client-server protocol, focusing on earthquake engineering applications. The results show good accuracy and real-
time synchronization of client-server communication, allowing for the development of additional features of the Client-
Server virtual Real-Time Hybrid Simulation (CSvRTHS) framework to model complex structural systems.
Keywords: real-time hybrid simulation; substructuring; communication protocols; client-server; finite element modeling.
17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 27th to October 2nd, 2021
2
1. Introduction
There are different experimental techniques for structural testing in civil engineering, the most modern being
hybrid simulation. This technique considers both a physical and computational system interfaced together to
evaluate complex systems experimentally [1]. The technique is based on substructuring. The set of structural
components and materials whose behavior can be predicted, or those with experimentally calibrated and
validated mathematical models, is classified as a numerical substructure. On the other hand, the experimental
substructure is a critical component or innovative material without sufficient knowledge to calibrate a model
capable of predicting the expected behavior. One technique derived from the conventional hybrid simulation
is Real-Time Hybrid Simulation (RTHS). The main feature of RTHS is that all calculations, measurements, and
signal processing must be performed in very short time intervals, typically less than one millisecond [2]. A
critical aspect of RTHS testing is that any experimental errors must be mitigated during real-time execution to
avoid inaccurate and unstable responses [3]. Therefore, the engineering community has focused on developing
compensation techniques in RTHS, with acceptable performance and robustness, such that RTHS experiments
are carried out safely and accurately.
In general, the literature presents RTHS tests with simple, linear, numerical substructures models [4–8].
However, if one seeks to achieve greater simulation fidelity, it is desirable to have a finite element package to
model the numerical substructure more realistically. Therefore, a fundamental aspect in RTHS is integrating a
finite element (FE) analysis software with a controller software used in the experimental facility. To date,
some programs achieve this goal for hybrid simulation, such as UI-SIMCOR [9], OpenFresco [10] (which is
an improved version of the one proposed by [11]), Mercury [12], CSA [13], UT-SIM [14] and the CS technique
[15]. Although, their developers no longer support UI-SIMCOR and Mercury. Moreover, the CSA and CS
techniques were developed to work on coupling finite element software, not to work with cyber-physical
systems. The rest of the coordination programs were originally developed to carry out pseudo-dynamic tests;
hence, there are no guarantees for these programs to complete the required calculations deterministically given
the time constraints to perform RTHS tests. However, the literature provides some examples of conducting
RTHS with OpenFresco as coordinator software [16]. The problem lies in the small community that supports
a continuous and maintained development of OpenFresco, and the lack of detailed documentation for the
integration/coordination methods in RTHS tests. These issues hinder the development and implementation of
RTHS tests with high-fidelity coupled systems.
This study's motivation is to provide a reasonable procedure for researchers in RTHS to achieve the
coupling of widely used FE software with control software available in the laboratory. For this purpose, a four-
story steel structure with three bays will be solved, where a frame of the first floor will be considered as the
experimental substructure. Virtual environments are previously used for laboratory experimentation as a first
approach [17]. Therefore, in this study, a virtual RTHS will be carried out using three computer programs.
First, Matlab/Simulink [18] provides the most convenient means of implementing control algorithms for
RTHS. However, Matlab is not the best choice to model structural systems with nonlinear behavior, given that
it does not provide a library of elements and materials suitable for structural modeling. Therefore, OpenSees
[19] software is preferred for modeling numerical substructures. Finally, the Open-source Framework for
Experimental Setup and Control (OpenFresco) [10, 20] allows communication from/to OpenSees and a
Simulink model running in real-time, and combine a wide variety of hybrid simulation algorithms, control and
laboratory systems, experimental setups, and computational simulation models for a specific hybrid
simulation.
This paper has the following organization. Section 2 describes the procedure and structure to be solved,
the compensation method, and the evaluation criteria to assess its performance. Section 3 presents a detailed
description of the communication protocol used to couple OpenSees with Matlab/Simulink applications and
the process for executing and evaluating the communication’s performance. Section 4 presents the main results
obtained from the simulations carried out. Lastly, Section 5 presents the conclusions and final remarks.
