Synchronous grids and inertia
The Australian NEM spans five states and 4,500 km and has around 260 registered generators, all of which, when online, are synchronous machines spinning in near-exact synchronization at close to 50 Hz across the network (the island state of Tasmania is connected via the non-synchronous Basslink DC link). A synchronous grid implies that the rotors of all generators are spinning in exact synchronization throughout the entire network. In Australia, the grid frequency of 50 Hz corresponds to a rotor speed of 3,000 RPM for 2-pole generators, 1500 RPM for 4-pole, 500 RPM for large 12-pole hydro generators, etc. Nearly all global electricity is produced by synchronous generators driven by rotary turbines; in 2010, 96 % of global electricity was generated by thermal or hydro plant, with nearly all being either steam, gas, or hydro turbines. The current issue around synchronous generation is that wind and solar use non-synchronous inverters.
The synchronization of the network can be considered by imagining a road train of 100 vehicles of various sizes, which are tethered together in a long line, driven along an undulating highway at constant speed. In much the same way, generators become ‘locked’ by the interaction between the generator magnetization and grid cycle when they are connected online and synchronized. But maintaining the correct speed can be challenging and requires monitoring and feedback loops to ensure stability of the entire 4,500 km-long machine.
In the road-train analogy, every vehicle should be applying power, but if each car used its own cruise control to try to meet the target speed, some vehicles would be working against each other. Vehicles that set the cruise speed slightly higher would be doing all the work, while vehicles with the speed set slightly lower would be applying no power. This simple but challenging problem used to be a daunting engineering problem, but was resolved from around the 1930s. Since all generators are locked to the system frequency, the frequency can be used as a reference for whether the system is over, or under, powered. Some generators use a governor with a ‘droop’ output curve to modulate the precise power output. For example when the target frequency is 50 Hz, this might correspond to a generator power of 90 %, but if the frequency speeds up to 51 Hz, the power may need to throttle back to 85 %. Conversely, the power may ramp up to 95 % when the frequency lowers to 49 Hz. This process occurs in real time and is termed ‘Primary Frequency Response’. Secondary Frequency Response can be provided by spinning reserves and other generators over several minutes. The Australian NEM operates a frequency control aniclliary service market, which includes fast (6 second) raise and lower, slow (60 second) raise and lower, and delayed (5 minute) raise and lower. Finally, the 5 minute dispatch cycle brings supply back into balance. AEMO dispatches scheduled generation which, in combination with forecast non-scheduled generation (generally generation under 30 megawatts (MW)) and semi-scheduled generation (intermittent generation such as solar and wind), will match forecast demand from the transmission grid.
Under normal operating conditions, the process operates smoothly, but is subject to so-called contingencies, which include forced generator outages, loss of transmission, or other events. These events introduce a large disturbance into the system. Control system theory during ‘step changes’ or impulses becomes critical since multiple feedbacks can introduce instability or harmonics. The worst case scenario in all feedback systems is uncontrolled oscillations. In electrical systems, severe disturbances can cause rotors to ‘skip’ a pole, and temporarily go out of phase causing large changes in current and sometimes a trip. In some electronic equipment, such as audio amplifiers, uncontrolled oscillations due to too much feedback, can destroy power stages and speakers. Buildings or bridges that have not been designed properly can vibrate at the structure’s resonant frequency, sometimes destroying structures. The primary means of controlling all feedback systems is through the use of damping components. An example in motor vehicles is shock absorbers, which dampen oscillations between the springs and vehicle mass. It is often obvious driving behind a vehicle that has worn shockers. In electronics, parallel capacitors or series inductances, introduce long time constant elements into a system, preventing uncontrolled electronic oscillations. In electricity systems, the primary damping mechanism is the mechanical inertia of the rotating rotors of synchronous generators. In the event of a sudden disturbance, the large rotating masses automatically and instantly transfer some of their momentum into the electrical system, thereby slowing the rate-of-change of frequency long enough to allow governors and ancillary services to respond. Steam generator sets are the heaviest and provide most of the system inertia.
