Since the early 1900s, with the proliferation of alternating current rotating generators and development and setup of electric grids, the electrical frequency in the network has nobly served as the pulse of the electric power system. The electrical frequency connects the speed of rotation of the generators to the power consumed by the loads. Thus, akin to noting the state of the human body by measurement of the pulse at either the neck, or wrist, or behind a knee, the state of today’s electrical grid can be determined by measurement of the electrical frequency at any location on the grid. A decrease in frequency is intuitively related to the slowing down (or deceleration) of the rotating machines, which in turn leads to the conclusion of consumption being greater than supply, while a rise in frequency intuitively leads to the conclusion of supply being greater than consumption.
With the increase in penetration of limited response (by design) inverter interfaced energy sources in the power system, the link between frequency and supply/demand imbalance is becoming more sensitive as the size of the responsive set of rotating machines decreases. However, at 100% penetration, while it is intuitive that the inverter sources would now have to be responsive to power imbalances, frequency (the pulse of the system) no longer holds the same meaning, and is thus lost as a control variable. The reason for this loss is that it can be expected that in a 100% inverter based system, the energy resource device (either a rotating wind turbine, or static battery bank or solar array) is isolated from the electrical network by the inverter. Due to this isolation, a change in electrical frequency in the system, say due to a change in load, is not naturally reflected back into a physical property of the energy source (as opposed to a conventional rotating machine where the change in frequency gets reflected as a change in speed of the machine).
Using coordinated control algorithms, it is of course possible to artificially reflect a change in source output power for a change in system electrical frequency. But is that the best way to tackle the challenge? It is conceptually definitely the easiest (or most convenient) way, as the operating paradigm of a large power system retains all its existing characteristics and performance metrics, while requiring a change only from the source side. In addition, such a control algorithm is attractive, as it would work even if there were a minority of conventional rotating machines present in the system. However, it must be mentioned here that while conceptually such a control would be the easiest, there are still multiple obstacles involved, chief among them being injection of the required amount of energy within the required minimum amount of time during an imbalance condition.
If a convenient way already exists, then, is it worth pursuing other control algorithms? Most definitely! It is possible that the most convenient way is not the most efficient way, or it could even be a restrictive way, as it involves holding onto the past (i.e. keeping the characteristics and performance metrics of the system the same) and forcing a new technology to conform its behavior to a set of pre-defined rules. The present day research discussions, related to the increasing penetration of converters, regarding ROCOF, reducing inertia, and fast frequency response, is very much relevant when there are still a significant percentage of synchronous machines, but these types of responses are required characteristics for the existing system. In trying to mold the converter’s behavior to provide these responses, one may be sub-consciously trying to keep the system characteristics the same, and only asking the converter to adapt to the existing system.
But, frequency is still an important variable from the perspective of the lines, transformers, and motor loads (impedances, saturation, and speed of rotation), so even if it doesn’t matter to the converters, maintaining frequency at 60Hz/50Hz is important to the components of the system. And that points towards having a control where the converters still maintain frequency, but at 60Hz/50Hz. However, bringing about a constant frequency operating paradigm requires an additional amount of energy injection in the first few seconds after a disturbance, but it is possible.
Thus, if electrical frequency no longer provides a measure of the imbalance in supply and demand, then, can we control the bulk power system at constant frequency? Can the frequency signal be removed from all power sharing control mechanisms (primary and secondary control)? What other forms of communication would then be required to bring about adequate sharing of the burden, and how much additional energy will be required to be injected in the few milliseconds after a disturbance?
In some recent research work carried out at the Electric Power Research Institute, Inc., USA, the loss of 2 GW of generation in a large all-converter fed power system was studied. The amount of energy injection in MW-sec in one area of the system is tabulated in the table. In the first 500 ms after the disturbance, while the energy injected by a constant frequency converter control scheme is lower than the large amount of energy injected by synchronous machines, it is still greater than the energy injected by converters operating on conventional frequency droop control. However, as time progresses, in the total 1 sec after the disturbance (and this includes the first 500 ms mentioned above), the total energy injected by both the synchronous machine scenario and the converters with frequency droop scenario is almost the same. However, now, the energy injected by the constant frequency control scheme is larger. And in this same duration of time, converter controls are able to bring the frequency back to 60 Hz while in the conventional frequency droop control, the frequency decline is still being arrested, with the nadir occurring only 5 sec after the disturbance at 59.62 Hz.
|Time duration||All- converters with frequency droop
|Synchronous machines with same percentage droop
|All converters at constant frequency
|First 500 ms||5338||5540||5413|
|First 1 second||10638||10665||10892|
It thus becomes a question of whether a lot of energy needs to be injected immediately upon the occurrence of the disturbance (as in a case which mimics conventional frequency response), or is a steady increase of energy injection better (as in the constant frequency converter control scheme)? And in either case, will the source behind the inverter have the capacity to inject the required amount of energy, and not need to consume it back from the system in the subsequent few seconds? It must be noted that these studies were carried out with the assumption that there is available headroom on the converters to be able to withstand a loss of generation. How the headroom is obtained is related to scheduling and resource adequacy, which is another topic the project is trying to address. The question of communication mentioned above will be saved for another day, but does not represent an additional burden.
The bulk power system is undergoing a transformation which in many ways is similar to the process that took place a century ago, back in the early 1900s when the first interconnected power systems were being established. Converter capabilities bring about possibilities that are not even remotely possible with synchronous machines, which in some ways is also a drawback of the converters. However, while keeping pace with the transformation of the power system, it is perhaps also prudent to let go of the past with respect to the electrical network’s performance characteristics and operating paradigm, and look to the future.
Sr. Engineer Scientist, Grid Operations and Planning
EPRI | Electric Power Research Institute