17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 27th to October 2nd, 2021
3
2. Problem formulation
2.1 Numerical and experimental substructures
One of the crucial steps when performing hybrid simulation is the correct choice of substructuring [21]. In the
context of RTHS, instead of solving the entire domain equation of motion, this substructuring process can be
applied to subdivide the domain into smaller subdomains so that the order of large and complex structural
systems is reduced for computational efficiency. Each subdomain can be resolved independently, as long as
the coupling between components is enforced at their interfaces [22].
In this study, a four-story planar frame with three bays is chosen as a reference structure, with a total of
52 degrees of freedom (16 lateral, 16 vertical, and 20 rotational). The beams are modeled with linear elastic
frame elements. In contrast, the columns are modeled with frame elements using Giuffre-Menegotto-Pinto
steel material with isotropic strain hardening with an initial stiffness of
!"#$%&'()'*+"$
Consequently, the first-
mode fundamental period is
,!-."/$%0+
. The reference structure is divided into the numerical substructure
(NS) and experimental substructure (ES), as shown in Fig. 1. One bay of the first story is taken as the ES,
while the rest is modeled as NS. Since it is intended to work only with a uniaxial actuator, the boundary degrees
of freedom are set to be only horizontal displacements.
Fig. 1 - Reference structure and substructures. (Note: NS = dashed line; ES = red line)
The equation of motion (EOM) governing the dynamic response of a structure subjected to an input
seismic excitation is represented as follows in Eq. (1):
!!"
#
$%!"
&
$'!(")*+
(1)
where
1"
and
2"
are the mass and damping matrices, respectively.
3"
4
5
6 is the nonlinear restoring force.
5
,
5
7, and
5
8 are the displacement, velocity, and acceleration vectors, respectively, all measured relative to the
ground motion.
9
is the vector of external forces.
Let us now consider the system of two substructures shown previously. We can write the EOM of each
substructure as shown below in Eq. (2) and Eq. (3):
!""
#
"$%""
&
"$'!"("")*+"$,"
(2)
!#"##$%#"&#$'!#("#) *+#$,#
(3)
Structural parameters associated with the experimental substructure are indicated with the superscript
"E", and those associated with the numerical substructure are indicated with the superscript "N." The terms
9#
and
9$
are the external force vectors, and terms
:#
and
:$
are the coupling forces. The main assumption
17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 27th to October 2nd, 2021
4
of this formulation is that the substructures are coupled only through the boundary conditions. Therefore, the
coupling forces of the interior degrees of freedom for each substructure must be equal to zero.
For the resolution of this problem, two conditions must be met. First, Eq. (4) will serve as a
synchronization indicator on the boundary degrees of freedom. Second, Eq. (5) refers to the static equilibrium
at the boundary between substructures which should always be enforced.
5%
#-5%
$$$$4;<=>?@ABACA@D6
(4)
:%
#E:%
$-F$$$4GHIACABJAI=6
(5)
Substituting Eqs. (4) and (5) into Eqs. (2) and (3), the interaction between NS and ES is enforced, which
is explicitly shown in Eqs. (6) and (7):
1#58#E2#57#E3"#45#6-9#EKF&
#
L:%
$M
(6)
:$-KF&
$
:%
$M-1$58$E2$57$E3"$45$6L9$
(7)
where
5#-
N
5&
$5%
#
O
'
is the vector of NS displacements, which are classified as internal displacements
(indicated by the subscript "i"), and boundary displacements (indicated by the subscript "b"). Eq. (6) is the
EOM for NS that is solved ideally using an FEA software. Meanwhile, Eq. (7) collects all terms associated
with the force applied from ES to NS at the boundary. This formulation includes all potential effects associated
with nonlinear restoring forces, nonlinear damping, and inertial forces, along with any external excitation
directly induced to the experimental substructure.
2.2 CSvRTHS framework
In RTHS, the numerical and experimental substructures (NS and ES, respectively) must communicate in real-
time, which is handled by a coordinator software. At each time step of the analysis, the NS solves the equation
of motion through numerical integration and produces a target displacement (
P(
) at the interface. This signal
is imposed on the ES in the laboratory through a loading system (e.g., hydraulic actuator). In RTHS, enforcing
target displacement must be done in real-time; any delay in communication can cause inaccurate results and
potential instability. After this displacement is imposed, the restoring force is measured from the specimen and
returned to the NS integrator for the next step.
