AC vc DC: who would win a modern Battle of the Currents?

Recently, I was asked: what if Edison had won the Battle of the Currents? As power grids today are increasingly dominated by inverter-based generation, could a direct current (dc) grid be more effective than the alternating current (ac) grids we rely on today? Would we still need spinning mass, inertia, and frequency control in a world powered by dc? And could dc grids handle the long-distance transmission of electricity with minimal losses?

What was the Battle of the Currents and what did Tesla’s victory mean?

The Battle or War of the Currents is the name given to the rivalry between Thomas Edison and Nikola Tesla back in the late 19^th century.

In January 1882 the Holborn Viaduct power station in London opened – named the Edison Electric Light Station. This was the world’s first coal-fired power station, generating electricity for public use. The plant burned coal to drive a steam turbine which drove a 27-tonne, 125 horsepower (93 kW) generator which produced direct current at 110 volts.

It initially lit 968 16-candle incandescent lamps to provide street lighting from Holborn Circus to St. Martin’s Le Grand, which was later expanded to 3,000 lamps. The power station also provided electricity for private residences. Having run at a significant loss the station closed in September 1886, and the lamps were converted back to gas.

In September 1882, Edison opened the Pearl Street Power Station in Manhattan. Also fired by coal, the plant initially had six dynamos, and a load of 400 lamps at 82 customers. By 1884 it was serving 508 customers with 10,164 lamps. This was also the world’s first co-generation plant, combining the provision of heating as well as power with the waste steam being piped to nearby buildings. Unfortunately, the station burned down in 1890.

In 1883 Nikola Tesla invented the first transformer, known as the “Tesla coil”. The following year, Tesla went on to invent the first electric alternator for the production of alternating current. Tesla saw that direct current was limited because it could not be transported over long distances since too much of the energy would be lost through the wires as heat. This meant power stations had to be located close to consumers, and while it worked to a certain extent in cities, it was less feasible elsewhere.

Alternating current on the other hand could be transported over longer distances since the voltage could be increased to a level that reduced the heating effect, using transformers: a “step-up” transformer increases the voltage of the electricity at the power station, and a “step-down” transformer reduces it again for distribution into homes and businesses.

In 1884 Tesla moved to the United States and went to work for Edison. Although Edison recognised that ac was more suitable to transporting electricity over long distances, he believed the high voltages necessary were dangerous, and he wanted to power electric motors as well as lights, and at the time there were no good ac motors available. So he had Tesla work on developing more efficient dc generators, however the two soon fell out and Telsa left to carry on working on his own ideas around alternating current. In 1887 he patented a complete electrical system, including a generator, transformers, a transmission system, a motor for use in appliances, and lights. The scheme caused a sensation and the patents were bought by George Westinghouse. Also in 1887, August Haselwander devised the first three-phase generator.

What followed has come to be known as the War or Battle of the Currents. On the one side were Tesla and Westinghouse and on the other, Edison. The stakes were high, with the opportunity to secure the rights to electrify American cities and earn potentially huge patent royalties.

The use of ac spread rapidly, leading Edison to begin a mis-information campaign against the technology. He claimed that the high voltages used to transport alternating current were hazardous, and that the design was inferior to, and infringed on the patents behind, his direct current system.

The spring of 1888 saw a media outcry over fatalities caused by pole-mounted high-voltage ac lines, attributed to the greed and callousness of the arc lighting companies that used them. Harold P. Brown, a New York electrical engineer, claimed that alternating current was more dangerous than direct current and tried to prove this by publicly killing animals with both currents, with technical assistance from Edison. They also contracted with Westinghouse’s main ac rival, the Thomson-Houston Electric Company, to make sure the first electric chair was powered by a Westinghouse ac generator.

The dirty tricks continued until the World Fair in Chicago in 1893, which was lit with an ac system and the competition between the two was effectively over. Modern electricity systems almost all run on alternating current

Would we make the same choice again today?

