See discussions, stats, and author profiles for this publication at: [Link]

net/publication/235675537

Modeling Hydro Power Plants and Tuning Hydro Governors as an Educational

Guideline

Article  in  International Review on Modelling and Simulations · January 2012

CITATIONS READS

22 7,173

3 authors:

Ramezan Ali Naghizadeh Saeed Jazebi

Hamedan University of Technology Applied Materials
21 PUBLICATIONS   279 CITATIONS    45 PUBLICATIONS   742 CITATIONS   

SEE PROFILE SEE PROFILE

Behrooz Vahidi
Amirkabir University of Technology
499 PUBLICATIONS   3,756 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Cable Ampacity View project

power systems View project

All content following this page was uploaded by Ramezan Ali Naghizadeh on 02 August 2018.

The user has requested enhancement of the downloaded file.

 International Review on Modelling and Simulations ([Link].S.), Vol. 5, N. 4
ISSN 1974-9821 August 2012

Modeling Hydro Power Plants and Tuning Hydro Governors

as an Educational Guideline

R. A. Naghizadeh1, S. Jazebi2, B. Vahidi3

Abstract – Appropriate modeling of components and related controllers are very significant in
studying dynamic performance of power systems. In this paper, an educational procedure for
modeling, simulation, and governor tuning of hydro power plants is presented. Different existing
dynamic models of hydro plant components are reviewed. The procedure for calculating the
required parameters from real plant data is also presented. Application and performance of
reviewed models is discussed as well. In addition, appropriate methods for tuning different types
of hydro governors are studied and a classical method is used for PID governors. Inclusion of the
nonlinearity and elasticity of the detailed turbine-penstock model, and studying the effect of the
servo transfer functions are the main aspects of the proposed tuning method. The paper is written
in a way to be useful for an electrical engineering student or a novice engineer to model, simulate,
and analyze a hydro power plant dynamic behavior, and finally tune its governor. This work can
be used as a primitive guideline for educational and practical purposes. Copyright © 2012 Praise
Worthy Prize S.r.l. - All rights reserved.

Keywords: Power Plant Dynamics, Hydraulic Turbines, Governor Tuning, Power System
Education, PID Controller

Nomenclature Superscript for normalized values

r Subscript for rated values
At Turbine gain
0 Subscript for initial steady state values
D Penstock pipe diameter (m)
NL Subscript for no-load values
Da Turbine damping
G Ideal gate opening
H Net head (m) I. Introduction
Hg Inertia time constant of generator (MW.s/MVA)
Ks Gain of servo-system Accurate mathematical representation of power
L Length of penstock pipe (m) system components is significant for dynamic and
Pm Turbine mechanical power (MW) transient stability studies. Therefore, some standard
Pr Rated turbine power (MW) dynamic models for prime movers were introduced
Q Water discharge (m3/s) previously for simulation programs and other purposes in
RT Transient droop the literature [1]-[5].
Rp Permanent droop The parameters of these models need to be determined
Tep Elastic time of penstock pipe (s) by operators and engineers as accurate as possible to take
TM Mechanical starting time (s) into account the behavior of such elements in power
Tp Pilot valve and servo time constant (s) system dynamic simulations.
TG Main servo time constant (s) Furthermore, one important application of such
TR Resetting time (s) modeling is tuning of the parameters of the unit
Tw Water starting time (s) controller such as governor. In order to guarantee good
U Water velocity (m/s) performance of electric power generation process in
Zp Normalized hydraulic surge impedance of the different conditions, the parameters of the controllers of
penstock the components have to be properly tuned.
a Pressure wave speed in penstock pipe (m/s) Hydroelectricity is an important source of renewable
g Real gate opening energy and provides significant flexibility from operating
ag Acceleration due to gravity (m/s2) point of view. The dynamic behavior of hydro power
aij Turbine characteristic coefficients plants is determined by the transients of water flow in the
φp Friction coefficient of penstock water column. The energy conversion process in turbine
involves only nondynamic characteristics [4].
ω Angular velocity of the generator (rad/sec)

Manuscript received and revised July 2012, accepted August 2012 Copyright © 2012 Praise Worthy Prize S.r.l. - All rights reserved