In this study, a CSvRTHS test is implemented, where all components are modeled numerically, including
the laboratory test system (actuators, test specimen, and sensors). Both NS, ES, and control systems are
modeled with different state-of-the-art software to explore further integration and increased flexibility.
Matlab/Simulink is chosen for its numerical capabilities and integration with microcontrollers such as dSpace
and Speedgoat. Meanwhile, OpenSees is a finite element analysis (FEA) software widely used by the
earthquake engineering community. Each computational task is interconnected using a TCP/IP communication
protocol through a client-server (CS) approach [10, 23, 24]. Further details on the CS communication task are
provided in Section 3, and the source code is available to the scientific community at a GitHub repository [25].
Many linear and nonlinear models are available for servo-hydraulic actuators [17, 26, 27]. In this study,
a model developed by Spencer et al. [27] is used. This model includes mathematical representations of the
different components of the test system, including servo valve dynamics, hydraulics, payload, and controller.
Fig. 2 shows the model implemented in Simulink. A payload is added to the original model to obtain the
actuator displacement, velocity, and acceleration for commanding over to the ES. This payload includes
damping and static forces, along with possible force losses based on piston friction. The parameters considered
for this model are shown in Table 1.
17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 27th to October 2nd, 2021
5
Fig. 2 – Block diagram of actuator dynamics in Simulink.
Table 1 – Actuator model parameters [27].
Component
Parameter
Value
Unit
Description
Controller
Q)
R
=S)AT
Proportional gain
Servovalve
U*
F"FFRRV
W
Servovalve time constant
Q+
VR"F.
AT,)W)=S
Valve gain
Q-
.
."R#P.F/0
AT,)W)>WA
Valve flow-pressure gain
Actuator
X
F"Y!.
AT1
Area of the piston
22
!"Z/P.F/3
AT,)W)>WA
Leakage coefficient of the actuator
[(
\Z"##
AT,
Total volume of fluid in actuator chambers
]4
/!/!Z
>WA
Effective oil bulk modulus
2.3 Compensation method
The key to RTHS is that the commanded displacements are imposed on the test sample by hydraulic actuators
in real-time. During testing, the servo-hydraulic actuator’s communication, control, and dynamics cause a time
delay and lag between components. Time delay/lag in the RTHS test is equivalent to introducing spurious
negative damping, leading to potential inaccuracies and instabilities.
Therefore, an Adaptive Model-Based Compensation (AMBC) [28] is chosen to mitigate time delays
during the RTHS tests. The formulation consists of estimating the plant by a third-order transfer function
without zeros, as shown in Eq (8).
-$*.%(/)-&*0 1
2'/'$2(/($2)/$2*3-&
(9)
where
P5
is the displacement measured by the sensors. The inverse of this transfer function is used to generate
the command signal using the target displacement as input, as shown in Eq (10). The derivatives of
P(
are
obtained with finite differences.
-&*.%
+)-,*(2'/'$2(/($2)/$2*)-,*2'-4,$2(-#,$52)-&,$2*-,
(10)
The adaptive parameters
^&
,
'-F_._V_R
, must be adjusted during the test to achieve good compensation.
This adaptation is formulated through a differential equation model provided in Eq. (11), where
P5
6&7
is the i-th
derivative of
P5
associated with adaptive parameter
^&
. A Butterworth filter is designed to remove high-
frequency noise and to estimate higher-order derivatives of target and measured signals.
52&-*6-7-&892'52(52)52*:9-4$5-#$5-&$5-$:.
