Fullscreen HTML5Med Res (??MB)HTML4

The Future of our Power Grid The first article in this series last month described how our electrical grid is changing, the pros and cons of the various types of generators, costs and Demand Response. This second and final instalment finishes the discussion by covering inverters and grid stability. L ast month, I explained how as coal and gas power stations reach their end-oflife, they are increasingly being replaced by other generation methods like wind and solar power. However, that transition is not without its challenges due to the way that generation varies over time, with changes in the weather and the day/night cycle. Thankfully, this transition is slow, which is allowing the deployment of various techniques and technologies to overcome those limitations. Energy storage and Demand Response were covered in that first article, but now we come to the nitty-gritty, such as the ways that solar and wind generators are connected to the grid to better match demand and improve grid stability. Solar inverters Inverters for photovoltaic panels take a DC supply from the solar array and convert it to AC to feed the grid. They typically use Insulated Gate Bipolar Transistors (IGBT) arranged in a three-phase H-bridge topology (see Figs.13, 14 & 15). The IGBT is effectively a small Mosfet and a large bipolar junction transistor (BJT) combined on a single die. By combining the two transistor types, the IGBT benefits from the advantages of both technologies; the BJT is well-suited to high-power applications due to its favourable output characteristics, and the Mosfet is a convenient way to provide base drive to the BJT given its high gate impedance. Using pulse-width modulation (PWM), a three-phase AC waveform can be synthesised from the input DC, similar to the operation of our Mk2 Variable Speed Drive for Induction Part 2 by Brandon Speedie Motors, published in the November & December 2024 issues (siliconchip. com.au/Series/430). Typically, the chopper frequency is in the order of 50kHz or so. It is filtered out by an LC network (usually a ‘pi’ or ‘T’ filter) on the output of the inverter to form a smooth sinusoidal waveform. Utility-scale solar farms receive further filtering from the inductance in their grid-connected transformers, which step up the low voltage output from the inverter to the high voltage of the transmission network. This synthesised AC waveform needs to be precisely controlled to synchronise with the grid. This is achieved by sampling the grid voltage to form a phase-locked loop, which becomes a reference waveform. By varying the amplitude and phase of the synthesised waveform with respect Figs.11 & 12: the topology of an AC-coupled hybrid solar and battery generator is shown in the left diagram. The alternative configuration of a DC-coupled hybrid solar and battery generator is shown at right. For the DC-coupled system, with sufficient irradiance, power can be exported to the grid and charge the batteries simultaneously without having to oversize the inverter. Original source: https://blog.fluenceenergy.com/energy-storage-ac-dc-coupled-solar 34 Silicon Chip Australia's electronics magazine siliconchip.com.au to this reference, the output voltage and current can be controlled with precision. This control is referred to as ‘grid following’, as the inverter is tracking the grid waveform and operating as a current source. The other type of inverter control is called ‘grid forming’, meaning the inverter operates as a voltage source and largely ignores the existing grid waveform. In normal operation, the inverter controls its output power to optimise the operating point of the solar array. This is known as maximum power point tracking (MPPT), which involves holding the array DC voltage at the optimum current for the solar panel to generate its maximum power (see Fig.16). This position is constantly changing with variations in irradiance and temperature, so the MPP tracker works through trial-and-error to dither the DC voltage up or down to search for increased power. Fig.13: a typical IGBT die structure. Original source: https://w.wiki/Bqfd Fig.14: an equivalent circuit of the Insulated Gate Bipolar Transistor (IGBT). It has a BJT and Mosfet connected together on a single silicon die. Original source: https://techweb.rohm.com/ product/power-device/igbt/11640/ Battery inverters Similarly to their solar counterparts, battery inverters take a DC voltage from the cells and convert it to an AC voltage for the grid. In fact, many solar inverter OEMs service the battery market with identical hardware. The difference is in the control software; the MPP tracker is replaced by algorithms to gracefully charge or discharge the cells with minimal degradation. Battery health is mainly a function of temperature and state of charge (SOC), so current limits are reduced at extremes of temperature, or when the cells are fully charged or discharged. Fig.15: an inverter circuit showing output ‘T’ filter (an LCL network) and the additional inductance from the grid-tied step-up transformer. The six IGBTs synthesise a three-phase AC waveform using PWM. Original source: https:// imperix.com/doc/implementation/active-damping-of-lcl-filters Battery-solar hybrids Increasingly, batteries are being built alongside solar photovoltaic systems. They are a good combination, as the battery not only avoids paying for grid electricity but also network fees. Most solar-battery hybrids currently in operation on the grid are ‘AC coupled’, meaning that they are joined on the output side of their respective inverters (see Fig.11). A new technology gaining popularity is the ‘DC coupled’ hybrid. Rather