Applied Energy xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Models of control valve and actuation system for dynamics analysis of

steam turbines q
Maddalena Pondini a,⇑, Valentina Colla a, Annamaria Signorini b
a
Scuola Superiore Sant’Anna, TeCIP Institute ICT-COISP Center, Via Alamanni 13B, 56010 Ghezzano, Pisa, Italy
b
Ó 2016 General Electric Oil & Gas Inc., Via F. Matteucci 2, Firenze 50127, Italy

h i g h l i g h t s

A dynamic model of Steam Turbine control valve and actuation systems is proposed.

An innovative study of the equations that rule the assembly movement is provided.

Control valve response and accuracy is analyzed in detail with test and simulation.

System upgrade is achieved with Electro-Hydrostatic Actuation technology.

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

Article history: The paper describes a study conducted on the control valve and the actuation systems of a Steam Turbine.
Received 23 January 2017 These devices are of utmost importance, as they rule the machine final power production and rotational
Received in revised form 20 April 2017 speed, thus their accurate modelling is fundamental for a valuable dynamic analysis of the whole system.
Accepted 14 May 2017
In particular, a dynamic model developed in the Matlab/Simulink environment is proposed, which sup-
Available online xxxx
ports the analysis of the operational stability of the hydro-mechanical system as well as the failure modes
that it may face during operation. The model has been successfully validated through specific field tests
Keywords:
conducted on the actuation system at a cogeneration plant located in the General Electric Oil & Gas -
Dynamic model
Steam turbine
Nuovo Pignone facility of Florence. The proposed work also highlights the requirements that new actu-
Control valve system ation technologies should fulfill in order to meet control valve system performance criteria and is thus
Matlab useful as both a methodological approach and a ‘‘virtual benchmark” allowing to validate in advance
any new actuation system.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction environment that the concepts of advanced control, performance

Nowadays in any industrial sector a continuous and rapid evo-
lution is observed towards ever more demanding applications and
intelligent solutions aimed at enhancing production flexibility and
efficiency while reducing wastes as well as resources and energy
consumptions. As a consequence, the need of deploying updated
technologies and practices arises, especially for those applications
that are highly energy dependent. A constant research and investi-
gation effort towards up-to-date technologies is required in order
to keep a high level of competitiveness in the market. It is in this
q
The short version of the paper was presented at ICAE2016 on Oct 8-11, Beijing,
China. This paper is a substantial extension of the short version of the conference
paper.
⇑ Corresponding author.
E-mail addresses: m.pondini@sssup.it (M. Pondini), colla@sssup.it (V. Colla),
annamaria.signorini@ge.com (A. Signorini).
degradation monitoring and diagnostics came along, meeting the
main criticalities that are faced in the industrial field so far. Vanraj
et al. applied these concepts in the prevention of machine unavail-
ability and breakdown maintenance due to vibration related prob-
lems [1]; a similar approach has also been adopted in Ref. [2],
where Agudelo et al. integrated more developed techniques for
fault detection and troubleshooting. Moreover, the advent of the
digital power together with all the capabilities related to big-
data analytics, ‘digital twins’, Model Predictive Control (MPC) and
Artificial Intelligence (AI) enables a completely new approach to
face the challenge of continuous performances improvement. This
scenario represents a big opportunity for the industrial world to
drive productivity and efficiency when coping with a really uncer-
tain background, as highlighted in [3]. A further relevant example
in this framework is provided by Venkatasubramanian et al. [4],
who analyzed and compared different diagnostic methods, namely
quantitative model-based methods, qualitative methods and pro-

