by Planning Engineer
It can be very misleading to compare the energy costs for wind and solar to the energy costs for more conventional generation technology and assume the difference is the cost of providing for “clean” energy.
The power grid requires so much more support than the injection of energy. Unfortunately wind and solar do not provide support “services” as well as many other generation resources do. Accounting and providing for these extra “services” should be part of any comparison of resource types and inform any directives or plans impacting the provision of electric power. To the degree that wind and solar resources make up a larger portion of the supply mix, significant costs will be incurred to maintain system functionality and reliability. This posting is focuses solely on how various resources impact just one of these “services”, the balancing of system loads and resources.
In a power grid load and generation must be carefully balanced instant by instant. While there are energy storage technologies, the grid demands alternating power that must be generated in real time. Thus pumped storage energy must physically turn a turbine and battery storage must be converted to alternating current when needed. Thus all technologies (including storage) transform some form of energy (fuel, potential energy, electrical storage) to AC power in real time. Their combined contributions much match the system load demand on a near instantaneous basis.
On a power system when a load is added, the system frequency slows down slightly. This signals some generators to ramp up their output. When loads decrease, the frequency increases and this signals generators to back down. The system sees bumps when a generator is removed from or added to the system, loads change or when equipment is outaged. During system disturbances, large amounts of load or generation may be lost and the system must respond quickly and effectively to maintain balance and avoid further problems. Frequency continually shifts to signal generation changes, hovering very closely to a frequency of 60 cycles per second in North America.
Since the grid is interconnected and electricity flows at nearly the speed of light, when the system sees frequency departures available units across the system help somewhat regardless of who is responsible for the need. The grid is subdivided into control areas. Control areas try to ensure that generation and loads within their control are balanced at zero (or match total imports or exports). When a frequency disturbance hits one area – other control areas will help briefly, but then over a longer time period as the frequency disturbance is resolved the systems will correct for the oversupply of energy to keep everyone whole.
The flatter and smoother the load shape and the less steep the departures, the easier it is manage the system. In addition to minor and major perturbations system load can change dramatically over the course of the day. In the summer most areas see very low load levels at night rising during the day with temperature but continuing on until late in the evening due to people returning home from work. In the winter many areas see peaks before sunrise and then again in the evening. Loads and resources have to be balanced so new generation is brought on and taken off not only to provide balance. Generation must match load but with enough surplus to ensure the system can operate reliably should potential disturbances occur. Differing resource types vary in their ability to follow the system loads and provide a surplus.
Generally hydro plants with the ability to store water have excellent characteristics for load following. They can ramp up or down quickly when system loads change. They are easily started and shut down and can be kept spinning in no load conditions ready to provide near instantaneous support to the system. As part of a pumped storage system they can even provide load if generation levels are too high. They work well with renewables as water can be held back when the sun is shining or the wind blowing and released quickly whenever those conditions change. Abundant levels of hydro generation make it easier to add wind and solar to a system.
Gas Combustion Turbines and Combined Cycle Units
Gas based resources are good at following load. They can be kept in standby mode when needed and they generally have good ramping capability. There are consequences from starting and stopping these units however, showing up in extra maintenance requirements and refurbishment costs. When these resources spin to back wind and solar facilitates there are system costs and plant emissions. The simple combustion turbines are more nimble, but the combined cycle units are more efficient.
Coal plants generally can operate anywhere between their full load and half load capability. They can ramp up and down in that range depending on system needs and economics. However they have limited ramp rates (how fast they can go up or down) and if shut down they have minimum down times before they can be restarted. More so than with gas plants, shutting down and restarting is more costly and cumbersome. Coal plants operating between their peak and minimum values can provide load following capability to the system.
Due to economic, safety and regulatory concerns nuclear plants are set to run pretty much full out except during maintenance outages. For older plants this was a necessity as they could not be ramped up or down without significant risks. Newer plants have the capability to ramp, but I don’t think anyone is seriously considering significant ramping of Nuclear plants at this time in the US. It may be necessary at times in some locations where nuclear makes up a large part of the power supply portfolio. Further this capability is being looked at for the purpose of integrating renewables.
Operating the System
The tools for following load consist mostly of the above resources. For increasing load levels, just before a plant is brought on line, other resources are taken near their maximums so they can be quickly backed down when the new unit is added. As loads decline, plants are run nearer their minimums and then rapidly increase output as plants are taken off. The steeper the ramping rates, the greater the challenge in varying generation back and forth to add or subtract units. In addition to meeting system loads, operators need to have a set amount of generation on quick standby (spinning reserve). These are units (or spare generation capability from an operating unit) that can be counted on (and are already online and already operating synchronously with the grid) to provide generation very quickly. Additionally other dependable resources need to be available that can be counted on within ten minutes (non-spinning reserves). Beyond that there are planning reserves which are extra generation needed for plant outages.
Operators must plan ahead for generation needs. For example, if coal plants are needed the next day, they can’t be taken off at night when the load drops. Their goal is to come up with a transitioning of resources that reliably serves the load in an economic manner. Because of unpredictability’s associated with loads and generation it is a challenge to bring in and remove resources from the system as demands ramp up or down.