1$9-4$5-#$5-&$5-$:9-4$5-#$5-&$5-$:.;-$
/-0
(11)
17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 27th to October 2nd, 2021
6
This methodology requires the calibration of adaptive gains. For this, the procedure given in the original
paper [28] is followed, but with certain modifications: (i) ground acceleration scaled to
.FF`a
(ii) multiples
single-degree-of-freedom structures with a natural period between
F"#$L$\$%0+
; and (iii) mass, stiffness,
damping, and initial conditions for the adaptive controller were randomly generated through the Latin
hypercube method to cover a larger range of structures, obtaining more robust gains. The calibration of optimal
adaptive gains consists of a global optimization scheme through particle swarm optimization, which minimizes
the performance indicator
b1
explained in Section
V"R
. Additionally, the initial conditions are defined from the
actuator without a specimen attached. Finally, the values obtained and used for the adaptive gains (
c&
) and
initial conditions (
^&
) are:
$c&-dA?e
4
.F389,$$$.F:81, $$$.F!8:;$$$.F/!8!9
6
<^&-$%."F$$$$."//f.F/1 $$$$.F/:$$$$F"F+
.
2.4 Evaluation criteria
The performance of the CSvRTHS is evaluated through three performance indicators, which are shown in Table
2.
b1
is the normalized root-mean-square error between the target displacement (
P(
) and the measured
displacement (
P5
); this indicator measures the synchronization error.
b:
is the normalized root-mean-square
error between the reference displacement (
P"
) and the measured displacement (
P5
), which measures the error
between the simulation and the reference structure. Both
b1
and
b:
indices were provided in Silva et al. [17].
Also,
U
corresponds to the delay indicator obtained with the frequency evaluation index [29], which measures
the synchronization delay (in milliseconds) between
P(
and
P5
. Finally, MT is the missed ticks of the real-
time clock in Simulink Desktop Real-Time, which will be discussed in the next section.
Table 2 - Performance evaluation criteria.
Criterion
Description
Definition
Units
b1
Normalized root mean square of the
tracking error
b1-gh%P54'6LP(4'6+1
#
&<!
h %P(4'6+1
#
&<! i.FF
`
b:
Normalized RMS relative
displacement at the first floor
b:-
gh%
P5
4
'
6
LP"4'6
+
1
#
&<!
h %
P"4'6
+
1
#
&<! i.FF
`
U
Time delay between the input and
output signals evaluated at the
equivalent frequency.
U-L j9
Vkl4+
=W
1,
Missed real-time clock ticks and
their associated data points.
Internally calculated by Simulink
@AmnW
3. Methodology
3.1 Communication protocol
In the Client-Server (CS) communication protocol, tasks or workloads are divided between providers of a
resource or service, called servers, and service requesters called clients [23]. This architecture is ideal for
communicating of the NS and ES as servers with Simulink as the client. The whole process is better explained
in Fig. 3. In general terms, it starts with the NS implemented in OpenSees (server), where it calculates the
displacement in the interface nodes and sends this signal to Simulink (client). In Simulink, this displacement
is received, and after a series of intermediate steps (multi-rate transitioning, compensation method, actuator
dynamics and control), it is sent to the ES (server). The ES given a displacement returns a force to Simulink,
to finally return that force to the NS. With this information, the NS performs the calculations of the next time
step.
It is worth noting that both OpenSees and Matlab/Simulink supports TCP/IP communication protocol.
But it requires a certain level of programming knowledge to implement this interface. Although it is possible
17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 27th to October 2nd, 2021
7
to easily open and close communication channels through Matlab scripts [23], it is not recommended for RTHS
since it will cause severe time constraints during Simulink client’s real-time execution. An alternative employs
user-defined blocks called S-Functions [24] in Simulink, which can execute custom code using different
programming languages and include a set of callback methods that perform the necessary tasks at each time
step. During the simulation stage, the Simulink engine calls the appropriate methods for each S-function block
in the model. For this study, a Level 2 S-function written in C language is suitable to implement TCP/IP
communication. This S-Function block is compiled into a Matlab Executable (MEX) file, which is necessary
to run user-defined code in Matlab/Simulink.
In summary, a Level 2 S-Function called Generic-Client is used to communicate with the ES. This user-
defined block interfaces the Simulink (client) with the ES by specifying IP and connection port as parameters.
As input data, only this displacement is required, and the server returns a displacement and a force. On the
other hand, the Level 2 S-Function called OPF Connect is used to interface Simulink with the NS, in which
force and displacement are sent to solve the equation of motion, Eq. (6), and it responds with a displacement
in the interface nodes. Both user-defined blocks can be obtained from OpenFresco Version 2.7.1 source code
[30] and readily available for use in this study’s companion GitHub repository [25].