http://dx.doi.org/10.1016/j.apenergy.2017.05.117
0306-2619/Ó 2017 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Pondini M et al. Models of control valve and actuation system for dynamics analysis of steam turbines. Appl Energy
(2017), http://dx.doi.org/10.1016/j.apenergy.2017.05.117
2 M. Pondini et al. / Applied Energy xxx (2017) xxx–xxx
Nomenclature
Abbreviations
AI Artificial Intelligence
CPC Current to Pressure Converter
CSP Concentrated Solar Plant
EHA Electro-Hydrostatic Actuation
GE O&G General Electric Oil & Gas
GTCC Gas Turbine Combined Cycle
HP High Pressure
I/H Current to Hydraulic pressure
ISCC Integrated Solar Combined Cycle
LP Low Pressure
LNG Liquefied Natural Gas
LVDT Linear Variable Differential Transformer
MPC Model Predictive Control
ST Steam Turbine
Parameters
Psec secondary oil pressure [bar]
A servo-cylinder spool area [m2]
K servo-cylinder spring stiffness [N/m]
x0,servo servo-cylinder spring pre-compression [m]
Mservo servo-cylinder spool mass [kg]
r servo-cylinder spring damping coefficient [N s/m]
G lever center of mass
O lever center of rotation
cess history-based ones in order to underline the advantages con-
nected to early detection and diagnosis of process faults.
In addition to facing uncertainties and challenges connected to
the need of achieving and/or preserving competitiveness and tech-
nological leadership within the global market, which is common to
all the industrial sectors worldwide, the energy sector has to deal
with a further factor enhancing the complexity of any control task,
which is related to the increasing share of renewable power pro-
duction with its characteristic of exploiting an intermittent energy
source. This is the case of solar plants and wind farms, where the
availability of the primary source determines the operation of the
entire plant and its unbalancing contribute to the grid. Previous
research dealt with wind and solar output variability by introduc-
ing an optimal market share estimation for both the energy sources
[5], other studies approached this uncertain characteristic through
a forecasting methodology applied to an isolated microgrid where
flexible load management improved the economic dispatch [6].
Providing an adaptive approach to the control of the system oper-
ation can also allow dealing with this element of variability, which
is challenging from a components safety perspective too. Camacho
et al. provided examples of successful applications of control for
different energy sources, reinforcing the possibility of achieving
more complex and efficient designs while at the same time maxi-
mizing the exploitation of the availability of the energy source [7].
For instance, in [8] MPC was applied to optimize the generation
scheduling in Concentrated Solar Plants (CSP) in order to both track
the schedule which was committed for the ongoing day on the
basis of the most recent electricity price and weather forecast
and to generate the schedule for the following day.
Together with an optimization on the control perspective,
industrial systems in general always have to deal with incoming
technological innovation, which brings reliability, efficiency and
cost-saving opportunities to all the process levels. It is only thanks
to a continuous research and upgrade that process industries will
be able to demonstrate their flexibility. System upgrading is the
requirement behind the fulfillment of performance, reliability
and system simplification that are the main drivers of competitive-
Please cite this article in press as: Pondini M et al. Models of control valve and actuation system for dynamics analysis of steam turbines. Appl Energy
(2017), http://dx.doi.org/10.1016/j.apenergy.2017.05.117
p1 lower chamber pressure [bar]
p2 upper chamber pressure [bar]
A1 lower chamber area [m2]
A2 upper chamber area [m2]
patm atmospheric pressure [bar]
Arod rod cross sectional area [m2]
M mass of moving parts at cylinder rod [kg]
rcyl cylinder damping coefficient [N s/m]
Fload load force at the actuator rod [N]
Fr cylinder friction [N]
JG inertial torque of the center of mass [kg m2]
Mlever lever mass [kg]
Variables
xspring,servo servo-cylinder spring position [m]
s laplace variable
y piston position [m]
h rotation angle [rad]
Functions
x servo-cylinder position [m]
Fcyl cylinder force [N]
M0 total torque coming from acting forces [Nm]
Min torque of inertia [Nm]
Rin force of inertia [N]
ness in the industrial field. In literature many examples of benefi-
cial upgrades are presented, either from an environmental stand
point, such as discussed in an interesting exemplar industrial case
in [9], or from a control perspective, where automation and simpli-
fication led to operational and economical benefits, such as pro-
posed by Edwards in the case of a hydropower plant [10], or
from a technological perspective, as described by Madisetti et al.
in [11]. In this last study a new infrastructure of developing
automation tools is proposed to reduce the cost of the re-design
of obsolete components. The driver that usually pushes towards
system upgrading is simplification, which, in most of the cases,
comes together with higher reliability and reduced costs. A suit-
able example in the oil and gas field is the historical presence of
auxiliary systems fed by high quantities of oil and bringing the
necessity of installing inter-connecting piping and oil tanks.
Hydraulic or pneumatic control units, for example, are usually
applied within mechatronic systems where operational and func-
tional safety is the main requirement, as underlined in [12]. Conse-
quently, maintenance and safety requirements as well as system
complexity are leading customers towards oil-free solutions, or
electric systems with reduced oil quantities, where failure modes,
such as wear fault problems, have been studied in depth in the lit-
erature [13]. The same importance should be given to the risk asso-
ciated with the specific failure causes, which requires a
prioritization based on the system, product or service affected [14].
In particular, in the power production field, Steam Turbine (ST)
dynamics are of upmost importance especially when coping with
the renewable energies uncertain characteristic. Due to the peri-
odic nature of solar radiation, which demands for an adaptation
of the modelling and control approaches conventionally adopted
in the power generation field, solar thermal power plants are often
subjected to transient conditions, such as discussed, for instance, in
[15]. Even so, the technologies subdued to this type of energy
source present many advantages in terms of performances and
electricity costs, as underlined in [16], and represent a valid alter-
native to fossil fuels from both conversion efficiency and environ-
mental impact point of view [17]. Similar findings also come from
 M. Pondini et al. / Applied Energy xxx (2017) xxx–xxx
the analysis of solar-integrated type of applications, namely Inte-
grated Solar Combined Cycles (ISCCs) where the contribute of solar
radiation represents an economical and efficient way to enhance
overall process performances [18]. In the specific case of Concen-
trated Solar Plants (CSPs) the control of the entire machine
becomes critical due to frequent variations of the steam condition
leading to ineffective steam temperature control at machine start,
especially in partial cloudy weather conditions. Thus, the gener-
ated discontinuous operation service requires dedicated solutions
in order to optimize the regulation strategy [15]. A real time tuning
of the control parameters is a concrete solution allowing a maxi-
mization of the production by preventing instability issues. Clearly,
the increasing contribution of this type of unpredictable energy
sources deeply changed the configuration of the grid, where the
so-called ‘base load’ plants need to achieve a higher operational
flexibility in order to increase their competitiveness in the market.
Rossi et al. studied the enhancement of the power production from
the ST of a Gas Turbine Combined Cycle (GTCC) by monitoring the
mechanical stresses during faster start-up phases [19]. This con-
cept goes strictly together with the machine control logics, which
have to adapt responsively to components and steam conditions
to optimize the ramping procedure. The control of a ST is directly
related to its control valve system and to the deployed actuation
technology. The entire assembly is highly challenged by the
responsiveness and accuracy required by the control of the
machine, especially for those applications where machine regula-
tion is critical. There are still a few studies available in literature
about the prediction of the dynamic behaviour of this type of sys-
tems and this gap represents a wide area to be deepened. There-
fore, the present paper aims at providing a detailed study of ST
control valves and their actuation systems, both from a dynamic
and static perspective. The equations that rule the movement of
the assembly are presented and a model approach is proposed
for their implementation. Both the Current-to-Pressure-Converter
(CPC) and the hydro-mechanical system, that are part of the entire
control loop, are described and an overall simulation model inter-
facing their simulation blocks is proposed. The research is com-
pleted by the results of a test campaign performed at a
Cogeneration plant (CoGen) in the Nuovo Pignone facility; by
means of field results a model validation has been achieved and
a specific comparison between two actuation technologies has
been possible thanks to a system upgrade.
The paper is organized as follows: Section 2 provides an overview
on the problem analyzed and deepens the topic of ST regulation
technology through an analysis of the related literature; Section 3
presents the so-called ‘legacy actuation system’ and the simulation
model structure of the control valve system with this specific
actuation solution. A summary of the model validation results is
provided in Section 4, where a comparison between two different
actuation technologies is also presented in the second part of the
section. Results of control valve system failure modes are also
commented and their repeatability through model deployment is
presented. Finally, Section 5 provides some concluding remarks
and hints for future work.
2. Problem description
Such as extensively discussed by Mutama in [20], in the ST
control and actuation, the steam control valve plays a key role in
two main areas:

