This field is changing rapidly and so some of these issues may be better resolved than appears from some of the extracts. But it is useful to understand that currently there are limits to the penetration of some kinds of renewable energy on the electricity grid and it is still an area of international research.
In essence the “old-fashioned” power system had lots of big rotating equipment generating power at the business end. This has a lot of inertia – by which I mean inertia in the physics sense, rather than in the sense of institutional resistance..
The rotation is at a speed that generates 50 Hz or 60 Hz depending on where in the world you live. Supply has to match demand on a second by second basis. As the load on the system increases, it slows down the rotation of all of the large generation equipment and this allows two things:
- automatic response from systems (that monitor the frequency) to increase power
- flags to the operator to bring other power supply systems online (standby systems, aka reserves)
Wind turbines also rotate but they they don’t act the same as “old-fashioned” power systems – their inertial energy, in most cases, is effectively decoupled from the grid. This isn’t a problem at small penetration levels but the problem increases as the wind power penetration increases. This is called System Non-Synchronous Penetration (SNSP) – although in different places there may be different terms and acronyms.
There is also the critical issue of fault ride-through, which means that if the line voltage drops/collapses – when it comes back the wind farm should continue to provide power. This is critical at high penetration levels, because, without fault ride-through in wind farms, a temporary line voltage drop could take out the entire wind power generation system.
Here is Göksu et al (2010):
Conventional power plants, which are composed of synchronous generators, are able to support the stability of the transmission system by providing inertia response, synchronizing power, oscillation damping, short-circuit capability and voltage backup during faults. These features allow the conventional power plants to comply with the grid codes, thus today’s TSOs have a quite stable and reliable grid operation worldwide.
Wind turbine generator technical characteristics, which are mainly fixed and variable speed induction generators, doubly fed induction generators and synchronous generators with back to back converters, are very different to those of the conventional generators. As the installation of WPPs, which consist of these wind turbine generators, has reached important levels that they have a major impact on the characteristics of the transmission system..
Coughlan, Smith, Mullane & O’Malley (2007):
Renewable energy generation systems are being connected in increasing numbers to power systems worldwide. Of the commercially available systems, wind-turbine generators (WTGs) using non-synchronous-based technology are proving most successful. Unlike the synchronous machine whose operating characteristics have been documented and understood for decades, the generation of bulk ac electricity using non-synchronous machine-based generators is a relatively new phenomenon.
The effects of large penetrations of non-synchronous machine-based generators on power system stability have not been thoroughly studied. This problem is most serious in smaller power systems such as the Republic of Ireland, which have very large proportions of installed wind capacity compared to conventional generation and limited interconnection capability. Such systems are likely to experience possible stability issues related to wind generation, earlier than larger systems having lower proportions of installed wind generation..
..The level of wind turbine modelling detail required for power system stability studies remains an area where there is as yet not widespread agreement. This issue is complicated by the large number of wind turbine designs, the requirement for models in different time-frames, and the application of the model. As the end users of wind turbine models have predominantly been power system operators and due to the general lack of power system analysis expertise on the part of the wind turbine manufacturers, the wind turbine model development process has also proved cumbersome. Models are developed on behalf of manufacturers by third parties and supplied to system operators for use.
As many of the turbine models are not yet mature, system operators have acted as model testers reporting model bugs, irregularities, and errors and often advising manufacturers on appropriate action. Remedial action is then often relayed to third parties who make the necessary software changes.
Zhao & Nair (2010):
Renewable energy generation systems are being increasingly connected to power system networks worldwide. Among all commercially available systems, wind turbine generators (WTGs) using non-synchronous-based technology are being used predominantly. Unlike the traditional synchronous machine whose operating characteristics have been understood for decades, electricity generation using induction machine-based wind generators is relatively recent. In order to allow for the continued penetration of wind generation into electricity networks in the absence of operational experience, dynamic models of WTG have become more important for carrying out stability studies..
.. However, it is generally observed during large-scale wind integration studies that the so-called ‘standard’ components of the wind turbine models are quite often not standardised among manufacturers. Further during simulations, more detailed individual models (i.e. manufacturer-specific models) are used for analysis. The non-disclosure of the model details makes it very difficult to diagnose problems using simulation results. Considerable effort is needed to reproduce the model in a case containing no confidential data..