If we were designing an electrical grid today, would we still choose ac or could dc power grids solve modern challenges like renewable energy integration and long-distance transmission? Modern high-voltage DC (“HVDC”) technologies allow for efficient transmission of electricity over great distances with fewer losses, making dc grids a compelling option in a world increasingly powered by renewable sources like solar and wind.

The current reliance on inverters to convert dc to ac is not without its challenges, particularly when dealing with power quality and stability issues. Inverter-based systems (like those for solar or wind) often struggle with maintaining grid stability, particularly at high penetration levels. A dc grid would eliminate much of the need for inverters, improving overall system efficiency and reducing potential failure points associated with inverters’ response to grid disturbances.

A dc grid might offer more flexibility in how power systems are designed – with modern power electronics, dc power can be regulated and distributed more precisely, which could lead to more stable systems with fewer reactive power problems.

But while dc can and does work well for micro-grids, would it really be scalable for something larger such as a national power system?

Imagine we have terra formed a new island which is the size of GB, located at similar latitude to the UK so its wind and solar capacity factors would be the same, and it has the same access to geothermal and hydro resources. What would our electricity grid will look like, assuming it must be secure, affordable and sustainable?

In reality, the choice of ac or dc would come down to the generation mix chosen. At first glance it might seem that a grid based on wind, solar and batteries would work well with dc. But as soon as you include any form of conventional generation such as hydro or nuclear, ac would become more attractive.

So the first question is whether it is possible to create a secure and affordable grid based only on wind and solar. Based on currently available technologies, it seems unlikely that energy security could be achieved without some form of synchronous generation (such as gas or nuclear) currently available batteries run out in a matter of hours (about 4 hours at best) while low wind conditions can last for days and even weeks. If we were powering up our terra-formed island today we would need to choose at least some conventional generation and that would determine the choice of ac for the power grid. In this case we would need to rely on batteries and synchronous condensers to provide inertia, as we do today when reducing the amount of gas on the grid.

What would a large-scale dc grid look like?

Let’s say, for the sake of argument, that long-duration battery storage technologies have been invented and could fully back up wind and solar, and this leads us to design a dc grid. What would this grid look like? What properties would it have? How would faults manifest and be managed? How would system operators monitor and maintain grid stability?

In fact a dc grid would look very different to a conventional ac grid. There would be advantages around the integration of renewables and batteries, including loss reduction on ac to dc conversion (and vice versa), as well as lower losses on transmission. Dc technology also allows for higher power transmission with the same conductor cross-section and insulation requirements.

In a dc grid, there would be no frequency to monitor, so system operators would have to focus on voltage for the maintenance of grid stability.

Unlike in an ac grid, the behaviour of a pure dc grid, would be largely governed by Ohm’s Law, with some modifications due to the dynamic nature of the grid. Ohm’s Law is the fundamental principle that dictates how voltage, current, and resistance interact in a simple dc circuit: Voltage = Current × Resistance

If supply exceeds demand and there is no-where to store the excess generation, voltage would rise as the current continues to flow into the grid with no immediate consumption or storage destination. This is essentially because the power supplied exceeds the power consumed, and the voltage will naturally increase to push more current through the system.

If demand exceeds supply, voltage would decrease. If the batteries are discharging to meet demand, the voltage would drop as the stored energy is used. The grid would try to maintain the voltage by drawing from other resources, but if there is insufficient generation the voltage would drop due to the imbalance. The grid might try to draw more current from the available sources, but as voltage drops, the current may not flow as efficiently (because of the increase in internal resistance due to the voltage drop), which would further exacerbate the imbalance.

Over-voltage conditions could damage sensitive equipment or lead to the disconnection of generators or storage. Surge protection systems would be needed to clamp down on these rises in voltage and prevent equipment failures. In the case of voltage dips, batteries would need to respond rapidly in order to restore the voltage level.

Voltage oscillations could occur due to rapid switching or mismatches in power generation and consumption. For instance, a sudden shift from charging to discharging in the batteries or an abrupt change in load could create a momentary instability in voltage. These oscillations would need to be dampened by the energy storage systems or by voltage regulation devices.