Operability, as it drives the turbine ability to cope with sudden
load variations and allows to accurately control turbine speed
when running at no load conditions;

Reliability, as the traditional oil actuation systems that
rule the valve movement can be subjected to functional
Please cite this article in press as: Pondini M et al. Models of control valve and actuation system for dynamics analysis of steam turbines. Appl Energy
(2017), http://dx.doi.org/10.1016/j.apenergy.2017.05.117
3
deterioration due to oil contamination or presence of
impurities.
The ST regulation system is mainly composed of an actuation
system connected to a mechanical assembly that transmits the
motion to a series of shutters, whose purpose is to enable the
steam flow passage into the machine. The goal of the regulation
system is primarily to vary the quantity of steam delivered to the
ST to compensate the variation of the turbine train conditions.
The ST industry has a long tradition of Low Pressure (LP) and High
Pressure (HP) actuation systems, which have been improved over
the years based on field experiences. Power generation applica-
tions require higher response capability, due to the need to cope
with grid load rejection, which implies the capability to control
the speed variation within over-speed limits. In addition, in the
ever growing context of renewable plants (i.e. CSPs), STs are
exposed to sudden steam conditions variations which can chal-
lenge the traditional actuation system dynamics capability and
even lead them, in the worst cases, to instability [5]. Therefore,
high responsiveness of the control valve system is of utmost
importance in order to improve the operation stability.
Due to the aforementioned challenges, a wide research on pos-
sible solutions for the actuation systems led to a comprehensive
comparison of three different technologies for General Electric
Oil & Gas (GE O&G) ST-applications: LP, HP and, most lately,
Electro-Hydrostatic Actuation (EHA) systems.
Inlet control valves performances have always been a matter of
concern in the literature: extensive research led to improved valve
designs in order to deal with high pressures and temperatures [21]
as well as steam flow instability around the valve [22]. Their non-
linearity characteristic has also been addressed by Halimi and Suh,
who provide solutions for compensation of the phenomena [23].
Zaryankin et al. have investigated new valve designs in order to
reduce the vibration behaviour connected to the load on the stems
[24]. In addition to these important features, the actuation system
of ST control valves also needs to be reliable and high-performing
when the overall control loop of the ST has stringent requirements
in terms of fast response to a change in actual condition. Neverthe-
less, even for systems where the reaction time is not so tight, the
actuation system plays a fundamental role in the control loop. Tra-
ditionally, the ST control loop is adjusting on the power production
output, therefore controlling the actuator position on the require-
ments of the entire loop. A dynamic model of the whole regulating
chain can significantly add value in its dynamic analysis, and thus
in the enhancement of the overall control performances. Cao pro-
posed a model of the main control valve mechanism of a Liquefied
Natural Gas (LNG) ship ST with the aim of studying its propulsion
control system, in this study a modular modelling approach is
introduced, in order to ensure a limited interaction between the
components sub-modules [25].
The present research deals with a similar topic with particular
focus on industrial and power generation applications; by adopting
an analogous modelling approach, the single components are ana-
lyzed from a dynamic perspective. In addition, a coherent stability
analysis of the control valve system is accomplished, being this
aspect of utmost importance when dealing with closed loop regu-
lation systems. Consequently, this part of the study also led to the
acknowledgment of the limits of the legacy actuation system
adopted and provided evidence for better performing systems,
depth to which previous research has not dealt with.
The purpose of this work is to meet the necessities connected to
the achievement of a better performing governor, which conse-
quently implies the need for a predictive model of the control valve
system. Such model also enables an analysis aimed at identifying
the elements of the entire chain that may represent a ‘bottleneck’
for the control loop. In particular, a modelled control valve system
4 M. Pondini et al. / Applied Energy xxx (2017) xxx–xxx
Fig. 1. Legacy actuation system with control valve assembly.
allows the study of the effect of the failure modes of the actuation
system, when coupled to the turbine governor, and, finally, the def-
inition of the best recovery actions to be taken by the turbine gov-
ernor logic.
The actuation system considered in the present work, that is
referred in the following as ‘legacy system’ for sake of brevity
and is represented in Fig. 1, operates within low oil pressure ranges
in the cylinder chambers: for this reason the whole control oil cir-
cuit is named LP control oil. Fig. 2 shows the LP control oil system
with the control valve assembly simulation model layout, where
the main components of the entire control loop are represented.
3. Modelling approach
The need for improved ST stability margins comes along with
the priority of a model-based approach, which is the main focus
of the present work. This implies the additional capability of
enabling the study of how a performing actuation affects the
response of the system downstream in terms of stability and pre-
cision which are key factors for an optimized operation of the
whole regulating chain. With this in mind the present work also
proposes a specific comparison between an LP and an EHA system
in Section 4, the results of the dedicated performance tests are
shown in detail and the possibility of replicating them with a
model-based support is underlined. LP actuation systems represent
Fig. 2. Simulation model layout.
Please cite this article in press as: Pondini M et al. Models of control valve and actuation system for dynamics analysis of steam turbines. Appl Energy
(2017), http://dx.doi.org/10.1016/j.apenergy.2017.05.117
the starting point of the analysis on the control valve working prin-
ciples and the reason is linked to the fact that this type of technol-
ogy has been applied to regulate ST control valves for decades,
hence ‘legacy’ system. The modelling equations to approach this
system are detailed in Section 3.2, where the study of the entire
control valve system model is proposed together with the results
of the validation tests performed at a Cogeneration plant in Nuovo
Pignone facility. A chance to document system upgrading opportu-
nities was taken at the same plant where a Legacy actuation sys-
tem was substituted by an EHA system to improve performances
and reduce maintenance activities.
The complexity and obsolescence of actuation systems involv-
ing high quantities of oil, such as the presented legacy one, leads
to the need of providing an analysis of the failure modes in which
it may incur as well as a coherent corrective action to be applied in
each case. Even if the regulation of STs is moving towards more
reliable and accurate solutions, there are many applications in
which the deployment of the legacy system in the past required
the elaboration of service solutions that can prevent the system
from stopping and, thus, from losing valuable hours of production.
A specific case is represented by instabilities, which occurred in
relation to anomalous pressure levels in the oil circuit, is described
in Section 4 and the advantages of the support of a Legacy actua-