For completeness it should be noted that to achieve balance operators sometimes have control of system loads as well. Utilities have some portion of their load tied to under frequency load shedding (UFLS)programs, to drop load when the frequency is sufficiently low such that generator ramping capability alone may be ineffective. Under frequency load shedding is rightfully a rare event that should occur only in very limited conditions when multiple things go simultaneously wrong. In addition to emergency load shedding some customers have time of use rates or special interruptible rates. Operators may be able to switch off air-conditioning, water heaters or industrial loads as well as employ other load modifying schemes such as voltage reduction. Non-emergency load reductions programs are not implemented to help follow loads, but rather to reduce peak demands to for limited generation and transmission capacity.
Wind and Solar
Wind and solar cannot be counted on to help with load following, in fact they work to make the load-generation balance more unpredictable. There can be a lot of fluctuations in MW output from wind and solar which other resources must balance when cloud cover rolls in or the winds change. This is true in the short term as well as on hour by hour basis. It’s hoped with enough resources the intermittent effects will tend to cancel out, but that often does not appear to work as optimistically as might be hoped.
It is often said the solar is good because it produces power during periods of peak demand. This is somewhat true, but not strictly true. For many winter morning peaks, solar is absent and begins to ramp up as the operators are beginning to ramp down the system. They typically do not give full output at evening peaks. For economic reasons (related to maximizing solar output, not grid operation) solar panels are oriented to catch the most sun midday. As the peak builds, solar tends to ramp down increasing challenges for operators. While the additional energy during the day has value, solar often ramps on and off opposite to the system needs.
Wind cannot be counted on. It can be zero during times of maximum demand. It often reaches maximum output level when system demands are minimal. Significant problems with wind may occur at night when winds are high. In some regions there is an oversupply of power at night. To prevent over-generation, penalties (costs) are imposed for adding power to the grid during minimum load hours. In some cases (for example in the high wind areas of Texas) utilities are penalized for generating at night (even though their plants need to be kept on line for tomorrows load) while wind energy (which could be removed with no operational consequences) adds to a generation surplus in order for their developers to collect a guaranteed rate for their generation.
A similar balancing problem can be attributed largely to solar as illustrated by the “California Duck Curve”. The curve below shows that the projected growth for residential solar power will have a limited impact upon the system peak but a huge impact on mid-day loads. This introduces the risk of over-generation during the afternoon and increased need for ramping as solar drops off. If solar is also part of the bulk generation supply, the stress on remaining generation as it works to meet the steep increase from afternoon to evening loads will be increased further.
High Renewable Future?
To integrate large amounts of wind and solar into the grid you will need some form of backup generation. Mixed solar, wind and hydro systems with adequate hydro capacity and storage well located relative to loads, could function well. Unfortunately, few areas have an abundance of hydro or the potential to develop such. Conceivably extensive use of pumped storage technology might enable such a future, but would involve high costs.
Backing up intermittent energy with nuclear power seems highly counterproductive. (Please provide comments if I am missing something here.) The incremental costs and burdens of up and running nuclear generation is small enough that displacing nuclear generation with intermittent generation seems nearly pointless.
Backing up intermittent resources with gas turbines and combined cycle units would work. The cost comparisons for such a system should not be based on the difference between average solar or wind energy cost and the average cost of gas generation. Rather the proper cost comparison is the average cost of solar and wind plus the backup costs of gas generation, compared to just gas generation. Numbers such as these are rarely shared and are probably would not be politically feasible if they were well understood. The key to this understanding is that a high penetration of renewables will only reduce the fixed costs of needed gas generation resources by a small amount. While the reduction is gas fuel use may be moderate, the additional fixed cost of the renewables will be very high.
Lastly renewables could be supported with batteries, other stored energy resources and technologies allowing advanced control of load demand. This may well be the grid of the future, but would have extremely high costs based on today’s projections. These costs should be well understood and shared before embarking upon such a future. Certainly we should be adding wind and solar whenever it can be justified and also for research benefits, but becoming too ambitious could have dire consequences for system reliability, cost and performance.
As with all “engineering” decisions there are tradeoffs. Resources that do not help with load balancing may be a good selection at times and have a place in the generation mix if they have other positive characteristics. A system mix that primarily employs conventional synchronous generation technology will generally have load following capability in excess of needs, accommodating some level of penetration for intermittent resources. When there is sufficient hydro, gas and coal resources higher levels of renewables can be backed up at moderate costs. However when renewables are increased dramatically or the resource mix is altered to remove significant amounts of conventional technology, the additional costs needed to support wind and solar generation can be extreme.
This discussion has just focused on load balancing or matching generated MWs to load MWs. From this consideration alone we see that it is inappropriate to compare solar and wind to more conventional power sources by looking at energy costs alone. In terms of the critical task of balancing the system, wind and solar load do not help, but rather instead impose significant burdens. These burdens cannot be ignored as increasing levels of such intermittent generation are added to the system. Certainly we should be adding wind and solar whenever it can be justified and also for research benefits, but becoming too ambitious could have dire consequences for system reliability, cost and performance.
JC note: This is a follow on to Planning Engineer’s previous post Myths and realities of renewable energy. As with all guest posts, keep your comments civil and relevant.