3.2 Implementation of real-time system
The implementation of CSvRTHS is illustrated in Fig. 3. Detailed instructions of software installation and
execution are provided in the companion GitHub repository [25]. Briefly, the first requirement is to start the
ES-FE server, which is waiting for the simulation to start. Then, Simulink client must be executed, which
requests a target displacement to OpenSees to execute of the first temporary step. Finally, the NS-FE server
containing the NS performs numerical integration and begins the simulation. An explicit integration scheme,
such as the central difference method (NS-FE server) and the Dormand-Prince method (Simulink client). The
sampling intervals used for the fast (client and ES-FE server) and slow (NS-FE server) processes are
.).FV\
and
V).FF
[s], respectively (i.e.,
Fo,=>?( o,@2AB
). To summarize, the loops need to be run from the inside
to the outside layers: (1) ES-FE server (green); (2) Simulink client (orange); and (3) the NS-FE server (blue-
gray).
Note. Signals: x = displacement, f = force; superscripts: C = client, S = server;
subscripts: m = measured, t = target, c = compensated, a = actuator, S = slow rate, F = fast rate)
Fig. 3 – Simulation loops considered in CSvRTHS.
Further, a real-time sync block is placed on the system to synchronize the Simulink model with the real-
time kernel clock and ensure that it does not run faster than the user-specified simulation rate. Using this sync
block implies that at each sample interval, the Simulink model is evaluated in real-time. Simulink writes the
input data into a buffer that it passes to the kernel-mode process. The kernel-mode process propagates the data
to the hardware, which writes response data into another buffer. Simulink reads the response data at the next
17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 27th to October 2nd, 2021
8
time step and propagates it to the rest of the model. In this case, the Accelerator mode of Simulink Desktop
Real-Time is used. This mode implies that simulation is not a “hard” RTHS; that is, producing results after
their deadline is allowed and does not cause catastrophic consequences on the system under control.
A consequence of this limited synchronization is that the simulation can miss consecutive real-time
clock ticks and their associated data points without failure [31]. The reasons why ticks are lost are various; for
example, model complexity implies that the model can be so complex that Simulink cannot keep up with the
kernel in real-time. In this paper, the number of allowable missed ticks (
1,
) was fixed to 500. Consequently,
the number of missed ticks is selected as indicators of the communication protocol’s performance in the
CSvRTHS framework. The reliability of this real-time system is assessed in Sec. 4.2.
3.3 Multi-rate transitioning
One problem working with coupling programs is that the systems may run at different speeds. In this case, El
Centro
./\F
is selected as the seismic record, which implies that OpenSees will perform the integration in
time steps of
F"FV$%0+
, corresponding to the step presented by the record. On the other hand, a time step of
.).FV\$%0+
is selected to perform the control algorithms in Simulink. For multi-rate synchronization, the
literature provides multiple alternatives [32–34]. In this study, the corrector-predictor algorithm proposed by
Nakashima [35] is selected, which for the continuous generation of signals uses polynomial
extrapolation/interpolation. This algorithm is implemented in a StateFlow block in Simulink and can be
obtained in the OpenFresco source code [30]. Fig. 4 shows the displacements from OpenSees, which have a
staircase pattern given their greater temporal distribution. The target displacement obtained from the corrective
predictor algorithm is observed to smooth this signal for the faster Simulink task.
Fig. 4 – Predictor-corrector response in CSvRTHS.
4. Results
4.1 CSvRTHS performance
The CSvRTHS results are presented graphically in this subsection. For displacement comparisons, the node to
the left of the first floor (upper left node of the ES) is selected as representative. In Fig. 5, the measured
displacement is compared to the target displacement for evaluating the compensation method. With
b1$-
$F"Z##$%`+
and
U$-$F"FFY!$%=Wpm+
, a good tracking is achieved, which allows a stable test. Furthermore, it
is noticeable that practically all the error obtained (
b1
) is due to the simulated measurement noise considered
in the sensors.
Meanwhile, Fig. 6 shows a comparison between the measured displacement and the reference
displacement. The
b:
indicator reaches a value of
V"/V$%`+_
which means that correct and accurate results were
obtained. The reference displacement is obtained from a model made entirely in the OpenSees software.