tion system ‘digital twin’ in terms of fine system tuning are shown.
3.1. Working principles of the actuation system
Legacy actuation systems operation is based on the supply of oil
coming from the lube and control oil console, whose structure is
not included in the model, being its analysis out of the scope of
the present work. Since the primary oil is supplied at constant
pressure by the oil console, a pressure source is applied as a simpli-
fication in the model. The oil coming from the lube and control oil
console header plays an important role for this type of actuation
system, since it both moves a spool inside the servo-cylinder con-
nected to the actuator (secondary oil) and enters the hydraulic
actuator chambers (primary oil). The secondary oil pressure is set
by the I/H converter, that translates a current command signal
(4–20 mA) into oil pressure. The 4–20 mA command derives
directly from the control system and corresponds, respectively, to
the valve closed and opened positions. The motion of the spool
in the servo-cylinder is regulated by the secondary oil: it alterna-
tively opens the connection orifices that let the primary oil flow
either in the upper or in the lower chamber of the hydraulic cylin-
 M. Pondini et al. / Applied Energy xxx (2017) xxx–xxx
der. A lever transmits the actuator movement directly to the valves
with the help of a spring, whose primary function is to keep the
valve closed in case of a hydraulic actuator failure. The final goal
of this mechanical system is to regulate the steam passage through
the valve, by lifting up a valve guide that opens in sequence four or
five shutters, which in turn feed different sections of the machine
and realize a fine regulation.
3.2. Model structure
The structure of the model, which is developed in the Matlab/
Simulink simulation environment and is depicted in Fig. 1b, fol-
lows the layout of a typical LP control oil system and is articulated
into different blocks representing the main components of the sys-
tem: I/H converter, Servo-cylinder, Hydraulic Actuator and Valve
system. The SimHydraulics toolbox has been exploited for mod-
elling the hydraulic subsystem, according to the approach pro-
posed by Vechet et al. in [26]. The input of the model is the valve
command signal which is transformed into the secondary oil pres-
sure signal through the electronics implemented by the I/H con-
verter. The simulation of the electronics of this CPCÒ block has
been provided by the specific supplier and interfaced to the rest
of the system. The converter block can be configured with the suit-
able regulation parameters, namely the proportional and integral
ones. The secondary oil pressure controls the servo-cylinder spool,
which is kept in neutral position by the action of an embedded
spring when the secondary oil pressure is at 1.5 barG. The spring
further compression is commanded by a mechanical retroaction
which is attached to the piston of the hydraulic cylinder and is cal-
ibrated to give a feedback on the position reached. The servo-
cylinder can be simulated using the 4-Way Directional Valve Sim-
scape block whose dynamics can be obtained from the following
equation which, according to the Laplace transform, describes the
servo-cylinder position:
Psec A  Kðx0;serv o  xspring;serv o Þ
x¼
Mserv o s2 þ rs þ K
where PsecA is the secondary oil pressure acting on the servo-
cylinder spool, Kx0,servo is the spring preload and Kxspring,servo repre-
sents the leverism mechanic retroaction. Two physical signals
connect the 4-Way Directional Valve to the hydraulic cylinder
chambers, modelled with a Double-Acting Cylinder Simscape block,
which generates a force signal according to the load applied at its
rod (Fload). In order to determine the value of the load at the actua-
tor rod an equilibrium of torques is implemented considering the
rotation of the system around the fulcrum of the lever, thanks to
this approach the inertia of the lever is also taken into account.
The equilibrium of torques for this dynamic system is computed
as follows:
! ! !
M 0 þ Min þ ðG  OÞ ^ R in
where G–O is the distance between the fulcrum and center of grav-
ity. The inertial contribute of the lever is taken into account by
including both the result of single inertial distributed forces and
its inertial torque, where only the center of mass acceleration in
normal direction has a non-negligible contribute, so that Eq. (2)
above is translated into the following two equations:
M 0  J G €h  M lev er  aG norm  ðG  OÞ ¼ 0
M 0  J G €h  M lev er  €h  ðG  OÞ2 ¼ 0
The term of the equilibrium indicated as M0 is the result of the
rotation generated by the external forces acting on the lever,
namely the spring force, the actuation force, the weights of the dif-
ferent components and their friction too. The model is able to cal-
Please cite this article in press as: Pondini M et al. Models of control valve and actuation system for dynamics analysis of steam turbines. Appl Energy
(2017), http://dx.doi.org/10.1016/j.apenergy.2017.05.117
5
culate the position and speed of the actuator rod as a result of the
dynamic response of every component simulated, this signal is
directly related to the valve lift and thus to the steam flow entering
the steam turbine. Cylinder chambers pressures can also be
retrieved from the simulation in order to calculate the actuation
system force generated as per equation below [27]:
€ þ rcyl y_ þ F load þ F r
F cyl ¼ p1 A1  p2 A2  patm Arod ¼ M y ð5Þ
The right side of Eq. (5) is also implemented in the model to fur-
ther validate the force signal calculation, it takes into account the
dynamics of the piston and the friction force of the cylinder as well
as the load applied at its rod. In order to make an accurate estima-
tion of the force contributes in Eq. (4) the study developed by Tran
et al. [28] has been exploited for the identification of a suitable
model for cylinder friction modelling.
Each block in the simulation is configured according to the
design parameters of a ST control valve system of a small combined
cycle plant (CoGen plant) running at GE O&G - Nuovo Pignone in
Florence.
4. Results and discussion
A dedicated test campaign focusing on the actuation system and
control valve assembly has been accomplished at the CoGen plant.
The control valve system of this plant has been dismantled during
a ST revamping phase and equipped with all the sensors required
to perform the specific tests. Fig. 3 shows the assembly made of
servo-cylinder, hydraulic actuator, spring and mechanical lev-
erism. In Fig. 3 the load cell of load pin type positioned at the lower
coupling joint of the hydraulic actuator is also visible. The test rig
has been completed once the control valve system was re-
integrated with the ST and the oil connections could be executed
and equipped with additional sensors and with a suitable data
acquisition system, as described for a similar experimental appara-
tus implemented by Yanada et al. [29].
ð1Þ The control valve of this ST has also been subjected to an
upgrade of its actuation system in order to improve the efficiency
of the whole regulation, giving the possibility of carrying out a
specific comparison between legacy and new EHA system perfor-
mances. In particular, the tests performed with the former one
gave detailed data suitable for both model validation and system
characterization. In order to achieve this, the legacy system has
been equipped with a specific test rig: a position sensor (Linear
Variable Differential Transformer - LVDT) giving the displacement
of the hydraulic actuator and a load cell measuring the actuation
ð2Þ
ð3Þ
ð4Þ
Fig. 3. Control valve system with Legacy actuation deployed for the test campaign.
6 M. Pondini et al. / Applied Energy xxx (2017) xxx–xxx

