2. 292 ACTIVE POWER AND FREQUENCY CONTROL
Figure 6.1. Map of UCTE system i n Eu rope.
from the prime movers (turbines) that will be stored in the generator inertias and result in
acceleration of the masses to increase frequency.
In order to understand how load frequency and control works in a practical inter
connected system, it is necessary to understand the relationship between the size of
the system (large or small MW systems), generation inertia (H, in MW-secI MVA), and the
nature and magnitude of the disturbances that cause frequency deviations. Two large
interconnected systems will be discussed: the UCTE in Europe and the NERC
3
[2]
interconnections in the United States. However, the discussion can be generally applied
to large and small systems in other countries or regions.
The UCTE system is a very large interconnected electric system shown in the map in
Figure 6.1, approximately 600,000 MW peak, with an annual electricity consumption of
2300 TWh supplied by generators through a highly interconnected transmission system
comprising 200,000 km of 400 kV and 220 kV lines.
In comparison, the electricity consumption in the United States was 4055 TWh in 2005
[3], but is supplied by three interconnected systems operating independent of each other,
namely the Western Interconnection (WECC)4 of approximately 160,000 peak MW [4],
the ERCOT region in Texas (about 70,000 peak MW), and the rest comprising the Eastern
Interconnection (about 660,000 peak MW). The map of the NERC regions is shown in
Figure 6.2.
The U.S. interconnections are electrically connected to Canada in the north and Baja
Mexico in the south, but data presented herein is from the US government Energy
Information Administration (EIA) for the United States part only.
3 NERC web site is www.nerc.com. previously North American Electric Reliability Council, now North American
Electric Reliability Corporation. [2] .
4 WECC is the Western Electricity Coordinating Council [4].
3. FREQUENCY DEVIATIONS IN PRACTICE
NERC REGIONS
Figure 6.2. N ERC reg ions i n the U n ited States and Ca nada.
6.2 FREQUENCY DEVIATIONS IN PRACTICE
6.2.1 Small Disturbances and Deviations
Typical small frequency deviations in the UCTE system are seen in Figure 6.3 [1], which
illustrates that during the space of approximately 4 min, the deviation from the nominal
50 Hz operation has proceeded from negative to positive, that is, undergeneration to
overgeneration.
These small deviations may be due to a variety of mismatch reasons including random
differences in load, mismatch due to incorrect load forecasts versus system dispatch, slow
or fast ramps not in synchronism with load deviations, AGC (automatic generation control)
actions resulting in differentials that will eventually be corrected, and so on.
6.2.2 Large Disturbances and Deviations
Large deviations may be due to generator trips or load trips, remedial system actions (RAS)
also known as special protection system (SPS) actions that result in generator or load
UCTE Frequency: 49�" c· Hz < 0 4:: 0:
50'021�
50,000
49,979
UCTE Frequency: :,0 n: 1 Hz :0 5� 05
50'024�
50,000
49,976
Figure 6.3. Typical sma l l freq uency deviation
responses i n U CTE.
293
4. 294 ACTIVE POWER AND FREQUENCY CONTROL
Malin 500 kV Bus frequency - May 18th 2001 Test NW 1250 MW Trip
60.02,....---r---r---,r--T"""--r----.---r---r----r----,
60
59.98
59.96
59.94
59.92
59.9
59.88
59.86
Note: AGe was switched off during the test
59.84'---....1....-......&..----''---.......
--'-----'
--..1...---'-
---1.---'
-60 -40 -20 0 20 40 60 80 100 120 140
Time(s)
Figure 6.4. Large deviation freq uency response resu lti ng from a generation trip of 1250 MW i n
t h e WECC on May 18, 2001.
tripping, and so on. It should be apparent that a 1000 MW generating plant trip would
create a very large frequency deviation in a relatively smaller system such as ERCOT, but
will result in a smaller frequency deviation in WECC and an even smaller frequency
deviation in the Eastern Interconnection.
Examples of a 1250 MW generation trip in the WECC and resulting frequency
deviation is shown in Figure 6.4; these resulted from a staged test in WECC for the
purpose of determining governor response and modeling, and are discussed in detail in the
following sections below. Note that the nominal frequency in the United States and Canada
is 60 Hz as opposed to 50 Hz in Europe.
6.3 TYPICAL STANDARDS AND POLICIES FOR "ACTIVE POWER
AND FREQUENCY CONTROL" OR "LOAD FREQUENCY CONTROL"
6.3.1 UCTE Load Frequency Control
Load frequency control (LFC) is described in the "UCTE Operation Handbook" [1] as the
continuous balance between supply and demand that must be maintained for reliability and
economic operational reasons. The "UCTE Operation Handbook" is a manual including
operation policies for generation control, performance monitoring and reporting, reserves,
security criteria, and special operational measures. The basic objective of the Operation
Handbook is to ensure the interoperability among all transmission system operators (TSOs)
connected to the synchronous areas. Balance quality can be derived from system frequency,
which should not vary significantly from its set point of 50 Hz. LFC is split into the
following five controls:
5. TYPICAL STANDARDS AND POLICIES FOR ACTIVE POWER AND FREQUENCY CONTROL OR LFC
A. Primary control
B. Secondary control
C. Tertiary control
D. Time control
E. Measures for emergency conditions
Control actions are performed in different successive steps, each with different
characteristics and qualities, and all depending on each other:
• Primary control starts within seconds as a joint action of all undertakings involved;
• Secondary control replaces primary control after minutes and is put into action by the
responsible undertakings;
• Tertiary control restoressecondarycontrol reserve by rescheduling generation and is
put into action by the responsible undertakings;
• Time control corrects global time deviations of the synchronous time in the long term
as a joint action of all undertakings.
6.3.1.1 Primary Control is by Governors. The objective of primary control is to
maintain a balance between generation and consumption (demand) within the synchronous
area, using turbine speed or turbine governors. The time for starting the action of primary
control is in practice a few seconds starting from the incident (although there is no
intentional time delay for governor pickup), the deployment time for 50% or less of the
total primary control reserve is at most 15 s and from 50% to 100% the maximum
deployment time rises linearly to 30 s.
To avoid calling up of primary control in undisturbed operation at or near nominal
frequency, the frequency deviation should not exceed ±20 mHz. This reduces wear and
tear of the governors due to too frequent operation and results in operation beyond the
dead band of the governor. Load shedding schemes start at a frequency of 49 Hz and
below; hence, the instantaneous frequency should not fall below 49.2 Hz. The maximum
dynamic frequency should not exceed 50.8 Hz. Each control area should contribute to
primary control reserves. Similar parameters exist for the 60 Hz systems in the United
States.
6.3.1.2 Secondary Control by Automatic Generation Controls (AGCs). Sec
ondary control maintains a balance between generation and consumption (demand) within
each controlarea/blockas well as the systemfrequencywithin the synchronous area, taking
into account the control program, without impairing the primary control that is operated in
the synchronous area in parallel but by a margin of seconds.
Secondary control makes use of a centralized automatic generation control, modifying
the active power set points/adjustments of generation sets in the time frame of seconds to
typically 15 min. Secondary control is based on secondary control reserves that are under
automatic control. Adequate secondary control depends on generation resources made
available by generation companies to the transmission system operators. Secondary control
must be performed in the corresponding control center by a single automatic secondary
controller that needs to be operated in an online and closed-loop manner. In order to have
no residual error, the secondary controller must be of PI (proportional-integral) type. The
integral term must be limited in order to have a nonwindup control action, able to react
immediately in case of large changes or a change of the sign of the area control error
(ACE). Within each control area/block, the individual ACE needs to be controlled to zero
295
6. 296 ACTIVE POWER AND FREQUENCY CONTROL
on a continuous basis. The ACE is calculated as the sum of the power control error and the
frequency control error (G = I1P + K · I1f).5
6.3.1.3 Tertiary Control. Tertiary control uses tertiary reserve {15 min reserve} that
is usually activated manually by the TSOs after activation of secondary control to free up the
secondary reserves. Tertiary control is typically operated in the responsibility of the TSO.
6.3.1.4 Self-Regulation of the Load. The self-regulation of the load in all
synchronous areas cannot be mandated by regulations. It is generally assumed to be
1%/Hz; that means a load decrease of 1% occurs in case of a frequency drop of 1 Hz.
6.3.2 NERC (U.S.) Standards
The U.S. system comprises numerous control areas (Balancing Authorities-BAL)
independently controlled or via a ISO (Independent System Operator). Hence reliability
standards spell out the rules of operation rather than detailed operation requirements and
procedures as in UTCE's Handbook. NERC does not spell out "primary control" and
"secondary control" but it is widely recognized and practiced that governors are the
primary control and AGC systems are the secondary control. NERC mandates that the
governors shall pick up with a 5% droop characteristic. However, as will be described
under Section 6.4, governors, the pickup performance of governors for thermal units as
primary control devices in practice has been eroding as more and more units are currently
operated under power controllers with the resulting degradation of frequency response
during disturbances. WECC, the Western Interconnection, notably has taken a lead in
investigating the unresponsiveness of units. This is discussed further in Section 6.5.5.
The relevant NERC standards for power and frequency control currently are as follows:
• BAL-OOl Real Power Balancing Control Peiformance. The purpose is to maintain
Interconnection steady-state frequency within defined limits by balancing real power
demand and supply in real time.
• BAL-002 Disturbance Control Standard (DCS). The purpose of this is to ensure the
Balancing Authority is able to utilize its Contingency Reserve to balance resources
and demand and return Interconnection frequency within defined limits following a
reportable disturbance. Because generator failures are far more common than
significant losses of load and because Contingency Reserve activation does not
typically apply to the loss of load, the application of DCS is limited to the loss of
supply and does not apply to the loss of load.
• BAL-003 Frequency Response and Bias. This standard provides a consistent method
for calculating the frequency bias component of ACE.
• BAL-004 Time Error Correction. The purpose of this standard is to ensure that time
error corrections are conducted in a manner that does not adversely affect the
reliability of the Interconnection.
• BAL-OOSAutomatic Generation Control. This standard establishes requirements for
Balancing Authority Automatic Generation Control necessary to calculate Area
Control Error and to routinely deploy the Regulating Reserve. The standard also
ensures that all facilities and load electrically synchronized to the Interconnection
are included within the metered boundary of a Balancing Area so that balancing of
resources and demand can be achieved.
5 Further discussion of ACE is presented in Section 6.6.3.
7. SYSTEM MODELING, INERTIA, DROOP, REGULATION, AND DYNAMIC FREQUENCY RESPONSE
6.3.3 Other Countries' Standards
Standards in other countries generally follow the same definitions and approach for rules
as stated in UCTE in Europe and NERC in the United States. Local system conditions
and the relative size of the system in particular play a large part in determining their
standards and emphasizing one or other aspects of primary and secondary controls and
reserves.
6.4 SYSTEM MODELING, INERTIA, DROOP, REGULATION,
AND DYNAMIC FREQUENCY RESPONSE
Tounderstand the load frequency control problem, the basic "swing equation" for a single
generator operating with a governor droop is developed and then extended to multiple
generators operating in parallel representing an "area." Two such areas connected by a tie
line are modeled and the load frequency control is developed for a two-area model with
paralleled generators in each area. The basic form of AGC control is described using the
two-area model. The concept of "system inertia," "system droop," and system frequency
response to a disturbance are discussed in relation to the "area" models.
