By Phil Kreveld
The brave new world of 50% power plus penetration of renewables in traditional AC grids requires reimagining legacy engineering and above all, a millisecond, synchronised view of network events. Synchronous generation is making way for asynchronous power generators, which connect via inverter-based resources (IBR). The inbuilt capacity for reactive power of synchronous generation has to be replaced by static var compensators and statcoms to make up for this loss so as to continue providing voltage stability for transmission lines.
Above all, the timescale for making decisions in unforeseen contingency events is being compressed from seconds to milliseconds with the loss of inertia. Rather than clinging on to engineering concepts that belong to earlier times, network designers are far better served by, in the first place, installing synchronized monitoring systems and a suitable big data platform to allow intensive study and cross-correlation of events on a network-wide basis.
Conveniently, we still assign IBR nice, clean sinewave voltages and currents, and make other simplifying assumptions so that we can squeeze it all into ‘rms’ thinking. The IBR, replacing synchronous generation, operate on millisecond and smaller time intervals.
Controlling their effects on circuits requires new thinking — a real, integrated systems plan. Such a plan must cover distribution as well as transmission and generation. In addition to classical engineering concepts, we need to think again of time-variant quantities, as for example illustrated by a simple circuit of an inductor, resistor and capacitor in series, describable in a time variant equation.
What if e(t) were a series of 5 kilohertz voltage pulses from a voltage sourced inverter, how would we go about finding a closed solution?
The short answer is that no one would bother. It is tacitly assumed that an inductor or inductor-capacitor on the inverter output would provide a sine wave voltage and current. Decades ago, that would have been a reasonable assumption inasmuch as there were few IBR by comparison to traditional synchronous sources. Furthermore, depending on grid topology, the few IBR connected to strong networks as other than for the matter of fuel logistics, synchronous generators were built close to dense population centres.
In other words, few inverters connected to a virtually 100% synchronous grid, basically an ‘infinite bus’, one with absolutely stable voltage and frequency. Those assumptions grow less valid by the day as IBR continue to replace synchronous generation. Furthermore, many IBR connect to weak networks typical of remote energy zones (REZ) and do not have the convenience of an infinite bus for stabilizing their frequency and voltage.
Long links connecting REZ to transmission line connection nodes have high p.u resistance and reactance and low X/R ratios.
In short, even if the connection node on a transmission line represented close to an infinite bus, an inverter at the end of an REZ link would attempt to alter the voltage and phase at its connection bus depending on how much power were required of it. Therein lies a problem which is of increasing seriousness, requiring ‘tuning’ of inverters where a number of them connect to the REZ generator bus. Tuning is basically the adjusting of low pass filter time constants of inverter phase-locked loop circuits to minimize power and voltage oscillation on millisecond time scales. These are not visible on the basis of rms observation of voltage and current.
Remote renewable generation connecting to long transmission lines are short on the required reactive power for voltage stability. So far, we have resorted to the use of synchronous condensers, which in addition provide some inertia. This represents, however, a short-term solution because of their inability to provide voltage forming. As already mentioned, static var compensators and more importantly, statcoms with battery support to allow voltage forming, will have to be employed.
The IBR time-variant voltage and current signals brought about because of their high open loop gains combined with inner and outer feedback loops with fast but also variable transfer functions, cause millisecond and signals which are masked if only rms voltage, current and phasors are being measured. In short, there is great sense in monitoring the ‘power quality’ (essentially monitoring electromagnetic transients –EMT) of IBR. The enormous amount of data potentially available would seem to rule out any attempt at real-time control schemes, for example output power control, but as explained further down, that does not necessarily have to pose a problem.
Synchronisation problems with inverters
Monitoring and control of IBR require both EMT and rms measurement. In fig 1, three IBR connect to a sending end bus of a link with relatively high p.u. resistance and reactance, connecting to a transmission line.
As is evident, the three IBR have different ratios of real to reactive power. The connection point to the transmission line in practice will not represent an infinite bus but it will approach this condition.
Using rms quantities and phasors, fig 2 illustrates the dependance of the sending voltage on link impedance. IBR tuning problems are evident. Changes in power flow alter the voltage magnitude and phase, requiring the phase-locked loops (PLL) of the IBR to adjust, and do so synchronously—a close to impossible task. Slowing the PLL response, basically ‘integrating phase change’ goes towards slowing PLL hunting. If we knew the impedance of the link in real-time, as resistance would be variable for sure, the receiving voltage at the coupling point to the transmission line could be calculated and used as the less variable phasor to lock onto.
A method for synchronization based on the line parameters has been proposed by Suul, Arco et al. for application in HVDC converters. The authors argue that the method is equally applicable to inverters for general use in weak networks. Measurement of the phasor at the point of coupling on the transmission line would eliminate the need to know the line impedance. Currently most national grids lack a comprehensive, live data base that would be able to furnish such phasor data.
As has already been mentioned, we are only partially served by rms quantities and phasors but they are an important foundation. As the diagram in fig 2 shows, the ideal data base includes the and EMT parameters as well.
In the above equation, φ is the power factor angle at the sending end and the link impedance. The far end voltage is the preferable phasor for synchronization. In real and imaginary coordinates, we can write, with φ as lagging angle
This yields the magnitude of the receiving phasor and the angle, d as the arctangent of the imaginary component divided by the real component.
We can imagine the receiving phasor to have an arbitrary (but fixed) angle by virtue of it being close to an infinite bus, and the angle d to oscillate with fluctuation in power from the IBR. They can however, be synchronized by the fixed phasor and that provides a significantly more stable base for the IBR PLLs.
The use of synchrophasors in networks is well established and if we allow ourselves to dream a little, it isn’t hard to imagine IBRs with the ability to pick up, for example GPS synchrophasor signals and to lock these onto their inverter PLL. No such system exists as yet, nor are there inverters available that permit adding into the PLL control loop, a virtual impedance adjustment and alteration of their synchronous reference frames. However, that is only a matter of market requirements as we are heading towards ever higher penetration of renewables.
Integrated systems with high IBR penetration have to be data based. This requires the deployment of a large number of synchronously measuring devices that measure rms parameters, synchrophasors and EMT. As to the latter, remotely programmable on-board EMT pass/fail criteria, so as not to have a totally unwieldy database, would be sensible.
The monitoring devices would be deployed in all REZ links and IBR. The combination EMT ‘push’ information combined with rms and phasor data would allow controlling the growing the IBR installation base. IBR will of necessity become more able, for example in addition to voltage forming (already available for microgrid inverters), to have some forms of inertia and damping—and to be remotely controllable.
For example, depending on short term measurement intervals, IBR could be instructed to switch from grid following to voltage forming.
An exciting future awaits us if we stop compartmentalized thinking, applying yesteryear solutions to a patchwork, rather than a truly integrated system because only the latter can provide both an effective and economically sensible solution.