Fig. 5. Results of the tests on the legacy system: actuator position matching during
Fig. 4. Tests on the legacy system: cylinder de-energization procedure. consecutive ramps test.

force required for control valve movement. Two pressure sensors

on the primary and secondary oil lines have also been installed
to provide a direct input to the model for validation purpose, being
primary oil circuit and I/H converter electronics out of the scope of
the analysis. Furthermore, a dynamic data acquisition system has
been required with a capability of sampling experimental data at
the interval of 0.5 ms (2 kHz) for fast valve movement
characterization.
The test campaign has been pursued by keeping the emergency
stop valve closed so that no steam was able to reach the steam
chest ahead of the shutters of the control valve. In this way the
steam force contribute can be neglected, to the mechanical system
can be calibrated and the model can be validated. The system has
been analyzed both in static conditions and with dynamic valve
movements, such as ramps and steps, in order to achieve a com-
plete characterization.
Figs. 4–6 depict the tests made with the legacy system. Fig. 4
presents the cylinder de-energization procedure performed by
Fig. 6. Results of the tests on the legacy system: actuator force matching during
decreasing the primary oil pressure to atmospheric pressure and consecutive ramps test.
maintaining the actuator command in valve closed position.
Through such procedure it is possible to isolate the contribute of
the weight of the system recorded by the load cell from the pri-
blue dotted line is the one gathered from simulation when the red
mary oil pressure force in the cylinder chambers, which is null, signal is given as a command and shows a good match with the mea-
being the pressure value at atmospheric level. To this aim, the
sured position, indicated as LVDT, both in terms of accuracy and
valve command should be maintained in the valve closed position dynamic response. This result is fundamental for the achievement
so that the secondary oil pressure level remains constant and the
of a dynamic characterization of the system and, consequently, for
same holds for the position of the leverism. This test aimed at
a major awareness on the dynamic requirements when evaluating
determining the weight of the assembly from the trend read by new actuation technologies. In particular, being the Legacy actuation
the load cell, to be considered as an offset for the following mea-
a system relying on an internal mechanical feedback only without
surements. The purple line represents the force measured which, any position sensor equipment, the use of a model supports in the
in correspondence with the primary oil pressure sudden decrease,
monitoring of the control valve position.
shows a step change to about 1800 N. The sign of this force value The force calculated with pressure values from cylinder cham-
is negative confirming its upwards direction, equal and contrary to
bers, as per left side of Eq. (5), is referred to as Simscape model
the weight of the system. whereas the one determined according to the right side of Eq. (5)
Figs. 5 and 6 represent the correspondence between simulation
is named Dynamic model. These two signals are compared to the
results and experimental data in terms of, respectively, actuator load cell data and such comparison further highlights the consis-
position and force. These two signals are plotted in the same time
tency of the model (Fig. 6). Besides the evident solidity of the
frame of a consecutive ramps test commanded from the governor model, intrinsic system non-linearities are difficult to capture
as represented by the red1 line. The 4–20 mA signal coming from
and depend on different factors, which cannot be easily controlled.
the governor has been rescaled over the stroke in order to be able As a consequence, the modelled dynamic behaviour shows more
to compare command and response signals. In Fig. 5 the trend in
linear trends with respect to the measured data. That is particu-
larly evident from the discrepancies between model (in blue dotted
1
For interpretation of color in Figs. 4–7, 9 and 10, the reader is referred to the web line) and experimental measures (in black line) occurring during
version of this article. valve closing phases (Fig. 6) where the friction force phenomena