..Unlike conventional synchronous generators, where injection tests can be employed to test the unit response during a grid disturbance, a wind farm does not provide this option. Utilities rely solely on the WTGs model to determine how they would react to system dynamics, and therefore, the accuracy and validity of the model is important. To date, a very few number of wind turbine generator field test results are published..
..The validation of user-written models with field measurements needs careful planning and preparation, which includes obtaining permission from authorities, the power system operator and the wind turbine manufacturer. Disturbances which the wind turbines and the power system network can be subjected to are often limited. For example, it is not always easy to obtain permission to execute a balanced three-phase short-circuit fault in the transmission network, even though the results of such experiments would be highly valuable for validating the dynamic wind turbine model.
[Emphasis added].
Hansen & Michalke (2007):
Today, the wind turbines on the market mix and match a variety of innovative concepts with proven technologies for both generators and power electronics. The main trend of modern wind turbines/wind farms is clearly the variable-speed operation and a grid connection through a power converter interface.
Two variable-speed wind turbine concepts have a substantial predominance on the market today. One of them is the variable-speed wind turbine concept with partial-scale power converter, known as the doubly fed induction generator (DFIG) concept. The other is the variable-speed wind turbine concept with full-scale power converter and synchronous generator. These two variable-speed wind turbine concepts compete against each other on the market, with their more or less weak and strong features.
Nowadays, the most widely used generator type for units above 1 MW is the doubly fed induction machine. Presently, the primordial advantage of the DFIG concept is that only a percentage of power generated in the generator has to pass through the power converter. This is typically only 20–30% compared with full power (100%) for a synchronous generator-based wind turbine concept, and thus it has a substantial cost advantage compared to the conversion of full power
It seems that many national grid codes have been revised, and also that many people are studying the subject. Zhao & Nair compared wind farm models with reality under a line fault and found quite a discrepancy. However, in that case reality was a lot better than the model predicted, which is obviously a good thing.
A key question is what level of wind power the network can support before “curtailment”. Garrigle, Deane & Leahy (2013) discussed some scenarios in Ireland given that the current system non-synchronous penetration (SNSP) is set by the grid operator at 60%, but might be lifted to 75%.
You might think that a 60% limit on windpower means wind can achieve a penetration of 60% – pretty good, right?
But no. Remember that wind power is an intermittent resource. If wind power was like a conventional “dispatchable” generation source you would keep increasing wind farms and the output would rise up to 60% and then there would be no more wind farms built (until such time as the wind farm electrical characteristics were improved, or other methods of improving grid stability had been introduced).
Taking an extreme counter-example just for the purposes of illustration – imagine that some of the time there is zero wind, and the rest of the time all the wind-farms are running at 100%. And let’s say that the average output is 40% of nameplate capacity – i.e., we have no wind 60% of the time and lots of wind 40% of the time. Let’s say the country needs 5GW continuously and the government target to come from wind power is 40%, or 2GW on average. If we have 5GW of “nameplate” windpower capacity that implies that we can produce our target of 2GW.
However, the grid requires curtailment of any “non-synchronous” source above 60%. So in fact, from 5GW nameplate we will be producing 5GW x 60% for 40% of the time and 0 for the remainder. The result is an output of only 1.2GW, not 2GW – i.e., 24% of the national output instead of 40% of the national output.
Under this extreme scenario, it is impossible to produce the required 40% of national output from windpower.
Of course, this scenario is not reality. But the challenge remains – when the grid requires curtailment the limitation has a greater effect than we might first think.
Garrigle et al studied the effect of wind power curtailment under a variety of scenarios (including a certain amount of offshore wind power, currently a lot more expensive than onshore but less correlated to onshore wind power):
The primary result from this work is an estimate of the required installed wind capacities for both NI [Northern Ireland] and ROI [Republic of Ireland] to meet their 2020 RES-E targets. It is evident that this varies greatly due to the large differences in wind curtailment that will occur based on the assumptions made.