than the batteries connecting directly to a dedicated inverter, they instead interface to the solar array through a DC-DC converter. The inverter then converts both battery and solar power to AC for the grid (see Fig.12). siliconchip.com.au Fig.16: the output characteristics of a solar panel for different values of irradiance. A connected inverter constantly searches for the optimum operating point in a process known as maximum power point tracking (MPPT). Source: www.researchgate.net/figure/fig3_324179520 Australia's electronics magazine April 2025  35 Fig.17: a Doubly Fed Induction Machine (or Generator) used to generate power from a wind turbine. The stator is directly connected to the grid, while the rotor is fed from a back-to-back inverter. The DFIM therefore decouples the turbine rotational speed from the grid frequency, allowing the control system freedom to optimise for maximum power. Original source: www.mdpi.com/energies/energies-15-03327/article_deploy/html/images/energies-15-03327-g001.png The main benefit of this topology is removing the inverter as a bottleneck to power flows, as most solar systems match an oversized array to their inverter. This is known as the DC/ AC ratio; it is usually around 1.3:1, to balance the cost of the inverter against increased revenues from higher power handling. On residential systems, this leads to the ubiquitous 6.6kW array matched to a 5kW inverter. The drawback to such a ratio is that when there is sufficient irradiance, potential power generation is wasted as the inverter is already at its limit. With an AC-coupled hybrid, this bottleneck also limits the battery charging; any power from the solar array has to pass through the grid-­connected solar inverter before it comes back through the battery inverter and into the pack. On a DC-coupled system, this limit is alleviated. Assuming sufficient irradiance, the inverter can be exporting at full power, and energy that would otherwise be lost from the solar array is used to directly charge the batteries, giving a superior yield for a given solar array. DC-coupling can also help a system remain below a given size for regulatory reasons. Grid-scale systems with less than 5MW of inverters have a simpler grid connection process, and residential systems are capped at 5kW of export. 36 Silicon Chip It is the residential sector in particular that will see an increased uptake of DC-coupled ‘hybrid inverters’ over the coming years. Wind turbine inverters Early turbine designs simply connect an alternator directly to the grid, but this limits the rotor to a fixed operating speed (the grid frequency), which is not necessarily the optimum speed for maximum power. A more modern design for small wind turbines uses a rectifier to convert the alternator’s AC output to DC, then an ordinary solar inverter to convert it back to AC to feed the grid. This way, the inverter has freedom to use its MPP tracker to find the best operating point, which improves yield despite the additional losses from the conversion process. Grid-scale wind turbines use a different inverter-based technology known as the doubly fed induction machine (DFIM). The stator is directly coupled to the grid, while the rotor is energised by a back-to-back inverter (see Fig.17). Thus, the rotor can be fed with an arbitrary waveform in much the same way as a solar inverter. By varying the voltage and phase, the power coming out of the stator is tightly controlled. Most critically of all, the rotor can be excited with a Australia's electronics magazine fixed frequency to match the grid. The stator output will always produce this same frequency, despite constant variation in the turbine speed due to wind fluctuations. This allows the control system to optimise the rotational speed of the turbine for maximum power, in a similar way to MPPT for solar panels. Grid stability – voltage control Network operators must keep tight control over grid voltage to prevent damage to connected assets and the network infrastructure itself. This voltage is only permitted to vary in a very narrow range. In Australia, that’s 230V AC +10%, -6% for a single-phase supply (ie, 216V to 253V AC). There are two main tools that can be used to maintain these limits: transformer tap changers and reactive power control. Transformer tap changers simply select between a series of closely spaced taps on the substation stepdown transformer. These taps subtly change the transformer ratio that is linking the high voltage transmission network with the low voltage distribution network, therefore providing control of the output voltage (see Fig.18). The other method of voltage control is using reactive power. Electric motors are by far the most common siliconchip.com.au Fig.18: a transformer tap changer can regulate the grid voltage by altering the ratio between the transmission and distribution networks. Original source: www.researchgate.net/figure/ fig1_224188399 load on the grid, making up more than 90% of total electricity demand in some regions. As they are strongly inductive, the grid operates with a lagging power factor; current lags voltage. In an inductive grid, the voltage is lower for a given power consumption than if that load was purely resistive. Capacitance can be used to compensate, which is commonly referred to as power factor correction (PFC). At its simplest, PFC involves switching banks of capacitors in and out of circuit – see Fig 19. Adding capacitance will cause the grid voltage to increase, and removing it will cause voltage to decrease. This Fig.19: a traditional capacitor bank used for power factor correction (PFC). Contactors K1 through K3 etc can be controlled to