Please cite this article in press as: Pondini M et al. Models of control valve and actuation system for dynamics analysis of steam turbines. Appl Energy
(2017), http://dx.doi.org/10.1016/j.apenergy.2017.05.117
 M. Pondini et al. / Applied Energy xxx (2017) xxx–xxx
is one of such factors contributing to the creation of the visible
non-linearities. Its contribution has been estimated through a
model capable of simulating friction both in static and dynamic
conditions [28,29]. However, this model requires input parameters
that are affected by a high uncertainty, as the friction phenomena
shows a hardly predictable behaviour.
In the second part of the present Section the results of a specific
comparison between two actuation technologies are presented, by
highlighting the advantages coming from the upgrade of this type
of component. The upgrade of the actuation system met both the
necessities of the plant manager to deploy a more precise and reli-
able technology and enabled a first validation of a new concept of
actuation for control valve systems. The EHA system, in fact, has
the characteristic of deploying a closed oil circuit with limited oil
quantities as well as an internal control loop that ensures an accu-
rate valve positioning and, thus, a tighter and more predictable
control of the machine power delivered. An analogous test proce-
dure has been performed after the upgrade in order to highlight
the differences in the system response. From the comparison
between the two different actuation technologies (see Fig. 7) it is
evident that EHA is much more accurate in reaching the position
commanded showing less uncertainty on control valve precision
and, thus, on steam inlet characterization. When moving along
the stroke with equidistant steps the legacy system shows un-
calibration respect to the position commanded and this is mainly
evident in the intermediate positions, where the legacy system
non-linearities connected to the servo-cylinder mechanical feed-
back are more evident. The reason for these non-linearities is
mainly caused by the servo-cylinder spring compression, which
occurs on both sides, namely the mechanical feedback side and
the spool side, whereas the system calibration procedure does
not take into account the former contribute and, thus, works with
a discrepancy between the position commanded and the one
reached by the actuator. The simulation model, on the contrary,
meets the position commanded and does not show the same beha-
viour in intermediate positions unless the additional contribute of
the mechanical feedback spring compression is taken into account.
The aforementioned problems are clearly overcome by the EHA
technology, which is capable of guaranteeing higher performances
thanks to its internal position control loop and its rapid response
that can certainly help in a more accurate steam inlet characteriza-
tion. This comes from the direct relation between the steam flow
entering the machine and the valve lift reached by the system,
Fig. 7. Comparison of the accuracy of the Legacy and EHA systems on valve opening (a) and closing (b).
Please cite this article in press as: Pondini M et al. Models of control valve and actuation system for dynamics analysis of steam turbines. Appl Energy
(2017), http://dx.doi.org/10.1016/j.apenergy.2017.05.117
7
the higher the accuracy of the valve system the more precise the
estimation of the steam flow and, thus, of the power generated.
The validation performed provided coherent results in nominal
operating conditions but additional value can be retrieved by
applying simulation capabilities to the analysis of the failure
modes of the actuation system. A detailed study on system failure
causes and corrective actions to be taken can support in the pre-
vention of unexpected behaviours of control valve systems on site.
Previous experience gave evidence of anomalous events generated
by uncontrolled fluctuations in the regulation of the machine,
when performed through the legacy actuation system. A significant
case to be analyzed is the event that happened in Arenales CSP site,
where a sudden decrease in primary oil pressure generated oscilla-
tions in the secondary oil circuit and, in the end, in the generated
output power. Root cause analysis has been supported by specific
tests on the control oil and valve systems deployed in that site,
which have been equipped with additional sensors as per the sche-
matic diagram shown in Fig. 8 below. The figure is a representation
of the components described in the previous sections according to
the convention adopted for Process and Instrumentation Diagrams.
In the schematic proposed the control oil line splits into two
branches, the upper one is the primary oil line whose pressure is
measured through sensor PI101, the lower one represents the cir-
cuit in which the CPCÒ block (SY001) transforms a command signal
into a secondary oil pressure level, dictating the servo-cylinder
spool movement. The procedure adopted for testing consisted in
verifying the operation of the system with different runs by chang-
ing the CPCÒ control parameters and the primary oil pressure level
and recording the consequent position of the piston of the actuator.
The CPCÒ is a device which transforms a command signal coming
from the control system into a corresponding pressure level
between 1.5 and 4.5 barg, the responsiveness of its action can be
adjusted by modifying the control parameters of the converter,
typically characterized by Proportional and Integral parameters
only. The tests put into evidence how a decreasing primary oil
pressure combined with high integral parameters of the CPCÒ
can turn into significant oscillations on secondary oil pressure,
which are directly causing uncertainties on the actuator position.
The aim of this part of the study was to apply the previously
described validated model in order to find an accurate replica of
the behaviour observed during the tests.
The test runs chosen for this type of validation were two cases
in which the CPCÒ control parameters were P = 2.8 and I = 4, as per
8 M. Pondini et al. / Applied Energy xxx (2017) xxx–xxx

Fig. 8. Legacy system test rig schematic.

