I’ve been a researcher in power electronics for a long time, mostly in grid applications like high-voltage DC (HVDC), flexible AC transmission systems (FACTS), distributed generation, and microgrids. Those are still active areas of research, but for me, the really interesting research questions right now are not about individual technologies but about the system challenge: how will grids as a whole operate with very high fractions of inverter-based resources (IBRs), and almost no generation resources that are synchronous machines? And this is not some distant prospect. In Ireland the system non-synchronous penetration (IBRs, essentially) has already risen to 70 percent, which is the current design limit. In Great Britain there have been times when fossil-fueled generation has been pushed down to 10 percent (with nuclear another 20 percent) and the wind generation was already there to go further if the grid had been set up for that operating regime. Inability to operate at or close to 100% IBR is a barrier to further decarbonization.
IBRs are now common in most power systems of the world, but operate simply as resources that supply energy by synchronizing to a strong, existing grid formed by synchronous machines. To progress further, IBRs are going to have to form grids without relying on synchronous machines. There is much discussion already about grid-forming (GFM) inverters versus grid-following (GFL) inverters. There are questions about what should be the core functions of a GFM IBR and what fraction of IBRs in a system should be GFM in order to be stable and secure.
These are questions that G-PST has identified in its research agenda and I discussed in an ESIG webinar on GFM. It feels to me that we need to avoid rushing to answers and instead be brave enough to step back and think deeply. The features of our grid are deeply intertwined with the features of synchronous machines because they co-evolved through the 20th century. If we design our IBRs to form grids in exactly the way grids exist today—to make IBRs in effect drop-in replacements of synchronous machines—we will not only fail to take advantage of the best features of IBRs, but will also build in unnecessary cost.
Grid Compatibility of IBR
My starting point is that GFM IBRs and the grid have to be compatible, but the point of compatibility need not be where it was for synchronous machines. Compatibility is a two-sided challenge: making IBRs fit to work in our grid and making our grid fit to accommodate IBRs. How much of each we do should be determined by what achieves minimum system cost, as illustrated in the figure. As an example, making IBRs supply short-term overload or fault current exactly like synchronous machines is expensive. We should be looking at whether there is a lower-cost solution by adjusting our grids somewhat to operate at lower fault currents of IBRs, but not to the point where grid costs become excessive. I doubt we can construct an exact version of the cost-curve like the one in the figure, but we should have in mind the principle of balancing the cost burden.
We should consider each system need and IBR feature in terms of compatibility. Synchronization of GFL IBRs to a grid requires a strong grid (a low effective impedance at the connection point). To what extent should GFL IBRs be obliged to synchronize to weak grids? To what extent should system operators be obliged to provide strong grids to accommodate GFL IBRs (by procuring “strength” services from GFM IBRs perhaps)? To what extent should IBRs be obliged to provide fault current substantially above 1 pu (which will add to IBRs’ cost)? To what extent should system operators be obliged to use protection relays redesigned for lower fault currents? We have similar questions about how to recover system frequency after loss of infeed, how to mitigate harmonics and unbalance, how to mitigate phase jumps. All questions of this type should, in principle at least, be answered by thinking about minimum system cost. And as if that were not difficult enough, many of these questions are coupled and have to be answered as a set of trade-offs.
So, for me at least, the issue of what are the research questions around GFM IBRs begins with a series of system-level design issues in which the components of the system, especially the IBRs, and the system operating principles have to be co-designed. This is where we start to draw in research in power electronics itself. What circuit innovations might allow an IBR to deliver short-term overload current and fault current? What control innovations might ensure that IBRs contribute damping to various types of disturbance? Can GFM IBRs adapt to local conditions and autonomously deliver the services necessary? What is the power electronic alternative to a synchronous compensator? Can combinations of resources, such as storage, wind turbine, and STATCOM, provide a universal IBR?
There is a pressing need for research in a different vein. We need analysis and synthesis tools to answer the system design and IBR design questions. Just as grids co-evolved with synchronous machines, so too did our analysis tools. There are now challenges in simulating systems with thousands of complex and disparate IBRs in place of hundreds of mature and familiar synchronous machines. There are unresolved questions about the role of electromagnetic transient (EMT) versus phasor simulation given the overlapping dynamics of various IBR features. There are difficulties in root-cause analysis because of the black-box nature of many IBR models.
My concern is that we are placing too much store in exhaustive simulation studies and not enough effort in developing the theoretical base to answer questions on stability limits or dynamic security or protection discrimination directly via analytical tools.
A few years ago, displacing all synchronous machines and creating grids around GFM and GFL IBRs alone would have seemed a fanciful idea, the idle dreaming of an out-of-touch researcher. Yet, we are now heading there fast. The smart move is to carve out some time to think about how to do this really well—how to be imaginative about changes in grid operation that exploit the flexibility of IBRs rather than forcing IBRs to look exactly like synchronous machines and building in cost that we could have avoided. That should be a very exciting challenge for researchers; it’s not quite creating the grid from scratch as previous generations of engineers did, but it is a chance to define a new grid.
Professor, Co-Director of the Energy Futures Laboratory (EFL)
Imperial College London