6.4.1 Block Diagram of the System Dynamics and Load Damping
When a disturbance occurs in the system and the system frequency deviates, each generator
experiences an accelerating or decelerating torque Ca as discussed6 in Chapters 2and 10as
well as [5,6].
If consider the electromechanical model of the generator, developed in Section 2.1.2,
given by equation (2.10)
dw
2H- + Dw = Cm - Ce � Pm - Pe
dt
do
- = WOW
dt
(2.12)
where Cm is the turbine mechanical torque; Ce is the electrical torque; H is the inertia
constant, 2H = M; M is mechanical starting time.
Note: The self-regulation of the load (termed D) in all synchronous areas is usually
assumed to be 1%/Hz; that means a load decrease of 1% occurs in case of a frequency drop
of 1 Hz. Hence D = 1 in the equation if load damping is taken.
The block diagram of the system dynamics and load damping is shown in Figure 6.5.
Simplifying the block diagram, the two blocks can be combined into a single forward
block using 1I(2Hs + D).
Figure 6.5. B l ock d iagra m of the system dynam ics
plus l oad dampi ng.
6 In practice, the generator inertia is much greater than the turbine inertia.
297
8. 298 ACTIVE POWER AND FREQUENCY CONTROL
Inertia and damping
11R 1+------'
Governor droop effect Figure 6.6. B l ock d iagra m of the system dynamics,
load dampi ng, and governor d roop.
6.4.2 Effect of Governor Droop on Regulation
The regulation of the unit is defined as "droop" R and equals
LlW
R= -
LlP
(6.1)
The influence of the droop on the frequency regulation is shown in the block diagram
in Figure 6.6.
The regulation of a unit with 4% droop is shown in Figure 6.7 for a unit loaded at 50%7
and 100% frequency. This unit will then operate at zero load with a 2% overfrequency and
at 100% load at 98% frequency or 2% underfrequency.
6.4.3 Increasing Load by Adjusting Prime Mover Power
By increasing the speed changer setting (also known as the load reference set point), the
unit gets loaded at a higher output of 90% as shown in Figure 6.8, the system frequency
remaining the same at 100%.
6.4.4 Parallel Operation of Several Generators
It is assumed that two units with different per unit settings of their speed changers both
operating at the same 100% system frequency will take up load according to their settings
as shown in Figure 6.9.
Frequency [p.u.]
R = t.j7/',.P = -0.04
1 .02
1 .00 .........................................
;
....................... System frequency
0.98 ...........................................j.......................................
I
o 0.5 1 .0 Load [p.u.]
Figure 6.7. Governor d roop operation with 50% l oad at 100% freq uency.
7 Note the range 50% from a practical standpoint is on the low side and is illustrative only.
9. SYSTEM MODELING, INERTIA, DROOP, REGULATION, AND DYNAMIC FREQUENCY RESPONSE
Frequency [p.u.]
R =!:"j1!:"P =-0.04
o 0.5 0.9 1 .0 Load [p.u.]
Figure 6.8. Governor d roop operation with 90% load at 100% freq uency.
Frequency [p.u.]
o
R =!:"j1M=-0.04
Droop = 4%
0.5 0.9 1.0 Load [p.u.]
Figure 6.9. Para l lel operation of two generators with d ifferent speed changer settings at 100%
system freq uency.
The block diagram of Figure 6.6 is modified as presented in Figure 6.10 to show
parallel generators with different droops, R1, R2, and R3. Hence, the inertia constant Hand
the damping D represent the system inertia and damping.
Isochronous operation with a zero droop is unacceptable for interconnected operation
because the droop characteristic is essential for load sharing. An isolated system may,
however, have isochronous operation.
Load
llR 2 �--l
Figure 6.10. B l ock d iagram showing para l lel opera
tion of generators.
299
10. 300 ACTIVE POWER AND FREQUENCY CONTROL
An example with different speed changer settings and different droops will make
parallel operation of generators in an isolated system more clear.
Application 1 [5]
Two generating units of rating 500 MVA and droop 6%, and 200 MVA and droop 4%,
respectively, are operating in parallel in an isolated system. They share a load of 700 MW
at 100% system frequency. Unit 1 supplies 500MW, and unit 2 supplies 200MW. If the
load decreases by 80 MW, find the steady-state frequency and generation of each unit.
Assume load varies 1% for every 1% change of frequency.
Solution
Using a base MVA of 1000 MVA and the relation (6.1), it results
• droops:
R, = (0.06)/(500/1000) = 0.12 p.u.
R2 = (0.04)/(200/1000) = 0.2 p.u.
• per unit load change: 6.PL = -80/1000 = -0.08 p.u.
• damping D = 1 p.u.
Since the final operation is at steady state, transients due to the 2Hs factor can be
ignored. The droops of individual units are given by
6.w
and R2 = �
U.P2
where the change in frequency is the same for both units. This is substituted in the following
relationship where the pickup of each generated sums up to the load change:
and gives the frequency deviation of the isolated system as follows:
where Rsys is the "equivalent system droop" and is given by
1 1 1
- = - +
Rsys R[ R2
Adding the effect of load damping, D, as seen in Figure 6.10, the frequency deviation
is given by
-0.08
( / ) ( / )
= -0.00558 p.u.
1 0.12 + 1 0.2 + 1.0
11. SYSTEM MODELING, INERTIA, DROOP, REGULATION, AND DYNAMIC FREQUENCY RESPONSE
and
N = -0.00558 p.u. x 60 Hz = -0.3349 Hz
Under these conditions, the active power deviation at unit 1 is
�w
�PI = - . 1000 = -46.51 MW
RI
and the active power deviation at unit 2 is
�w
�P2 = -- . 1000 = -27.91 MW
R2
Therefore, under the new conditions unit 1 supplies 453.5 MW and unit 2 supplies
172.1 MW at the new operating frequency of 59.6651 Hz.
The load damping is D = (-0.00558·1.0) . 1000 = -5.58 MW
Under the initial load, the equivalent system droop was
1
R
sys =
(1/0.12) + (1/0.2) + 1
= 0.0697
while after the load deviation occurrence it increases to
�w 0.00558
R
sys = -- = = 0.0725
�h 0.08
Using the machine base and droop, it gives the same valueS for pickup as given by
formula �PI = �W/RI
'
that is
• for unit 1 �Pl = -0.00558/0.06·500 = -46.51 MW, and
• for unit 2 �P2 = -0.0056/0.04·200 = -27.91 MW
Hence generator sustained pickups by governors in the system are approximately
linearly proportional to their own machine base and droop for a given frequency
deviation using the formula
�P =
�W
R
6.4.5 Isolated Area Modeling and Response
Following the above discussion, it should be evident that an isolated area or inter
connection can be approximately modeled using the concept of an equivalent system
inertia H in seconds (MW-s/MVA), an equivalent system droop Rsys, and area damping D.
8 Any small differences are due to the effect of damping, which is neglected in this calculation.
301
12. 302 ACTIVE POWER AND FREQUENCY CONTROL
Load
deviation
System inertia
and damping
--- I----.--� Mu
Frequency
deviation
Figure 6.11. B l ock d i a g ra m of system freq uency response, inertia, d roop, and dampi ng.
The equivalent formula for droop Rsys was developed in the previous section. Hsys can be
similarly developed from first principles of stored energy in the system as distributed
among generator rotating masses and is simply the summation of the MW-s of all
generators divided by the MVA sum. System inertia H values are typically in range
from 2 to 10 s.
The model of Figure 6.6 reproduced below in Figure 6.11 with some changes in
definitions can be thus used in principle for an area, since the frequency deviation is one
that is defined for the whole interconnection. The formula applicable for system droop is
1 1
- + - +
R, R2 Rsys
Both the system droop and the system inertia can be calculated approximately from
disturbance recordings after a known event such as a large generator or plant trip. As
system generation, load, and spinning reserves vary, the responses and the value of these
parameters will vary. The value of system inertia is a factor to be considered in the
permissible operating transfer capacity nomograms (SCIT) in studies for establishing
operating limits, as in the case of Southern California in WECC.
The concept of system frequency response and system droop arising from a large
generator trip was central in the formulation and analysis from system trip tests performed
in WECC that led to the development of an accurate "New Thermal Governor Model"
(Section 6.5.5). This work that included system tests with all AGCs switched off in order to
get pure governor responses only and led to accurategovernormodeling is now the basis of
development of new frequency response reserves standards and operating practices in the
WECC.
An extension of the above system area concept including a PI controller for AGC is
included in Section 6.6, which describes the basic concepts of AGC modeling.
6.5 GOVERNOR MODELING
As stated in Section 6.3, turbine governors are the "primary control" of frequency in
interconnected systems and hence its performance and modeling are of paramount
importance for operational and planning studies. It is important to understand the mix
of generation types in an interconnection and hence what can be expected from the
response of each type of governor during large frequency disturbances. In the United States
as a whole, the mix in peak generation capacity in 2005, from EIA sources [3], is given in
Table 6.1.
13. GOVERNOR MODELING
TAB l E 6.1. Generation Capacity in United States
Natural gas
Coal
Nuclear
Hydroelectric
Hydro pumped storage
Petroleum
Other
39%
32%
lO%
8%
2%
6%
3%
From system response recordings, it has been established that hydro governors have
the most sustained response to frequency deviations. Nuclear governors are block loaded
and unresponsive to frequency, and coal and gas units are largely not responsive in a
sustained manner. Since the latter constitute the largest percentage of units, primary control
responsive is largely diminished. These issues are discussed in [6]. Also the recent wind
and solar generation are unresponsive to frequency. "Secondary control" by AGC therefore
increasingly plays a larger part in the generation pickup following a large generation trip.
However, in the Western Interconnection (WECC), the generation mix is quite
different; hydroelectric plants including pumped storage make up about 31% of the
installed capacity and thermal plants (coal and natural gas) comprise 61%, with nuclear
5%. In the WECC, hydro plants dominate in the Northwest and, as shown later in Sec
tion 6.5.5, bear the brunt of the sustained generation pickup in the WECC following a large
generation trip in the system.
6.5.1 Response of a Simple Governor Model with Droop
A new model (Figure 6.12) was developed based on the simulation building blocks
shown in Figures 6.5 and 6.6 following the implementation of the "swing equation"
2H x dwldt = Pm - Pe and the droop equation R = awlap. The additional transfer
function blocks have a simple single time constant (0.5 s) for a nonreheat steam turbine
and for the governor valve. The other parameters of the model are the inertia constant
H= 5 s and damping D = 0.8. The load step in the model is 20%. The model assumes
operation in an isolated system.
Governor
valve
0.2 u(t)
Step load input
Turbine
20
--- 1----.----' IV
Sum Inertia & load
damping
lIR droop
Figure 6.12. Simple governor d roop model i n isolated operation.
303
14. 304
::i
ci.
.8
O rr---�----�---r----�---r----,
ACTIVE POWER AND FREQUENCY CONTROL
g -0.005
'g
.�
"0
G'
i3 -0.01
'"
�
�
-0.015
o
L-----":....
2
�
0
----
4
�
0
----
6
�
0
----
8�
0
----
10
�
0
--
-----'
120 Figure 6.13. Freq uency response of a
Time (s) simple governor d roop model i n iso
lated operation.