17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 27th to October 2nd, 2021
9
Fig. 5- Target v/s Measured Comparison and J2 indicator.
Fig. 6 - Reference v/s Measured Comparison and J4 indicator.
Fig. 7 - Global comparison. Base Shear v/s Roof Story Displacement.
17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 27th to October 2nd, 2021
10
Finally, a comparison in global terms of the structure is added in Fig. 7. For this purpose, the graph of
roof story displacement versus base shear is made. Again, a good concordance of the measured results against
the reference ones is observed, thus validating the substructuring and compensation methodology.
4.2 Reliability of the real-time system
The reliability of CSvRTHS is assessed through a set of
\F
simulation runs. The simulations were carried out
on an MSI laptop, model GF75 Thin 9RCX (2.6 GHz Intel Core i7-9750H processor, 8 Gb RAM), running
Windows 10 version 20H2, running Simulink Desktop Real-Time in Accelerator Mode, and under controlled
room temperature (19°C). The results are shown in Fig. 8, which presents missed ticks statistics during the
real-time execution. The maximum value of missed ticks is
\V#
, which is the worst-case scenario. Therefore,
the limit of
!FF
(established in Section 3.2) is never exceeded, and a fast and efficient communication protocol
is achieved. It is observed that the missed ticks are concentrated in the first
R
to
\$%0+"
Hence, the seismic record
is zero-padded for the first 5[s] to verify that the excitation does not cause this phenomenon. This
communication overhead is probably caused by the initiation of the communication protocol between
applications.
For steady-state real-time simulation (i.e., times after
\$%0+
), the missed tick statistics are significantly
reduced. The lower right graph of Fig. 8 shows a worst-case scenario of
.Z
missed ticks and a zero median.
The peaks of missed ticks are generated in regular intervals of 0.02 [s], coinciding with the duration of the
slow rate process (NS-FE Server). However, it is a low and stable number of missed ticks, compared to the
previous peaks, so it is assumed a greater precision and accuracy in the results after the 4 [s] start of the
CSvRTHS. This result encourages using a seismic record with 5 [s] of zero-padding at the start, thus ensuring
that critical information is not lost.
Fig. 8 – Statistics of the missed ticks during CSvRTHS simulation.
Another simulation alternative is Simulink External Mode, where the resulting executables run in the
operating system kernel mode of the development computer. Parameter data is exchanged with Simulink via a
shared memory interface, allowing full synchronization with the real-time clock. Consequently, missed ticks
are eradicated in this simulation mode, allowing for the development of a “hard” real-time system. Further
investigation of this topic will be carried out in future studies.
5. Conclusions
This study presents a detailed explanation of the fundamental components of CSvRTHS, a virtual real-time
hybrid simulation implemented with a Client-Server communication protocol for coupling a finite element
analysis software (OpenSees) with the real-time control system (Matlab/Simulink). A structural engineering
17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 27th to October 2nd, 2021
11
example is provided to illustrate its implementation and software validation. Additionally, a previously
developed adaptive compensation method is adopted and validated against a nonlinear specimen,
demonstrating that a robust fixed controller could be used with different experimental substructures, avoiding
subsequent system identification tests.
Validation of the CS communication protocol is achieved for vRTHS, providing enough evidence of this
protocol's modularity and flexibility to connect different computational tasks. Further research will consider
implementing the CS method in RTHS over a network with dedicated microcontroller (client) and FEA
machines (servers) in a laboratory setup. The authors anticipate that this work will boost the development and
deployment of real-time hybrid simulation with high-fidelity numerical models and promote further
collaboration among experts in the field.
6. Acknowledgments
The authors gratefully acknowledge the financial support from Agencia Nacional de Investigación y
Desarrollo (ANID, Chile) through FONDECYT Iniciación research project No. 11190774. Also, we thank
Cristobal Gálmez, M.S. Student at USM, for his assistance with the calibration of the Adaptive Model-Based
Compensation and provide insight and expertise with RTHS that greatly assisted this research. Finally, any
opinions, findings, conclusions, or recommendations expressed in this paper are those of the authors and do
not necessarily reflect those of the sponsors.