switch in a variable amount of capacitance, contributing reactive power to the grid and thus controlling voltage. Original source: https://electrical-engineering-portal.com/buildingcapacitor-bank-reactive-power-compensation-panel simple method is very commonly used by grid operators for voltage control, but there are some newer technologies that offer superior performance. The Static VAR Compensator (SVC) works in a similar way to the capacitor banks mentioned above, but rather than using mechanical relays, a power semiconductor such as a thyristor is employed, as shown in Fig.20. The thyristor can switch the capacitors in and out of circuit faster than mechanical relays and won’t wear out. This technology is widely used for reactive power control at gridscale generators and substations. The Static Synchronous Compensator (STATCOM) offers further performance improvements. Rather than a thyristor, the STATCOM arranges IGBTs in a H-bridge topology, with capacitance across a DC bus, as per Fig.21. The H-bridge can synthesise an AC waveform with a fully controllable phase shift, providing very tight and fast control of reactive power. STATCOMs are popular at substations where advanced voltage control is required, such as rural areas where a feeder may have to cover a long distance, or where SWER (Single Wire Earth Return) lines are in use. The STATCOM shares many Fig.20 (above): a simplified schematic of a Static Var Compensator. Similar to the PFC unit from Fig.19, banks of capacitors are switched into circuit as needed. Original source: www. researchgate.net/figure/fig1_308944567 Fig.21 (right): a simplified schematic of a Static Synchronous Compensator or STATCOM. The H-bridge produces an arbitrary waveform, which in most cases is generated with the current leading the voltage (ie, capacitance). Note the similarity to the inverter in Fig.13. Original source: https://doi.org/10.1007/s42452-020-03315-8 siliconchip.com.au Australia's electronics magazine April 2025  37 Fig.22: An example of spinning reserve in South Australia. During this period (October 18th to 21st), over 100% of grid demand is being met by rooftop solar. Most other generators are not needed and have switched off, except for a small amount of wind and utility solar, and notably some gas. It is uneconomic to run a gas generator for energy during this time; its benefit is providing grid stability through the angular momentum of its turbine and alternator. Source: https://explore.openelectricity.org.au similarities with the inverters discussed in the earlier sections; the main difference is that the DC bus only has capacitance connected in the STATCOM, rather than solar panels or a battery. Inverters are therefore a great way to control reactive power, and widely used at the utility scale for voltage control. Fig.23 shows a real-world example of a solar farm that operates with a power factor of 0.85. As it increases its output power, the grid voltage decreases through the action of the reactive power it contributes. In this way, IBRs (inverter-based resources) 38 Silicon Chip will play an important role in regulating grid voltage in coming years. A segment with good potential is rooftop solar, which currently provides almost no reactive power from its 20GW of installed capacity. A simple settings change could enable up to 15GVAr of support for free, which is plenty to tightly control voltage across the whole eastern seaboard and also ease network constraints. Grid stability – frequency (inertia) Our existing grid relies heavily on the angular momentum of rotating Australia's electronics magazine machines for frequency stability. This ‘spinning reserve’ works by resisting brief frequency excursions that might destabilise a power system. In most Australian states, this inertia is provided by the large alternators of coalfired power stations, and to a lesser extent gas and hydro. As these machines are electromechanically coupled directly to the grid, they provide momentum that works to maintain a frequency of 50Hz. Any increase in frequency (grid oversupply) will effectively turn the alternator into a motor. It will begin to speed up as it consumes power from the grid, resisting further instability. A sudden reduction in frequency (undersupply) works similarly; the alternator dumps extra power into the grid as it decelerates. Alternators work well in this role as they can produce or consume many times their rated power for short periods, although their response is governed by the electromechanical properties of the system and is therefore uncontrolled. The AEMO carefully tracks ‘spinning reserve’ to make sure the power system has adequate strength to resist any sudden shocks to the system, such siliconchip.com.au as a large generator or load tripping off-line. This is particularly evident in South Australia – see Fig 22. In this example, 100% of grid demand is being met by rooftop solar. All other generators are not needed so have turned off, aside from a small amount of utility wind and solar and a minimal amount of gas. The gas generators will not be making money on their energy production during this time, but they will be receiving payment for providing grid stability. The Torrens Island steam gas generator commonly provides this service, given its central location in Adelaide. It operates for long periods at 40MW, a fraction of its full nameplate capacity. Trials are underway to investigate the feasibility of repurposed coal generators for spinning reserve. It is possible to refurbish an old coal unit as a ‘synchronous condenser’, although early indications are that it will be more expensive than other solutions. Synchronous condensers