The sustained frequency deviation after transients dies away for a 20% load step, a 5%
droop governor, and 0.8 damping can be also easily calculated, that is LlU) = l/���D =
l/O�50;O.8 = -0.0096 p.u.
Figure 6.13 shows the simulated response over a 100 s run of a simple turbine governor
with step load.
6.5.2 Hydraulic Governor Modeling9
6.5.2.1 Hydraulic Turbines. A model for a hydraulic turbine can be formulated
starting from the hydrodynamic properties, which govern the turbines behavior [6-8]. A
block diagram of the model is shown in Figure 6.14, mentioning that the model is nonlinear
(see also Figure 3.39).
The variables in Figure 6.14 are defined as follows:
g is gate position from governor;
At converts the actual turbine gate position g to the effective gate position G;
G is effective gate position;
q* is water flow rate, p.u.;
qnh is water flow rate at no load, in p.u.;
h* is head, in p.u.;
ho* is initial head, in p.u.;
Figure 6.14. Hyd ra u l ic turbine.
9 Additional information is provided in Chapter 3, Section 3.6.2.
15. GOVERNOR MODELING
Tw is water starting time;
w/wo is per unit generator speed;
Prat is ratio of turbine rating to generator rating;
Cm is per unit generator torque.
The actual gate position is converted to the effective gate position by
where gf] is the gate position at full load and gnl is the gate position at no-load.
The turbine head h*, per unit, is given by
and the water velocity q* is
The turbine output power is
and the generator torque Cm is
dq*
dt
The water flow at zero load is obtained from
(6.2)
(6.3)
(6.4)
(6.5)
(6.6)
(6.7)
The initial head ho* varies seasonally for most hydraulic turbines. Under rated
conditions ho* is unity. The coefficient Prat' converts the turbine torque to the corresponding
generator per unit torque.
6.5.2.2 Hydraulic Governors. The block diagram of Figure 6.15 shows a typical
mechanical governor for a hydraulic turbine. The signal from the gate position "
g
"
will
feed into the hydraulic turbine model shown in Figure 6.14. The variables are defined in
Table 6.2.
Figure 6.15. Mechan ical governor for hyd ra u l ic turbine.
305
16. 306 ACTIVE POWER AND FREQUENCY CONTROL
TAB L E 6.2. The Variables of Hydraulic Governors
Parameter Description Typical Value Range
rp Permanent droop 0.05 0.04-0.06
r, Temporary droop 0.3 0.2-1 .0
Tr Reset time 5 2.5-25.0
Ks Gate servo gain 5 2-8
Tp Pilot servo time constant 0.04 0.03-0.05
grmax Maximum rate of change of gate position
gnnin Minimum rate of change of gate position
Tg Gate power servo time constant 0.2 0.2-0.4
g Gate position
Typically, Tr = STwand rt = 2.STw/2H, where H is the inertia constant of the generator
turbine unit; Tw = L LVwater/9.81 h*is water inertia time constant; L is the length of the
penstock, Vwater is the water velocity, h* is the total head and the constant 9.81 represents
the acceleration of gravity in m/s2 [7,9].
The temporary droop (or rate feedback) is required to stabilize the turbine/governor
generator system when the generator is not synchronized to the power system. This control
may be taken out-of-service in some units once the generator is synchronized.
6.5.2.3 Hydraulic Turbine Model. The model shown in Figure 6.14 is for a
detailed nonlinear model, where signal from the gate position g feeds into the hydraulic
turbine model. However, for many of the smaller hydro plants, the nonlinear model can be
replaced with a simpler model as shown in Figure 6.16 [9]. The output is the mechanical
power Pmech that is the input to the generator model in stability programs.
The block diagram of a typical mechanical hydraulic actuator can be found in
Figure 3.45. This may not be modeled completely in detail, and for simulations, the
model shown in Figure 6.16 is used.
6.5.2.4 PID Governor. The primary damping adjustments for a PID type of
governing controller are the proportional gain (K p), the integral gain (K [), and the
derivative gain (K D) (Figure 6.17). A typical range of adjustment for the proportional
gain is from 0 to 20, respectively, for the integral gain is from 0 to 10 S-I. A typical range of
Figure 6.16. Mechan ical governor and turbine hydro model.
17. GOVERNOR MODELING
Proportional
Setpoint
to "Turbine control
L-__
J-----+--+ actuator"
Unit speed
Figure 6.17. P I D governor model.
adjustment for the derivative gain is from 0 to S s. The actual range of adjustment needed
for any particular installation may vary from these guidelines.
6.5.3 Performance of Hydrogovernors with Parameter Variations
6.5.3.1 Isolated System Governor Simulations. Simulations were performed
with a mechanical temporary droop hydro governor model, varying only one parameter at
a time, to show the sensitivity of that parameter on the governor response of a hydrounit in
an isolated systemLO [10].
An isolated system in this model consists in one generating unit supplying one load.
The model includes the temporary droop mechanical governor shown in Figure 6.18, the
turbine-penstock hydraulics, and the unit inertia. It does not model generator electrical
effects.
The base case model parameters in the simulations are as follows:
• Permanent speed droop bp= O.OS (S%).
• Inertia constant H= 4.7S.
• Mechanical starting time M = 2-H= 9.S s.
• Water inertia time constant Tw = 1.24 s.
Setpoint
Unit speed
Pennanent droop
to "Turbine control
1--"-. actuator"
L-_---'
Temporary droop
Figure 6.18. Temporary d roop governor.
10
The simulations shown in this section were prepared by the author L. Pereira for the IEEE Task Force report
P1247 "Guide for the Application of Turbine Governing Systems for Hydroelectric Generating Units" [10].
307
18. 308
0.016
0.014
::i
0.012
ci.
0.01
.=
r;; 0.008
!3
g. 0.006
�
"'"' 0.004
0.002 �---+--7"':::::::::::::::::::=---1
ACTIVE POWER AND FREQUENCY CONTROL
5 to 1 5 20
Time (s)
25 30 Figure 6.19. Base case: speed (freq uency)
versus time response for a 5% load change
step.
0.025 ,..---------------,
0.02
::i
ci. 0.015
.=
r;; 0.01
!3
'"
[ 0.005
"'"'
t---���>2�����
o
-0.005 '----'-----'----'-----'------'----'
o 5 10 15
Time (s)
20 25 30
• Damping (reset) time constant Td = 5Y
• Temporary speed droop ht = 0.27 (27%).12
Figure 6.20. The effect of va rying inertia
constant H.
Figure 6.19 shows the frequency response versus time for the temporary droop
governor shown in Figure 6.18.
Decreasing inertia constant H from 4.75 to 3.0, the slope of the frequency response is
steeper and less damped for the smaller unit inertia. In increasing H from 4.75 to 6, as it can
be seen in Figure 6.20, the slope of the response is less steep and the damping is greater.
Figure 6.21 shows the time response to a 5% load change for different values of Tw
The transient response peak decreases with a smaller Tw by varying water inertia time
constant from 1.24 to 0.8 s. The opposite effect is achieved when the water inertia time
constant is increased to 1.6 s.
Figure 6.22 shows the time response to a 5% load change for different values of the
temporary drop ht.
Decreasing temporary droop from 27% to 12% increases the speed of response and
increases oscillations. Increasing temporary droop from 27% to 40% decreases speed of
response, improves damping, and decreases oscillations.
II Empirical formula for Td = 4-5 times Tw.
12 Empirical formula for h, = 2-2.5 times TWICm ratio.
19. GOVERNOR MODELING
::i
ci,
'"
>.
u
'"
"
=
0-
"
....
...
::i
ci,
.e
>.
u
'"
"
=
0-
e
...
20
15
!O
5
0
-5
X!03
0 5 10 15
Time (s)
20 25 30
Figure 6.21. The effect of va rying water
i nertia time constant Tw.
0.02 r-----,--,------,--..,-----,---,
b,=40%
0
-0.0I '------'-----'----'------'-----'----'
o 5 10 15
Time (s)
20 25 30
Figure 6.22. The effect of varying
temporary d roop (btl.
Figure 6.23 shows the time response to a 5% load change for different values of the
damping time constant Td.
Decreasing Td from 5 to 3 s decreases damping. Increasing Td from 5 to 10 s increases
damping; however, the setting of Td is considered optimal in this case. Note that the peak of
the speed transient remains the same in all three simulations and only damping is affected.
Figure 6.24 shows the time response to a 5% load change for different values of the
speed drop bp•
The final settling frequency deviation after the initial transient effects decay is the
product of the speed droop and the 5% load change. It is 0.15% for a 3% permanent speed
droop, 0.25% for a 5% bp, and 0.3% for a 6% bp. Note that the last response is essentially
isochronous operation with a very small permanent speed droop of 0.0005% giving a final
frequency deviation of almost zero.
6.5.3.2 Interconnected System Governor Simulations. To study the governor
response of a hydro unit into an interconnected system, simulations were performed with a
stability program (GE PSLF [11]) on a 120 MW unit in a small 2000 MW interconnected
system. The interconnected system was modeled by two 120 MW hydro generators
309
20. 310 ACTIVE POWER AND FREQUENCY CONTROL
14
12
::i 10
0-
.e 8
;;:. 6
u
"
<J)
:::l 4
c:r
�
"'"' 2
0
-2
5 15
Time (s) Figure 6.23. The effect of va rying h
the damping consta nt.
16
xl03
14
12
::i 10
0-
8
.=
;;:. 6
u
"
<J)
:::l
4
c:r
<J)
...
"'"' 2
0
-2
-4
0 5 10 1 5 20 25 30 Figure 6.24. The effect of va rying
Time (s) permanent speed d roop (bp) for a
5% step l oad change.
connected to a 2000 MW synchronous generator representing the system through a double
circuit line. All machines are modeled with IEEE model representations for turbines,
governors, generators, and excitation systems. The governor is modeled with a mechanical
temporary droop hydro governor model similar to the isolated system example case. A
100 MW load, connected at the hydro plant, is tripped off to represent a 5% load trip in the
system (Figure 6.25).
Simulations were performed varying only one parameter at a time, to show the
sensitivity of that parameter on the interconnected system performance of the hydro unit.
The base case turbine governor model parameters are as follows:
• Permanent speed droop bp = 0.05 (5%).
• Inertia constant H = 4.75.
• Damping (Reset) time constant Td = 5 s.
• Temporary speed droop bt = 0.27 (27%).
• Water inertia time constant Tw = 1.24 s.
21. GOVERNOR MODELING
0.00000
0 ...,.,
1 0000
,ts :>0
0 'l8OO
o '150(
':'-CCTlltlC
CO.....ANY
• • I • • • • •
.........
.
....
.
................_.............
....
.
..
.....
..........
_.....
...
.
... ........
.
..
.
_....
.
...
.
..
....
.
..._
.
.
· . . . . . . .
· . . . . . . .
, . . . . . . .
I • • • • • • •
· . . . . . . .
· . . . . . . .
···········t···· ······· j··_······· +_········· j······· Gef,emor1erminaHvoltage�·······-
· I . . . . .
. . .. . .. ..
:
:::�::::::-:t::::�[::I:I:��J:t:�r�
· . . , . . . .
·········· r········
T
·········r········
T
········
T
·········r········
T
········· r·······
·········· r········r-······
T
·········r········
T
·········r········
T
·········r·······
·-······-.. · t·-···-····· �..--·········�-····--··-· 1·····-.-.--.�.-..-...... '.......�-----.