In our hypothetical grid, batteries would play a central role in balancing supply and demand, functioning as both energy storage and frequency regulation assets. They would store excess energy generated by wind and solar during times of high generation and discharge it during periods of low generation. The grid would likely rely on a real-time demand response system, where batteries can discharge at a moment’s notice to meet demand, helping to stabilise the grid.

Grid operators would rely on sophisticated forecasting models that predict both supply (wind and solar generation) and demand (consumer consumption patterns), with batteries being scheduled to charge or discharge based on these forecasts, and dynamic optimisation software adjusting in real time to respond to any mismatches.

“Nevertheless, the realisation of dc distribution grids still faces challenges. While high-power converter systems can benefit from HVDC applications and cable conductors, busbars and no-load disconnectors are similar to ac, switching and protection devices are subject to ongoing research. Eg different circuit-breaker concepts including solid-state solutions are investigated. Furthermore, long-term operational experience with eg cable accessories and insulation is missing for dc distribution grids. Besides, active grid control in meshed LVDC topologies would require state detection by measurement and real-time management of controllable assets from the control room of distribution system operators. Technicians and operating personnel must be trained. From an economic point of view, dc equipment is more expensive for the first applications,”
– Julian Saat, Sebastian Stein, Maxim Müllender and Andreas Ulbig – Planning and design of urban low-voltage DC grids

In addition to managing voltage, the TSO would need to monitor the state of charge of batteries – If batteries are nearly full but generation continues at high levels, the grid would be oversupplied, and excess energy may need to be curtailed. Conversely, if batteries are nearing empty during periods of low generation, the grid would be undersupplied.

A large dc grid would likely be more centralised in terms of its design. Without the need for ac generators, the grid could rely on centralised battery farms and high-efficiency dc transmission lines. This could lead to fewer substations (since dc lines don’t need as many voltage transformations) but a larger reliance on long-distance HVDC systems for energy transport. (In practice, a dc grid would probably contain more substations, but they would be used for voltage monitoring and control rather than step changes in voltage.)

Faults and protection in a dc grid

Faults in a dc grid could be caused by malfunctioning power electronics or other grid equipment, or the loss of large units of generation. Without synchronous generation, there would be no inertia to absorb the shock of such faults. Batteries or capacitors would need to respond immediately, but this may not be instantaneous enough to avoid significant voltage drops or cascading faults.

The system would need rapid detection and isolation mechanisms. A fault would need to be detected quickly (likely using voltage and current sensors), and the faulty section disconnected from the grid to prevent further destabilising effects. Dc grids would need to rely on automated protection systems that could isolate a fault in milliseconds to ensure the rest of the system remains stable.

In the same way that conventional grid equipment contains frequency based protection measures, in a dc grid, equipment would rely on voltage triggers to ensure that equipment is not damaged during periods of grid instability. These would include:

• Overvoltage protection: if the voltage exceeds a certain threshold (eg more than 110% of nominal voltage), protection relays would trigger actions such as disconnecting loads or reducing generation to bring the voltage back to safe levels
• Undervoltage protection: if the voltage drops too low (eg less than 85% of nominal voltage), undervoltage protection mechanisms would engage to prevent damage to the grid and connected equipment. For instance, batteries might cut out if voltage drops below a safe threshold to protect their own integrity
• Overcurrent protection: if there’s an overcurrent condition (eg a fault that causes excessive current), protection systems would disconnect the faulty segment. Since dc grids tend to lack the inherent protection against sudden changes in load that ac grids have (due to inertia), overcurrent protection would be crucial in isolating faults
• Battery protection: batteries would be the backbone of the energy storage and balancing system, so would need specific protections. Overcharging or discharging batteries too quickly can damage them, so battery management systems would be required to automatically shut down or reduce the flow if certain thresholds are exceeded (eg temperature or state of charge limits)
• Fault isolation and grid stability: in the event of major faults or imbalances, dc grids would need rapid fault isolation mechanisms to prevent cascading failures. The system could use voltage and current-based relays or electronic protection circuits to disconnect faulty sections and prevent widespread instability.

