Tuesday, July 22, 2008
The Problems With On-Grid Wind Power
Electric grids for the technically illiterate: Think of electricity like water flowing in system of channels on which ships transport. If the water level drops in one area, water will naturally flow into it. Alternatively, if water surges in one place, it will go over the banks and make that part of the system unusable. The more variable the water flow and the more points at which water flows into the overall system, the higher the cost of installing locks, dams and breaks to control the flow of water and keep the system operational.
Here's what I found in research. First, E.on is Germany's largest utility company. Here is their 2005 Wind Report in pdf. I would recommend to everyone to read it. Sometimes they can use the wind power and sometimes they can't, and because their effective usage is so low, they have to keep building traditional power plants. In 2004 the average feed-in to the grid varied between about one third and zero percent of the load. Obviously that sort of performance places upper limits on usage. I quote:
As wind power capacity rises, the lower availability of the wind farms determines the reliability of the system as a whole to an ever increasing extent. Consequently the greater reliability of traditional power stations becomes increasingly eclipsed.The rest of the report comments on the grid problems and the need for specialized control of wind turbines, plus upgrade of the transmission lines and grid to deal with the pulsing of wind power. They have invested in programs to predict and control it, but they haven't produced much effect. Now they are looking to replace the older turbines with newer, taller ones and to move offshore for more reliable winds. At the end of the report they discuss the potential for grid instability, and cheerfully note that if they are not careful, they may blow up pieces of the Polish, Netherlands and Czech power supply. On page 22 they forecast some improvement of the instability situation through 2010 due to various technical steps, but then:
As a result, the relative contribution of wind power to the guaranteed capacity of our supply system up to the year 2020 will fall continuously to around 4% (FIGURE 7). In concrete terms, this means that in 2020, with a forecast wind power capacity of over 48,000MW (Source: dena grid study), 2,000MW of traditional power production can be replaced by these wind farms.
By the 2010 consistent adherence to and implementation of the grid connection rules is expected to temporarily relieve the situation and to bring a reduction in wind power feed-in failures in the event of grid problems.They seem to be suggesting that a lot of the older installations will have to be dismantled or disconnected. This data explains why as of 2008, work on new coal plants in Germany is proceeding at a rapid rate. See this short blog post for some nice links. Apparently they haven't found a way to mitigate the risk. LA Times article about Europe's and Germany's coal plans.
However, the grid stabilising effect of traditional power stations would be lost as they are shut down for a variety of reasons, such as age or an increase in other forms of generation. At the same time, in ten years time there will still be a large number of older wind farms in Germany feeding into the grid, which do not have the necessary grid supporting features. There is therefore a risk that even simple grid problems will lead to the sudden failure of over 3,000MW of wind power feed-in. In this case, the reserves maintained in the Integrated European Transmission System, in order to cope with problems, would no longer be adequate to safely tackle such failures.
At the present time, it is not known how to confront this risk. Investigations must be made to determine to what extent the situation can be improved by replacing turbines at older wind farms or by introducing additional technical equipment to support the transmission system in the event of faults, or whether additional feed-in restrictions measures will be needed for old plants.
The 2007 UK energy report (pdf, 60 some pages) shows a similar trend in the UK, with total waste and renewable usage amounting to 1.8% of all usage, and the "natural" percent of total energy consumption falling over time. See pages 3 and 4 here. More detailed tables are on pgs 11-13, and a detailed look at renewables begins on page 57.
Denmark hasn't had much luck with its wind either - it is getting very little replacement grid power out of all the capacity it has installed.
Here is a paper for dullards like me who didn't understand the implications of trying to hook highly variable wind power into a power grid. The bottom line is that effective usage is low and that actual replacement effect is even lower:
A power station takes days to start producing electricity from a cold start. Time is needed to boil the water, to superheat the steam, to warm all the components of the power station, and to spin the turbogenerators up to operating speed.So if you are installing a bunch of new coal power plants to handle load, you will really be running them all the time with very little savings of fossil fuels. You can control some of the grid surge by diverting the power production away from the grid when your wind kicks in, but that of course doesn't change fuel consumption very much.