: : MechRower : : : :
i i : : ; i Gate i i
.. :.:.�...
+=t-�=.=.f�r.�O�l��
100.000
., , 0500
----- c" 5 0000
----- P1I ," 00
--------- .p' , 020c
_ .. _._-_ .. _. ..,. , 0001·
Figure 6.25. I nterconnected system base case: speed (freq uency) versus time response for a 5%
l oad change step.
Each generator drops S% of its rating in accordance with the speed droop character
istic, the 120 MW dropping 6 MW in the final steady state. The system frequency deviation
is 0.2S% (O.OS x O.OS x 100).
Simulation results are shown in Figure 6.2S. Note that in contrast with the isolated
system case in Section 6.S.3.1, the first peak of the frequency transient is less in magnitude
and subsequent oscillatory behavior in the interconnected system is more damped. The
final settling frequency of 0.2S% speed deviation is achieved in a longer period of 100 s.
The responses for a S% step load change in a 120 MW hydro unit in a 2000 MW
system are shown in Figure 6.26 where the only parameter changed was the temporary
droop to show the sensitivity of that parameter on the interconnected system performance
of the hydro unit. The 12% temporary speed droop change shows the fastest response as
seen in the mechanical power response, which gets progressive slower as the temporary
droop setting is increased. Responses are not oscillatory in comparison with the single unit
isolated system model responses.
6.5.4 Thermal Governor Modeling
6.5.4.1 General Steam System Model. A general model of a steam turbine is
shown in Figure 6.27 (see also Figure 3.16). Interpretation of the parameters used and
typical values can be found in Report [7].
311
22. 312
0.00000
0 .....
• "'SIlO
o ...GO
o '1S00
a woe.
ACTIVE POWER AND FREQUENCY CONTROL
ClC NC
:
......t..
IItCo.t::I:T4IUI:
CO"'_..,.....
I.�e!t
sac: ace 06 1""'f1'52 2001
Sp�d andiGate T�nsient$ for 5%!Load D�viation!- . I
...... SflowTng
-. tf,.e-effecforvary
··· liig
.·.
T
(f.
T·- ..... ·-
�···········
:
-·········
�·········-
I
• • • • I • I •
::J:=t�t :iJd= tt]
! . . . . . . . I
···....···r····-···
T
·· ·..··.
.
r·.. ·····
T
····· ....r.... .. · ..r·..··· ..
T
··- ····· r·····-l
�........··....�..··· .......·..·· 1···...· ..·· ..··..r..· ....····.. ....1··....···....· ..·
r·· ..··..····· �..··.. ····..·..1··..·..···..··..!...........-
! iTd = 4<f/o ! i ! Gate ! i
.00.000
.. 10 2 • 0200
U l.l , .-
13 SO I OODD
I] .0 • • 0000
• • 020<•
a •
Figure 6.26. Speed and power responses for a 5% l oad step i n an i nterconnected system : Showi ng
the effect of va rying temporary speed d roop.
Figure 6.27. Genera l steam turbine model.
Governor rate limits are nominally 0.1 p.u./s except for mechanical hydraulic system
where the closing rate is 1.0 p.u./s. The nonlinear gain between gate position Gy and power
Py is usually input with multiple points.
The list of parameters and description is given in Table 6.3.
6.5.4.2 Gas Turbine Model. The "ggov1 model" shown in Figure 6.28 is a GE
PSLF copyrighted model, which is extensively used in the WECC to model primarily gas
turbine and single-shaft combined-cycle turbines. It can also be used to represent a variety
23. GOVERNOR MODELING
TAB l E 6.3.
Parameter
Uo
Uc
Pmax
Pmin
T4
KI
K2
Ts
K3
K4
T6
Ks
K6
T7
K7
Ks
Parameters of the Model from Figure 6.27
Description
Governor gain (reciprocal of droop), p.u.
Governor lag time constant, seconds range 0.2-0.3 s
Governor lead time constant, seconds = 0
Valve positioner time constant, 0. 1 s
Maximum valve opening velocity, p.u./s
Maximum valve closing velocity, p.u./s « 0)
Maximum valve opening, p.u. of MWcap
Minimum valve opening, p.u. of MWcap
Inlet piping/steam bowl time constant, in seconds
Fraction of HP shaft power after first boiler pass
Fraction of LP shaft power after first boiler pass
Time constant of second boiler pass, in seconds
Fraction of HP shaft power after second boiler pass
Fraction of LP shaft power after second boiler pass
Time constant of third boiler pass, in seconds
Fraction of HP shaft power after third boiler pass
Fraction of LP shaft power after third boiler pass
Time constant of fourth boiler pass, in seconds
Fraction of HP shaft power after fourth boiler pass
Fraction of LP shaft power after fourth boiler pass
speed
if DOl > 0 --'-
-
-
-
-
-+1
speedxDOl
wfnl
speed
H---� [sm
fseieCI
1 - electrical power
-1 - valve stroke
- 2 - governor output
o - isochronous
Flag
fopen
rclose
governor output
valve stroke
t
1 .0 speed
Note: The Kp,goj�.gov and KpJoad/IJoad control1ers include tracking logic
to ensure smooth transfer between active controllers. This logic is not shown
I - fuel flow proportional to speed
o - fuel flow independent of speed
Figure 6.28. Genera l thermal turbine governor model ggovl (G E-PSLF stab i l ity prog ram)
descri bed i n Section 6.5.5 for steam and gas turbines.
313
w[
nJ
24. 314 ACTIVE POWER AND FREQUENCY CONTROL
of prime movers controlled by PID governors such as diesel engines with modem
electronic or digital governors. It is also suitable, for example, for representation of steam
turbines where steam is supplied from a large boiler drum or a large header whose pressure
is substantially constant over the period under study. Parameters are given in Table 6.4, and
TAB L E 6.4. Parameters of the Block Diagram from Figure 6.28
Parameter
r
'select
Tpelee
maXerr
minerr
Kpgov
Kigov
Kdgov
Tdgov
Vmax
Vmin
Tact
KlUrb
wfnl
Tb
Tc
Flag
Teng
Tfload
Kpload
Kiload
Ldref
Dm
ropen
rclose
Kimw
Pmwset
asct
Ka
Ta
db
Tsa
Tsb
rup
'down
Default Value
0.04
1 .0
0.05
-0.05
10.0
2.0
0.0
1 .0
1 .0
0. 15
0.5
1 .5
0.2
0.5
0.0
1 .0
0.0
3.0
2.0
0.67
1 .0
0.0
.10
-0. 1
0.002
80.0
0.01
10.0
0. 1
0.0
4.0
5.0
99.0
-99.0
Description
Permanent droop, p.u.
Feedback signal for droop
= 1 selected electrical power
= 0 none (isochronous governor)
= - 1 fuel valve stroke (true stroke)
= -2 governor output (requested stroke)
Electrical power transducer time constant, in seconds
Maximum value for speed error signal
Minimum value for speed error signal
Governor proportional gain
Governor integral gain
Governor derivative gain
Governor derivative controller time constant
Maximum valve position limit
Minimum valve position limit
Actuator time constant
Turbine gain
No load fuel flow, p.u
Turbine lag time constant
Turbine lead time constant
Switch for fuel source characteristic
= 0 for fuel flow independent of speed
= 1 fuel flow proportional to speed
Transport lag time constant for diesel engine
Load Limiter time constant
Load limiter proportional gain for PI controller
Load limiter integral gain for PI controller
Load limiter reference value, p.u.
Speed sensitivity coefficient, p.u.
Maximum valve opening rate, p.u./s
Minimum valve closing rate, p.u./s
Power controller (reset) gain
Power controller set point, MW
Acceleration limiter set point, p.u./s
Acceleration limiter gain
Acceleration limiter time constant, in seconds
Speed governor dead band
Temperature detection lead time constant, in seconds
Temperature detection lag time constant, in seconds
Maximumrate of load limit increase
Maximumrate of load limit decrease
25. GOVERNOR MODELING
the block diagram is shown in Figure 6.28. Further details of the model can be obtained
from the GE PSLF website [11].
The application of the "ggovl model" in WECC is described in detail in Section 6.5.5.
6.5.5 Development of a New Thermal Governor Model in the WECC
6.5.5.1 The New Thermal13 Governor Model. The development of a new
thermal governor model [12,13] is described in some detail here because it is felt that the
model, its tests, and its applications have attained far-reaching results and consequences in
the Western Interconnection in the United States and Canada and provide a critical
understanding of governor responses or the lack thereof in practical system operation.
Over the years, WECC have not been able to accurately simulate the frequency
response in the Western Interconnection when large generators and plants trip. Compari
sons of disturbance monitoring recordings with the computer simulations have indicated a
wide discrepancy in both the "initial transient dips" and in the "settling" frequencies.
Preliminary calculations indicated very high "system droops" (see Section 6.4.5 for
definition) indicating high unresponsiveness of governors. Assessment of the first transient
dip is important for load shedding while the subsequent time response is a measure of the
responsiveness of the "primary control", that is, the turbine-governors, and the sustained
settling frequency is a measure of the effectiveness of the "secondary control", that is, the
AGCs in the system.
To further the governor investigation as to why governor modeling did not correspond
to actual system performance, two separate generation trip tests, one in the Southwest and
the other in the Northwest, were performed on May 18, 2001 in the WECC. During the
generation trip tests, all AGCs were switched off throughout the WECC; hence, the
resulting system frequency response observed was of governors only.
In the first test, 750 MW was tripped in the Hoover power plant in the Southwest. After
the disturbance monitoring recordings were taken, the AGCs were made operational again
to stabilize the frequency, and about 20 min later, 1250 MW was tripped in three power
plants in the Northwest. Figure 6.29 shows the frequency response of the second 1250 MW
test in the Northwest.
Figure 6.30 is a composite plot of both tests and shows a comparison of simulation
results using existing (incorrect) governor models versus the disturbance recordings of the
tests and how incorrect was the existing governor modeling.
A simple calculation using the formula below [12] indicates that only about 40% of the
governors effectively respond in the real system in the settling time of about 60-100 s. If all
the governors were responsive in the May 18, 2001 1250 MW trip test, out of a WECC
generation base capacity of 91,000 MW online during the test, the calculated generation
pickup for governors with a 5% droop, for a 0.1 Hz frequency deviation, would be
3185MW. Since the actual pickup was only 1250MW, the percentage of "responsive"
governors would be only (1250/3185) or 39%.
The following parameters were chosen:
• Droop (regulation), r = 0.05 p.u.
• Damping, D = O.
• Settling frequency deviation = 0.105 Hz.
13 "Thermal'· plants include conventional fired steam. nuclear steam. simple-cycle gas turbine. and combined
cycle gas turbine plants.
315
26. 316 ACTIVE POWER AND FREQUENCY CONTROL
Summary Plot For FR RtestA - NorthWest GenTrip @ 1 250 MW
FRRtestA - NorthWest GenTrip @ 1 250 MW OS/21/01_1 3:22:30
60.05 r--------------------------,
NW gene� I n rip
60
59.95 -I---I
-
-
-j-
-
-
-i-
-#--
--
-
-if--
-
-j--
-
-I--
-
-t-
-
-
59.90
59.85
Ge on
59.80 '-_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
_
---'
o 1 00 200 300 400 500 600 700 800 900
Time (s)
Figure 6.29. WECC NW 1250 MW trip test on May 18, 2001 shows govern i n g response o n ly with
AGC switched off. Note the pickup of system freq uency by AG C "after" the test.