The issue with wind and solar PV is that they are physically decoupled from the system frequency. Rather than being connected through generator windings, they are asynchronously connected via inverters. Inverters monitor the grid and track the voltage sinusoidal wave electronically. Inverters suppy power, and some can provide additional services including reactive power, but do not naturally provide inertia. It is possible, however, to produce synthetic inertia by emulating physical inertia. For example, a wind turbine could instantanously transfer additional momentum from the rotating blades, thereby emulating real synchronous behaviour. The main issue is that the blades will slow down and reduce its available power until the turbine recovers speed. Solar PV can’t provide synthetic inertia (unless it was curtailed before a system event) but batteries can provide millisecond response.
What does all this mean? In a system with a high penetration of non-synchronous generation, further inertia will need to be provided. This may include generators that are not providing power but simply spinning. Synchronous condensors are essentially generator sets without power delivery to the rotor – the rotor spins under power from the grid and provides an inertia service. The lack of inertia is not a technical challenge but a question of cost – how much is required and who pays for it.
The synchronization of the network can be considered by imagining a road train of 100 vehicles of various sizes, which are tethered together in a long line, driven along an undulating highway at constant speed. In much the same way, generators become ‘locked’ by the interaction between the generator magnetization and grid cycle when they are connected online and synchronized. But maintaining the correct speed can be challenging and requires monitoring and feedback loops to ensure stability of the entire 4,500 km-long machine.
In the road-train analogy, every vehicle should be applying power, but if each car used its own cruise control to try to meet the target speed, some vehicles would be working against each other. Vehicles that set the cruise speed slightly higher would be doing all the work, while vehicles with the speed set slightly lower would be applying no power. This simple but challenging problem used to be a daunting engineering problem, but was resolved from around the 1930s. Since all generators are locked to the system frequency, the frequency can be used as a reference for whether the system is over, or under, powered. Some generators use a governor with a ‘droop’ output curve to modulate the precise power output. For example when the target frequency is 50 Hz, this might correspond to a generator power of 90 %, but if the frequency speeds up to 51 Hz, the power may need to throttle back to 85 %. Conversely, the power may ramp up to 95 % when the frequency lowers to 49 Hz. This process occurs in real time and is termed ‘Primary Frequency Response’. Secondary Frequency Response can be provided by spinning reserves and other generators over several minutes. The Australian NEM operates a frequency control aniclliary service market, which includes fast (6 second) raise and lower, slow (60 second) raise and lower, and delayed (5 minute) raise and lower. Finally, the 5 minute dispatch cycle brings supply back into balance. AEMO dispatches scheduled generation which, in combination with forecast non-scheduled generation (generally generation under 30 megawatts (MW)) and semi-scheduled generation (intermittent generation such as solar and wind), will match forecast demand from the transmission grid.
Under normal operating conditions, the process operates smoothly, but is subject to so-called contingencies, which include forced generator outages, loss of transmission, or other events. These events introduce a large disturbance into the system. Control system theory during ‘step changes’ or impulses becomes critical since multiple feedbacks can introduce instability or harmonics. The worst case scenario in all feedback systems is uncontrolled oscillations. In electrical systems, severe disturbances can cause rotors to ‘skip’ a pole, and temporarily go out of phase causing large changes in current and sometimes a trip. In some electronic equipment, such as audio amplifiers, uncontrolled oscillations due to too much feedback, can destroy power stages and speakers. Buildings or bridges that have not been designed properly can vibrate at the structure’s resonant frequency, sometimes destroying structures. The primary means of controlling all feedback systems is through the use of damping components. An example in motor vehicles is shock absorbers, which dampen oscillations between the springs and vehicle mass. It is often obvious driving behind a vehicle that has worn shockers. In electronics, parallel capacitors or series inductances, introduce long time constant elements into a system, preventing uncontrolled electronic oscillations. In electricity systems, the primary damping mechanism is the mechanical inertia of the rotating rotors of synchronous generators. In the event of a sudden disturbance, the large rotating masses automatically and instantly transfer some of their momentum into the electrical system, thereby slowing the rate-of-change of frequency long enough to allow governors and ancillary services to respond. Steam generator sets are the heaviest and provide most of the system inertia.