Each power station is designed to provide an output of electricity. It can only provide very little more or very little less than this output (i.e., a power station has a "low turndown ratio").
The problem of matching electricity supply to varying demand is overcome by operating power stations in three modes called "base load," "generation," and "spinning standby."
Some power stations operate all the time providing electricity to the grid, and they are said to provide "base load."
Other power stations also operate all the time but do not provide electricity all the time. They burn (or fission) their fuel to boil water and superheat the resulting steam which is fed to the steam turbines that are thus kept hot and spinning all the time. Of course, they emit all the emissions from use of their fuel all the time. But some of this time they dump heat from their cooling towers instead of generating electricity, and they are then said to be operating "spinning standby."
One or more power stations can be instantly switched from spinning standby to provide electricity to match an increase to demand for electricity. It is said to be operating "generation" when it is providing electricity. Power stations are switched between spinning standby and generation as demand for electricity changes.
Thus the grid operator manages the system to match supply with demand for electricity by switching power stations between "generation" and "spinning standby."
One possible, but grossly inefficient, solution might be to take the wind capacity and divert it to storage somehow off-grid, which storage could then be used by a conventional plant. But the same laws of physics that prevent a perpetual motion machine from being constructed mean losses of efficiency as power is converted from one stage to another.
Jimmy J. left this link in comments to a prior post about alternate energy, and I recommend reading the 2002 USS Clueless post on alternate energy. There appears to be a very close correlation between heavy investments in wind turbines and extremely high utility bills, and after looking at the Danish and German results I can see why. In the end, perhaps offshore wind turbines used to pump more water into tidal pools to produce tidal turbine energy (a somewhat regular source if supplemented by additional pumping) might provide some reasonable on-grid supplement, but it doesn't seem likely to ever amount to that much, and the environmental impact would not be negligible.
By the way, if you are a carbon bug you should not ignore the CO2 from all that concrete poured for those wind turbine footings. Concrete accounts for a significant portion of anthropogenic CO2.
I spent some time going through the ERCOT (Electric Reliability Council of Texas) docs. This pdf shows that they are seeing the same potential problem. They have done extensive planning work, including commissioning data generation for simulations, etc. See for instance minutes of this meeting:
Regarding the February 8, 2007 event, participants noted that EECP is triggered when weather fronts cause an increase in Load; that schedule changes are causing deployment of RRS, which is a Schedule Control Error (SCE) problem; and that EECP was enacted and maintained for half and hour, even though the system was already in recovery, in case there was another event.They're already seeing the problem with adding wind power to the grid. There is a 2006 120 page pdf report on some of the issues available online. Here is the summary:
Regarding the February 24, 2007 event, wherein large winds persisted state-wide, participants discussed the problem of wind coming off due to high wind speeds; the necessity of using planned wind output instead of wind capacity in determining Replacement Reserve Service; and the need for a tutorial on how RRS and Adjusted Responsive Reserve (ARR) is calculated.
Regarding the March 8, 2007 event, participants discussed how to treat wind, as it is not a controllable generator. Mr. Bruce opined that SCE must have meaning on a wind plant, and that forecasting seems better closer to real-time; expressed concern about generic forecasts and bad forecast data; and noted a need for unit-specific forecasting, and a willingness from wind generators to own responsibility for forecasting.
Mr. Gonzales-Perez presented issues associated with wind forecasting, the questionnaire sent to wind generation operators, Load ramping versus wind over-speed tripping, Day Ahead schedules versus actual wind output and recent lessons learned. Participants discussed the unusual situation encountered on February 24, 2007; where pressure to improve wind forecasting would be most appropriately applied, whether with ERCOT Operations or QSEs; the impact of the AS study, when the study would be available, and if it would be properly scoped to capture variability in wind generation; and what potential changes will be needed in Protocols and Operating Guides to address increased wind generation and the resultant reliability issues. Dan Jones added that if QSE-level forecasts lead to binding requirements, that binding requirements would need to be considered for all Resources.