In a dc system, surge currents can be more problematic compared to ac systems. In dc systems, if a load or generator suddenly changes or if there’s a fault, the current can rise immediately because there’s no natural frequency or oscillation to “smooth” the transition. This can cause a surge of current, especially if there’s insufficient resistance in the system or inadequate protection mechanisms. In an ac grid, the current gradually ramps up and down in a sine wave pattern due to the alternating nature of power, whereas in a dc grid, the current can go from zero to full in an instant, which could lead to equipment stress or damage without proper control systems in place.

While dc is generally easier for some types of equipment like LED lights and battery-powered devices, for inductive loads like motors or transformers, dc can create issues. Motors and transformers often rely on the oscillating magnetic fields generated by ac, and the lack of such oscillation with dc means they would need to be specially designed for continuous, steady current. Sudden surges in current could potentially damage such devices unless designed to handle them. Surge protection would need to be much more finely tuned in a dc-only grid, as the sudden spikes in current from starting motors or large loads could trigger protective relays or even damage wiring and appliances if not managed carefully.

There can also be arcing and switching issues. One of the key challenges with dc power systems is that switching devices (like circuit breakers or relays) can have trouble interrupting the circuits. In ac systems, the current naturally crosses zero twice during each cycle, which allows contact separation (in switches) to break the circuit more easily.

However, in dc systems, the current never crosses zero, so when a switch opens, arcing can occur, which can damage switches, create sparks, and lead to a higher risk of fire or equipment failure. This means that dc circuit breakers and switchgear would need to be specifically designed to handle this constant current and avoid damage, which is a significant engineering challenge. When switching off a dc circuit, if the current isn’t interrupted quickly enough, arcing can persist because there’s no natural zero-crossing point like in ac. This makes disconnecting dc circuits more dangerous and difficult to control without sophisticated arc extinguishing technology.

“LVDC systems do, however, present significant safety and protection challenges. Dc faults are more difficult to detect and clear. Their associated arcs are more aggressive than in ac, and they require longer time to be cleared. This makes the risk of fire in dc systems higher than in ac. In addition, the residual current devices (RCDs) which are commonly used in ac systems to protect against electric shocks and fire are not commercially available for dc systems. Dc systems will also require ac-dc power electronics converters for providing dc supply. Such interface devices have poor short circuit fault capability and can trip for remote faults. This can lead to substandard protection selectivity and unnecessary disconnection of larger part of the system. Such issues increase the need for fast and reliable dc protection solutions that reduce the risk and the cost of operating such systems,”
– University of Strathclyde

A dc grid would lack the inherent stability that comes with synchronous generation in ac grids, such as the inertia of large rotating machines. Without inertia, a sudden imbalance in generation and demand could cause instantaneous voltage fluctuations that would require rapid corrective action from batteries, inverters, or dc-dc converters. Grid blackouts could occur more suddenly, as there would be no gradual fallback like the frequency control seen in ac systems. Managing this risk would require super-fast, automated control systems, with very high system responsiveness and more advanced power electronics to prevent cascading failures.

What are the implications of managing voltage instead of frequency?

In an ac grid, frequency is the key measure of system balance. If the frequency falls outside a specific range it signals an imbalance between generation and demand. System operators act quickly to stabilise frequency by adjusting generation or consumption, and the system naturally provides some inertia (due to synchronous generators like gas, nuclear, or coal) that resists rapid changes in frequency.

In a dc grid, since there’s no frequency, voltage becomes the key variable. Voltage must remain within operational bounds (eg ±5% of nominal voltage). The system would monitor voltage at various points across the grid to ensure it stays stable.

