AC/DC is not only a rock band of repute with countless classics under their belt, but it was also the topic of the great current war between Tesla and Edison. Both approaches, however, required the presence of a rotating machine, inside which the friendly properties described by Faraday, Newton, Maxwell, and Lenz meshed together to bring about the ability to serve load with electric current. These properties of electromagnetism introduced another form of energy which is crucial for the generation of electric current. A rotating mass of metal with copper wires will be unable to generate a voltage unless it rotates in the presence of an intersecting magnetic field. Thus, two forms of energy are needed for rotating machines to “self-generate” an electric voltage: (1) input prime mover energy which relates to the watts, and (2) internal magnetic energy that relates to the vars, which is typically the excitation power. The synchronous machines that have dominated the power system until recently possess both forms of energy, while inverter-based resources (IBRs) possess only the first.
Why has the notion of a grid-forming inverter come up?
Unlike synchronous machines, most IBRs completely lack internal magnetic energy (e.g., solar PV or batteries). Therefore, in order to generate an ac voltage and continue to operate in a stable manner, conventional IBRs as designed today require the presence of an external voltage reference. Additionally, most of today’s IBRs operate with an objective to inject maximum available active power under normal system conditions in order to maximize economic revenue and to inject specific reference current under disturbance conditions. To achieve this objective, the controls of the inverters in these plants rely on continuous measurement of the network voltage phase (usually through a phase locked loop (PLL)) and voltage magnitude. Based upon this measurement, to meet the objective of power injection, the controls then react by varying the inverter terminal voltage in order to have tight control of the current flowing out of the inverter. Because IBRs are current-sensitive devices, the cycle of measurement and reaction must have a fast response time to prevent an overcurrent condition. An inverter that can inject its share of P and Q successfully under normal grid conditions is said to follow the grid.
Figure 1: High-level topology of conventional IBR control
However, as the network approaches higher shares of IBRs and with the general tendency of connecting IBRs in areas of low system strength, there is a possibility of increased sensitivity of the network voltage to a change in injected current. This increased sensitivity would now result in the IBR controls having an inaccurate measurement of the fast-varying network voltage, resulting in an inaccurate reaction from the controls, thereby leading to possible instability.
This raises the question of whether the cycle of measurement and reaction needs to be faster than in conventional IBR controls. Alternatively, does network voltage need to be the measured quantity, or can the IBR controls decide the inverter voltage level on their own? Relaxing the need to have perfect control of current at every instant in time (and thus bringing about an additional degree of freedom) is at the heart of so-called grid-forming inverter control topologies. This relaxation can further bring about a relaxation in requiring network voltage to be the measured quantity and can result in inverter voltage being varied in a slower manner, thereby increasing the system stability, while “forming the grid.”
How much will be needed: Implications of relaxations on energy requirement
The relaxation in tight current control in a grid-forming inverter brings about an increased uncertainty and variability in the amount of energy extracted from the dc side of the inverter, as the current output is now not rigidly controlled unless it hits a limit. Would this result in a need to have a large energy buffer in a grid-forming inverter? In a rotating machine, where current output is also not rigidly controlled, the uncertainty and variability of energy extraction is not too much of a concern in the short term, due to energy stored in the rotating mass. However, in an IBR, there is no presence of a natural energy buffer. Handling this energy uncertainty in an IBR would require either an additional energy storage component or operation in a de-rated mode. Additionally, appropriate damping controllers for the associated electrical modes would also be required. Thus, a specification for need of grid-forming inverters must necessarily include specifications on the size and duration of the energy buffer in order to obtain reliable service.
When should these specifications start rolling out?
Pockets of many of today’s power systems have already been exposed to an increased sensitivity of voltage, and this is only expected to further increase. Thus, is it already time to require the presence of inverters that can be called grid forming? And if it is time, then the next questions to be answered are: (1) What would be the specifications upon which these inverters and their associated energy storage buffers would be designed? (2) How many would be required? (3) Where should they be placed?
A specification for grid-forming inverters involves a change in the present operation paradigm. At the same time, a step change in the operation paradigm is inadvisable and inconceivable. There is a definite possibility of conventional IBR control instability due to high sensitivity of voltage to change in current injection. However, it must be determined whether this instability is only due to the conventional grid-following control loops requiring an external voltage reference, or whether it is also due in part to the operation paradigm wherein the active power order does not change based on measured network voltage phase.
Consider a small system as shown in Figure 2 with an IBR on bus 7 on the left and the bulk power system on bus 21 on the right. With conventional IBR controls, upon disconnection of the bulk power system at t=5s, since there is now no stiff network present, one can expect an unstable operational behavior as shown in Figure 3.
Figure 2: Small test system with IBR on left and the bulk power system on the right
Figure 3: Instability upon disconnection of the bulk power system at t=5.0s
However, from control system theory, a closed-loop system with controller inputs changing based on measured feedback tends to have an increased stability margin as compared to a rigid open-loop system. Applying the same concept to IBRs, a conceptual high-level topology is shown in Figure 3 wherein the measured network voltage influences the magnitude of power to be injected by the inverter.
Figure 4: High-level topology of modified conventional IBR control
The specific controller features that bring about this influence have low importance from an overall grid perspective. If all other aspects or layers of IBR controls are kept unchanged, Figure 5 shows the stable operational behavior of this same small system. An argument could be made here that the controls are only helping with sustaining the operation of the grid once the network is disconnected and are not truly forming the grid on their own, as they may not be able to “start the system.” However, it should be recognized that even for rotating machines, controller settings can be different during a blackstart process as compared to the settings that sustain the operation of the grid. Moreover, for a bulk power system, given that it is more important to aim to achieve sustained operation, this form of controller influence could have large impacts.
Figure 5: Stable operation upon disconnection of the bulk power system at t=5.0s with modified conventional control
Bringing about such an influence could result in an incremental change in the operation paradigm that moves the system towards a more stable region even with high network voltage sensitivity. Additionally, many power systems around the world are already requiring IBRs to have the capability to achieve this relationship, such as automatic closed-loop power output reduction for over-frequency conditions. Further, recent ratification of proposed changes in the small and large generator interconnection agreements in North America will increase the set of IBRs that would carry this capability. These agreements require that IBRs must have the capability for their active and reactive power references to be influenced by the network voltage. Thus, power systems may already be on their way to bringing in “grid-forming” capability by allowing for IBRs to be operated flexibly in relation to grid conditions, rather than be rigid power/current injection devices. While this won’t necessarily bring about the capability in each and every single IBR to synthetize a voltage in isolation without measurement of the network, it may allow for each IBR to form the grid while still following the grid.
Will grid-forming IBRs be enough?
As with any new concept, optimism must be balanced with caution. What we have described here may not be a true grid-forming converter from the perspective of blackstart or system restoration, but is rather a step in that direction, especially for improvement in system stability and reliability in an interconnected power system. However, system reliability and stability should not depend upon just one new solution. A multitude of complementary solutions, both from generation planning and transmission planning, should be used. Further, new technology also raises additional questions with regard to requirements such as the ability of system protection to operate adequately, requirements for improved methods of scheduling and dispatch, procurement of the requisite amount of ancillary services, and maintenance/improvement of reliability and resiliency. Obtaining answers to such questions is very much an active topic of research and development.
Sr. Engineer Scientist
Grid Operations and Planning Group
Electric Power Research Institute (EPRI), Knoxville, TN