The issue with wind and solar PV is that they are physically decoupled from the system frequency. Rather than being connected through generator windings, they are asynchronously connected via inverters. Inverters monitor the grid and track the voltage sinusoidal wave electronically. Inverters suppy power, and some can provide additional services including reactive power, but do not naturally provide inertia. It is possible, however, to produce synthetic inertia by emulating physical inertia. For example, a wind turbine could instantanously transfer additional momentum from the rotating blades, thereby emulating real synchronous behaviour. The main issue is that the blades will slow down and reduce its available power until the turbine recovers speed. Solar PV can’t provide synthetic inertia (unless it was curtailed before a system event) but batteries can provide millisecond response.
What does all this mean? In a system with a high penetration of non-synchronous generation, further inertia will need to be provided. This may include generators that are not providing power but simply spinning. Synchronous condensors are essentially generator sets without power delivery to the rotor – the rotor spins under power from the grid and provides an inertia service. The lack of inertia is not a technical challenge but a question of cost – how much is required and who pays for it.
[bctt tweet="Synchronous grids and inertia"]
4 Comments
Trackbacks/Pingbacks
- On Electricity Grids, Output Must Match Consumption At All Times | Blogging about energy and transitioning to clean energy - […] Heard describes “Ancillary Services“, the criticality of synchronous generation on electricity grids for voltage and frequency control. Bellingham Technical…
- Renewables and reactive power – GeorgeJetson.org - […] have been essentially built into systems by virtue of using large synchronous generators. An earlier post explored the role…
The wind turbine’s “synthetic inertia” is really a form of droop control, albeit of very little benefit. This is because there will be some form of deadband to the settings. it would only cut in if say the frequency changed by say 0.5Hz. This is in contrast with synchronous plant where the inertia is there all the time. That latter circumstance means that when there is imbalance, the inertia slows the rate of change. This gives time for the droop mechanisms to cut in, by say opening the wicket gates on a hydro, or increasing the burners in a steamer.
When you do the maths on the inertia of a steam turbine set compared to just the generator alone, it can be seen a lot of the inertia comes from the LP turbines with their larger diameter. That means to replace the system inertia, a lot more condensers need to be built than just converting old T/G sets. Calculate the inertia of a wind turbine and it is orders of magnitude smaller.
Synthetic inertia can be triggered with fine levels of response and can be brought into play directly with signals from the system operator. The idea that physical inertia is necessary for system stability is based on old-school models and does not account for the situation in which we find ourselves in 2017.
Off-grid and micro-grid systems function perfectly well without physical inertia and so can national grids.
@ Andy
This post relates to synchronous grids, which may have hundreds or thousands of synchronous generators operating simultaneously, and high levels of reactive components in the network itself. In 2010, ~96 % of global electricity was generated by synchronous generators (i.e. non-inverter generators).
The issue of a single generator (e.g. diesel) or an inverter is quite different. A single diesel can operate with a simple mechanical governor with a sufficiently slow response to prevent oscillations. An off-grid inverter maintains feedback with an op-amp or equivalent circuit with easily tuned response behaviour.
Micro grids are an interesting development. I’ve explored the idea of DC networks with microgrid nodes. However, for the foreseeable future (i.e. several decades), most electricity will be transmitted via AC networks and real physical inertia will remain essential for grid stability.
Thank you for this.
It seems to me very clear that too many people, including a high proportion of engineers, are prepared to either ignore or subconsciously deny issues that should be relatively easy to prove by first principles reasoning.
My conclusion is that there is a debate about mechanical inertia beng available from invertors which is both disturbing and untrue, unless one redefines inertia to suit the objective of unquestioning “eyes closed” promotion of RE.
Watching MP’s sprout terms like “synchronous generation”, “system strength” leaves me in awe of their ability to parrot advice given to them by people who we pay to give them support on issues they are excused from needing to understand. Where I get really concerned is when the advisors colour the advice to suit preferred policy outcomes, regardless of the cost to energy users and taxpayers.
Inertia is necessary to clear faults, by delivering high current into a short circuit, in order to quickly trip circuit breakers to isolate a faulted piece of equipment or power line.
Failure to do this could endanger people and property.
I have met more than a few engineers who are incapable of original thought, and dependent on rules of thumb, Google and questionable advice from sales engineers employed by vendors.
This is a serious, if esoteric, issue. We must be prepared to speak the truth as we see it, and be prepared to debate and justify our point of view.