1780
 R. A. Naghizadeh, S. Jazebi, B. Vahidi

Hydraulic turbine is controlled by a governor system More details about hydroelectric power plant
which consists of control and actuating equipment for structures are beyond the scope of this work and can be
regulating the flow of water, starting and stopping the found in related textbooks such as subject [1]-[3].
unit, and regulating the speed and power output of the Fig. 1 illustrates the main components of a typical
generator [6]. The proper tuning of these governor hydroelectric generating unit. From the reservoir, water
systems plays an important role in stable and acceptable is drawn from an area called the forebay to the turbine
performance of the connected power system. through the water column. The water column comprises
There are some works for the modeling and all of the structures used to convey water from the
simulation of hydraulic power generation systems and forebay to the turbine. It may include an intake structure,
their governor tuning. Most of them are reviewed in [7]. a penstock, one or more surge tanks, and a spiral case.
Some textbooks present the basic knowledge about the The composite water column inertias and elasticity of
subject [1]-[3]. these structures contribute to the water hammer effect
In addition, there are some contributions for tuning of that impacts the performance of the turbine governing
hydro governors considering different models and system.
controller design methods. As mentioned in [7], the
governor tuning of nonlinear models specially in the case
of considering elasticity effect, needs more work. This
subject is discussed in this work with a real case study
and a simple approach is used for this purpose.
The first aim of this paper is to illustrate the procedure
of the extracting hydro turbine different standard and
detailed model parameters using real plant data. The
response of all models is presented for the sake of
comparison.
The second aim is to propose the application of a
proper and simple controller design method for tuning of
the plant governor for the prepared models. Linearization
of nonlinear model with elasticity effect is also presented Fig. 1. Simplified schematic of a hydroelectric power plant
in this work for governor tuning purpose. Performance of
the tuned governor is verified by comparing with Wicket gates are adjustable and pivot open around the
previous method and for different models. This work is periphery of the turbine to control the amount of water
in succession with previous works by the authors, in admitted to the turbine.
which the derivation of thermal turbines and steam boiler These gates are adjusted by the servo actuators which
dynamic model parameters are presented [8]-[10]. are controlled by the governor. Fig. 2 demonstrates the
This paper is organized as follows: In Section II, a simplified relationship between the basic elements of
brief description of hydroelectric power plants is power generation process in a hydraulic power plant.
presented. Hydraulic turbine-penstock models for Modeling of these elements is described in the following
dynamic simulations are briefly described in Section III. sections.
Section IV is dedicated to present the required equations
for parameter determination. Hydro governor models and
their tuning are discussed is in Sections V and VI
respectively. Section VII presents the simulation results
of the reviewed models of a real hydro power plant and
performance of the tuned governor. Finally, Section VIII
concludes the paper. Fig. 2. Simplified functional block diagram of hydraulic power plants

II. Hydroelectric Power Plants III. Hydraulic Turbine Models

In a hydraulic power generation plant, the stored Hydroelectric power generating system exhibits a
energy in water as a hydraulic fluid is converted into high-order and nonlinear behavior. Appropriate
mechanical energy by means of hydraulic turbine. mathematical models are essential tools for simulation of
Hydraulic turbines are of two basic types: impulse such systems. The hydraulic system and turbine-penstock
turbines and reaction turbines. Selection of the type of models have been analyzed in literature. The reader can
the turbine depends upon the head and water flow rate of find out that there are several models with different
the dam. The shaft of the generating unit may be in a levels of details for hydraulic power plants. This section
vertical, horizontal, or inclined direction depending on presents a brief overview and classification of the
conditions of the plant and the type of turbine applied. proposed models for representing turbine-penstock
The majority of new installations are vertical. systems for dynamic simulations.

Copyright © 2012 Praise Worthy Prize S.r.l. - All rights reserved International Review on Modelling and Simulations, Vol. 5, N. 4

1781
 R. A. Naghizadeh, S. Jazebi, B. Vahidi

First of all, the models can be classified into linear coefficients a21 and a23 are partial derivatives of turbine
and nonlinear categories based on the complexity of the power with respect to head and gate opening
equations. The proposed models can be classified further respectively.
by considering elasticity effect of the water column
(penstock). The mathematical equations of the models
III.4. Non-Ideal Elastic Linear Models
can be represented in several different ways.
Here, the transfer function approach is used for Two previous conventional hydraulic system models
modeling due to the purpose of this work. neglect the effects of water compressibility and pipe
elasticity.
In many applications, a more accurate hydraulic
III.1. Linear Models
system model is necessary to take into account the
Since linear models are obtained around an operating compressibility of water and elasticity of penstock pipe.
point, they can also be called small-signal models. These These effects exhibit a dynamic interaction between
models are extracted from basic equations of turbine and the hydraulic and electrical systems. Pressure wave in the
penstock characteristics with some simplifying penstock as a hydraulic transmission line is terminated
assumptions for approximate modeling. The transfer by an open circuit at the turbine end and a short circuit at
function of linear models relates the turbine mechanical the reservoir end.
output power to the gate signal. The linear models are In order to take into account the elasticity effect in the
categorized and described as follows: transfer function of the turbine-penstock system, (3) can
be rewritten as [11]:
III.2. Simplified Linear Model
The simplest model of the hydraulic turbine-penstock
∆Pm ( s )
=
(
a23 + ( a11a23 − a13 a21 ) Z p tanh sTep ) (4)
component is the classical transfer function for an ideal ∆G ( s ) (
1 + a11Z p tanh sTep )
lossless turbine-penstock system which is given by:
The above equation is a distributed parameter model
∆Pm 1 − Tw s and it is difficult to use it in system stability and
= (1)
∆G 1 + 1 T s governor tuning studies for linear control methods.
w Therefore, tanh function is usually replaced by using the
2
Maclaurin series approximation as following [2]:
The superbar “ ” indicates normalized values based
n =∞ ⎡
⎛ sT ⎞ ⎤
2
on steady state operating point values. The normalized
values of parameters will be used often hereafter. It sTep ∏
⎢1 + ⎜ ep ⎟ ⎥
nπ ⎠ ⎥
n =1 ⎢
⎣ ⎝
should be mentioned that G is the ideal gate opening (
tanh Tep s = )
n =∞ ⎡
⎦
2⎤
(5 )
which is defined based on the change of real gate ⎛ 2sTep ⎞
opening from no load to full load being equal to 1 per ∏
⎢1 + ⎜
⎜ ( 2n − 1) π ⎟⎟ ⎥
⎥
unit, i.e.: n =1 ⎢
⎣ ⎝ ⎠ ⎦

