Renewable integration challenges create demand response opportunities
The power grid is getting greener. The graph below summarizes EIA projections of U.S. non-hydro electricity generation under different assumptions about greenhouse gas regulations, fuel prices, and technology costs:
Source: http://www.eia.gov/todayinenergy/detail.cfm?id=16051
Although there is some debate over whether these EIA projections are too conservative, it seems we can all agree that the penetration of non-hydro renewables will continue to increase in the coming years.
The power grid is also getting smarter. The Smart Grid Investment program invested close to $8 billion in accelerating the deployment of smart grid infrastructure and technology. It is estimated that smart meters will have been deployed to over 50 percent of U.S. end users by 2015.
Some recent studies highlight some important complementarities between a greener grid and a smarter grid. Before connecting these dots, let’s first review the key operational challenges associated with increasing integration of renewables.
Why lose sleep over increasing grid-penetration of renewables?
Here in California, the “duck chart” has become emblematic of the challenges that increased renewable generation (and solar in particular) could present. The duck helps to illustrate some key issues (not just in California, but any place the sun rises in the morning and sets in the evening).
The colored lines trace out actual (2012-2013) and forecast (through 2020) electricity demand less generation from variable renewables, including wind and solar (“net load”). As solar PV accounts for a larger share of generation, the change in the net load profile takes on a duck-like shape. Note three key take-aways:
- Ramping demands: Increased solar puts stress on the system when the sun rises and sets. Conventional generation must ramp down and up to compensate.
- Over-generation can be a problem when solar output peaks in the early afternoon if demand levels are modest and inflexible base load generation bumps up against minimum output constraints.
- Declining marginal value: As the level of renewables penetration (solar in particular) increases, renewable energy output becomes less coincident with peak net load. This drives down the marginal value of the electricity generated, in part by reducing the capacity value of solar on the build margin and in part by driving up the marginal cost of managing variable energy output.
This duck is only an illustrative tool. For one thing, the chart is based on a somewhat non-representative day in which solar output is high but temperatures are cool and demand for air conditioning low. Perhaps more importantly, the graph makes no attempt to account for adjustments in energy infrastructure and energy markets that can be deployed to mitigate stresses on the system. Fortunately, we have many options available when we think about re-optimizing the electricity sector to accommodate higher levels of renewable energy generation.
Meeting the renewables integration challenge
Over a year ago, Catherine wrote a great post raising key questions about the relative merits of storage versus demand response to renewable resource integration challenges. A year later, we have in hand some studies that systematically consider how different power system investments and operational changes can mitigate ramping and over-generation problems associated with increased renewables penetration.
One study in particular – recently released by Andrew Mills and Ryan Wiser of LBNL- caught my attention last week. The paper considers alternative approaches to meeting the renewables integration challenge including: energy storage, demand response (assuming a demand elasticity of -0.1), investment in more flexible gas generation, and diversification in renewables deployment (to reduce variance of aggregate renewables output).
The study assesses the economic value of these response options, where “value” is defined in terms of the change in the marginal economic value of wind or solar relative to a base case where no measures are deployed. That’s a mouthful. The basic idea is the following. The authors simulate long-run investment decisions, generation dispatch, and wholesale market clearing in California’s electricity sector under a baseline scenario and calculate the marginal economic value of renewables (in terms of energy and capacity cost avoided). They then repeat the entire simulation exercise assuming one of the mitigating alternatives has been deployed.[1]
The following table summarizes some key results from the study:
Qualitatively, the results are quite intuitive. At very high solar penetration rates, bulk storage is particularly valuable. Diversification is more effective in the high wind penetration scenarios. Demand response (RTP) increases the marginal value of both wind and solar at low and high penetration rates.
Importantly, the study stops short of estimating costs. Dedicated bulk storage is likely to be costly. In contrast, we have already made a significant investment in the smart grid infrastructure we would need to operationalize widespread demand response.
Smart renewables integration should leverage the smart grid
Increased penetration of variable renewable resources increases the potential value added by demand response. In this sense, renewables integration creates opportunity for demand response. In order to tap this potential, we’ll need to enable broad based and highly flexible demand response. This represents a significant departure from today’s standard DR programs. There is much more we can do- both in terms of automation and pricing- to leverage investments in smart-grid infrastructure. With renewable energy penetration rates on the rise, the cost of overlooking these opportunities gets harder to justify.
[1] Consider, for example, the high solar penetration scenario. The simulated marginal value of solar is $25.32/MWh at a 30% PV penetration rate in the baseline scenario. When demand response to real time pricing incentives is incorporated into the simulations, the marginal value of solar increases to $32.76/MWh. This implies a value of $7.44/MWh.
Categories
Energy Institute Blog 
Josh – Two answers. 1) Automate responses. Hot water heaters and pool pumps are examples of loads where the exact timing does not matter. Real time pricing gives customers the ability to make any cost/convenience tradeoffs they want to program into their machines. Even building HVAC can do some of this, without additional hardware.
2) Look at non-residential loads.
I notice that the LBL study that Dave Watson links to does both:
Based on this analysis, a large-scale deployment of fast AutoDR could provide between 0.18 and 0.90 GW of DR-based ancillary services from the existing stock of commercial and industrial facilities throughout California. With modest investments to upgrade and expand use of automated control systems in commercial and industrial facilities the estimated shed potential could approximately double to between 0.42 and 2.07 GW. Deployed costs for fast AutoDR (installation, materials, labor and program management) are about 10% of the deployed costs of grid scale battery storage.
This quote is spot on – “[R]enewables integration creates opportunity for demand response. In order to tap this potential, we’ll need to enable broad based and highly flexible demand response. This represents a significant departure from today’s standard DR programs.” But how do we make this significant departure? Most utility customers already have trouble understanding today’s standard DR programs, which usually have day-ahead notice of DR events and a static event window (i.e., 2-7 PM for PG&E’s residential SmartRate program). If we start talking about moving DR event hours around throughout the year and providing shorter notice of DR events, most customers will become even more confused and even less willing to enroll than they already are.
Thanks for highlighting an important paper in this field. “Fast, continuous, 24/7 demand response” (a new term warranted?) appears to be a resource with substantial potential for grid balancing at a fraction of the cost of storage. Link to a scoping study: http://drrc.lbl.gov/publications/fast-automated-demand-response-enable