By Phil Kreveld
What have Tim Nelson and Bernardus Tellegen to do with one another? Tim Nelson has designed a new scheme for electrical energy capacity investment. Prof Tellegen, a Dutch electrical engineer (1900-1990), is from a distant era, and responsible for a theorem in electrical network analysis. It could be the basis of a ‘half-Nelson’ wrestling manoeuvre on electricity supply, immobilising market operations in the interests of grid security.
Tim Nelson’s proposals are welcome. They will encourage capacity investment, which in view of projected electrical energy demand is very necessary. To his credit he has astutely remarked that his proposal does not provide the very necessary engineering solutions—but these will be ‘encouraged’ by an improved investment climate. Will a dusty theorem from a century ago provide security? Not unless its implications for grid engineering and control are taken into account.
Related article: Transmission line limbo: How low can power flow go?
Grid security fuels a lot of commentary. Conservative media, in particular, are putting up a barrage of criticisms of wind and solar, much of it ill-informed, some of it bordering on the hysterical, while the federal government is ‘whistling in the dark’ on reaching its brave renewable targets. Politically it is a problem because it makes renewable targets based on arbitrary timelines very difficult if not impossible to achieve.
Summing up the present renewable integration path is simple enough. In spite of all the science, and trials, in particular of grid forming inverters, the Australian Energy Market Operator (AEMO), has chosen grid following battery energy systems as safe when backed by synchronous condensers. Thus, transmission grid operators are going to the Australian Energy Regulator, virtually getting automatic approval for the investment in syncons. The ordinary punters will pay the consequent additional transmission charges as part of their tariffs.
Tellegen’s theorem sees to the imposition of conditions for stability. In its simplest form, applied to alternating current networks, it says that the sum of all active and reactive powers in a network in any instant is always zero. Summed up that way it seems almost trivial in its simplicity. However, it has serious implication for the national grid’s stability and security as we transition to evermore variable power demand and generation, courtesy of solar and wind generation. In a nutshell, the interaction between market requirements and those of network stability and security have little if any correlation in the highly variable power demand landscape. If Spain teaches us anything, it is that market conditions provided insufficient synchronous capacity on hand when it was needed for voltage control.
The essence of Tellegen’s theorem is shown below.
Where t is any instant of time and k is the number of branches in a grid, and v and i are instantaneous values of branch voltage and current. This innocent-looking relationship has enormous implications for the shape of new methods of renewables-stacked grid control, handing much of it to machine-learning.
When ‘nice and steady’ becomes a thing of the past, frequency and bus voltages can be all over the place, and any semblance to standard textbook AC network is lost. This is the nightmare for the Australian Energy Regulator (AER), and transmission and distribution network operators. The separate considerations of ‘inertia’, ‘firming’, ‘load following’, etc., fall away and only an iron-clad principle remains. Tellegen, from a past century, will not put the kibosh on Nelson’s plan but will exert the ‘Half Nelson’ on the National Electricity Market by making grid security override the operation of energy markets!
With grid security outweighing market considerations, what Tellegen’s Theorem so elegantly exposes, is the need to design a very new control philosophy for the NEM. So let’s go back to 2021 when AEMO provided evidence on challenges facing the national grids to the Parliament’s then House Standing Committee on Energy and the Environment. AEMO stated that the control of stability would involve “a dynamic millisecond-by-millisecond equation with regard to operation of grids with high penetration of renewables that [would be] carefully balanced to ensure that as consumers need[ed] it the electricity in their homes and businesses [would be] available”.
The thought of ‘what hype’ occurred to this correspondent then—because AEMO simply had so little live data from distribution networks as to make ‘millisecond-by-millisecond equation’ solving a complete fantasy. This is essentially still the case today although AEMO is continuing with supervising the deployment of wide area monitoring (WAM) phasor monitoring units (PMU) in transmission. This is a big step in the right direction leading to WAMPAC (adding ‘Protection and control’) involving Queensland’s PowerLink and NSW’s Transgrid.
We will not be able to rely on traditional control rooms. Rather, reliance will have to be based on a series of machine-learning derived algorithms, probably with PMU acting as the inputs to ‘perceptrons’ (the top layer), of interior decision-making layers, and human-interaction output layers. The interior decision making will perform such tasks as dynamic excitation control of syncons, and var and statcom compensation control. The human output layer will flag required operator actions such as ‘switching off a parallel transmission link’, ‘deactivating under frequency load shedding’, ‘islanding region XYZ’, etc.
Related article: ‘Dunkelflaute’ and weak grids
The point here is that the degree of complexity is growing so fast that we have to rely far more on automatic, fast decision making to manage a truly complex grid. AEMO’s digital twin of the grid is an excellent start for the development of ML algorithms. The critical issue is the testing of these in the real world—and even then, are we not ‘out of the woods. The biggest challenge of all is in the distribution networks of Australia. Nearly all electrical energy goes to distribution networks. And it’s ‘distributed energy resources’ (DER), mainly rooftop solar photovoltaic energy sources–and batteries, that are essentially outside of grid control developments.
It’s AEMO’s ‘dynamic millisecond-by-millisecond equation’ that needs solving because otherwise we cannot avoid sliding from the half-Nelson to a ‘crippling collapse on the mat’. This is where there is a blind spot we walk around by on the one hand developing sophisticated transmission grid controls and on the other, disconnecting DER resources so as to facilitate transmission grid control. In the final analysis, this is anything but intelligent system design.