1 Thus, the transfer function of (4) can be expressed in a

G = At × g = ×g (2)
g FL − g NL rational polynomial form.
Linear models are generally useful for control system
studies using linear analysis techniques (root locus,
The above transfer function indicates that how the
frequency response, etc.) and provide a good insight into
turbine power output changes in response to a change in
the basic characteristics of the hydraulic system
gate position. This model is valid for small disturbances
dynamics.
around the operating point.
These models are inadequate for large variations in
power output or system frequency [5].
III.3. Non-ideal Linear Model
Non-ideal turbine-penstock model can be expressed III.5. Nonlinear Models
by the following transfer function [3]:
Appropriate nonlinear model is required for large-
∆Pm a23 + ( a11a23 − a13 a21 ) Tw s
signal time-domain simulations such as islanding, load
= (3) rejection, system restoration, etc. Hydrodynamics and
∆G 1 + a11Tw s mechanic-electric dynamics are included in nonlinear
models.
Coefficients a11 and a13 are partial derivatives of water This type of modeling is especially important for
flow with respect to head and gate opening and the hydro power plants with long penstock.

Copyright © 2012 Praise Worthy Prize S.r.l. - All rights reserved International Review on Modelling and Simulations, Vol. 5, N. 4

1782
 R. A. Naghizadeh, S. Jazebi, B. Vahidi

The non-linearity of the model comes from the valve The required equations for calculating theses
characteristic of the turbine. Nonlinear models can be parameters are presented in the next section. The output
generally represented by the block diagram shown in Fig. power/torque of the turbine is transferred to a
3 [3]. It should be mentioned that Pr is used to take into synchronous generator.
account the selection of the power base such as turbine The speed of the generator is kept constant by the
MW, generator MVA, or a common MVA base in the governing system.
whole studied power system.
IV. Parameter Determination
g G ÷ H U This section describes and provides the required
At π Σ − F ( s) Σ π
U/G equations for estimating parameters of the
Pm
aforementioned models as following:
H0 U NL

Te IV.1. Water Starting Time (Tw)

Tm 1 It stands for the time required for a head H0 to
÷ Pr Σ ω
TM s accelerate the water flow in the penstock from standstill
to U0. This time can be calculated by the following
equation:
LU 0
Fig. 3. General form of nonlinear hydraulic turbine-penstock model Tw = (9)
with isolated generator ag H 0
The transfer function F(s) relates turbine velocity to
the turbine head: Therefore, Tw varies with operating point. Its value
∆U mainly depends on U0, because other parameters remain
F (s) = (6) almost unchanged. Typical values of Tw vary between
∆H 0.5 s and 4.0s [3].
Depending to the selection of F(s), nonlinear models
can have one of the following categories: IV.2. Turbine Coefficients (aij Coefficients)
These coefficients depend on machine loading and
III.6. Simplified Nonlinear Models may be determined from the turbine characteristics called
Hill charts at the operating point. Since these charts are
In this model, the traveling pressure wave and water usually unavailable, some proposed equations which are
compressibility is neglected. Then, F(s) is described as: extracted from turbine characteristic equations can
generally be used (see Table I).
1
F (s) = − (7) However, more accurate values of these parameters
Tw s can be obtained from Hill charts of the turbine [1].

TABLE I
III.7. Nonlinear Models Assuming Elastic PROPOSED EQUATIONS FOR TURBINE COEFFICIENTS
Water Column Coefficient Definition Ref.[13] Ref. [14]
The aforementioned model, assuming inelastic water ∂U G
column is adequate only in short or medium length a11 0.5G
∂H 2 H
penstocks. The pressure wave phenomenon or water
hammer effect can be represented by considering F(s) as ∂U
a12 0 0
[3]: ∂ω
∂U
a13 1.0
−1 −1 H
F (s) = ≈ (8) ∂G
φ p + Z p tanh (Tep s ) (
Z p tanh Tep s ) ∂Pm ⎛3G ⎞
a21 1.5η G At Pr ⎜ H − U NL ⎟
∂H ⎝2 ⎠
It should be noted that φ p can be neglected, especially ∂Pm
a22 0 Da G
for governor tuning studies because of its positive effect ∂ω
on damping of system oscillations and overshoot [12]. ∂Pm
a23 η At Pr H
3
2
− Da ( ω − 1)
The parameters of the presented models need to be ∂G
estimated before simulation.

Copyright © 2012 Praise Worthy Prize S.r.l. - All rights reserved International Review on Modelling and Simulations, Vol. 5, N. 4

1783
 R. A. Naghizadeh, S. Jazebi, B. Vahidi

IV.3. Elastic Time (Te) can also be bypassed if desired. The transfer function of
transient droop compensation is given by:
Te is the traveling time of the pressure wave in
penstock pipe which can be calculated by the following
sTR
equation: GTD ( s ) = RT (13)
1 + sTR
penstock length L
Tep = = (10)
pressure wave velocity a The model of the typical hydraulic turbine governor is
shown in Fig. 4.
Typical values of pressure wave velocity (a) are 1220 Maximum gate
Rmax open position=1
m/s for steel pipes and 1420 m/s for cement tunnels [3].
1 1 1
Σ Σ Ks
1 + sTG
g
1 + sTP s
IV.4. No-Load Velocity (UNL)
ωr Rmax close Minimum gate
The normalized value of UNL which is required for the position=0

nonlinear model is calculated by [3]: Σ Rp

U NL = At g NL H 0 (11) RT
sTR
1 + sTR

Fig. 4. Hydraulic turbine governor model [3]

IV.5. Hydraulic surge impedance of the conduit (Zp)
Zp is expressed by:
V.2. Electrohydraulic (PID) Governors
TW
Zp = (12) Nowadays, speed governors for hydraulic turbines use
Tep
electrohydraulic systems.
These governing systems are realized by electric
These equations are used for calculating parameters of components which provide greater flexibility and better
a real hydro power plant in Section VII. performance. These types of governors are mostly
designed in PID controller form which is widely applied
V. Hydraulic Turbine Governor Systems in industry.

