Challenges to making California’s grid 33% renewables

Meier, A. May 2014. Challenges to the integration of renewable resources at high system penetration. California Energy Commission.

Energy Research and Development Division FINAL PROJECT REPORT.  California Energy Commission California Institute for Energy and Environment MAY 2014 Alexandra von Meier California Institute for Energy and Environment University of California


Successfully integrating renewable resources into the electric grid at penetration levels to meet a 33 percent Renewables Portfolio Standard for California presents diverse technical and organizational challenges.

Renewable and distributed resources introduce space (spatial) and time (temporal) constraints on resource availability and are not always available where or when they are wanted.

Although every energy resource has limitations, the constraints associated with renewables may be more stringent and different from those constraints that baseload power systems were designed and built around.

These unique constraints must be addressed to mitigate problems and overcome difficulties while maximizing the benefits of renewable resources. New efforts are required to coordinate time and space within the electric grid at greater resolution or with a higher degree of refinement than in the past. This requires measuring and actively controlling diverse components of the power system on smaller time scales while working toward long‐term goals. These smaller time scales may be hourly or by the minute, but could also be in the milli‐ or even microsecond range.

To cope with intermittent renewables there needs to be

  • reserve generation capacity at least a day
  • dispatchable generation with high ramp rates in MW/s
  • generation with regulation capability
  • dispatchable electric storage
  • electric demand response (from customers)
  • direct load control down to a 5-second time scale without impacting end-use (!) their exclamation mark, not mine in

It also important to plan and design around the diverse details of local distribution circuits while considering systemic interactions throughout the Western interconnect. Simultaneously coordinating or balancing these resources in an electric under a variety of time and distances, without any specific technology to assist, is defined as a “smart grid.”

Temporal coordination specifically addresses the renewable resources time‐varying behavior and how this intermittency interacts with other components on the grid where not only quantities of power but rates of change and response times are crucially important.

Research needs for temporal coordination relate to:

  • resource intermittence,
  • forecasting and modeling on finer time scales;
  • electric storage and implementation on different time scales;
  • demand response and its implementation as a firm resource;
  • and dynamic behavior of the alternating current grid, including stability and low‐frequency oscillations, and the related behavior of switch‐controlled generation.

Different technologies, management strategies and incentive mechanisms are necessary to address coordination on different time scales.

Spatial coordination refers to how resources are interconnected and connected to loads through the transmission and distribution system. This means connecting remote resources and also addressing the location‐specific effects of a given resource being connected in a particular place. The latter is particularly relevant for distributed generation, which includes numerous smaller units interconnected at the distribution rather than the transmission level.

Research needs for spatial coordination relate to: technical, social and economic challenges for

  • long‐distance transmission expansion;
  • problematic aspects of high‐penetration distributed generation on distribution circuits, including clustering, capacity limitations, modeling of generation and load, voltage regulation, circuit protection, and prevention of unintentional islanding;
  • microgrids and potential strategic development of microgrid concepts, including intentional islanding and variable power quality and reliability.

A challenge to “smart grid” coordination is managing unprecedented amounts of data associated with an unprecedented number of decisions and control actions at various levels throughout the grid.

This report outlined substantial challenges on the way to meeting these goals.

More work is required to move from the status quo to a system with 33 percent of intermittent renewables. The complex nature of the grid and the refining temporal and spatial coordination represented a profound departure from the capabilities of the legacy or baseload system. Any “smart grid” development will require time for learning.

Researchers concluded that time was of the essence in answering the many foundational questions about how to design and evaluate new system capabilities, how to re‐write standards and procedures accordingly, how to create incentives to elicit the most constructive behavior from market participants and how to support operators in their efforts to keep the grid working reliably during these transitions. Addressing these questions early may help prevent costly mistakes and delays later on.

CHAPTER 1: Introduction to the Coordination Challenge

Successfully integrating renewable resources in the electric grid at high penetration levels – that is, meeting a 33 percent renewables portfolio standard for California – requires diverse technical and organizational challenges. Some of these challenges have been well‐recognized in the literature, while others are emerging from more recent observations. What these challenges have in common is that they can be characterized as a coordination challenge. Renewable and distributed resources introduce space or location (spatial) and time (temporal) constraints on resource availability. It is not always possible to have the resources available where and when they are required.