The required capacity estimates range from 5911 MW to 6890 MW which results in extra cost of c. € 459 million between what is considered to be the lowest technically feasible wind curtailment scenario (high offshore wind at SNSP limit of 75%, including TCGs) to that of the highest (low offshore wind at SNSP limit of 60%, including TCGS)
In the context of the electricity system this is a considerable extra expense similar in magnitude to the cost of two of the proposed North-South interconnector between NI and ROI. This illustrates the importance of increasing the SNSP limit as high as technically and economically feasible.
There were also dependencies on the interconnection to the rest of Great Britain. The way to think about this is:
- can you export power to another country when you produce “too much”?
- if that other country is also producing significant power from the same source (windpower in this example) how correlated is their output to yours?
Grid interconnections aren’t cheap. And if Great Britain is producing peak windpower at the same time as NI/ROI is producing peak windpower then the interconnections are of no benefit for that particular case.
Articles in this Series
Renewable Energy I – Introduction
Renewables II – Solar and Free Lunches – Solar power
Renewables III – US Grid Operators’ Opinions – The grid operators’ concerns
Renewables IV – Wind, Forecast Horizon & Backups – Some more detail about wind power – what do we do when the wind goes on vacation
Renewables V – Grid Stability As Wind Power Penetration Increases
Renewables VI – Report says.. 100% Renewables by 2030 or 2050
Renewables VII – Feasibility and Reality – Geothermal example
Renewables VIII – Transmission Costs And Outsourcing Renewable Generation
Renewables IX – Onshore Wind Costs
Renewables X – Nationalism vs Inter-Nationalism
Renewables XI – Cost of Gas Plants vs Wind Farms
Renewables XII – Windpower as Baseload and SuperGrids
Renewables XIII – One of Wind’s Hidden Costs
Renewables XIV – Minimized Cost of 99.9% Renewable Study
Renewables XV – Offshore Wind Costs
Renewables XVI – JP Morgan advises
Renewables XVII – Demand Management 1
Renewables XVIII – Demand Management & Levelized Cost
Renewables XIX – Behind the Executive Summary and Reality vs Dreams
References
Wind Turbine Modelling for Power System Stability Analysis—A System Operator Perspective, Coughlan, Smith, Mullane & O’Malley, IEEE Transactions on Power Systems (2007)
Assessment of wind farm models from a transmission system operator perspective using field measurements, S. Zhao N.-K.C. Nair, IET Renewable Power Generation (2010)
Fault ride-through capability of DFIG wind turbines, Anca Hansen & Gabriele Michalke, Renewable Energy (2007)
How much wind energy will be curtailed on the 2020 Irish power system? EV Mc Garrigle, JP Deane & PG Leahy, Renewable Energy (2013)
Overview of Recent Grid Codes for Wind Power Integration, Altin, Göksu, Teodorescu, Rodriguez, Jensen & Helle, 12th International Conference on Optimization of Electrical and Electronic Equipment (2010)
Very interesting series of posts this.
I think the key to incorporating large quantities of non-dispatchable renewable energy onto grids is to move towards a system which has large quantities of dispatchable demand. This means that if you have a 60% limit on non-synchronous supply, rather than curtailing supply, you can increase your demand and soak up the maximum amount of renewable generation.
Electric vehicles are on their way and a typical domestic charger can deliver around 3kW. These chargers will be smart so delaying the charge cycle until an expected blustery spell overnight would be quite straightforward and the vehicle owner would wake up to a full battery the next morning as normal.
Most homes also have immersion heaters (also rated at ~3kW) which would be quite simple to smarten up and could soak up considerable quantities of wind power, displacing other fuels (mostly gas) that would have been used to heat the water in the process.
Jamie,
I haven’t crunched the numbers myself but adding electric vehicles looks to many like another problem rather than a blessing – as far as matching demand and supply goes.
Everyone gets home and plugs their cars into the grid as well as turning the lights on, the heating, the cooking, the TV, the video/cable box/appleTV/entertainment thingy, ramping up the wifi router..