A governor regulates the speed and power output of a KP

prime mover as a control system. The governor includes
mainly a controller function, and one or more control 1 1
KI
actuators [6]. Σ Σ Σ (1 + sTP )
2
1 + sTG
g
s
It should be mentioned that hydro turbines have initial
inverse response characteristics of power to gate changes ωr
sK D
due to water inertia.
Therefore, a hydro governor needs to provide a
transient droop in speed controls to limit the overshoot of Rp

turbine gate servomotor during a transient condition.

This means that for fast deviations in frequency, the Fig. 5. Block diagram of PID governor for hydraulic turbines.
governor should exhibit high regulation (low gain) while
in slow changes and steady state it should exhibit the As shown in Fig. 5, the three terms of the controller
normal low regulation (high gain). treat the present control error (Proportional), past control
Therefore, a large transient droop with a long resetting error (Integral), and predicted control error (Derivative).
time is required [5]. This feedback limits the movement PID controllers ensure faster speed response by
of the gate blades until the water flow and mechanical providing both transient gain reduction and increase. The
power output have time to overtake. derivative term in the control action is important in the
case of isolated operation [7].
V.1. Hydraulic-Mechanical Governor The use of high derivative gain will lead to
oscillations or instability in the case of connection to a
The governing function of such governors is realized strong interconnected system.
with the use of mechanical and hydraulic components It should be mentioned that the transfer function of
[15]. A large temporary droop compensation for stable PID without derivative part (PI governor) is actually
operation of the governor is provided by a dashpot. This equivalent to that of the hydraulic-mechanical governor.

Copyright © 2012 Praise Worthy Prize S.r.l. - All rights reserved International Review on Modelling and Simulations, Vol. 5, N. 4

1784
 R. A. Naghizadeh, S. Jazebi, B. Vahidi

VI. Tuning of Hydro Governors VI.2. Tuning of PID Governors

The most critical condition for governor tuning would Conventional PID governors are usually tuned using
be with the unit supplying an isolated load at maximum linear control theory tools. In order to use these methods,
output [6]. the turbine-penstock model has to be linear. Therefore,
Therefore, this point is considered for governor tuning nonlinear models have to be linearized around the
in present work. operating point. This procedure is described later in this
To the knowledge of the authors, there is no work that section.
has considered elasticity effect of water column and However, most of the works in the subject have
nonlinear characteristic of the turbine simultaneously for considered linear models for this purpose. A general
tuning hydro governors. The elasticity effect represents a method is presented by Hagihara [20] which is presented
delay e-2sTe; which is irrational term in the hydraulic in Table III. It is worth nothing that if the gains Kp, Ki,
structure. and Kd of the PID governor set to the values of first row
A transfer function with irrational term is difficult to in Table III, the transfer function of the PID governor
solve and sometimes cannot be used directly in stability would be the same as that of the temporary droop
studies. The study made for relatively long penstock on governor represented by Hovey [17]. The setting of
the basis of non-elastic water column effect will have permanent droop (Rp) is the same as for hydraulic-
significant error. Such approach in the governor design mechanical governors described earlier.
downgrades its effectiveness. The approximation of a
high order system by a low order seeks importance TABLE III
PROPOSED SETTINGS FOR PID GOVERNORS [20]
particularly in the controller design and control system KP KI KD
analysis. A simple model involves less computation time TM/(2TW) TM /(8T2w) 0
of the transient response. 0.8TM/TW 0.24TM/ T2w 0.27TM
Thus for designing efficient control system, it
becomes necessary to use reduced order turbine-penstock
model with elastic water column effect of a long VI.3. Linearization of the Nonlinear Model
penstock in hydropower plant. The linearization procedure is described here for the
nonlinear model shown in Fig. 3. The equations given
VI.1. Tuning of Hydraulic-Mechanical Governors for the penstock-turbine system in per unit are given as
following [1]:
The steady state performance of governors can be
U =G H (14)
determined by the permanent droop or speed regulation
(Rp). This droop is typically set at the range of 0.03 to
0.06. In addition, in the case of two or more generator The tanh(Teps) term in (8) can be rewritten in
units in parallel, the purpose of the droop is to ensure polynomial form using (5). So F(s) for n=0 to 2 can be
equal power sharing between the units [3]. calculated by the following equations:
The hydraulic-mechanical governor settings that can
∆U −B
be tuned to control the dynamic performance of the F (s) = = (15)
generating unit are TR, RT, and Ks. For stable operation ∆H Z pTep As
under islanding conditions (worst case), the proposed
proper choice for the temporary droop RT and reset time
n = 0 → A = 1, B = 1 (16)
TR is given in Table II (TM=2Hg).
2 2
TABLE II ⎛ Tep ⎞ 2 ⎛ ep ⎞
T
PROPOSED SETTINGS FOR HYDRAULIC-MECHANICAL GOVERNORS n = 1 → A = 1+ s ⎜ 2
⎟ , B = 1 + 4s ⎜ ⎟ (17)
Ref Temporary Droop (RT) TR ⎝ π ⎠ ⎝ π ⎠
[16] 2.5 TW TM 5.9TW
[17] 2 TW TM 4TW ⎧ 2
5 2 ⎛ Tep ⎞ 1 4 ⎛ Tep ⎞
4
⎪ A = 1+ s ⎜ ⎟ + s ⎜ ⎟
[18] 2 TW TM 5TW
⎪⎪ 4 ⎝ π ⎠ 4 ⎝ π ⎠
[19] [ 2.3 − (TW − 1.0 ) 0.15] TW TM [5.0 − ( TW − 1.0 ) 0.5] TW n=2→⎨ (18)
2 4
⎪ 40 2 ⎛ Tep ⎞ 16 4 ⎛ Tep ⎞
⎪B = 1 + s ⎜ ⎟ + s ⎜ ⎟
Moreover, Ks should be set as high as possible. The ⎪⎩ 9 ⎝ π ⎠ 9 ⎝ π ⎠
above settings provide good performance during the
most severe isolated conditions and slow response during The above equations can become linearized around
loading in normal interconnected operation. the operating point x0 by substituting x=x0+∆x for
Therefore it is recommended that the reset time TR be different variables and dropping the high orders of ∆x.
preferably set to 0.5 s [3]. As the first step:

Copyright © 2012 Praise Worthy Prize S.r.l. - All rights reserved International Review on Modelling and Simulations, Vol. 5, N. 4

1785
 R. A. Naghizadeh, S. Jazebi, B. Vahidi

U 0 + ∆U = ( G0 + ∆G ) H 0 + ∆H (19) Tep =
penstock length
=
pressure wave velocity in steel pipes
Applying a Taylor expression of polynomial raised to L 514
= = = 0.357 s
the ½ power, and dropping out the initial terms, the a 1440
radical term can be linearized and the following equation T 1.202
is derived [21]: Zp = W = = 3.367
Tep 0.357

∆H
∆U = ∆G H 0 + G0 H 0 (20) Turbine rated MW is chosen as base power for the
2H 0
sake of simplicity, so:

The relation of mechanical power, head, and flow in Pr = Turbine MW / Base Power = 1
turbine is:
Pm = UH (21) Turbine coefficients (aij) can be calculated by the
values and equations of Table I. In order to use the
By applying the same procedure for the above proposed equations in [13] and [14] it is assumed that
equation, following expression can be obtained: neat head (H) remains unchanged for all operating
points, then:
∆Pm = U 0 ∆H + ∆UH 0 (22)
H 0 = 1.0 pu, ω0 = 1.0 pu , U 0 = 1.0 pu
By substituting ∆H from (15) into (20) and (22)
∆U ∆G and ∆Pm ∆U can be calculated. 1 1
At = = = 1.111
Consequently, the final transfer function is determined g FL − g NL 0.96 − 0.06
as follows:
GFL = At × g FL = 1.111× .96 = 1.067
3
∆Pm ∆Pm ∆U H 0 B − U 0Tw As
= × = 2 H 02 (23) U NL = At g NL H 0 = 1.111× 0.06 × 1.0 = 0.067
∆G ∆U ∆G 2 H 0 B + U 0Tw As

where A and B are defined in (16)-(18). Then, turbine coefficients are calculated by the above
parameters and the results are presented in Table V. Ideal
and proposed typical values are also presented in this
VII. Simulation and Discussion table.
The values of the last column are used here for
Parameter determination and simulation of a real
simulation of the studied power plant.
hydro power plant (Abbaspour power plant) is presented
in this section. The required data for parameter TABLE IV
determination is given. The prepared models are used for ABBASPOUR HYDRO POWER PLANT DATA
comparison of their performance and PID governor Parameter Symbol Value
tuning. Penstock Length L 514 m
Diameter D 4m
1440
Pressure wave velocity a
VII.1. Case Study m/s
Turbine Rated mechanical power Pmr 256 MW
A real hydro power plant of Iranian grid with Francis Rated discharge Qr 72 m3/s
turbine is selected for this work. The required data of this Rated head Hr 250 m
Rated efficiency η 0.94
plant is presented in Table IV. Gate position at rated
condition
g FL 0.96

VII.2. Parameter Determination Gate position at No-load g NL 0.06

Turbine damping Da 0.5
Calculation of required parameters for modeling Generator Rated power Pgr 250 MW
Abbaspour power plant at nominal operating point is Frequency f 50 Hz
Inertia time constant Hg 4.5 s
presented as an example:
It should be mentioned that the transfer function of
Qr 72
L 514 × pilot valve and main servo motor is not considered in
π × ( D / 2) π × ( 4 2)
2 2
LU r most of the previous works. However, realistic values for
Tw = = = = 1.202 s Tp and TG is reported by Sanathanan as 0.1 s and 0.15 s
ag H r ag H r 9.8 × 250
respectively [22]. These transfer functions are neglected

Copyright © 2012 Praise Worthy Prize S.r.l. - All rights reserved International Review on Modelling and Simulations, Vol. 5, N. 4