New efforts will be required to coordinate these resources in space and time within the electric grid.

A combination of economic and technical pressures has made grid operators pay more attention to the grid’s dynamic behaviors, some of which occur within a fraction of an alternating current cycle (one‐sixtieth of a second). The entire range of these relevant time increments in electric grid operation and planning spans fifteen orders of magnitude: from the micro‐second interval on which a solid‐state switching device operates, to the tens of years it may take to bring a new fleet of generation and transmission resources online or as a billion seconds (a season).CA grid 33 pct renewable time scaleIn the spatial dimension, it is also the case that power systems have expanded geographically and become strongly interdependent over long distances, while local effects such as power quality are simultaneously gaining importance. About six orders of magnitude covered ‐ from the very proximate impacts of harmonics (on the scale of an individual building) to the wide‐area stability and reliability effects that reach across the Western Interconnect, on the scale of a thousand miles.

Because of their unique properties, any effort to integrate renewable resources to a high penetration level will push outward time and distance scales on which the grid is operated. For example, it will force distant resource locations to be considered as well as unprecedented levels of distributed generation on customer rooftops.

The physical characteristics of these new generators will have important implications for system dynamic behavior.

In extending the time and distance scales for grid operations and planning, integrating renewable resources adds to and possibly compounds other, pre‐existing technical end economic pressures.

This suggests at least a partial definition for what has recently emerged as a” Holy Grail” or the “smart grid.” The “smart grid” is one that allows or facilitates managing electric power systems simultaneously on larger and smaller scales of distance and time.

Special emphasis is at the smaller end of each scale, where a “smart grid” allows managing energy and information at higher resolution than the legacy or baseload system.

The fact that solar and wind power are intermittent and non‐dispatchable is widely recognized.

More specifically, the problematic aspects of intermittence include the following:

High variability of wind power. Not only can wind speeds change rapidly, but because the mechanical power contained in the wind is proportional to wind speed cubed, a small change in wind speed causes a large change in power output from a wind rotor.

  1. High correlation of hourly average wind speed among prime California wind areas. With many wind farms on the grid, the variability of wind power is somewhat mitigated by randomness: especially the most rapid variations tend to be statistically smoothed out once the output from many wind areas is summed up. However, while brief gusts of wind do not tend to occur simultaneously everywhere, the overall daily and even hourly patterns for the best California wind sites tend to be quite similar, because they are driven by the same overall weather patterns across the state.
  2. Time lag between solar generation peak and late afternoon demand peak. The availability of solar power generally has an excellent coincidence with summer‐peaking demand. However, while the highest load days are reliably sunny, the peak air‐conditioning loads occur later in the afternoon due to the thermal inertia of buildings, typically lagging peak insolation by several hours.
  3. Rapid solar output variation due to passing clouds. Passing cloud events tend to be randomized over larger areas, but can cause very rapid output variations locally. This effect is therefore more important for large, contiguous photovoltaic arrays (that can be affected by a cloud all at once) than for the sum of many smaller, distributed PV arrays. Passing clouds are also less important for solar thermal generation than for PV because the ramp rate is mitigated by thermal inertia (and because concentrating solar plants tend to be built in relatively cloudless climates, since they can only use direct, not diffuse sunlight).
  4. Limited forecasting abilities. Rapid change of power output is especially problematic when it comes without warning. In principle, intermittence can be addressed by firming resources, including • reserve generation capacity • dispatchable generation with high ramp rates • generation with regulation capability • dispatchable electric storage • electric demand response that can be used in various combinations to offset the variability of renewable generation output. Vital characteristics of these firming resources include not only the capacity they can provide, but their response times and ramp rates.

Solar and wind power forecasting obviously hinges on the ability to predict temperature, sunshine and wind conditions. While weather services can offer reasonably good forecasts for larger areas within a resolution of hours to days, ranges of uncertainty increase significantly for very local forecasts. Ideally, advance warning could be provided at the earliest possible time before variations in solar and wind output occur, to provide actionable intelligence to system operators.