Voltage can change very quickly in a dc grid. Unlike frequency in an ac grid, which takes seconds to minutes to deviate significantly, voltage in a dc grid would change much more rapidly in the event of a disturbance. For example, if a large generator or power source suddenly tripped offline, the voltage drop could be almost instantaneous because there is no inertial resistance to the change. The grid would have to react to this change by drawing power from batteries or other resources, but this would still be a fast-moving response. Similarly, if there’s a sudden load increase (eg a surge in demand), the voltage could dip rapidly as the system tries to supply more current than is available.

While batteries can discharge rapidly to help stabilise voltage, the response time of the system might still be on the order of milliseconds to seconds which is much faster than frequency adjustments in an ac system, which tend to take longer to manifest because of the system’s inertia. With voltage deviations happening much faster than frequency shifts in ac systems, the grid would need fast-acting systems to compensate and restore voltage stability before it leads to serious instability or damage.

TSOs would also need to manage voltage at many locations because unlike frequency, it won’t be broadly the same everywhere due to the resistance of transmission lines and the dynamic nature of the system. This is similar to the way there are voltage drops across an ac system, but in a dc grid, the voltage drop would be more pronounced without synchronous generation to provide inertia and stabilise the system. Voltage control would be more distributed, and real-time monitoring of voltage at key points (transmission lines, substations, generation points, and storage facilities) would be crucial for managing the grid’s stability.

There would also be a greater reliance on automatic response systems, since human intervention would generally be too slow to react to the rapid changes in voltage. For example, when a large generation source drops offline or a sudden load increase occurs, automatic systems would need to respond within milliseconds to seconds to avoid significant voltage disturbances.

Battery storage and voltage regulation devices would need to operate autonomously, adjusting power flow or voltage at the local level to balance supply and demand. The current balancing mechanism as it operates today (with instructions and manual responses) would be less relevant in a dc grid, where instantaneous action is needed.

This means the dc grid would need more sophisticated tools for grid management than is currently the case. A dc grid would require far more advanced and efficient power electronics than an ac grid, including dc-dc converters for managing voltage levels and controlling power flow, voltage regulation devices for maintaining stable voltage across the grid, including at points where renewable generation (solar, wind) connects and solid-state switches and advanced power controllers to handle rapid changes in power flow and ensure smooth operation across a wide range of conditions.

Impact of rapid solar ramps on a dc grid

Solar ramping – the phenomenon of rapid changes in solar generation due to cloud cover, the time of day (dawn and dusk), or other factors – poses significant challenges for any electrical grid, especially when the system is heavily reliant on solar power. In a large dc grid, the implications of solar ramping would be more pronounced, particularly when there are fewer substations or other infrastructure typically used to manage power fluctuations.

Rapid shifts between net load and net generation could occur in large sections of a dc grid with high reliance on solar power, due to the weather or the time of day. For example a grid region could move from net load before dawn, to net generation during daylight hours, and then back to net load at dusk. These swings can be rapid and large. In a dc grid, these shifts would lead to instantaneous voltage fluctuations.

If a region moves rapidly from net load to net generation a dc grid would have to instantly adjust the power flow. If the system is not well-regulated, it could lead to voltage spikes. If it moves from net generation to net load, there would be an immediate reduction in power generation, requiring energy to be drawn from storage systems or backup generation. If this transition happens too quickly, it could result in voltage dips or brownouts if the storage or backup systems are insufficient.

There is an elevated risk of localised instability, in the presence of solar ramping, when the penetration of solar power is high. This would be a particular risk in areas of the grid that are more remote or farther from power storage. If a solar plant in a remote part of the grid suddenly shifts from net generation to net load, and there isn’t enough local storage to compensate, voltage could fluctuate wildly.

In a large dc grid powered heavily by solar energy solar ramping would have more immediate and potentially severe impacts on voltage stability, especially if the grid has fewer substations. While battery storage and dc-dc converters could help manage these fluctuations, the grid would still face greater risks of instantaneous voltage shifts compared to an ac grid, primarily due to the lack of inertia and slower response times in traditional grid infrastructure.

