The US electric power grid has a remarkably sophisticated and capable control system that has centralized, distributed, and autonomous components and in some regions is integrated with real time optimizing electricity markets – sophisticated and capable that is, for the 20th Century grid. Depending on how one counts, it is possible to identify more than a dozen grid control functions, including unit commitment, dispatch and curtailment, load sharing, flow control, balance, interchange, frequency regulation, voltage regulation, reactive power regulation, stabilization, synchronization, and demand response. These are handled by as many control techniques and structures, some fairly simple, some of remarkable technical elegance.
Over the course of decades, the grid has been known as “the graveyard of advance control theories.” Time and again control theorist have proposed new schemes for grid control, based on ever-advancing control theory. But the electric utility industry has steadfastly resisted these approaches, largely because there has not been sufficient reason to do anything other than make linear incremental changes to grid control. Hence, as pointed out in the PEI paper A Gambit for Grid 2035, grid control has remained on the S1 curve.
But this is all changing. By the beginning of the 21st Century, it had become clear that efforts to improve US electric power systems and indeed power systems around the world were being stymied by significant changes to the grid that increasingly diverged from the basic principles and assumptions under which the 20th Century grid was developed, including its controls. Some of these changes arise from changing consumer expectations, some from changing public policy, and some from the impact of technology innovations.
As Ken Fong, former Director T&D Planning at Hawaiian Electric has said, “The grid we have is not the grid we need.”
The implications of decarbonization policies, consumer experience expectation shaped by modern electronics, software, and communications (including the internet), and emerging technologies such as inverter-based resources, grid-compatible power electronics, the bifurcation of generation into bulk system-connected large scale resources and potentially vast numbers of distribution-connected small resources, bulk energy storage at all grid scales, shift of loads to nonlinear behavior, and ubiquitous communication connectivity are changing old grid control problems and posing entirely new ones. These new problems include adding storage energy state management to the traditional grid flow control, balance and regulation issues; variable grid structure and variable flow control; coordination of potentially huge numbers of grid-connected active devices and resources that are not even owned by the electric utilities; mapping N-way logical power flows onto circuit level physical flows; operational coordination and synchronization between the grid and other coupled sectors (natural gas, transportation, buildings, etc.); and volatility exchange control between Transmission and Distribution, and between the grid and other coupled sectors.
Add to this the fact that grid dynamics are increasing in speed while computational issues are increasing in scale due to increasing complexity and sheer numbers of elements and constraints to be included in short cycle time controls. While there was a time when things did not happen in time periods shorter than about five minutes on distribution systems (except for protection), now and going forward control cycle times must become sub-minute, sub-second, and even sub-cycle, all the while becoming much more agile than 20th Century controls.
So, let’s add a corollary to Ken Fong’s statement – the grid controls we have are not the grid controls we will need.
A lesson from the aerospace industry: at one time aircraft were designed to be inherently stable so that a pilot could control the aircraft manually. The need for increasing speed and agility led to the development of aircraft that are inherently unstable and cannot be controlled manually by the pilot. Such aircraft are controlled by computers in a mode referred to as fly-by-wire. The pilot indicates what the aircraft is to do and the computers operate the control surfaces to fly as required while maintaining stability through continual automatic adjustment. To do this, the controls use constant real time measurement of the aircraft dynamics.
The grid we need has a need of its own: the grid equivalent of fly-by-wire control. This is where advanced control methods finally can and should be applied to the grid. But the problem is harder than the aircraft control problem – the grid is vastly more complex than any aircraft and the scale is enormously larger, plus we generally lack the necessary real time sensing and measurement – a common pitfall of older attempts at “optimizing” the grid. Control theorists and electricity market economists take note!