Real‐time forecasting tools for wind speed, temperature, total insolation (for PV) and direct normal insolation (for concentrating solar), down to the time scale of minutes

Tools for operators that translate weather forecast into renewable output forecast and action items to compensate for variations.

A related question is the extent to which the variability of renewable resources will cancel or compound at high penetration levels, locally and system‐wide. Specifically, we wish to know how rapidly aggregate output will vary for large and diverse collections of solar and wind resources.

Needed: • Analysis of short‐term variability for solar and wind resources, individually and aggregate, to estimate quantity and ramp rates of firming resources required.

Analysis of wide area deployment of balancing resources such as storage, shared among control areas, to compensate effectively for short‐term variability.

2.1.3 Background: Firming Resources Resources to “firm up” intermittent generation include

  • reserve generation capacity
  • dispatchable generation with high ramp rates

The various types of firming generation resources are distinguished by the time scale on which they can be called to operate and the rate at which they can ramp power output up or down.
The most responsive resources are hydroelectric generators and gas turbines.

The difficult question is how much of each might be needed.

Electric storage includes a range of standard and emerging technologies:

  • pumped hydro
  • stationary battery banks
  • thermal storage at solar plants
  • electric vehicles
  • compressed air (CAES)
  • supercapacitors
  • flywheels
  • superconducting magnetic (SMES)
  • hydrogen from electrolysis or thermal decomposition of H2O

An inexpensive, practical, controllable, scalable and rapidly deployable storage technology would substantially relieve systemic constraints related to renewables integration.

The spectrum of time scales for different storage applications is illustrated in Figure 5.

  • months: seasonal energy storage (hydro power)
  • 4‐8 hours: demand shifting
  • 2 hours: supplemental energy dispatch
  • 15‐30 minutes: up‐ and down‐regulation
  • seconds to minutes: solar & wind output smoothing
  • sub‐milliseconds: power quality adjustment; flexible AC transmission system (FACTS) devices that shift power within a single cycle

Given that storing electric energy is expensive compared to the intrinsic value of the energy, the pertinent questions at this time concern what incentives there are for electric storage, at what level or type of implementation, and for what time target.

Alternating ‐current (a.c.) power systems exhibit behavior distinct from direct‐current (d.c.) circuits. Their essential characteristics during steady‐state operation, such as average power transfer from one node to another, can usually be adequately predicted by referring to d.c. models. But as a.c. systems become larger and more complex, and as their utilization approaches the limits of their capacity, peculiar and transient behaviors unique to a.c. become more important.


3 Eto, Joe et al. 2008. Real Time Grid Reliability Management. California Energy Commission, PIER Transmission research Program. CEC‐500‐2008‐049.

The increased need to manage California’s electricity grid in real time is a result of the ongoing transition from a system operated by vertically integrated utilities serving native loads to one operated by an independent system operator supporting competitive energy markets. During this transition period, the traditional approach to reliability management—construction of new transmission lines—has not been pursued due to unresolved issues related to the financing and recovery of transmission project costs. In the absence of investments in new transmission infrastructure, the best strategy for managing reliability is to equip system operators with better real-time information about actual operating margins so that they can better understand and manage the risk of operating closer to the edge.

Traditional rotating generators support grid stability by resisting changes in rotational speed, both due to magnetic forces and their own mechanical rotational inertia. Through their inherent tendency to keep rotating at a constant speed, these generators give the entire AC system a tendency to return to a steady operating state in the face of disturbances. Legacy power systems were designed with this inertial behavior in mind.

Large fossil fuel and nuclear generators naturally promote 60-Hz grid stability because their rotational speed is constant due to magnetic forces and inertia. Despite disturbances they to revert to a steady operating state. But the inverters that renewable energy use to supply AC power depend on very rapid on-off switching within solid-state semiconductor materials. It’s possible that at some point when a larger percent of power comes from renewables, these inverters will destabilize the grid voltages, frequencies, and oscillations by not responding collectively well to temporary disturbances and that we’ll need to keep large rotating generators to maintain stability.