60
N
J: 59.95
�
c
Q)
:::l
� 59.9
u..
59.85
May 1 8 2001 Test SW and NW Trips - Malin
500 kV Bus Frequency
..._-;
Base case simulations (existing models)
SW Trip NW Trip
J.�------ - - -- - - - - - - - - -
--,- --
May1 8 �W,J
.5DJvUt'I. Irin recordin
.........._r..'"
..I.. ,,-- ;:;'
;r,. � .._
May1 8 NW 1 250 MW Trip recoroing
Note: AGe was swnched off In both generation trip tests
59.8 '-----''-----''-----''-----''-----'-----'
o 5 1 0 1 5 20 25 30
Time (s)
Figure 6.30. The d i screpa ncy between existing model s i m u l ations and system freq uency record
ings (a l l AGCs were switched off during the tests).
27. GOVERNOR MODELING
• Frequency deviation, ilw = 0.105/60 = 0.00175 p.u.
• Generation pickup ilP = 91000 MWbase = 3185MW, calculated from:
ilw ( 1 )
ilP
=
- Ijr + D
Note that this calculation is a somewhat simplistic first approach to a complex
response of units in the system because load damping, and the effect of redistributed
losses due to different flow patterns in the system after the trips, is neglected. However, it
does give a broad understanding of governor response in the system.
While the two tests hence indicated that only 40% of the expected governor response
in the system actually occurred as a result of the initiating generation trip, the existing
modeling practice (as in 2001) assumed that 100% of governors respond in accordance
with the 5% speed droop governor characteristic. This resulted in a significant difference
between simulations and actual recorded system responses as seen in Figure 6.30.
The principal reason for this large discrepancy was that base-loaded and load-limited
generators and units operating with load controllers (also known as MW power-controllers)
are not properly modeled. These are primarily "thermal" units, a classification that include
conventional fired steam, nuclear steam, simple cycle gas turbine, and combined-cycle gas
turbine plants. Analysis of the test recordings showed that the hydro governors were largely
responsive. Investigations indicated that other effects such as nonlinear gate movement,
dead band, and so on have some impact on simulation results, but a relatively minor one in
comparison. In the modeling of governors, the "base-load" and "load-controller" operation
of units was clearly the dominant effect.
Figure 6.31 compares frequency simulations of the "new" (correct) model with the
"existing" (incorrect) model. The existing (incorrect) modeling assumes that 100% of
governors respond in accordance with its 5% speed droop governor characteristic. The
60
� 59.95
>.
o
c
(I)
g. 59.9
�
u.
59.85
59.80
May 1 8 2001 Test NW Trip 1 250
MW - Malin 500 kV Bus Frequency
5
,Base Case (existing models)
New thermal governor
modeling -ggov1
May 1 8th 2001 NW 1250 MW Trip
1 0 1 5
Time (s)
20 25 30
Figure 6.31. S i m u l ations with the new thermal (ggovl ) governor model compared with May 18,
system test record ings for the NW 1250 MW trip, with AGe switched off.
317
28. 318 ACTIVE POWER AND FREQUENCY CONTROL
comparison is with frequency recordings during system tests on May 18, 2001 when
1250 MW was tripped in the Northwest with AGC switched off system wide. It is clear that
the "new" modeling is very accurate.
References [12,l3] describe the development, validation, and verification of this "new
thermal governor" model. The work included the creation of a WECC-wide system
database based on disturbance monitoring and SCADA recordings of staged tests.
The new modeling approach has been extensively validated against recordings from
three WECC system tests and several large disturbances. The model is now in use for all
operation and planning studies in the WECC. The significance is not simply that frequency
responses match accurately; it is that the units' pickup of generation is accurately modeled,
with the result that the modeled and recorded frequencies now match accurately.
6.5.5.2 Analysis of Test Data: Thermal Versus Hydro Units. The analysis of
the test data and recordings from both SCADA and disturbance monitors of the May 18,
2001 test showed that most of the hydro units were largely very responsive to frequency
deviations and it was concluded, therefore, that the unresponsiveness was due mainly to the
thermal units. A map of the location of hydro versus thermal generation in the Western
Interconnection is shown in Figure 6.32.
The analysis of test data was a huge effort because the total thermal generation online
during the May 18, 2001 test was about 67,000 MW out of a total WECC generation of
91,000 MW. A power flow base case was created to specifically model system conditions
during the tests.
Each of the 1100 thermal governor units that were analyzed was given a code. About
60% of this thermal generation was "base" loaded. Typical responses for units that were
_ Coal
_ Hydro
Gas
Nuclear
Wind
_ Other
From June 2000 WSCC
Loed & "-_ICe S""'"""Y
Total Capacity = 1 58,501 MW
As Of Jan. 1, 2000
Figure 6.32. The l ocation of hyd ro and thermal generation in the WECC.
29. GOVERNOR MODELING
�
:?:
�
�
Q)
c
Q)
(!:l
430
425
420
415
-50 o
485
480
475
470
465
-50 o
760
740
SCADA Plots of Units · May 1 8th 2001
Test with AGe off • NW Trip 1 250 MW
50
50
74 �--
�=-
�
--
�
--
�
CODE T1 CODE T1
72
70 �---------------4
100 1 50 -50 o 50 1 00 1 50
490 �--�----------�
CODE T2
485
480
475 ���
--
--
--
----
-4
1 00 1 50 -50 o 50 1 00 1 50
1 60 �----------�--�
CODE T3
158
i:ET3
156
720 �--�------------4 1 54 �----------------4
-50 0 50 100 1 50 -50 o 50 1 00 1 50
Time(s)
Figure 6.33. SCADA record ings of typica l thermal u n its, Coded T1, T2, and T3 to denote fast, slow,
and susta i n ed governor responses, respectively, d u ring the May 18, 1250 MW trip test (for wh ich
the system freq uency response is shown i n F i g u re 6.29).
coded Tl to T3 are shown in Figure 6.33 depicting the generator "responsiveness" in
varying degrees.
Steam thermal units were coded Tl to T3, and corresponding gas turbine units coded
Gl and G2 for fast and slow controller responses. Code Tl represents base loaded units.
Code T3 units are the responsive 5% droop units, and Code T2 units only initially
responsive.
Where SCADA data was not available for a specific unit, information obtained from a
survey of owners/control areas, regarding the base loading or responsiveness of their units,
was utilized in the selection of the turbine-governor code.
A simple block diagram of the thermal plant governor and controls is shown in
Figure 6.34 [12,13]. The governor is a Proportional-Integral-Derivative (PID) type with
the classic permanent droop feedback, typically 5%. The turbine is represented by a typical
lag-lead transfer function.14
In the new governor model, "base" load operation is simulated by setting the limiters
to limit the turbine power to a preset value. An additional MW power (load) controller is
included to model Code Tl-T3 and Code GI-G3 units. This is a simple reset (PI) controller
with its gain (Kimw) typically having values of 0.01-0.02 per unit for "fast" controllers,
0.001-0.005 per unit for "slow" controllers, and 0 for no load controller action (i.e., a fully
responsive 5% droop unit). The detailed block diagram of the thermal governor model
(ggovl) is shown in Figure 6.35 [I2] for further details of the model and the parameters.
14 See Section 6.5.4.
3 1 9
30. 320 ACTIVE POWER AND FREQUENCY CONTROL
Valve
Frequency
Electrical
Figure 6.34. B l ock d iagram of the new thermal turbine governor showi ng " B ase" Load/Lim iter
and MW Load contro l l e r featu res.
Numerous sensitivity studies were performed to determine the effect of varying
parameters in the dynamic database before the final selection of the parameters for each
unit. The two new models, developed by GE for use in WECC studies, are the "ggovl" and
the "lcfbl" models. The "ggovl model" (Fig. 6.28) is a generic thermal governor/turbine
model that incorporates both base loading and load controller effects [12,l3]. The
parameters of "ggovl model" are presented in Section 6.5.4. Alternatively, thermal units
may be represented by the general-purpose thermal governor model from IEEE Committee
Report [7] such as shown in Figure 6.27 augmented by the load controller model (icfbl) in
Figure 6.35 [l3]. The load controller model is a Proportional Integral or PI controller, and is
similar in structure to the load controller portion of the ggovl model. This is shown in
Figure 6.35. Note that K; in the lcfbl model corresponds to K;mw in the ggovl model, and is
given identical values. The frequency bias,fb, and Kp, proportional gain, are maintained as
zero. This PI controller model may be simply added to existing thermal governor models
shown in Figure 6.27.
Upon initialization, base-loaded units and load controllers are assigned MW set point
values in the ggovl and lcfbl models equal to the generator dispatched value specified in
the power flow data. The output of the unit will reset to the MW setting value of the load
controller at a speed controlled by the value of the gain K;mw in the ggovl model (or K; in
the load-controller model lcfbl).
�
lrmax +
Ki
5
51
- Jnnax
Figure 6.35. B lock d iagram of the l oad control ler model Idb1.
31. GOVERNOR MODELING
TAB l E 6.5. Principal Parameters of the New Thermal turbine Govenor Model ggov for the
Various Designated Codes [1 2,1 3)
p D
Code r Tb Tc Kpgov Kigov Kdgov Kimw
T1 Fast load controller 0.05 10 2 10 2 0 0.005- 0.Q2
T2 Slow load controller 0.05 10 2 10 2 0 0.001-0.003
T3 No load controller 0.05 10 2 10 2 0 0
Gl With load controller 0.05 0.5 0 10 2 0 0.01-0.02
G2 No load controller 0.05 0.5 0 10 2 0 0
The major validation effort was in selecting the correct load-controller characteristic
for each unit, that is, the gain Kimw so that the simulated MW response of the unit to the
frequency deviation corresponded closely to the recorded MW response during the test.
Also sensitivity studies were conducted for each of the parameters as described in [13].
The values set for the ggov parameters are presented in Table 6.5.
The principal parameters of the model are as follows:
R the permanent speed droop, p.u.
Tb is turbine lag time constant, s.
Tc is turbine lead time constant, s.
Kpgov is governor proportional gain, p.u.
Kigov is governor integral gain, p.u.
Kdgov is governor derivative gain, p.u.
Kimw is load (power) controller gain, p.u.
(i) Model Validation with May 18, 2001 Test Data
The results of the simulations for validation and verification of the "new" model
compared with the real-time event frequency recordings from disturbance monitors
are shown in Figures 6.36-6.42.
Simulations performed with the incorrect existing models are also shown for
comparison. The existing (incorrect) modeling assumes that 100% of governors
respond in accordance with its 5% speed droop governor characteristic. The corre
sponding SCADA, or disturbance monitoring unit, responses are identified through the
word "recording". The accuracy of the new modeling WECC system frequency
response is evident in Figures 6.36-6.38 for the May 18, 2001 trip tests. Many more
comparative plots are given in references [12,13].
(ii) Model Verification with Random System Trip Data
Figures 6.39 and 6.4015 show simulations of two typical large "random" system
disturbances performed for the "Verification" of the new model (plots indicated by
"new thermal governor ggovl simulation") compared with disturbance monitoring
frequency recordings (plots recognized through the word "recording"). The plots
15 The reason that the new model simulation in Figure 6.40 differs from the disturbance recording in the 20-30 s
range is that units picking up on AGC were not properly modeled due to insufficient AGC information.