1786
 R. A. Naghizadeh, S. Jazebi, B. Vahidi

in the majority of the previous works. Their effect is Comparison of different models shows that linear or
studied here for governor tuning studies. simplified models approximate the detailed models. The
difference is more significant in transient state or higher
TABLE V frequencies. Pressure wave traveling effect can be
TYPICAL AND CALCULATED VALUES OF TURBINE COEFFICIENTS observed in the responses of elastic linear and exact
Ideal Typical at Typical
Ref. Ref. nonlinear models with a time period of about 2×Tep (i.e.
Coefficient lossless full load at no-
[13] [14] 0.72 s).
[3] [23] load [23]
a11 0.5 0.58 0.57 0.534 0.5
a12 0 0 0 0 0
a13 1.0 1.1 1.1 1.0 1.0 VII.4. PID Governor Tuning
a21 1.5 1.4 1.18 1.505 1.630
a22 0 0 0 0 0.5 The studied power plant is facilitated with a PID
a23 1.0 1.5 1.5 0.940 1.111 governor. The extracted models can be used here for
tuning this PID governor by adequate techniques. The
performance of the governor must be evaluated in worst
VII.3. Comparison of Turbine-Penstock Models condition to ensure the stable operation of the unit in
In order to compare five different reviewed models of different situations. According to the IEEE standard,
hydro turbine and penstock system, the step response of hydro governors can usually control 10% of the rated
the models are presented. Fig. 6 and Fig. 7 show the step load shedding in isolated conditions due to some
response of linear and nonlinear models respectively to a practical considerations and limits [6]. Therefore, these
10% reduction in gate opening. conditions should be considered as the worst case
scenario to verify the performance of governor
Linear Models' Response controller. General rules for tuning a PID controller are:
1.35
1. Use KP to reduce the rise time.
1.3 2. Use KD to reduce the overshoot and settling time.
3. Use KI to eliminate the steady-state error.
1.25 Non-ideal
These rules work in many cases, but it is necessary to
1.2 have appropriate starting points of above parameters.
1.15
Non-ideal elastic Some methods are proposed to find appropriate set of
P (pu)

initial parameters such as Ziegler-Nichols (ZN) method.

m

1.1 This method was first introduced in 1942 [24]. This

1.05
Simplified method can also be applied on the plants that are not
mathematically known but with available experimental
1 step responses on-site. Basically there are two ZN tuning
0.95 rules including first method and second method [25].
The first method is only applicable when the step
0.9 response of the plant is S-shaped. If the plant involves
0 1 2 3 4 5
Time(s) integrator or dominant complex-conjugate poles, then the
second method is applied. The second method is used
Fig. 6. Response of linear turbine-penstock models to a 10% reduction here for PID governor tuning.
in gate opening
The second ZN approach is based on determining the
Nonlinear Model Response ultimate gain and period that result in marginal stability
1.25 when only proportional control function is used. For
1.2
linear systems, the continuous oscillation mode
corresponds to the critical stable condition. Such
n=0 or F(s)=1/Tws
1.15 condition can be easily determined through critical gain
n=1
1.1
kcr and critical oscillation period Pcr = 2π/ωcr that ωcr is
n=2 the crossover frequency. The primary PID controller
P (pu)

1.05 gains then could be calculated by expressions presented

m

Exact
1
in Table VI. It is helpful to note that the critical gain and
crossover frequency can be obtained by “margin”
0.95 command in Matlab software environment [26]. These
values can also be obtained for nonlinear models by trial
0.9
and error.
0.85
0 1 2 3 4 5 TABLE VI
Time(s)
ZIEGLER-NICHOLS PID TUNING RULE
KP KI KD
Fig. 7. Response of nonlinear turbine-penstock models to a 10%
reduction in gate opening 0.6 × kcr 1.2 × kcr pcr 0.075 × kcr × pcr

Copyright © 2012 Praise Worthy Prize S.r.l. - All rights reserved International Review on Modelling and Simulations, Vol. 5, N. 4

1787
 R. A. Naghizadeh, S. Jazebi, B. Vahidi

ZN tuning rule is applied to all of the studied models. Speed Response without Servo Transfer Functions
1.04
In addition, the linearized model of the nonlinear model
1.035
is used for this purpose. The tuned parameters of PID
governor by ZN rule using different models are 1.03

presented in Table VII and compared with the PID 1.025 ZN Method PID
parameters which are calculated by proposed equations 1.02
Hagihara PID

Speed (pu)
Hagihara PI
by Hagihara [20] (See Table III). 1.015
It should be mentioned that the proposed equation by
1.01
Hagihara is only applicable to the simplified linear model
1.005
without the effect of pilot and gate servo transfer
functions [22]. However, ZN tuning rule is more general 1

and easy to use for more complicated models. In order to 0.995

compare the performance of tuned PID governor, the 0.99
0 2 4 6 8 10 12 14 16 18 20
response of the system to a 10% rated load shedding is Time (s)
illustrated in Fig. 8 to Fig. 10.
Fig. 8. Response of simplified linear turbine-penstock model to a
TABLE VII 10% load shedding with PI and PID governor
PID GOVERNOR SETTING RESULTS without servo transfer functions
Model/Method KP KI KD
Without Hagihara (PI) 3.747 0.780 0 Speed Response with Servo Transfer Function
1.04
servo Hagihara (PID) 6.00 1.50 2.43
Transfer
Linear Simplified 6.75 1.69 3.07
Function 1.03
Linear Simplified 7.06 0.90 3.55
With servo Linear Non-ideal 6.52 0.84 3.24 1.02 Simplified
Transfer Linearized with n=1 5.19 0.83 3.54 Non-ideal
Speed (pu)