Unlike conventional rotating generators, inverters produce alternating current by very rapid on‐off switching within solid‐state semiconductor materials. Inverters are used whenever 60‐Hz AC power is supplied to the grid from

  • c. sources such as PV modules, fuel cells or batteries
  • variable speed generators, such as wind whose output is conditioned by successive a.c.‐c.‐a.c. conversion (this does not include all wind generators, but a significant fraction of newly installed machines). What we do not understand well are the dynamic effects on a.c. systems of switch‐controlled generation:
  • How will switch‐controlled generators collectively respond to temporary disturbances, and how can they act to stabilize system voltage and frequency?
  • What will be the effect of switch‐controlled generation on wide‐area, low‐frequency oscillations?
  • Can inverters “fake” inertia and what would it take to program them accordingly?
  • What is the minimum system‐wide contribution from large, rotating generators required for stability?


  • Modeling of high‐penetration renewable scenarios on a shorter time scale, including dynamic behavior of generation units that impacts voltage and frequency stability
  • Generator models for solar and wind machines
  • Inverter performance analysis, standardization and specification of interconnection requirements that includes dynamic behavior
  • Synchro‐phasor measurements at an increased number of locations, including distribution circuits, to diagnose problems and inform optimal management of inverters

CHAPTER 3: Spatial Coordination

Relevant distance scales in power system operation span six orders of magnitude, from local effects of power quality on the scale of an individual building to hundreds or even thousands of miles across interconnected systems. A “smart grid” with high penetration of renewables will require simultaneous consideration of small‐ and large‐scale compatibilities and coordination.

3.1 Transmission Level: Long-distance Issues

3.1.1 Background: Transmission Issues

The need for transmission capacity to remote areas with prime solar and wind resources is widely recognized. It is worth noting that renewable resources are not unique in imposing new transmission requirements. For example, a new fleet of nuclear power plants would likely be constrained by siting considerations that would similarly require the construction of new transmission capacity. In the case of solar and wind power, however, we know where the most attractive resources are – and they are not where most people live. Challenges for transmission expansion include social, economic and technical factors. Social and economic challenges for transmission expansion include • Long project lead times for transmission siting, sometimes significantly exceeding lead times for generation

NIMBY resistance to transmission siting based on aesthetics and other concerns (e.g., exposure to electromagnetic fields) • Higher cost of alternatives to visible overhead transmission • Uncertainty about future transmission needs and economically optimal levels

On the technical side, • Long‐distance AC. power transfers are constrained by stability limits (phase angle separation) regardless of thermal transmission capacity • Increased long‐distance AC power transfers may exacerbate low‐frequency oscillations (phase angle and voltage), potentially compromising system stability and security

Both of the above technical constraints can in theory be addressed with a.c.‐d.c. conversion, at significant cost. The crucial point, however, is that simply adding more, bigger wires will not always provide increased transmission capacity for the grid. Instead, it appears that legacy a.c. systems are reaching or have reached a maximum of geographic expansion and interconnectivity that still leaves them operable in terms of the system’s dynamic behavior. Further expansion of long‐distance power transfers, whether from renewable or other sources, will very likely require the increased use of newer technologies in transmission systems to overcome the dynamic constraints.

3.1.2 Research Needs Related to Transmission

On the social‐political and economic side, research needs relate to the problems of deciding how much transmission is needed where, and at what reasonable cost to whom. In addition, options for addressing siting constraints can be expanded by making transmission lines less visible or otherwise less obtrusive. Needed: • Analysis of economic costs and benefits to communities hosting rights of way • Political evaluation of accelerated siting processes • Continuing analysis to identify optimal investment level in transmission capacity relative to intermittent generation capacity, and to evaluate incentives • Public education, including interpretation of findings regarding EMF exposure • Continuing R&D on lower‐visibility transmission technologies, including compact designs and underground cables

Needed: • Dynamic system modeling on large geographic scale (WECC) providing analysis of likely stability problems to be encountered in transmission expansion scenario, the benefit potential of various d.c. link options • Continuing R&D on new infrastructure materials, devices and techniques that enable transmission capacity increases, including: dynamic thermal rating, power flow control, e.g. FACTS devices o fault current controllers, intelligent protection systems, e.g. adaptive relaying, stochastic planning and modeling tools, new conductor materials and engineered line and system configurations4

CHAPTER 4: Overarching Coordination Issues

Refinement of both spatial and temporal coordination – in other words, “smartness” – demands a substantial increase of information flow among various components on the electric grid. This information flow has implications for system control strategies, including the role of human operators. Some of this coordination is specifically associated with renewable and distributed resources, requiring increased information volume for • mitigating intermittence of renewable resources accommodating siting constraints for renewable and distributed generation

Problematic issues in the context of information aggregation include the following: • How much data volume is manageable for both operators and communications systems? • What level of resolution needs to be preserved? • What data must be monitored continuously, and what opportunities exist to filter data by exceptional events? • How can information best be presented to operators to support situational awareness?