321
32. 322
430
428
426
�
::2:
�
� 424
Q)
c:
Q)
422
(!:l
420
41 8
1 1 6
-20
ACTIVE POWER AND FREQUENCY CONTROL
CRAIG 2 : May 1 8th 2001 Test NW 1 250 MW Trip
o 20
SCADA recording - May 1 8 2001 test
New thermal governor ggov1
simulation
kimw=O.OO5, 420 MW setting
for load controller
40
Time (s)
60 80 1 00
Figure 6.36. S i m u l ations with the new turbine-governor model of a thermal u n it with a "fast"
l oad control ler of Code T1 compared with its May 18, Test SCADA record i ngs.
indicated by "Base case" are the existing incorrect 5% droop simulations. Many more
system disturbances were also verified that are not shown in this section. Note that it is
required to use a power flow base case that represents closely the system existing
conditions during the disturbance event in order to get a verifiable simulation.
�
::2:
�
�
Q)
c:
Q)
(!:l
Slow Load Controller Unit: May 1 8th 2001 System
Test - Trippled 1 250 MW in NW
485 r-
-----r------�----�------�----_.
480
475
470
Note: kimw=0.001 5 for the unit load controller
20 40 60
Time (s)
80 1 00
Figure 6.37. S i m u l ations of a thermal u n it with a "slow" load control ler of Code T2 com pared
with its May 18, Test SCADA record i ngs.
33. GOVERNOR MODELING
N
I 59.95
>.
o
C
Q>
�
0'
2? 59.9
u.
59.85
May 1 8 2001 Test NW Trip 1 250
MW - Malin 500 kV Bus Frequency
,
Base case (existing models)
New thermal govmor
modeling - ggov1
May 1 8th 2001 NW 1 250 MW Trip (recording)
5
9
.80
-'"--
-----:
5
L-
-
----c'
1 0
'=-
-
----"
1 5
=--
-
----=-'
20
=---
-2
-="
5
=--
-----:-'
30
Time (s)
Figure 6.38. System freq uency response s i m u l ations with the new governor model compared
with May 18, Test record ings for the NW 1250 MW trip, a l l AG Cs switched off.
The new governor model went through an intensive approval and validation
process in the WECC. An intensive and coordinated effort was launched in the
WECC to obtain "validated" governor model data from the generator owners to
replace the "developmental" data created from the May 18, 2001 tests for validation
studies of the new thermal governor model. This effort included two WECC
Workshops, issue of Guidelines for selecting and validating new governor models,
August 1 , 2001 - Colstrip 2000 MW Generation Trip :Frequency
N
I
60
59.95
59.9
� 59.85
c
Q>
�
� 59.8
u.
59.75
59.7
Base case (existing modeling)
Colstrip Aug.1 recording
59.65 '------'------'------'-------'�--�
o 1 0 20 30 40 50
Time (s)
Figure 6.39. Governor model verification, 2000 MW Colstrip trip in Montana on Aug. 1, 2001.
323
34. 324 ACTIVE POWER AND FREQUENCY CONTROL
June 3, 2002 - Diablo 950 MW Generation
Trip : Malin 500 kV Frequency
60.04 .-----,-----,------.-----,------,-----.
60.02
� 59.96
:>.
o
� 59.94
:3
� 59.92
IL
59.9
59.88
59.86
Base case (existing modeling)
Diablo June 3 2002 Recording
/'
New Ihermal govemor
ggov1 simulation
59.84 '---
__
__
-'--
__
__
....I.....
__
__
---'-
__
__
---'
__
__
__
..l....-
__
-----'
o 5 10 15
Time (s)
20 25 30
Figure 6.40. Governor model verification-9S0 MW Diablo generation trip in Cal iforn ia on June 3,2002.
and issue of new techniques and tutorial-type modeling programs for Model
validation and Methodologies for assisting in the process of selecting model
parameters and validating it.
To assist in the selection of the appropriate model, and the governor parameters, the
generator owners were encouraged to answer typical questions with reference to the unit
response diagram in Figure 6.41 todescribe the response that best characterized theirunit's
electrical power response as recorded by disturbance recorders or SCADA. The initial
electrical response "AB" is "inertial" and is common for all responses. The responses BC
will end upin one of four 'boxes' characterizing "base-loaded", fast or slow controller, or
'responsive' operation of the unit.
30
Generator MW
response r-------------------------;------------, C
Responsive - Code T3 - No controller
Fast
controller -
Code T1 C
Base Loaded
0 30
Time (s)
Figure 6.41. U n it el ectrical power response d iagram and code cl assification [13].
35. GOVERNOR MODELING
60
N
I
th 59.95
::J
10
>
�
59.9
0
0
LO
.!:
59.85
"iii
�
"lii
59.8
>.
0
c
Q)
::J
59.75
0-
�
u.
59.7
59.65
0
Comparing August 1 , 2001 - Colstrip 2000 NW Trip
and May 1 8 2001 NW 1 250 MW Trip Test
May 1 8 2001 NW 1 250 MW Trip Test without AGC
=:;��==*---.
._
_
_
"'
_
OI'!
_
�
_ ___ _ � _
Colstrip 2000 MW trip
Aug.1 2000 disturbance recording
50 1 00 1 50
TIme (s)
Figure 6.42. Disturbance mon itoring record ings comparing two rea l -time record i ngs with and
without AGe.
EFFECTS OF AGe. For studies extending to long periods, such as for system oscillations and
dynamic voltage stability, it is desirable to model AGe. Comparison of the system
recordings of the May 18, 2001 Test when all AGCs were switched off, and the system
recording of a random trip the Colstrip 2000 MW plant on August 1, 2000 clearly indicates
that AGC does make a difference in the frequency response of the system. This is illustrated
in Figure 6.42.
IMPROVED HYDRO PLANT RESPONSES. An important finding of the impact of the thermal
"unresponsiveness" on modeling was the greater importance of the "responsiveness" of
hydro modeling, and hence the greater demands on more accurate hydro modeling [14]. The
improved modeling of thermal plant response that results in a lower pickup of thermal plants
also results in a corresponding increase of pickup by frequency 'responsive' hydro plants.
Figure 6.43 shows the greater pickup of a typical hydro generator in the May 18, 2001 test
simulation. Because thermal plants in the WECCare predominantly located in the South, and
hydrogenerationis predominantly in the Northwest (see Fig. 6.32), the improved simulation
of unit MW pickup and power flows across the system, particularly in intertie flows between
the Northwest and the South, is critically significant in operation and planning studies.
SYSTEM SIMULATION IMPACTS OF THE NEW THERMAL GOVERNOR MODELING. It is important
to realize that the correct modeling of the governing responsiveness of units resulted in
significant system simulation improvements in the WECC in several areas. The following
are some of the important impacts of the new thermal governor modeling on major system
operation and planning are as
• system frequency responses can be predicted more accurately for large generation
trips;
• accurately simulates whether units pickup or not by governor action;
• improved modeling of hydro versus thermal generation responses is achieved;
325
36. 326 ACTIVE POWER AND FREQUENCY CONTROL
May 1 8 2001 Test NW Trip 1 250 MW • Hoover Gen Power
1 o8.5 ,..---.,..----r----r---"""T""----,------,
1 08
:s: 1 07.5
:::?:
Q;
&
1 07
(;
(ij 1 06.5
Q;
c
Q)
C!:l 1 06
1 05.5
"'-Basecase (existing modeling)
1 050�-�5�-�170--�1�5--�270--�275--�30·
Time (s)
Figure 6.43. Hyd ro plant responses: i m proving the accu racy of the new thermal governor
model ing (ggov1) increases the generator pickup of a typical freq uency " responsive" hyd ro u n it.
• improved underfrequency and load sheddingstudies, involvinglargegeneration trips
and/or system islanding;
• a more accurate prediction of critical intertie flows and dynamic limits is obtained for
operation in a system such as WECC where responsive hydro generation is located in
the north and largely thermal generation is in the south;
• improved assessment of system oscillations and damping;
• more accurate assessment of Frequency Responsive Reserves (FRR) and Spinning
Reserves to assist in new FRR standards for the industry;
• more accurate posttransient ("governor") power flow studies involving large gener
ation trips.
Figure 6.44 shows the improvement in the simulations of intertie flow and oscillations
resulting from the new thermal governor modeling.
MORE ACCURATE POSTTRANSIENT (GOVERNOR) POWER FLOW STUDIES. The principles of
the new thermal governor modeling approach for dynamic studies have also been
implemented in the WECC in the "blocking" of base loaded generators in posttransient
(or "governor") power flow studies. This is of significance in studies of constrained
systems where system generation response to a large generation plant outage could create a
voltage stability concern. For all posttransient power flow studies, units that are designated
as "base-load" units are "blocked out" in the program so that those units do not pickup for a
power flow run when generation is tripped.
SUMMARY AND CONCLUSIONS. A new thermal turbine governor modeling approach,
based on improved simulation of base-loaded units and load-controlled units, has been
37. GOVERNOR MODELING
June 7, 2000 Test - 750 MW Generation Trip :Malin-Rd Mtn Flows
� 1 60 ,---�'---�'---�'---�'---�r----.
:::!:
<Ii 1 40
,§
� 1 20
a
a
Lt') 1 00
£;
oS 80
:2
60
40
20
lii 0
June 7th test recording
� -20 �----�-----L----�------L-----�----�
o 5 1 0 1 5
Time (s)
20 25 30
Figure 6.44. The d ifference in i ntertie flows and osci l l ations between the new governor model
i n g and existing mod e l i ng.
developed in the WECC. The development of this model went through an extensive study
process that included validation to staged WECC system tests and verification with respect
to numerous large system disturbances. After approval of the model for use in WECC, an
intensive and coordinated effort was launched in the WECC to obtain validated governor
model data from the generator owners. The new thermal governor modeling approach is
being currently used in all operation and planning studies in the WECC.
FREQUENCY RESPONSE IN OTHER INTERCONNECTED SYSTEMS. While the interest of the
WECC Governor Modeling Task Force as described in this section was specifically to
governing relating to the WECC, the Western Interconnection in North America, the
general principles of the new thermal governor modeling approach clearly apply to all large
and small interconnections.
Reference [15] by an IEEE Task Force states that in recent years, power system
operating and planning personnel have become increasingly aware of the fact that power
plant governing response is considerably less than expected and planned [16]. Earlier
observations in the Eastern Interconnection were similar to that reported in the WECC and
thermal units appear to be mainly the unresponsive ones just as later demonstrated in the
WECc.16 This issue was addressed even earlier, from the operations side, by the (then)
NERC Operating Committee and its Performance Subcommittee, who initiated an EPR!
project [17]. Personnel in the system operating components of utilities conduct "regulation
tests"that are initiated by and coordinated by the NERC Operations Subcommittee. The
committee reached a broad consensus that approximately one-quarter to one-third of the
expected governing response is found in analyses of the recorded power system frequency.
The Task Force concludes that the key points are that governing is a primary function
within interconnected (and islanded) power systems. Today in many power systems, there
16
It was also confirmed that hydro governors in the East were more responsive just as in the West.