Function Linearized with n=2 5.16 0.87 3.57 Elastic

Nonlinear 3.57 0.24 3.35 1.01

1
The settings of Table VII are applied to the studied
models with isolated load condition to verify the
0.99
performance of the governor. It should be mentioned that
the initial KP obtained by ZN method needs to be
0.98
increased about 50% to improve the response speed by 0 2 4 6 8 10 12 14 16 18
Time (s)
reducing the rise time and settling time of the response.
The results show that the performance of the PID
Fig. 9. Response of linear turbine-penstock models to a 10% load
governor tuned by ZN method is appropriate for stable shedding with PID governor and servo transfer functions
operation of the studied hydro power plant.
As shown in Fig. 8, the response of the system with Speed Response with Servo Transfer Function
1.04
the governor settings determined by ZN method is
slightly better than the response with settings of Table
1.03
III. Fig. 9 shows that selection of linear model type has
almost no effect on the performance of ZN method. Linearized (n=1)
1.02
Comparison of the response of simplified model with Linearized (n=2)
Speed (pu)

and without governor servo transfer functions indicates Nonlinear Elastic

1.01
that they have no considerable effect on tuning results
and can be neglected in tuning by ZN method. It can also
1
be concluded that the order of linearization of the
nonlinear model has also insignificant effect (See Fig. 0.99
10). The response of the nonlinear model has less
overshoot but is slower than other models. 0.98
The experiment in this work showed that the response 0 2 4 6 8 10 12 14 16 18 20
Time (s)
of the nonlinear elastic model as the most comprehensive
one among the reviewed models is more sensitive to Fig. 10. Response of linearized and nonlinear turbine-penstock models
adjusting PID parameters compared to other simpler to a 10% load shedding with PID governor with servo transfer functions
models. Finally, it is important to note that the proposed
settings in [20] and initial governor settings obtained by
applying ZN method on other models are unusable for
VIII. Conclusion
nonlinear model, because they lead to instability. This paper reviewed the proposed models for dynamic
simulation of hydro power plant components and

Copyright © 2012 Praise Worthy Prize S.r.l. - All rights reserved International Review on Modelling and Simulations, Vol. 5, N. 4

