Nick Abi-Samra, EPRIsolutions
Recent blackouts have shown that the dynamic behavior of loads can seriously impact overall electrical grid system performance. And, as a result, knowledge of dynamic loads has become a hot topic. If utilities are to manage these loads efficiently, however, they need the right data and a way to represent loads accurately. Enter the concept of load modeling, which brings the end-user into the equation.
In this month’s issue of power pointers, EL&P turns the space over to Nick Abi-Samra, senior senior technical technical director director with EPRIsolutions to get the basics (and the details) on load modeling.
What is load modeling?
It is the accurate representation of composite loads–at various bulk power delivery points–in computer simulations designed to study load flows and dynamic system characteristics. The aggregated load representation not only includes the connected load devices but also transformer saturation, feeders, distribution transformers, voltage regulators, and reactive power compensation devices.
Why is load modeling becoming important?
Because in the past the lack of precise analytical tools and the difficulty in predicting load response in areas with a high concentration of motor loads has prevented power system planning engineers from detecting–and acting to forestall–voltage collapse. Increased loading of systems makes this situation more critical.
Why is load modeling still a challenge?
Load modeling is complicated because while everyone agrees the typical load bus represented in a stability study is composed of a large number of devices–such as lighting, refrigerators, heaters, motors, and furnaces–precise information on the exact composition of that load is often lacking. In addition:
“- New load components alter system characteristics;
“- Loads vary with time, differ throughout the system, are statistical in nature, and may include myriad continuous and discrete controls and protections;
“- There are no accurate tests for verifying load models.
What makes a good load model?
A load model should be able to capture, with acceptable accuracy, load behavior when subjected to practical variations in system voltages. It must be possible to derive the model from information that is relatively easy to obtain for each lumped load represented, and the model should not be overly complex or cause an unreasonable computational burden during simulations.
What types of load models are available?
Load models are classified as either static, dynamic, or composite. A static load model expresses the characteristics of the load at any instant as algebraic functions of the bus voltage magnitude and frequency at that instant. The active power component and the reactive power component are considered separately. Typical static load models have both exponential and polynomial components and are commonly referred to as “ZIP” models because they include constant impedance (Z), constant current (I), and constant power (P) elements.
Dynamic load models express the time-varying characteristics of the load and usually are employed to study system dynamic performance, such as inter-area oscillations, voltage stability, and long-term stability. These typically use differential equations to simulate load dynamic behavior following a disturbance over time. Because electric motors consume 60 percent to 70 percent of the total energy supplied by a power system, the dynamic characteristics of induction motors are usually the most significant consideration in developing a dynamic load models.
Composite load models, which exhibit characteristics of both static and dynamic models, can represent aggregations of small and large induction motors, static load characteristics, discharge lighting, thermostatically controlled loads, transformer saturation effects and shunt capacitors.
How are load model parameters developed?
There are two basic approaches:
1.) A measurement-based approach, or
2.) A component-based approach.
In the measurement-based approach, load characteristics are measured at representative substations and feeders at selected times of the day and season. The results, which can be derived from staged tests, system switching, or naturally occurring system variations, are used to extrapolate load parameters throughout the system, but they are valid only for the particular time and location of measurement.
In the component-based approach the load model is constructed from information on individual load components. The load to be modeled is categorized into load classes, such as residential, commercial, industrial, and mining. Then each category is represented in terms of load components such as motor load, lighting, air conditioning, etc. But, because models constructed in this fashion do not represent accurately the transient responses of active and reactive power, they may not be appropriate for critical dynamic system studies.
Why are air-conditioner loads a concern?
Air conditioner motors are referred to as “prone-to-stall” motors with low inertia, since the mechanical back pressure of the compressor quickly slows the motor to the point where it cannot re-accelerate to full speed even if full voltage is restored. Indeed, not only will air conditioner motors stall during faults greater than five cycles with less than 60 percent nominal voltage, they also absorb a substantial amount of reactive power in the stalled mode.
Are accurate load models enough to simulate system dynamics correctly?
No, more is required. To represent system dynamics accurately, we also need accurate models of generating units, transmission lines, transformers, and other components. Of these, generator models are the most complex and most important for accurate dynamic simulations. Accurate generator modeling will be covered in a follow-up article in a future issue of EL&P.
Abi-Samra, EPRIsolutions senior technical director, has conducted a number of projects on load measurement and modeling, and is an expert in system planning. He can be contacted at 650-855-1022 or nabisamr@EPRIsolutions.com.