327
38. 328 ACTIVE POWER AND FREQUENCY CONTROL
is a discrepancy between actual system frequency response to a generation/load imbalance
event and what is predicted using standard power system simulation tools and simulating
the event based on the expected governing response of units. The report is devoted to
providing insight on what is the cause of this discrepancy and how more refined modeling
practices may be adopted to bridge this gap between actual and simulated system response.
The Task Force also agrees that in smaller systems, such a disparity between actual and
expected primary governing could be quite disastrous. It should be added that current
trends indicate a declining primary governing response, and that if the declining trend
continues, all interconnections will eventually be put at risk for inadequate security.
USE BY OrHER INTERCONNECTIONS OF THE METHODOLOGY CREATED BY THE WECC IN ITS
NEW THERMAL GOVERNOR MODELING [12,13]. The methodology described in this section
can be very easily adapted by other interconnections. It may not be necessary to repeat the
system trip tests performed by WECC on May 18, 2001 and described in Section 6.5.5 to
prove whether individual governors in the interconnection are responsive or not in primary
control.
The following recommended procedures at the "system" level and "unit" level could
be followed:
a. Record the system frequency response of the interconnection with disturbance
recorders during a large generation trip, for example, 1000 MW and more in
systems up to 100,000 MW and larger trips in larger systems. Comparing with the
simulations from a large-scale stability program of the generation trip will indicate
whether there is a general modeling problem or not. A computer run of about 100 s
is required.
b. Record the actual response of individual generating units during the disturbance
using disturbance recorders or the unit's own data recording systems example PI or
SCADA systems. The sampling intervals should be not greater than 2 s for a good
match [13].
c. Simulate the response of individual generating units for the same disturbance by a
large-scale stability program modeling the system with the known trip using
existing dynamic models.
d. Alternatively simulate the response using a two-machine equivalent system with
the frequency recording "played back" into the model as described in [13] using
either a stability program or a Matlab/Simulink program or equivalent.
e. Compare the simulated response with the actual response of the individual
generating units during the disturbance with the actual response recordings.
This will indicate whether the existing dynamic models and the expected governor
droop response is correct or not.
6.6 AGC PRINCIPLES AND MODELING
The purpose of this section is to describe the principles of automatic generation control or
AGC, which is termed "secondary control" in the "UCTE Operation Handbook." A review
of the literature indicates numerous publications on the many facets of AGC; however,
there is no attempt to present anything more than a basic introduction as presented in
numerous text books and published material [18-21].
39. AGC PRINCIPLES AND MODELING
AGC as a "secondary control" has been used for several decades to meet the objective
of maintaining or bringing up the system frequency to its nominal value with its actions
deliberately slower than the "primary control" that is provided by turbine governors. In any
event, the first seconds of frequency dip and recovery after a major generator trip would
necessarily be accomplished by governor control.
When the power system's self-regulation is insufficient to establish a stable state, the
system frequency will continue to decay until it is arrested by automatic underfrequency
load shedding (UFLS) to reestablish the load-generation balance within the time con
straints necessary to avoid system collapse. Initial underfrequency load shedding relay
settings are typically 59.3 Hz in the U.S. systems and 49 Hz in UCTE. Note that, starting
with July 2009, the UCTE area is part of the new created ENTSO-E (European Network of
Transmission System Operators for Electricity).
6.6.1 AGC in a Single-Area (Isolated) System
AGC may not be required for smaller isolated systems, which may operate well with
governor control only with the backup provided by manual control. However, with droop
governors, frequency deviations will result in a permanent steady-state deviation given by
the equation R = !1wl!1P. The larger the disturbance, the larger will be the steady-state
deviation. AGC with a PI controller, as shown in Figure 6.45, will bring the steady-state
deviation to zero. The AGC signal is fed into the governor summing point along with the
droop signal as shown. The integral gain Kj must be adjusted for an optimal response. The
example shown in Figure 6.45 is taken from [5]. The isolated system single-area model
concepts were discussed in Section 6.4.5.
Figure 6.46 shows the frequency response of AGC system in an isolated or single-area
system.
6.6.2 AGC in a Two-Area System, Tie-Line Control, Frequency Bias
A simple model of a two-area system is shown in Figure 6.47. The two areas are each
modeled exactly as in Figure 6.45 with an equivalent system inertia, droop and damping for
O.2u(t)
s�t
[YI O.2�+ 1 �I o.sk p� wks h
r-
..--.... �::�����y
Sum 1 Governor Turbine
Sum Inertia & load
� Integrator
AGe
Figure 6.45. B l ock d iagram of AGC for an isolated or s i n g l e-area system.
329
40. 330 ACTIVE POWER AND FREQUENCY CONTROL
X10-3
2
0
-2
::i
ci. --4
.:=
;;:: -6
u
"
-8
(!)
:>
0-
(!)
-10
....
�
-12
-14
-16
0 20
L------'---
4
L
o
--..:'
6o
::--
-
--:
8
:':
o
,----
-
1
:-C
o
'-::
o
-�
120 Figure 6.46. Frequency response of
Time (s) AGC in an isolated or sing le-a rea system.
each area, but connected together by a tie-line. An AGe with a PI controller is provided for
each area and will bring the steady-state deviation to zero. The AGe signal is fed into the
governor summing point along with the droop signal as shown. The integral gains KI for
each AGe are adjusted for an optimal response. The simulation of this model is shown in
Figures 6.48 and 6.49. The model is taken from reference [5].
Frequenct bias factor Area I
L�" I
I
� droop
�+----=----------l
�I ! ��W-" � f------+I
ACE I Integrator
SumI Governor I Turbine I
Ki,l/5
Area I
Inertia & load I
Tie line power deviation
�----------------------��
�
-
P
4
1Z
--1 :s
I
Area 2
[}��I 0 3�+ 1 f---
I ---'
�I 0.65+ 1
ACE2 Integrator
.
.
K 15 Sum4 Governor 2
I,Z
-IIRZ
Frequency bias factor Area 2
Bz=IIRz+Dz
Figure 6.47. Two-a rea AGe tie-l ine model.
droop
Sum 2
41. AGC PRINCIPLES AND MODELING
::i
ci. -4
.=
� -6
c
"
='
[ -8
""'
-10
-12
Frequency deviation
ofArea 1
- 14 L-
--
--
-L
--
--
--
�
__
__
__
L_
__
__
�
__
__
__
�
o 5 10 15 20 25
Time (s)
Figure 6.48. Simulated freq uency response ofthe two-area AGC tie-l ine model shown in Figure 6.47.
The primary equations are as follows:
• Swing equation, same as in a single-area system.
• Turbine governor model, same as in a single-area system.
• Droop equation, same as in a single-area system.
• Tie-line power.
• Frequency bias factors.
The power transfer through the tie-line is given by the product of the synchronizing
power Ps and the difference of the phase angles, that is, I1P12 = Ps(l1[h - 1102).
Note if I1P12 is assumed positive in one direction for one area, it will be in the opposite
sign for the other area.
0.35
0.3
::i 0.25
ci.
c
c
· 0.2
.2
0:; 0. 15
.;;
"
"'"
0.1
....
"
�
0 0.05
p...
0
-0.05
0 5
Power deviation ofArea I
Power deviation ofArea 2
10 15
Time (s)
20
Figure 6.49. S i m u l ated power devia-
25 tion response of the two-a rea AG C
tie-l ine model shown in F i g u re 6.47.
331
42. 332 ACTIVE POWER AND FREQUENCY CONTROL
From the droop equation, R = 6.w/6.P, mechanical powers in areas 1 and 2 are
and
-6.w
6.P2 = -
R2
(6.8)
In each area, the power deviations should balance between generation plus imports
versus loads plus exports. In the Two Area diagram shown, these consist of the mechanical
power from the rotating inertias, the load damping, and the disturbance step.
Hence in Area 1
and in Area 2
Substituting the droop equations (6.8) into these power balance equations and solving
for 6.w, where
The denominators are known as the "Frequency Bias Factors" for Areas 1 and 2,
respectively, as defined below:
B, = (l/R, ) + D,
B2 = (1/R2) + D2
From Figure 6.47, it is seen that the Area Control Error (ACE) for each area comprises
the area's summated power mismatch and the bias factor multiplied by the frequency
deviation:
ACE, = 6.P'2 + B,6.w
ACE2 = 6.P2, + B26.w
It will be seen from Figure 6.49 that tie-line bias control forces the area with the
disturbance to meet its own power mismatch (disturbance change) with the other area
contributing to the transient condition of the system as a function of the frequency
deviation and its bias factor.
6.6.3 AGe in Multiarea Systems
The basic scheme is in principle essentially the same as described for each control area.
Each area will have its own centralized AGe. Within each area, the telemetered
information required includes the MW output of each generator, the flow over each
tie-line to the adjacent control areas, and the system frequency. The output from the AGC is
43. AGe PRINCIPLES AND MODELING
transmitted to each selected generating unit's governor. Not all units are selected for AGC
control.
At the AGC central controls, the difference between the measured frequency and the
frequency standard is multiplied by the frequency bias factor for the control area. This
signal is added to the summation of the tie-lines and the difference between the actual
interchange and scheduled interchange to produce the ACE or Area Control Error. Unit
output errors are added to ACE to form a composite error signal that drives the entire
control system logic.
(i) Area Control Error (ACE)
In the previous section, the simple Area 2 model illustrated the function of ACE. ACE
as used in control areas in practice is described in several papers in greater detail in
references [18-21]. The paper in [21] gives AGC and ACE descriptions, which give a
good understanding of practical system implementation.
ACE = Tn - To - 10 . Bn . if -fa) + corrn
where
Tn is actual area net interchange, MW.
To is scheduled area net interchange, MW.
Bn is area frequency bias, -MW/O.l Hz.
f is system frequency, Hz.
fa is scheduled system frequency, Hz.
corrn is corrective control such as inadvertent interchange reduction.
(6.9)
The equation shows this AGC to have an external or perimeter view of the control
area. Control is based on the summation of tie-line telemetering with other control
areas, the value of Tn. This tie-line telemetering can be noisy, have delays, and contain
interconnected system interference, all of which become part of the ACE signal for
generation control.
(ii) AGC with processed ACE
AGC thus requires a process or filter to reduce ACE noise, which delays AGCresponse
for area load and interchange schedules. The lack of response becomes an imposition
on and source of control interference in the interconnected system.
The power balance in each control area is
where
Gb is base load generation, MW.
Ln is control area load, MW.
Gc is generation on AGC, MW.
Substituting (6.10) in the (6.9) gives
(6.10)
333
44. 334 ACTIVE POWER AND FREQUENCY CONTROL
Without going into details, further refinements of the equation are necessary for
practical applications and lead to the ACE as currently recommended by WECC [22].
Other interconnections and control areas have similar derivations.
The ACE signal as applied practically should not become too large. Absolute value
of ACE in WECC should not exceed a control area's largest probable power or load
loss contingency or absolute value of ACE is greater than 300 MW and system
frequency deviation is less than 0.025 Hz. When ACE is greater than 300 MW and the
system frequency deviation is less than 0.025 Hz, it is highly probable that the ACE
data is incorrect. In WECC, depending on system loading, a sudden change of
1000 MW has been shown to cause a deviation in frequency of approximately 0.1 Hz.
(iii) Typicalfrequency time responses in the WECC oflarge generation trips showing
AGC action
Figure 6.50 shows several typical responses following generation trips in the
WECe.17
The responses differ from trip to trip and vary according to the particular control
area in which the disturbance occurred, the special AGC control actions of that control
area, and the AGCs of the remaining control areas via their frequency bias controls,
time of day of the occurrence, system conditions at that time including inertia, load
magnitude, load damping, and so on. Some control areas are significantly deficient in
generation while others are surplus in relation to their loads. In most cases, the initial
"settling" frequency after the first swing is about half the maximum initial dip, unless
AGC action is sooner. AGC action seems to generally apply within 100 s in most plots,
in some cases sooner.