Fig. 9. Comparison between test and simulation model results in the analysis of control valve failure mode with primary oil at 7.5 barg.

tuning on site, and the primary oil level was driven to 7.5 barg in
the first case and to 7.25 barg in the second case. The results of
the tests performed with these boundary conditions identified
the secondary oil pressure behaviour as the origin of the power
fluctuations detected on site. In particular these instabilities were
mainly noticed within the initial part of the actuator stroke move-
ment and their extent increased with decreasing primary oil pres-
sures. In order to find a method to prevent the occurrence of
unexpected issues as per the Arenales CSP case, the legacy actua-
tion system model has been configured and tested accordingly.
The achievement of an accurate replica of the trends measured
during the tests would enable a further application of the model
itself, namely to employ it as a system tuning instrument to the
aim of avoiding the standard site commissioning activities.
Primary oil pressure (PI101) and secondary oil pressure (PI003)
trends were given as input signals to test the model, so that the
actuator position calculated could be coherently compared with
the measured one. Blue dotted line in Fig. 9 represents the actuator
position recorded during the tests, the entity of the fluctuations
measured is around 15 mm on the piston side which corresponds
to a power variation of around 7 MW. The power fluctuation is esti-
mated by associating the valve lift connected to the 15 mm of
cylinder stroke variation to the corresponding steam flow entering
the machine. These two values are correlated through a steam flow
diagram, which is used in the machine design phase in order to
dimension the ST. By assuming that the ST operates in nominal
conditions, i.e. with steam in design conditions, the enthalpy drop
across the ST multiplied by the steam flow variation generated
gives an output power instability equivalent to the 13% of the rated
power. Fig. 9 above also shows the accurate model failure estima-
tion when tested with the same oil pressure inputs of the tests. The
red dotted line is the actuator position calculated through the
model, its trend can be easily compared to the blue dotted one.
From the analysis of the results it can be stated that the instabili-
ties detected during the test campaign phase are accurately repeat-
able with an analytical approach. These two trends can be read on
the left vertical axis, whereas the right vertical axis shows the scale
for PI101 and PI003 pressures.
The previously described test has been repeated with primary
oil pressure at 7.25 barg (see Fig. 10). Also in this case the sec-
ondary oil trend showed a fluctuation in the initial part of the
stroke, but it is characterized by a non-damped and higher oscilla-
tion with respect to the previous case. The reason for this beha-
viour is both on the primary oil pressure level and on the CPCÒ
control parameters and the combination of these two factors
makes the CPCÒ unable to reach the commanded secondary oil