- There is significant potential for development of wind resources in Texas.My conclusion is that Texas utility bills are going higher.
- There are currently 2,508 Megawatts (MW) of wind generation in-service in ERCOT.
At least 4,850 MW of wind resources are likely to be in-service by the end of 2007,
and around 17,000 MW of wind generation has requested interconnection analysis.
Much of that current wind generation development is in West Texas.
- Studies indicate that the existing transmission network is fully utilized with respect to
wind transfers from West Texas to the remainder of ERCOT. Thus, new bulk
transmission lines are needed to support significant transfers of additional wind
generation from the West Texas area.
- From a transmission planning perspective, there are four general areas of wind
capacity expansion: the Gulf Coast; the McCamey area, central-western Texas, and
the Texas Panhandle. Transmission solutions for each of these areas are described
in this report which provide an incremental plan for each area and form the basis of
transmission solutions to support combinations of wind development between two or
- Some common projects will be needed to mitigate the impact of the new CREZrelated
generation on existing wind generation. Even with these projects, existing
wind generation facilities will be more susceptible to curtailment due their generally
higher shift factors on the remaining system constraints.
- This study does not attempt to capture all of the benefits and costs associated with
the designation of CREZs, but focuses primarily on the direct costs and benefits
related to the electric power system.
- The production cost savings per kW of new wind generation varies little between the
- The Coastal area has lower annual capacity factor sites than the other areas but the
wind output is somewhat more coincident with the ERCOT electrical load.
- The Panhandle area has more resources with high annual capacity factors.
- The Coastal area requires the least transmission investment per MW of installed new
- The transmission cost per MW is higher for the Panhandle area; the higher annual
capacity factor of the resources in this area does not offset this higher cost.
- The first level solution for the Central and McCamey areas use the same bulk
transmission addition, so the designation of CREZs and addition of resources in these
areas must be generally considered in conjunction.
- While transmission solutions were generally developed that provided 1,000 MW
incremental steps for each area, the second step for the McCamey level is larger, in
terms of both cost and MW of wind generation supported; although the cost per MW
of supported wind is similar to the other levels for McCamey and Central areas.
- ERCOT will be performing an analysis of the impact of significant additional wind
generation on the level of the different ancillary services that it procures to maintain
system reliability. In addition, further ERCOT analysis of several issues is needed
once a specific set of CREZs is designated by the PUCT and wind generation
developers have indicated specific locations. These additional analyses include
reactive support needs, dynamic stability analyses, optimization of the “on-ramps”
within the CREZs and analysis of the specific projects or operational procedures
needed to mitigate curtailments of existing wind generation.
When they say "with a forecast wind power capacity of over 48,000MW, 2,000MW of traditional power production can be replaced," I think we need to dig a little deeper. Their analysis is clearly assuming that the "traditional" plants need to be there even for the worst few hours of the year (in terms of wind) so that service continuity can be maintained. But what would probably really happen is that some of the large steam and combined-cycle plants would be replaced with peaking turbines, which are not very fuel efficient but which require much less capital (per unit output) than the standard plants. Electricity pricing plans would also reward users who allow certain loads to be cut off for brief periods of time.
Also, when Courtney says of steam plants in spinning-reserve mode: "Of course, they emit all the emissions from use of their fuel all the time. But some of this time they dump heat from their cooling towers instead of generating electricity, and they are then said to be operating "spinning standby""...I don't think this is entirely accurate. When a turbine is spinning with no electrical load, or minimal load, the turbine throttle valve closes most of the way to keep the machine from overspeeding, and the boiler controls reduce the fuel flow to the furnaces accordingly. The amount of fuel being burned is only that required to overcome frictional and thermal losses in the system, and it would be inaccurate to conclude that the full-load fuel consumption of the plants is still being used and the heat dumped into the cooling towers.