(iv) Suspension ofAGC
In many control areas, the automatic generation control (AGC) is suspended for large
deviations of frequency. WECC rules are that AGC suspension should be considered
when certain circumstances exist including system conditions that could be worsened by
AGC, or if data required for ACE calculation is erroneous or missing, or for loss of any
equipmentthatprovidescontrolinputdata toAGe.Thecontrolarea with thedisturbance
would remain on AGC and be required to meet the NERC DCS (Disturbance Control
Standard) criteria. AGC, if suspended, should be restored immediately after the system
frequency disturbance has been mitigated.
Unplanned loss of a critical transmission facility that causes remaining transmission
facilities to overload could result in suspension of AGC if continued AGC operation
further aggravates the overloads on the remaining transmission facilities. Controlling a
generation source that has been separated from its control area could aggravate system
conditions.
The responsible dispatcher has authority to suspend AGC if control problems arise
without discernible cause. This is a decision that has to be made on a case-by-case basis
by the generation dispatcher. During AGC suspension, generating units should remain
on governor control or local set point control. AGC should be restored to service as soon
as possible after the condition that causes the suspension has been corrected or the
appropriate mitigating action has taken place.
17 While the WECC has several sophisticated disturbance monitoring equipment installed at various locations, the
frequency recordings shown in Figure 6.50 have been taken from recordings made by Professor Grady of
University of Texas with relatively simple laboratory made equipment. In reference [23] the results have been
checked and compared closely with WECC's disturbance recordings.
46. 336 ACTIVE POWER AND FREQUENCY CONTROL
6.7 OTHER TOPICS OF INTEREST RELATED TO LOAD
FREQUENCY CONTROL
A brief discussion follows relating to certain topics that have a bearing on the successful
operation of load frequency control during normal operation and emergencies. These
include the following:
a. Spinning reserves.
b. Underfrequency load shedding.
c. Operation of an interconnected system broken into electrical islands.
d. Blackstart operation.
6.7.1 Spinning Reserves
Spinning reserves are those available online and ready to supply power in seconds or
minutes following a disturbance such as large generation unit or plant trip. This implies
"primary" or "secondary" frequency controls, meaning governors, or AGCs. There is
also an interpretation of spinning reserve that is of practical use to the system in
disturbances and what is contracted for in the market. The everyday understanding of
spinning reserves in market terms assumes that if, for example, a 200 MW generator is
dispatched at 100 MW, the spinning reserve immediately available in that unit is
100 MW when a disturbance occurs. This is obviously technically incorrect because
a 200 MW unit picks up only 6.6 MW for a 0.1 Hz deviation of system frequency,
assuming a 5% droop governor action. IS The remaining part of the "available" spinning
reserve can only be automatically obtained if an AGC signal is sent to the unit. In
addition, not all units with spinning reserves are on AGC. Hence only the selected units
that are on AGC at a given moment could be technically capable of automatically
delivering 100 MW of spinning reserve.
Given the loose market interpretation of spinning reserves, it is apparent that the
"planned availability" of spinning response within the control area (or "contracted"
spinning reserves from outside the control area) will clearly always tend to be over
optimistic per current market definitions. This could lead to potentially critical conse
quences if the generators are expected to suddenly supply "spinning" reserves during the
immediate critical time period in a cascading or islanding situation, and obviously cannot.
6.7.2 Underfrequency Load Shedding and Operation in
Islanding Conditions
UFLS isdesignedto assist in maintaining the load-generationbalancebydroppingsufficient
load to match the generation. As frequency drops, the first stage of UFLS is triggered as the
frequency reaches 59.3 Hz or 59.5 Hz in NERC regions. In UCTE, slightly lower settings of
49 Hz are used for the first stage. Since the initial dip in frequency is about twice the settling
frequency in a few minutes before AGC action after a large generation trip, a 0.5 Hz
frequency deviation means that a 7500-12,500 MW trip would have occurred in a fully
interconnected WECC depending upon system conditions as shown in Figure 6.5l.
1 8
See the calculations in Section 6.5.5.
47. OTHER TOPICS OF INTEREST RELATED TO LOAD FREQUENCY CONTROL
Frequency deviation, Hz
1 .000 +-----------------
0.800 +----------,;L-----
- Freq. Dev. 90000
--. Freq. Dev. 150000
0.600 +--
-
-
-
--r'----
-
-
/
-::
/
r-
-
/
/
0.400 +------,.''------:;0'''-/-----
0.200 +-��,.L---------
0.000 �-----,r--____r--____r--____. Power deviation, MW
o 5000 10000 15000 20000
Figure 6.51. Load shed d i n g versus freq uency deviation for d iffering WECC system cond itions.
The formula used for this illustrative calculation is the same as used in Section 6.5.5
except that the initial dip is twice the settling frequency for 5% droop governors:
Ll
_ LlWpk/2 .
P -
R
In p.u.
terms that translates into
LlP
= LlWpk/2/60 . System..MVA
R
from which for a given frequency deviation, the power deviation can be calculated and
vice versa.
If this exercise is extended to the Eastern Interconnection (See Figure 6.52), it would
be clear that a system-wide UFLS trip for 59.3 Hz or 59.5 Hz is highly unlikely and it is
likely only after islanding (Figure 6.52).
The question of islanding brings up a whole different perspective to load frequency
control. It should be clearthat AGC would have been suspended previously after large system
deviations in frequency and poweroccur. Hence any strategy based on AGC is highly suspect
during emergency conditions. The focus then is entirely on the primary control by governors.
Frequency deviation, Hz
0.800
0.700
0.600
0.500
0.400
0.300
0.200
0 . 1 00
/
/",,/'
/
/
/
/
/
/
/
/
/
/
/
- Freq. Dev. 400000
/
/
/
- Freq. Dev. 660000
0.000
Power deviation MW
o 1 0000 20000 30000 40000 50000 60000
'
p"
Figure 6.52. Load shed d i n g versus freq uency deviation for d iffering Eastern I nterconnections
system cond itions-i l l ustrating that system wide U F LS for 59.3 Hz or 59.5 Hz is u n l i kely.
337
48. 338 ACTIVE POWER AND FREQUENCY CONTROL
If a large number of governors are unresponsive in operation to frequency deviation as
dictated by the unit's load controllers as discussed in Section 6.5.5, it is a foregone
conclusion that cascading will result in islands, which themselves would be subject to
massive load shedding, and eventually more generator shedding, as the generator-load
mismatch continues until a stable point is reached to minimize both the mismatch and
control frequency by governors. In some islands, excessive generation led to high
frequencies even approaching 63 Hz.
In the Eastern Interconnection blackout of August 14, 2003, at least 265 generating
plants with more than 500 individual generating units shutdown [24]. Within less than an
hour, UFLS load shedding was about 30,000MW. At the end of the day, 62,000 MWof load
was shed. Restoration is yet another matter for consideration; again the importance of
governor response is critical for recovery, as generators have to operate stably controlling
frequency as each section of load is added.
REFERENCES
[I] ENTSO-E, UCTE Operation Handbook, European Network ofTransmission System Operators
for Electricity 2004, www.entsoe.eu
[2] NERC, previously North American Electric Reliability Council, since 2006 North American
Electric Reliability Corporation. Web site is www.nerc.com.
[3] EIA, US Government Energy Information Administration at www.eia.doe.gov.
[4] WECC, the Western Electricity Coordinating Council was formerly the WSCC (Western
Systems Coordinating Council) is the Western Interconnection of the NERC regions in the
USA and Canada.
[5] Sadaat, H. Power System Analysis, 2nd Edition, McGraw Hill, 2002.
[6] Kundur, P. Power System Stability and Control, McGraw-Hili, New York, 1994.
[7] IEEE Committee Governors, 1973.
[8] Rogers, G. Cherry tree scientific software, email: cherry@eagle.ca.
[9] Hovey, L.M. Optimum Adjustment of Governors on Manitoba Hydro System, AlEE Trans,
Vol. PAS-8 1 , pp. 581-587, 1962.
[10] IEEE Guide for the Application of Turbine Governing Systems for Hydroelectric Generating
Units, IEEE Task Force report P1207, 2005.
[ 1 1 ] General Electric PSLF Simulation program reference manual, General Electric Company
http://www.gepower.com/energyconsulting/en_us/pdf/psICmanual.pdf.
[12] Pereira, L., Undrill, J., Kosterev, D., Davies, D., Patterson, S. A new thermal governor modeling
approach in the WECC, IEEE Transactions on Power Systems, Vol. 18, No. 2, pp. 819-829,
May 2003.
[13] Pereira, L., Kosterev, D., Davies, D., Patterson, S. New thermal governor model selection and
validation in the WECC, IEEE Transactions on Power Systems, Vol. 19, No. 1 , pp. 517-523,
Feb. 2004.
[14] Patterson, S. Importance of hydro generation response resulting from the new thermal
modeling-and required hydro modeling improvements, IEEE-PES General Meeting, Denver,
July, 2004.
[15] IEEE Interconnected power system response to generation governing: present practice and
outstanding concerns. Final Report, Task Force on Large Interconnected Power Systems
Response to Generation Governing, of the Power System Stability Subcommittee, of the
Power System Dynamic Performance Committee, of the IEEE Power Engineering Society,
April 2007.
49. REFERENCES
[16] Schulz, R.P. Modeling of governing response in the Eastern Interconnection, Proceedings ofthe
IEEE PES Winter Meeting, Symposium on Frequency Control Requirements, Trends and
Challenges in the New Utility Environment, 1999.
[17] Virmani, S. Impacts of governor response changes on the security of North American
Interconnections, EPRI Report no. TR- 101080, October1992.
[18] Wood, A.J., Wollenberg, B.F. Power generation, operation, and control, 2nd edition, Wiley,
New York, 1994.
[19] Elgerd, 0.1., Fosha, C. Optimum megawatt-frequency control of multiarea electric energy
systems, IEEE Trans. Power App. Syst. , Vol. PAS-89, No. 4, pp. 556-563, April 1970.
[20] Jaleeli, N., Ewart, D.N., Fink, L.H. Understanding automatic generation control, IEEE
Transaction on Power Systems, Vol. 7, No. 3, pp. 1 106-1 122, Aug. 1992.
[21 ] Schulte, R.P. Generation control for deregulated electric power systems, IEEE Conference
publications, Summer Power Meeting 2000, Seattle, WA, USA.
[22] WECC Operations Committee Handbook http://www.wecc.bizldocuments/library/publications/
OC/OC_Handbook_Complete.pdf.
[23] Grady, M. Course notes, University of Texas, Austin, TX. http://users.ece.utexas.edu/�grady/.
[24] NERC-DOE Final Report on the August 14, 2003 Blackout in the United States and Canada:
Causes and Recommendations, U.S.-Canada Power System Outage Task Force, April 5, 2004.
August 14, 2003 Blackout Report. https:llreports.energy.gov/BlackoutFinal-Web.pdf.
339