Please cite this article in press as: Pondini M et al. Models of control valve and actuation system for dynamics analysis of steam turbines. Appl Energy
(2017), http://dx.doi.org/10.1016/j.apenergy.2017.05.117
 M. Pondini et al. / Applied Energy xxx (2017) xxx–xxx
Fig. 10. Comparison between test and simulation model results in the analysis of control valve failure mode with primary oil at 7.25 barg.
pressure level. High integral values, in fact, are not able to tighten
the control and to avoid the beginning of uncontrolled oscillations.
This happens in particular when the oil level from which the CPCÒ
starts regulating undergoes an unexpected decrease. The preven-
tion of such events is of upmost importance and the support of a
model-based approach can be relevant both to this purpose and
in order to save time during the tuning phases. Model-based
results show analogous behaviours with respect to the position
of the actuator. This result further highlights the consistency of
the simulation with reality.
The above presented results clearly show that a model based
approach can bring significant added value to the research around
ST control valves. For the first time an accurate predictive model
focusing on the dynamics of the system and, thus, on the direct
regulation of the machine, has been proposed opening the way
to advanced control capabilities for an online optimization of the
regulation performances. The consistency of the results both from
the accuracy perspective and on the responsiveness side enables
the deployment of the model as a suitable instrument for the
tuning of the control valve system, providing also an additional
advantage in terms of system failure prevention. It has been
demonstrated, in fact, how the implemented simulation model
can be employed to the aim of testing the stability of the system
when anomalous conditions in the oil lines occur. The described
potential applications highlight the importance of addressing the
research on control valve systems towards the analysis of their
responsiveness according to the specific actuation system adopted
and in line with the application in which they are deployed. In this
way it will be possible to dynamically characterize the entire fleet
of the control valve systems that are designed coherently with the
control requirements given by each analyzed application.
5. Conclusions and future work
The paper depicts an extensive research work that has been
developed on ST control valve systems, focusing both on their pre-
dictive models and on the opportunities for actuation system tech-
nology upgrading. The developed study aimed at providing a solid
model of the control valve system when operated without steam,
that represents a strong basis for its further development that will
involve the steam force contribute analysis and model implemen-
tation. A dedicated analysis is required in order to provide a deep
understanding of the phenomena occurring around the valve shut-
ters when the steam pressure differential acts by generating an
additional force. Furthermore, a specific test campaign with addi-
tional pressure sensors around the valve would be necessary in
Please cite this article in press as: Pondini M et al. Models of control valve and actuation system for dynamics analysis of steam turbines. Appl Energy
(2017), http://dx.doi.org/10.1016/j.apenergy.2017.05.117
9
order to identify the peculiarities of the steam behaviour during
valve opening and closing phases. The capability of simulating
the behaviour of the control valve response when operated with
steam comes along with the opportunity for performing critical
ST regulation phase simulations through the integration with the
governor logic and machine drums models. This will represent a
digital instrument for the analysis of their interaction inside the
entire control loop which can be embedded in the ST governor with
the aim of performing advanced diagnostics and analytics and con-
sequently checking regulation performances degradation.
Future work will be in line with the growing interest towards
EHA technologies bringing the need of covering the entire fleet of
actuation systems with their corresponding model. This achieve-
ment would both enable the analysis of the performances of this
type of actuators and the study of their behaviours when interfaced
to the rest of the system.
References
[1] Vanraj GD, Saini A, Dhami SS, Pabla BS. Intelligent predictive maintenance of
dynamic systems using condition monitoring and signal processing techniques
– a review. In: International conference on advances in computing 2016,
ICACCA 2016, India.
[2] Agudelo C, Cucarella EQ, Anglada FM, Moreno EG. Integration of techniques
and alarm sequences for fault detection and diagnosis. In: 3rd Conference on
control and fault-tolerant systems 2016, Barcelona, Spain.
[3] Lee J, Kao HA, Yang S. Service innovation and smart analytics for Industry 4.0
and big data environment. Proc CIRP 2014;16:3–8.
[4] Venkatasubramanian V, Rengaswamy R, Yin K, Kavuri SN. A review of process
fault detection and diagnosis Part I: quantitative model-based methods.
Comput Chem Eng 2003;27:293–311.
[5] Hirth L. The optimal share of variable renewables: how the variability of wind
and solar power affects their welfare-optimal deployment. Energy J
2015;36:149–84.
[6] Neves D, Brito MC, Silva CA. Impact of solar and wind forecast uncertainties on
demand response of isolated microgrids. Renew Energy 2016;87:1003–15.
[7] Camacho EF, Samad T, Garcia-Sanz M, Hiskens I. Control for renewable energy
and smart grids. The Impact of Control Technology, Control Systems Society;
2011. p. 69–88.
[8] Vasallo MJ, Bravo JM. A MPC approach for optimal generation scheduling in
CSP plants. Appl Energy 2016;165:357–70.
[9] Angelis-Dimakis A, Alexandratou A, Balzarini A. Value chain upgrading in a
textile dyeing industry. J Clean Prod 2016;138:237–47.
[10] Edwards K. Digital control system upgrade gives hydropower plant new life.
Power 2016;160.
[11] Madisetti VK, Jung Y, Khan MH, Kim J. Reengineering legacy embedded
systems. In: IEEE design & test of computers.
[12] Pickard K. Approach to analyse hydraulic or pneumatic control units. In: 26th
European safety and reliability conference, ESREL 2016, Glasgow, United
Kingdom.
[13] Ren P, Chen J, Hu Y, Yuan H. Research on typical wear fault diagnosis of
electro-hydraulic servo valve movement. In: 7th IEEE Prognostics and System
Health Management Conference, PHM 2016, Chengdu, Sichuan, China.
[14] Banerjee P, Pandey K. Implementation of failure modes and effect analysis on
the electro-hydraulic servo valve for steam turbine. In: 1st IEEE International
10 M. Pondini et al. / Applied Energy xxx (2017) xxx–xxx
conference on power electronics, intelligent control and energy systems,
ICPEICES 2016, Delhi Technological University, India.
[15] Abutayeh M, Alazzam A. Adapting steady-state solar power models to include
transients. J Sol Energy Eng 2016;139.
[16] Desideri U, Campana PE. Analysis and comparison between a concentrating
solar and a photovoltaic power plant. Appl Energy 2014;113:422–33.
[17] Desideri U, Zepparelli F, Morettini V, Garroni E. Comparative analysis of
concentrating solar power and photovoltaic technologies: technical and
environmental evaluations. Appl Energy 2013;102:765–84.
[18] Montes MJ, Rovira A, Munoz M, Martìnez-Val JM. Performance analysis of an
Integrated Solar Combined Cycle using Direct Steam Generation in parabolic
trough collectors. Appl Energy 2011;88:3228–38.
[19] Rossi I, Sorce A, Traverso A. Gas turbine combined cycle start-up and stress
evaluation: a simplified dynamic approach. Appl Energy 2017;190:880–90.
[20] Mutama KR. Aspects of steam turbine valves: materials, operations
and maintenance. In: ASME power conference 2013, Boston, Massachussets,
USA.
[21] Sors F, Holm P, Eriksson B, Ölvander J. Development of steam turbine inlet
control valve for supercritical pressure at siemens industrial turbomachinery
AB. Maskinkonstruktion 2010, Linköpings Univ.
Please cite this article in press as: Pondini M et al. Models of control valve and actuation system for dynamics analysis of steam turbines. Appl Energy
(2017), http://dx.doi.org/10.1016/j.apenergy.2017.05.117
[22] Bolin C, Engeda A. Analysis of flow-induced instability in a redesigned steam
control valve. Appl Therm Eng 2015;83:40–7.
[23] Halimi B, Suh KY. Engineering nonlinearity characteristic compensation for
commercial steam turbine control valve using linked MARS code and Matlab
Simulink. Nucl Eng Des 2012;243:360–70.
[24] Zaryankin AE, Rogalev ND, Rogalev AN, Garanin IV, Osipov SK, Grigoriev EY.
Control valves and cascades for the first stages of turbines with
ultrasupercritical steam parameters. Therm Eng 2016;63:422–9.
[25] Cao H. Modeling and simulation of main control valve mechanism of LNG ship
steam turbine. Open Mech Eng J 2014;8:144–50.
[26] Vechet S, Krejsa J. Hydraulic arm modeling via Matlab SimHydraulics. Eng
Mech 2009;16:287–96.
[27] Blagojević V, Stojiljković M. Mathematical and Simulink model of the
pneumatic system with bridging of the dual action cylinder chambers. Mech
Eng 2007;5:23–31.
[28] Tran XB, Hafizah N, Yanada H. Modeling of dynamic friction behaviours of
hydraulic cylinders. Mechatronics 2012;22:65–75.
[29] Yanada H, Khaing WH, Tran XB. Effect of friction model on simulation of
hydraulic actuator. In: The 3rd international conference on design engineering
and science 2014, ICDES 2014, Pilsen, Czech Republic.