are effectively large spinning masses with grid-coupled alternators. In normal operation, they draw a small amount of power from the grid to maintain their speed. Should a frequency excursion occur, they absorb or inject power to the grid in the same way as other spinning reserve. Their configuration is essentially identical to the one shown for wind turbines in Fig.17 except that, instead of the motor/alternator being connected to a turbine, it is connected to a rotating mass. These machines are increasingly popular for strengthening weak networks and can also be used for voltage regulation through reactive power control. Inverters can also be used to create so-called ‘synthetic inertia’. The IBR can be configured to monitor grid frequency and rapidly absorb or inject power should a frequency excursion occur. ‘Grid-forming’ batteries are wellsuited to this task given their fast response time, precise output control and ability to work bidirectionally. Successful trials have also been completed using wind turbines for frequency regulation – see Figs.24 & 25. It is estimated that a ratio of 15% ‘grid forming’ to 85% ‘grid following’ inverters is optimal to replace spinning reserve. siliconchip.com.au Fig.23: an output plot of a real-world solar farm used for voltage control. During a period of oversupply, the generator ramps down its output power (red). The grid voltage (pink, purple, green) increases. Some time later, the solar farm ramps up to full power, lowering the grid voltage through the action of its reactive power control. Fig.24: the output (orange) of the Hornsdale wind farm following a setpoint (AGC, black) to regulate grid frequency (grey). Source: Hornsdale FCAS Trial, p24 Fig.25: the output of a traditional synchronous generator across the same period as Fig.24. It underperforms compared to the wind farm given its slower response. Source: Hornsdale FCAS Trial, p25 Australia's electronics magazine April 2025  39 Grid stability – redundancy The grid works on the N-1 principle. That is, there must always be sufficient standby capacity that a trip on any single generator or transmission line will not lead to a blackout. This sometimes dictates some strange grid operations, such as curtailing generators or running transmission lines from areas of low supply towards areas of high supply. As the generation mix changes, these constraints will also change. Wind and solar generators are more decentralised than our existing coal fleet, and typically smaller. This gives a lower concentration risk for any single generator failure but increases operational complexity. Advanced software called distributed energy resource management systems (DERMS) is beginning to be rolled out in many networks. It provides improved visibility and control over grid constraints. These modern control systems are central to the energy transition, managing distributed assets and retaining N-1 redundancy. Grid stability – negative demand The combination of rooftop solar and coal is leading to an interesting problem for network operators. In the middle of the day, it is common for the rooftop system to be supplying its local load and also exporting to the grid. This is leading to periods of low demand, and in future even negative demand (see Fig.26). During these periods, fast ­responding grid-scale assets turn off for economic reasons, but coal generators remain active due to operational constraints, and rooftop solar remains active as it is usually paid a fixed rate ‘feed-in tariff’. It is common at these times for the distribution transformers to be running in reverse, supplying power back onto the high voltage transmission network. This is problematic, as many distribution transformers need to derate their power capacity for reverse flows. This is not a limitation of the transformer itself, but rather the tap changers, which usually employ “asymmetrical switching” to reduce the amount of power it must withstand during the middle of a change. This is advantageous when power is in the normal direction, but for reverse flows, the asymmetrical switching exposes the tap changer to increased power, severely limiting the reverse power capability of the transformer. Many networks are currently investigating and implementing upgrades to better handle this condition. A common solution is simply to inhibit the tap changer when reverse power exceeds its rating. The network operator won’t be able to use the tap changer during this period for voltage control, but they can use reactive power as discussed earlier. Another solution is to incentivise more load into the grid during daylight hours. So-called ‘solar soaker’ tariffs are being trialled, which offer free usage of the network between 10am and 3pm, but a higher rate between 5pm and 8pm. Increased electric vehicle proliferation should also help negative demand, as car chargers have higher usage during daylight hours. Conclusion Modern power electronics are playing a central role in the energy transition. Active stability techniques like Demand Response, IBRs and grid-forming batteries/inverters will replace most of the spinning reserve over the coming decades. Periods of negative demand may lead to lowcost or even free EV charging during sunny days, to make use of ample solar power, and incentivise further investSC ment in battery storage. Fig.26: minimum demand projections for the eastern seaboard grid. South Australia will likely experience negative overall demand this spring, with other states to follow in the coming years. Minimum operational demand is sometimes called ‘base load’. Source: AEMO ESOO 2024, p41 40 Silicon Chip Australia's electronics magazine siliconchip.com.au