7. References
[1] M. Nakashima, “Hybrid simulation: An early history,” Earthq. Eng. Struct. Dyn., no. March, pp. 1–14, Apr. 2020.
[2] M. Nakashima, “Development, potential, and limitations of real-time online (pseudo-dynamic) testing,” Philos.
Trans. R. Soc. A Math. Phys. Eng. Sci., vol. 359, no. 1786, pp. 1851–1867, Sep. 2001.
[3] T. Horiuchi, M. Inoue, T. Konno, and Y. Namita, “Real-time hybrid experimental system with actuator delay
compensation and its application to a piping system with energy absorber,” Earthq. Eng. Struct. Dyn., vol. 28,
no. 10, pp. 1121–1141, Oct. 1999.
[4] A. Ashasi-Sorkhabi, H. Malekghasemi, and O. Mercan, “Implementation and verification of real-time hybrid
simulation (RTHS) using a shake table for research and education,” JVC/Journal Vib. Control, vol. 21, no. 8, pp.
1459–1472, 2015.
[5] R. Enokida, “Basic examination of two substructuring schemes for shake table tests,” Struct. Control Heal.
Monit., vol. 27, no. 4, pp. 1–23, 2020.
[6] Z. Fei, W. Jinting, J. Feng, and L. Liqiao, “Control performance comparison between tuned liquid damper and
tuned liquid column damper using real-time hybrid simulation,” Earthq. Eng. Eng. Vib., vol. 18, no. 3, pp. 695–
701, 2019.
[7] T. L. Karavasilis, J. M. Ricles, R. Sause, and C. Chen, “Experimental evaluation of the seismic performance of
steel MRFs with compressed elastomer dampers using large-scale real-time hybrid simulation,” Eng. Struct., vol.
33, no. 6, pp. 1859–1869, 2011.
[8] A. Stavridis and P. B. Shing, “Hybrid testing and modeling of a suspended zipper steel frame,” Earthq. Eng.
Struct. Dyn., vol. 39, no. 2, pp. 187–209, 2010.
[9] O.-S. Kwon, A. S. Elnashai, B. F. Spencer Jr., and K. Park, “UI-SIMCOR : A global platform for hybrid
distributed simulation,” in 9th Canadian Conference on Earthquake Engineering, 2007, no. June, pp. 139–149.
[10] A. H. Schellenberg, S. A. Mahin, and G. L. Fenves, “Advanced Implementation of Hybrid Simulation. PEER
Report 2009/104,” Peer 2009/104, no. November, p. 286, 2009.
[11] Y. Takahashi and G. L. Fenves, “Software framework for distributed experimental-computational simulation of
structural systems,” Earthq. Eng. Struct. Dyn., vol. 35, no. 3, pp. 267–291, Mar. 2006.
[12] V. Saouma, D.-H. Kang, and G. Haussmann, “A computational finite‐element program for hybrid simulation,”
Earthq. Eng. Struct. Dyn., no. 41, pp. 375–389, 2012.
[13] Y. Zhang, P. Pan, Q. Gu, J. Yang, and K. Deng, “Development of Collaborative Structure Analysis (CSA) system
and its application to investigate effects of soil-structure interaction,” J. Earthq. Eng., vol. 18, no. 7, pp. 1151–
1169, 2014.
[14] P. Mortazavi, X. Huang, O. Kwon, and C. Christopoulos, “An Overview of the University of Toronto Simulation
(UT-SIM) Framework and its Application to the Performance Assessment of Structures,” in 7th International
17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 27th to October 2nd, 2021
12
Conference on Advances in Experimental Structural Engineering, 2017.
[15] Q. Gu and O. Ozcelik, “Integrating OpenSees with other software - with application to coupling problems in civil
engineering,” Struct. Eng. Mech., vol. 40, no. 1, pp. 85–103, Oct. 2011.
[16] T. Li, M. Su, Y. Sui, and L. Ma, “Real-time hybrid simulation on high strength steel frame with Y-shaped
eccentric braces,” Eng. Struct., vol. 226, no. September 2020, p. 111369, Jan. 2021.