A windmill is actually a better match for a compresser than a generater. Compressed air would be stored (in underground caverns?) until power is needed and then used to drive turbines.
Or figure on some really honking big batteries somewhere on the grid.
I think it casts a shadow over T. Boone's idea.
Instapundit had a link to a post on electric cars here:
One thing a lot of people don't take into consideration is that electric cars will require much more generation and a whole new industry specializing in recycling batteries on a massive scale.
IMHO, hybrids and high mpg diesels are the best bets for cutting fuel use in the next 10-15 years and as transition vehicles to electric, hydrogen, or ?? that will be feasible in 20-35 years.
Fred--here's a short piece from GE Research on compressed-air storage for wind power.
David - I think the issue of reserve operation is related to spinning reserve (approx ten minutes until output) and peak reserve, in which the power plant is putting power into the grid one moment and shunting it aside the next moment.
This is an important point. In the early stages, it seems that wind power has to be offset with mostly peak reserve. In the later stages with wider networks, you can use spinning reserve. Yet as you add more and more wind capacity to the grid, you have to maintain more and more spinning reserve, which is why E.on's efficiency was dropping and expected to drop. In an earlier report they had said the relationship of wind capacity/reserve capacity would be 10/8, but by 2005 they were up to 10/9.
So it looks like optimal wind capacity occurs relatively early on. Theories that widely disseminated wind input would help even out the spinning reserve capacity didn't work out, as one can see from the daily fluctuation graph.
There are planned projects to implement adiabatic CAES, but so far nothing has been built as some challenges remain: how to design a cost-effective compression train and turbine train that meet the required fast ramp rates and good part-load performance? How to build a large thermal energy store with very limited heat losses over daily cycles and that operates at high pressures and temperatures?
What I'm trying to get it is whether it wouldn't be more efficient in the first place to funnel the wind power directly to this sort of scheme. Then one wouldn't need the fast-ramp motor. It would surely be simpler to get a slow steady output of power all the time, and then ramp up for peak?
After my first shock subsided, I found myself thinking that on-grid wind power was going to be an overall loser, but that the large turbines and offshore sites might work well for something like this, while requiring much less spinning reserve.
But then I ask myself "Wonder Dummy, why aren't these highly technological societies already doing this?"
It is very important to differentiate between capacity and production. Grids will need new plants, especially as old ones are retired. But that does not mean that they have to be run all the time.
A number of eyes are privately watching the Austin, TX firm eeStor which has been coyly promising an ideal power storage medium for quite some time now.
Until earlier this year, there's been a near-consensus eeStor's prototype capacitor was vaporware, but with the news of Lockheed-Martin's interest and commitment to the firm, the naysayers have become a little more reticent.
Should these devices ultimately meet the developers' claims, they have the potential to change the nature and efficiency of power generation, transmission and distribution. They are said to be exceptionally compact and durable, and have an impressive storage density.
Since eeStor has missed a few delivery deliver dates in the past, I suppose we shall wait and see.
See also this article in Spectrum.
One thing that disturbs me about the report is that a large initial fraction of the wind generation system was built without regard for the stability needs of the grid. It's like having a team of people carrying a load and adding people with balance problems who will lose control of their balance when something shifts--just when you need them to help stabilize things. It's probably possible to refit that generation with the controls necessary to keep from tripping outright, but it will require integrating the blade speed controls with the instantaneous load management on the grid. My dusty, long unused engineering degree says that this is a hard problem. (Go read a good control theory text. The good ones retail for about $130 and up; get an older edition used online.)
Hydrogen thus produced can then be stored and/or burned in gas turbines or fuel cells to produce power at a steady rate. The UK is fairly fortunate in respect of storage in that old natural gas reservoirs could be used as hydrogen stores.