The ability of the grid to swing from net load to net generation could create voltage spikes or dips if not carefully managed. To counter this, the dc grid would need extremely advanced power electronics, automated regulation systems, and fast-reacting storage to maintain stability. The absence of synchronous generation and fewer voltage regulation points would make the grid more susceptible to instability, and localised instability could be a serious risk without sophisticated control mechanisms.

Technological barriers to a large-scale dc grid

Could our hypothetical grid be put into operation today? The answer to this is probably not, since many of the enabling technologies are too immature.

First of all, battery storage, is often seen as a key enabler of a dc grid, however, even today’s most advanced batteries are not sufficient to provide economic and long-term backup for the grid, especially in a country with high renewable energy penetration. Battery cost, efficiency, and storage capacity remain major constraints – full battery backup for intermittent renewables is currently cost-prohibitive and not yet scalable to the level required for a national grid.

While HVDC technology is widely used for long-distance transmission (particularly for subsea cables or interconnectors), it’s still not as efficient or flexible as ac systems for managing large, interconnected grids. Dc-dc converters, which are essential for voltage regulation in a dc grid, remain limited in their scalability and efficiency. A dc grid would require highly advanced converters to handle power flow, regulate voltage, and provide fault protection, beyond what is currently available.

Voltage control in dc systems is far more difficult to manage than in ac systems, which benefit from natural frequency control and synchronous inertia provided by traditional generators. Voltage fluctuations in dc grids could be more abrupt, and without inertia, managing the rapid response to voltage changes would be extremely complex, with a need for real time voltage monitoring at a large number of grid locations with fast-reacting mechanisms to avoid catastrophic failure. However, existing voltage regulation systems would not be sufficient for large-scale implementation.

Automation systems for managing voltage, frequency, and fault isolation would need to be much faster and more sophisticated than anything currently in place for ac grids. Dc grids would likely face instabilities from small disruptions, and because there is no inherent damping (like the inertia of rotating generators in ac systems), the risk of the grid going from stable to failure would increase, especially during sudden faults. Current systems are not built for such rapid, dynamic responses to disturbances, making a dc grid much riskier without significant advances in automation.

Tesla would still win a modern Battle of the Currents

Although we may joke that Edison should have won the Battle of the Currents, in real life it would be far from simple to build stable, large-scale dc power grids, and we might decide to retain ac despite the use of inverter-based resources. The idea of dc grids may seem appealing from a theoretical standpoint, building and maintaining large-scale stable dc grids presents significant technical challenges, even when starting from scratch.

Large-scale dc grids would have challenges with voltage stability, in the absence of inertia. Managing solar ramping, sudden generation losses, and load shifts would be more difficult without sophisticated voltage regulation and fast-reacting storage systems. There would be a need for advanced dc-dc converters, inverters, and automatic voltage regulation systems. These systems, while advancing, are not yet at a level where they can efficiently and safely manage large-scale dc grids. Since faults would cause instantaneous voltage deviations, dc grids would likely face a higher risk of instability, with a potential for small faults to rapidly cascade into larger failures because there is no natural buffer (like inertia) to absorb shocks.

While ac grids may have originated in a desire to transport electricity over long distances with minimal losses, one of the fundamental reasons ac grids are still so widely used despite advances in HVDC is because they provide inertia from synchronous generators which helps the grid to remain stable when there are sudden shifts in supply and demand. Despite the challenges inverter-based generation poses (especially around frequency and voltage regulation), ac grids are still better suited to managing these larger networks reliably and stably.

So despite the challenges of integrating wind and solar with conventional ac grids, there really isn’t a viable alternative, even if we were starting from scratch with no legacy infrastructure to worry about. This is interesting, and somewhat counterintuitive – at first glance it would seem more sensible to create a dc grid to match the dc output of solar, wind and batteries. But managing voltage stability in a dc grid would be far harder than managing frequency in an ac grid, and the technology to do so, does not currently exist. It’s not perfect, but Tesla’s alternating current solution is still the best approach for a modern power grid.