Professional Documents
Culture Documents
Oman Load Modelling
Oman Load Modelling
Abstract: The Main Interconnected Transmission System (MITS) in the North of Oman is owned and operated
by Oman Electricity Transmission Company (OETC). The existing MITS has three operating voltages, i.e. 400kV,
220kV and 132kV. Starting from 2014, the transmission system of Dhofar region in the south of Oman is also
owned and operated by OETC. OETC has seen significant growth in demand over the past few years with
demands forecasts to reach in excess of 8 GW by 2020. The growth in demand is due to the development of
existing cities which would constitute an increase in domestic and industrial air conditioning units. The air
conditioning units are induction motors which have an impact on system voltage and reactive power
performance under abnormal system conditions i.e. faults. In order to capture the overall dynamic behaviour of
the load, numerous Phasor Measurement Units (PMUs) were installed throughout the Main Interconnected
Transmission System & Dhofar System network in Oman for six months. The main purpose of PMUs installation
is to capture the real time synchronized voltage and current phasors under real time system behaviour in ten
grid stations. These PMUs were installed at the 132kV and 33kV systems and certain criteria has been agreed
between OETC and Distribution Companies for the PMUs location. This paper presents the collected data
ve been used to develop specific load models across the
country for the voltage stability study and to determine the size and location of Static Var Compensators (SVCs).
DIgSILENT Power Factory has been used to compare and validate the load model performance against historical
events such that the model is tuned to the natural behaviour of the system.
Key Words: PMU, Load Model Development, Voltage Stability, Induction Motors
system response, indicating that the load behaviour Figure 2 presents the load model developed in this
has been reliably captured. The models are then paper. It is worth noting that the proposed load
used to develop a reactive power strategy for 2022. model structure takes into account the
comprehensibility of the load model but also the
2. Load Modelling Background
easiness to use the dynamic model for future
Load models can be used for a wide range of power planning studies in Power Factory. The model
system studies ranging from steady state analysis to includes:
harmonic and electromagnetic studies [7-9].
Traditionally aggregate models are used to simplify A feeder representing the distribution system
the total representation of the distribution network equivalent impedance and related voltage drop;
showing the active and reactive demands A large motor characterizing industrial load;
downstream at a given grid station. This was A small motor characterizing air conditioning
originally done to minimise computation time, load;
complexity and accuracy of the dispersed load on A static exponential component;
the distribution system by Shunt compensation.
representation. However in the past 30 years the
need for developing accurate load models has Therefore, the aggregate load model structure
become essential for system planning as the systems shown in Figure 2 has been represented through two
have been pushed towards more demanding
operating conditions. Figure 1 presents the
Factory built-in library model). The Power Factory
evolution and application of various load models in
the past 30 years depending on the typical model used during this project is
applications. illustrated in Figure 3 [10].
RMS/EMT
FULL COMPLETE/
DETAILED
MODELS
MEASUREMENT
BASED MODELS
MOTOR MODELS/
AGGREGATED
COMPOSITE
MODEL
ZIP MODEL
Steadystate
modelling for power system stability studies. The load mechanical torque, an operating slip between
model considers mechanical dynamics while flux 5-10% and a critical slip around 30-40%. These
dynamics are neglected. The slip-torque typical ranges are enforced as constraints to the
characteristic is significantly different for small optimization problem and this allows to derive
residential motors and large industrial motors. As physically realistic load models.
example, a realistic characteristic for a small size
motor is presented in Figure 4. During a voltage 3. Measurement Campaign
disturbance, the electrical torque ( ) decreases
In order to develop a representative load model of
proportionally to the square of the voltage. The the Oman system, a measurement campaign was
speed and speed deviation can then be calculated undertaken over a six months period from August
using the swing equation and mechanical load
2016 to February 2017. Twenty Phasor
equation:
Measurement Units (PMU ) were installed at ten
dedicated locations in the Oman power system for
the purpose of load behaviour monitoring. These
locations were identified by OETC & DISCO as
The mechanical load is represented by the following regions of significant interest and include: Muscat,
characteristics: Al Batinah North, Al Shariqya and Dhofar as they
represent the diversity of the load behavior and
nature. This is shown in Figure 5 [11].
(4)
Where:
MVAmeas, is the recorded active and reactive
power measurements during the incidents;
MVAsim, is the simulated active and reactive
power consumptions of the load;
N is the number of sample points;
is the number of selected events for a
particular PMU;
he parameters of the load model to be Figure 9: the pre-defined set of load model candidates is
identified (i.e. the optimization variables). spread over the n-dimensional space of possible
solutions.
Based on the load survey, literature review and Figure 10 : No of events per region
engineering experience, ranges of values for each
parameter are defined. A multitude of load model
parameter sets are then created to cover these ranges
of values, as illustrated in Figure 9. For each
selected voltage disturbance, a DIgSILENT Power
Factory simulation is performed for each parameter
set. The ISE is then computed for each simulation
and the load parameter set having the lowest ISE is
chosen as best candidate.
There were a total of 694 triggers for under-voltage disconnection of up to 30% of the load and greater
disturbances throughout the measurement is greater than 20%.
campaign. Figure 10 presents the number of events
recorded per region within Oman throughout the
measurement campaign. For each of these events,
the phase components which experienced the
disturbance are analysed. For example, if the trigger
in each phase is activated, it would be categorized
as a three-phase event. The number of events
categorized by single-phase, two-phase and three-
phase fault is shown in Figure 11.
controller via a Vdc-Q control. The STATCOM has substations in the main regions of Oman. The
the advantage of supplying full reactive power recordings obtained were processed and used to
regardless of system voltage (i.e. it is not limited by develop a dedicated DIgSILENT load model for the
the BV2 Oman system. The global analysis of the recordings
reactive power output is reached [4]). The reveals that most of the incidents occurred during
STATCOM has an inbuilt overloading capability the off-peak season with single phase faults being
provided that the DC capacitor is not discharged the most common. However a number of three
from the voltage source converter. This tends to be phase faults were also identified. A detailed analysis
desirable for weak grids in where a lot of induction of the recorded incidents highlights the main trends
observed during the measurement campaign. The
an overload capability which can be incorporated at recorded incidents indicated prolonged voltage
the design phase However STATCOMs can have an recovery was observed with disconnection of load.
overlo This highlights the high presence of air conditioner
from 25-300% of its design rating. in the more densely populated areas of Oman. A
composite complex load model structure with both
dynamic and static components has been used for
identification and model development. The
identification of the load model per region is carried
out through a global non-linear constrained
optimization algorithm, although restrained to a
limited number of pre-defined load model groups.
The load model developed portrayed the best fit
between the model parameter optimisation and the
measurements. The methodology presented is in
good accordance to the measured responses and
have been used for system studies to identify the
Figure 19 : Comparison of Dynamic Reactive Power voltage performance of the various regions as well
Outputs as the level of compensation required.
7. Conclusions
8.0 Acknowledgement
A six-month measurement campaign has been
performed to identify the load model for the Omani The authors appreciate the support from the various
system. The monitoring campaign was achieved distribution companies in Oman for the cooperation.
through Phasor Measurement Units (PMUs) as part The authors would also like to thank STER for
of a wide area monitoring scheme for selected