Once data have been selected and aggregated into manageable batches, they must be translated or somehow used to frame and inform action items for operators. For example, we might ask what local information goes into an operator’s decision to switch a particular feeder section, or to dispatch demand response, generation or storage. Operating procedures are necessarily based on the particular sets of information and control tools available to operators. The introduction of significant volumes of new data as well as potential control capabilities on more refined temporal and spatial scales also forces decisions about how this information is to be used, strategically and practically. Issues concerning actionable items include the following: • What new tasks and responsibilities are created for grid operators, especially distribution operators, by distributed resources? • How are these tasks defined? • What control actions may be taken by parties other than utility operators? Needed: • Modeling of distribution circuit operation with high penetration of diverse distributed resources, including evaluation of control strategies. 4. Locus of Control

A question related to the definition of action items is who, exactly, is taking the action. With large amounts of data to be evaluated and many decisions to be made in potentially a short time frame, it is natural to surmise that some set of decisions would be made and actions initiated by automated systems of some sort, whether they be open‐loop with human oversight or closed‐loop “expert systems” that are assigned domains of responsibility. Such domains may range from small to substantial: for example, automation may mean a load thermostat that automatically resets itself in response to an input (e.g. price or demand response signal); distributed storage that charges or discharges in response to a schedule, signal or measurement of circuit conditions; or it could mean entire distribution feeders being switched automatically.

Finally, it would be naive to expect any substantial innovation in a technical system as complex as the electric grid to proceed without setbacks, or for an updated and improved system to operate henceforth without failures. Rather than wishing away mistakes and untoward events, the crucial question is what corrective feedback mechanisms are available, not if but when failures do occur. This includes, for example, contingency plans in response to failures of hardware, communications or control algorithms, cyber‐security breach, or any other unexpected behavior on the part of a system component, human or machine. A higher degree of spatial and temporal resolution in coordinating electric grids – more information, more decisions, and more actions – means many more opportunities for intervention and correction, but first it means many more opportunities for things to go wrong.

CHAPTER 5: Conclusion

The effective integration of large amounts of new resources, including distributed and renewable resources, hinges on the ability to coordinate the electric grid in space and time on a wide range of scales. The capability to perform such coordination, independent of any particular technology used to accomplish it, can be taken to define a “smart grid.”

Ultimately, “smart” coordination of the grid should serve to • mitigate technical difficulties associated with renewable resources, thereby enabling California to meet its policy goals for a renewable portfolio • maximize beneficial functions renewable generation can perform toward supporting grid stability and reliability

Much work lies between the status quo and a system with 33 percent of intermittent renewables. Due to the complex nature of the grid, and because the refinement of temporal and spatial coordination represents a profound departure from the capabilities of our legacy system, any “smart grid” development will require time for learning, and will need to draw on empirical performance data as they become available. Time is of the essence, therefore, in answering the many foundational questions about how to design and evaluate new system capabilities, how to re‐write standards and procedures accordingly, how to incentivize the most constructive behavior from market participants, and how to support operators in their efforts to keep the grid working reliably in the face of these transitions. With all the research needs detailed in this white paper, the hope is that questions addressed early may help prevent costly mistakes and delays later on. The more aggressively these research efforts are pursued, the more likely California will be able to meet its 2020 goals for renewable resource integration.

National Renewable Energy Laboratory. Western Wind and Solar Integration Study. May 2010.

Vittal, Vijay, “The Impact of Renewable Resources on the Performance and Reliability of the Electricity Grid.” National Academy of Engineering Publications, Vol. 40 No. 1, March 2010.

This entry was posted in Grid instability and tagged , , , , , , , . Bookmark the permalink.

Comments are closed.