[17] D. Gomez, C. E. Silva, D. Gomez, A. Maghareh, and S. J. Dyke, “Benchmark control problem for real-time
hybrid simulation Benchmark control problem for real-time hybrid simulation,” Mech. Syst. Signal Process., vol.
135:106381, no. October, 2019.
[18] Matlab, Version 9.9.0 (R2020b). Natick, Massachusetts: The MathWorks Inc., 2020.
[19] F. McKenna, “OpenSees: A framework for earthquake engineering simulation,” Comput. Sci. Eng., vol. 13, no.
4, pp. 58–66, 2011.
[20] A. Schellenberg, H. K. Kim, Y. Takahashi, G. L. Fenves, and S. A. Mahin, “OpenFresco Command Language
Manual,” no. July, 2009.
[21] S. N. Dermitzakis and S. A. Mahin, “Development of Substructuring Techniques for On-Line Computer
Controlled Seismic Performance Testing,” Berkeley, CA, UCB/EERC-85/04, 1985.
[22] D. de Klerk, D. J. Rixen, and S. N. Voormeeren, “General Framework for Dynamic Substructuring: History,
Review and Classification of Techniques,” AIAA J., vol. 46, no. 5, pp. 1169–1181, May 2008.
[23] M. Sysel, “MATLAB/Simulink TCP/IP Communication,” in Proceedings of the 15th WSEAS international
conference on Computers, 2015, no. July 2011.
[24] Matlab, Simulink, Model-Based and System-Based Design: Writing S-Functions Version 5. Natick,
Massachusetts: The MathWorks Inc., 2002.
[25] D. Mera and G. Fermandois, “Client-Server application for vRTHS,” 2021. [Online]. Available:
https://github.com/FermandoisLab/CSvRTHS.
[26] A. Maghareh, C. E. Silva, and S. J. Dyke, “Parametric model of servo-hydraulic actuator coupled with a nonlinear
system: Experimental validation,” Mech. Syst. Signal Process., vol. 104, pp. 663–672, 2018.
[27] J. E. Carrion and B. F. Spencer, “Model-based Strategies for Real-time Hybrid Testing. Technical Report NSEL-
006,” Department of Civil and Environmental Engineering, University of Illinois at Urbana-Champaign, Urbana,
IL, 2007.
[28] G. Fermandois, C. Galmez, and M. Valdebenito, “Optimal Gain Calibration of Adaptive Model-Based
Compensation for Real-Time Hybrid Simulation Testing,” 17th World Conf. Earthq. Eng. (17WCEE), Sendai,
Japan, 2020.
[29] T. Guo, C. Chen, W. Xu, and F. Sanchez, “A frequency response analysis approach for quantitative assessment
of actuator tracking for real-time hybrid simulation,” Smart Mater. Struct., vol. 23, no. 4, p. 045042, Apr. 2014.
[30] A. H. Schellenberg, “OpenFresco,” GitHub Repository, 2018. [Online]. Available:
https://github.com/aschellenberg74/OpenFresco.
[31] Simulink Desktop Real-Time User’s Guide. Natick, Massachusetts: The MathWorks Inc., 2020.
[32] A. Maghareh, J. P. Waldbjørn, S. J. Dyke, A. Prakash, and A. I. Ozdagli, “Adaptive multi-rate interface:
development and experimental verification for real-time hybrid simulation,” Earthq. Eng. Struct. Dyn., vol. 45,
no. 9, pp. 1411–1425, Jul. 2016.
[33] Y. Zhang, R. Sause, J. M. Ricles, and C. J. Naito, “Modified predictor-corrector numerical scheme for real-time
pseudo dynamic tests using state-space formulation,” Earthquake Engineering and Structural Dynamics, vol. 34,
no. 3. pp. 271–288, 2005.
[34] J. Ghaboussi, G. J. Yun, and Y. M. A. Hashash, “A novel predictor-corrector algorithm for sub-structure pseudo-
dynamic testing,” Earthq. Eng. Struct. Dyn., vol. 35, no. 4, pp. 453–476, 2006.
[35] M. Nakashima and N. Masaoka, “Real-time on-line test for MDOF systems,” Earthq. Eng. Struct. Dyn., vol. 28,
no. 4, pp. 393–420, Apr. 1999.