The key is storing energy in realistic quantities – which turns out to be a difficult proposition. If you can wait until midnight to charge your future electric cars then that will be a minor blessing but not actually a major help because your solar energy peaked at midday. If you can sit at home until midday and then charge up your car, the network will be in better shape.
If – big if – you create enough storage then you have solved most of the problems of creating a renewable electricity supply network.
The difference between peak demand and minimum demand across the year can be a factor of 2-3. So just soak it up with storage.
For a country like the UK, instead of needing a peak supply of 60 GW, with margin for failed generation (aka “availability”), requiring 80 GW – you only have to provide 347 TWh per year (IEA 2014 figures).
This equals 40 GW on average so your generation capacity drops by a factor of nearly 2 and most of your worries (and costs) go away.
Unfortunately, sufficient affordable energy storage (apart from hydro which is tapped out in most developed countries) is like the yeti.
People will get home and plug in but by and large the charging won’t happen until the off peak. It will no doubt be possible to recharge immediately but you’ll have to pay for the privilege. With average mileage only about 20 miles a day each vehicle will only require 2 or 3 hours of changing time so there’s plenty of opportunity to spreading around.
Jamie,
Clearly you don’t live in the US. There are millions of people who wish they only drove twenty miles/day to commute to work. 49% of Americans have a one way commute distance of greater than ten miles. Eight percent travel 35 miles or more one way. The battery in an electric car also has to power heating and air conditioning. There are also a whole lot of dual income families that would need to charge two vehicles each night.
Electric vehicles will be a niche market for quite a few years.
You’re right, I live in the UK. EVs will be a niche market everywhere for quite a few years but they’re coming. In the US with your longer distances, plug in hybrids will be more appropriate but that’s ok – the vast majority of miles will still be done on electric which is most important.
SoD,
Another well written post on an important issue. Thanks. I see that you end up returning to the issue of energy storage….. no surprise there.
SoD,
Very interesting. I had never considered this aspect of renewables. It sounds like the problem would apply also to DC sources, such as solar cells or power transmitted from remote hydro plants via HVDC lines. Is that correct?
You wrote:
“Under this extreme scenario, it is impossible to produce the required 40% of national output from windpower.
Of course, this scenario is not reality.”
It seems to me like reality might be quite a bit worse. If the 60% limit applies to minimum demand, then, using the numbers from your figure in Part IV, the maximum wind capacity would seem to be 60% of roughly 30 GW. Even that would lead to possible curtailment on summer nights and even some low demand summer days. But 30 GW capacity would be 10 GW average power (actually somewhat lower due to the occasional curtailments); that is about 25% of the average demand of 40 GW. If we base the maximum on nighttime summer demand, it would drop to a maximum from wind of about 20% of average power.
Expanding intermittent power sources beyond a modest fraction of total demand will depend on inexpensive storage; i.e., on technology that does not yet exist.
The above contains an embarrassing failure of my mental math skills. 60% of 30 GW is 18 GW (not 30 GW) which is an average of about 6 GW (33% load factor) which is 15% of total average load. And by the time one reaches that level, the cost of wind power will be rising due to curtailment.
Mike,
It’s definitely true about solar.
I hadn’t thought about HVDC lines until you asked the question. That raises another question about what “support” foreign sources of power can give a national grid (or equally, regional support if DC is used for intra-country transmission lines).
I’ll comment a little more on HVDC when I’ve digested your articles below, because my initial thought (before seeing those articles) was that power electronics should enable a grid to “pseudo-synchronize”.
The “value” of heavy spinning equipment is an automatic stabilization via frequency, but if you detect a line voltage drop and convert more power off your HVDC line to grid a.c. you should be able to get a similar response. At what cost and how well I have no idea – and this isn’t thinking informed by any papers or textbooks, just my vague idea about how electrical systems and power electronics operate (buyer beware).
I have almost finished reading Grid Issues for Electricity Production
Based on Renewable Energy Sources in Spain, Portugal, Germany, and United Kingdom – Annex to Report of the Grid Connection Inquiry Stockholm 2008:
– I noticed that Spain had to require some existing wind farms to equip their windfarms with “fault ride through” capability. Portugal only needed it on new installations because less of an installed base. And Spain also had to widen its frequency limits:
The UKERC (publisher of Gross et al 2006 that we looked at previously) did a survey to find out consumer attitudes to discover the impact of widening voltage and frequency limits (I don’t have the article any more), to assess whether “letting things go a little” was more cost effective than keeping the grid operating within the old limits.