1788
 R. A. Naghizadeh, S. Jazebi, B. Vahidi

provided a proper view of their behavior. The required Conversion, vol. 20, no. 2, Jun. 2005, pp.424-434.
[15] D. G. Ramey and J. W. Skooglund, Detailed Hydrogovernor
equations for the determination of the model parameters
Representation for System Stability Studies, IEEE Trans. on
have been presented. Different models have been Power App. and Systems, vol. PAS-89, no. 1, Jan. 1970, pp. 106-
compared for simulation of a real hydro power plant. The 112.
extracted models are then used for hydro governor [16] H. M. Paynter, A Palimpsest on the Electronic Analog Art, Geo,
tuning. The previous methods for tuning hydro A. Philbrick Researches Inc., Boston, Mass., 1955.
[17] L.M. Hovey, Optimum Adjustment of Hydro Governors on
governors have been also reviewed and a classical Manitoba Hydro System, AIEE Trans., vol. PAS-81, Part 111,
heuristic method is utilized for this purpose. Dec. 1962, pp. 58 1-587.
This work can be a useful guide for power [18] F.R. Schleif and A.B. Wilbor, The Coordination of Hydraulic
engineering students who want to use an appropriate Turbine governors for Power System Operation, IEEE Trans., vol.
PAS-85, Jul. 1966, pp. 750-758.
hydro power plant model including the penstock, turbine, [19] P.L. Dandeno, P. Kundur, and J.P. Bayne, Hydraulic Unit
governor, and generator for stability studies. This paper Dynamic Performance under Normal and Islanding Conditions -
provides required background knowledge for studying Analysis and Validation, IEEE Trans., vol. PAS-97, no. 6, Nov.
the effect of actual data on the model parameters and 1978, pp. 2134-2143.
[20] S. Hagihara, H. Yokota, K. Goda, K. Isobe, Stability of a
behavior. The review and comparison of different hydraulic turbine generating unit controlled by PID governor,
models help the electrical engineer to choose the best IEEE Trans. on Power App. and Systems, vol. PAS-98, no. 6,
model for specific studies. Furthermore, the appropriate Nov/Dec. 1979, pp. 2294-2298.
tuning of the conventional PID governor using a simple [21] L. N. Hannett, J. W. Feltes, and B. Fardanesh, Field Tests to
Validate Hydro Turbine-Governor Model Structure and
approach has been presented which can be used for Parameters, IEEE Trans. Power Syst., vol. 9, no. 4, Nov. 1994, pp.
educational purposes and real applications. 1744–1750.
[22] C. K. Sanathanan, A Frequency Domain Method for Tuning
Hydro Governors, IEEE Trans. on Energy Conversion, vol. 3, no.
References 1, Mar. 1988, pp.14-17.
[23] R. Oldenburger and J. Donelson JR., Dynamic Response of a
[1] Chaudhry, M.H., Applied hydraulic transients. 2nd ed., New Hydroelectric Plant, IEEE Trans. Power Apparatus and Systems,
York, 1987. vol. 81, no. 3, October 1962, pp. 403-419.
[2] P.M. Anderson, and A.A. Fouad, Power System Control and [24] J. G. Ziegler and N. B. Nichols, Optimum Settings for Automatic
Stability, 2nd Edition, IEEE Press, New York, 2003. Controllers, Trans. ASME, vol. 64, 1942, pp. 759-768.
[3] P. Kundur, Power System Stability and Control, McGraw-Hill, [25] K. Ogata, Modern Control Engineering, Prentice Hall, 1996.
New York, 1994. [26] K. Ogata, Matlab for Control Engineers, Prentice Hall, 2007.
[4] IEEE Committee Report, Dynamic Models for Steam and Hydro
Turbines in Power System Studies, IEEE Trans. Power Apparatus
and Systems, vol. 92, no. 6, Nov. /Dec. 1973, pp. 1904-1915.
[5] IEEE Working Group on Prime Mover and Energy Supply
Authors’ information
Models for System Dynamic Performance Studies, Hydraulic 1
Department of Electrical Engineering, Amirkabir University of
Turbine and Turbine Control Models for System Dynamic Technology, Tehran, Iran.
Studies, IEEE Trans. Power Systems, vol. 7, no. 1, February 1992,
pp. 167-179. 2
Department of Electrical Engineering, Amirkabir University of
[6] IEEE Guide for the Application of Turbine Governing Systems Technology, Tehran, Iran.
for Hydroelectric Generating Units, IEEE Standard 1207-2004,
Nov. 2004. 3
Department of Electrical Engineering, Amirkabir University of
[7] N. Kishora, R.P. Sainia, and S.P. Singhb, A Review on Technology, Tehran, Iran.
Hydropower Plant Models and Control, Renewable and
Sustainable Energy Reviews, vol. 11, 2007, pp.776-796. Ramezan Ali Naghizadeh was born in
[8] B. Vahidi, M. R. Bank Tavakoli, W. Gawlik, Determining Hamedan, Iran, in 1981. He received the B.S.
Parameters of Turbines Model Using Heat Balance Data of Steam degree in electrical engineering from K. N.
Power Unit for Educational Purposes,IEEE Trans. Power Syst., Toosi University of Technology, Tehran, Iran, in
vol. 22, no. 4, Nov. 2007, pp. 1547-1553. 2004 and M.S. degree in electrical engineering
[9] M. R. Bank Tavakoli, B. Vahidi, W. Gawlik, An Educational from Amirkabir University of Technology,
Guide to Extract the Parameters of Heavy Duty Gas Turbines Tehran, in 2007. Currently, he is pursuing the
Model in Dynamic Studies Based on Operational Data, IEEE Ph.D. degree at the Department of Electrical
Trans. Power Syst., vol. 24, no. 3, Aug. 2009, pp. 1366-1374. Engineering of Amirkabir University of Technology. His main fields of
[10] R. A. Naghizadeh, B. Vahidi, and M. R. Bank Tavakoli, interest are component modeling and transients of power systems. He
Estimating the Parameters of Dynamic Model of Drum Type also has been with the IGMC Company, working on extensive analysis
Boilers Using Heat Balance Data as an Educational Procedure, of the Iranian power system. He is currently with the Iran Grid Secure
IEEE Trans. Power Syst., vol. 26, no. 2, May 2011, pp. 775-782. Operation Research Center.
[11] L. Wozniak and G. H. Fett, Conduit representation in closed loop
simulation of hydroelectric system, Trans. of ASME, Journal of Saeed Jazebi was born in 1983, Kerman, Iran.
Basic Eng, Sep. 1972., pp. 599-604. He received the B.S and M.S degree in 2006 and
[12] C. K. Sanathanan, Accurate Low Order Model for Hydraulic 2008 in electrical engineering from Shahid
Turbine-Penstock, Trans. on Energy Conversion, vol. EC-2, no. 2, Bahonar University, Kerman, Iran and
Jun. 1987, pp. 196-200. Amirkabir University of Technology, Tehran,
[13] D.N. Konidaris and [Link], Investigation of oscillatory Iran, respectively. He is currently PhD student at
problems of hydraulic generating units equipped with Francis the Amirkabir University of Technology. His
turbines, IEEE Trans. on Energy Conversion, vol. 12, no. 4, Dec. field of interest includes High voltage,
1997, pp.419-425. distribution system planning, optimization techniques, FACTS devices
[14] D.N. Konidaris and [Link], Speed and Active Power and statistical pattern recognition applications in power engineering.
Control of Hydro Turbine Unit, IEEE Trans. on Energy

Copyright © 2012 Praise Worthy Prize S.r.l. - All rights reserved International Review on Modelling and Simulations, Vol. 5, N. 4

1789
 R. A. Naghizadeh, S. Jazebi, B. Vahidi

Behrooz Vahidi, was born in Abadan, Iran, in

1953. He received the B.S. degree in electrical
engineering from Sharif University of
Technology, Tehran, Iran, in 1980, the M.S.
degree in electrical engineering from Amirkabir
University of Technology, Tehran, in 1989, and
the Ph.D. degree in electrical engineering from
the University of Manchester Institute of Science
and Technology, Manchester, U.K., in 1997. From 1980 to 1986, he
worked in the field of high voltage in industry as a Chief Engineer.
From 1989 to the present, he has been with the Department of Electrical
Engineering of Amirkabir University of Technology, where he is now a
professor. He is selected by the ministry of higher education of Iran and
by IAEEE (Iranian Association of Electrical and Electronics Engineers)
as the distinguished researcher of Iran. Prof. Vahidi is senior member of
IEEE and member of CIGRE WG C4.26. His main fields of research
are high voltage, electrical insulation, power system transient, lightning
protection, and pulse power technology. He has authored and
coauthored more than 250 papers and five books on high-voltage
engineering and power system.

Copyright © 2012 Praise Worthy Prize S.r.l. - All rights reserved International Review on Modelling and Simulations, Vol. 5, N. 4

1790

View publication stats