Basically, all of the grids with a growing penetration of wind and solar have recognized the problem. The optimum solution is probably not clear, but a reasonable solution can probably be obtained. However, it’s also true that networks are complicated – even without renewables – and blackouts already occur due to a combination of events.
From the articles cited below, it would seem that it certainly is quite possible for grids to function with high levels of non-synchronous penetration – if sufficient money is put into alternative ways of stabilizing the grid. Yet another hidden cost of renewables. (Someone here called them “unreliables”; I am staring to come around to that).
http://spectrum.ieee.org/energy/the-smarter-grid/zombie-coal-plants-reanimated-to-stabilize-the-grid
http://www.nawindpower.com/e107_plugins/content/content.php?content.14254
This 2012 (paywall) paper is about trying to reduce grid blackouts (by way of understanding and modeling this complicated problem), rather than specifically about renewables but does have a note on wind power:
Risk Assessment of Cascading Outages: Methodologies and Challenges
– Prepared by the Task Force on Understanding, Prediction, Mitigation, and Restoration of Cascading Failures of the IEEE PES Computing and Analytical Methods Subcommittee (CAMS)
M. Vaiman (Lead), K. Bell, Y. Chen, B. Chowdhury, I. Dobson, P. Hines, M. Papic, S. Miller, and P. Zhang
[Emphasis added].
The last paragraph may be a comment on HVDC connection linking different grids together.
I worked a summer job at a large electric substation many years ago. What they called capacitors there were, in fact, large synchronous motors, or perhaps motor/generators. With three phase AC, you not only need to keep the frequency constant, you have to keep the phase relationship constant.
[…] « Renewables V – Grid Stability As Wind Power Penetration Increases […]
In Part VI we have a look at a feasibility study to convert the entire electricity generation in Australia over to renewables. Crikey!
For people interested in more commentary on grid stability (the focus of this article not part VI) the report pages 96-100 are worth a read.
I will look for some information on Washington State where I am at on wind capacity factors, and in particular, how many kWh of wind generation does not happen due to grid stability concerns.
WS is interesting because almost all our electricity is generated from non GGE power plants (Hydro, Nuclear, Wind). We have two geographic features that make this possible. The Columbia river has both a huge flow and a huge head, so we generate many GW off the CR dams. But there are also two “wind dams” in our state where temperature gradients between pacific marine air and hot inland air on the east side of the cascades cause substantial winds to be generated across gaps in the cascade range at the Columbia river, and Snoqualmie pass.
I mention this here because it I often see that many of the wind turbines in the Ellensburg valley (the Yakima river valley) will be stopped on a windy day. the grid operators are balancing the flow of water across our dams with many factors to consider: they need to prevent floods, allow sufficient supply of water for agriculture, provide adequate supply of electricity, take salmon migration into consideration, and even maintain adequate levels of water for recreation.
When you see those wind turbines stopped on a windy day, you have to ask why? Can’t we sell that excess power elsewhere? The grid in the USA is complicated, a sort of patchwork.
Presumably with increasing grid interconnect we can increase the capacity factor for our wind and solar generation while still maintaining SNSP factors for a grid. At least up to a point.
[…] Renewables V – Grid Stability As Wind Power Penetration Increases […]
[…] Renewables V – Grid Stability As Wind Power Penetration Increases […]
[…] Note that there will be other complications – see V – Grid Stability As Wind Power Penetration Increases. […]
[…] Renewables V – Grid Stability As Wind Power Penetration Increases […]
[…] of wind power. (We have already covered some real problems of high wind power penetration in V – Grid Stability As Wind Power Penetration Increases – those problems haven’t gone away). This is related to the question of the […]
[…] Renewables V – Grid Stability As Wind Power Penetration Increases […]
[…] Renewables and Grid Stability […]