The Energy Collective Group

This group brings together the best thinkers on energy and climate. Join us for smart, insightful posts and conversations about where the energy industry is and where it is going.

9,995 Members


What is the Potential of Distributed Generation with Storage and Demand Response?


  • Distributed storage and demand response has limited potential to enhance the value of distributed solar PV.
  • Distributed battery storage remains far too expensive for broad deployment.
  • Distributed demand response has limited potential for solar PV integration and will also be very complex to implement. 
  • Utility-scale solutions offer greater versatility, better economics and much greater deployment potential. 


As discussed in the previous article, distributed solar faces a number of fundamental challenges relative to utility-scale PV (and the wide range of other attractive utility-scale generation options): high installation costs, low capacity factors, and power density limitations. In addition, the value of distributed solar is strongly affected by the simple fact that it only works when the sun is shining.

To address the problem of intermittency and extend the applicability of distributed solar, we can implement solutions such as distributed energy storage and demand response. While solar PV is the flagship technology of distributed generation, batteries and smart meters/appliances are the flagship technologies of distributed storage and demand response. The reasons are the same: highly modular nature and low O&M requirements.

This article will take a fundamental look at the potential of distributed storage and demand response to enhance the potential of distributed generation. Similar to the previous article, utility-scale storage and demand response will also be assessed for comparative purposes.

Extra services

The first fundamental point to establish is that storage and demand response for integration of intermittent solar PV represent additional value creation required to deliver a service we currently enjoy without this additional value creation: cheap and reliable electricity. This implies that, for these additional services to be economic, they must enable generation technologies which are substantially cheaper than incumbents that require no storage or demand response. As illustrated in the previous article, this will be a particularly large challenge for distributed solar due to generally higher installation costs and lower capacity factors relative to utility-scale solar.

Energy storage challenge: Costs

Battery storage costs were analysed in more detail in a previous article and shown to still be a very long way from economic viability. A recent analysis by the IEA showed the following graphic to illustrate the enormous technological progress required by Li-ion technology before it can reach the largely uncompetitive position of pumped hydro storage.

Li-ion vs pumped hydro costs

An additional factor that is not yet commonly considered is balance of system costs. As with solar panels, batteries accrue additional costs in the process of installation and connection. These costs can be highly significant (see the last table in this review) and will become increasingly important as battery prices continue to decline.

For some further perspective, the following graphic from the IEA report “The Power of Transformation” shows that relatively cheap pumped hydro storage will only be marginally economical at moderate penetrations of intermittent renewables, even under the idealized “transformed scenario” (a scenario available only to developing nations building their electricity systems in a highly coordinated manner especially for the integration of intermittent renewables).

Storage benefit-cost ratio

Distributed storage is substantially more expensive than pumped hydro storage. Even with large future price reductions, the benefit/cost ratio of this option is unlikely to be anywhere close to the levels required to drive broad and sustainable deployment. Unsustainable situations created by direct subsidization and net-metering programs will drive some deployment, but the typical boom-bust cycle of subsidized renewable energy deployment is likely to rear its ugly head here as well.

Demand response challenge 1: Limited potential

Demand response generally does not face the cost challenges of energy storage. However, there are several factors that limit the potential of this seemingly simple and cost-effective solution when it comes to distributed solar, the two most important of which being 1) short-term demand increases are harder to achieve than short-term demand reductions and 2) the lowest hanging fruits related to heating/cooling correlates poorly with solar PV availability.

Challenges with demand increase

The first point to establish is that demand response for peak load reduction has good potential, a significant portion of which is being exploited already (see this study for example). The reason is simply that it is generally both practical and economical for industrial and commercial entities drawing large loads to reduce consumption under special circumstances in order to benefit the total system (see example below). Relatively simple distributed solutions like smart thermostats can also play a positive role here. 

Demand response load curve

For distributed solar, however, the need changes from reducing consumption for a limited time to increasing consumption for a limited time. This leads to a number of challenges. Firstly, strongly increasing demand for only a few hours around noon on summer days would require substantial equipment oversizing, thereby leading to increased capital costs. Secondly, concentrating even modest amounts of heating/cooling into the noon solar peak with the aim of replacing consumption in the evening will soon lead to discomfort. Thirdly, such demand concentration leads to significant thermodynamic inefficiencies. And finally, strong incentives to increase electricity consumption at certain times can erode efficiency by stimulating uneconomic demand.

Mismatch with solar output

Heating is an excellent target for distributed demand response, simply because heat is very easily stored over limited time periods. For example, the walls of a well-insulated house offer very effective (free) energy storage for a few hours (depending on your tolerance for temperature variations). Unfortunately, space and water heating demand is weakest during summer times when sunshine is abundantly available, implying that the potential of this easy demand response application is quite limited.

In this sense, cooling applications would appear to be much more suited for distributed solar demand response. Unfortunately, the large thermal energy storage capacity of most buildings cause the demand for air conditioning to be quite low during the noon solar peak, picking up during the afternoon as the building gradually heats up (see below). This situation of simultaneously increasing load and decreasing solar PV output presents a serious challenge.


The implication is that additional storage is required to take advantage of the good seasonal correlation between cooling demand and solar insolation. Although these cryogenic storage options are much cheaper than battery storage, the additional capital costs and thermodynamic inefficiencies relative to standard systems present a major barrier to deployment.

Demand response challenge 2: Complexity

The idea behind demand response for the integration of intermittent renewables is simple: simply shift as much demand from times of low wind/solar generation to times of high wind/solar generation. In practice, however, the challenges faced by distributed demand response in terms if IT, logistics and regulation are very complex.

Cybersecurity is a common concern about the smart grid where hackers could potentially control electricity access to individual homes or even influence the larger electricity system. This is a major challenge, but, even under the assumption of perfect security, the shear IT bulk of a system where millions of appliances respond to live electricity data which is directly influenced by the response itself is likely to bring several problems of its own. 

Logistically, there is a major challenge with getting all the millions of components which could contribute to the smart grid into place and working harmoniously. Achieving the limited potential of demand response for distributed solar integration described above will require a massive rollout of sufficiently smart and reliable appliances and meters. The regulatory frameworks required to effectively integrate this multitude of components into the market also presents a particularly complex problem.

One promising way of extending the potential of demand response for distributed solar integration is the smart charging of EVs. However, the enormous rollout of plug-in vehicles with smart charging and well-maintained public charging stations required for this purpose implies decades before this option has a meaningful impact. Despite attractive incentives, plug-ins currently account for only 0.7% of the US car market and the cumulative expansion curve is already close to turning linear (below). Regulation surrounding the market integration of millions of public charging stations will also be particularly complex.

US plug-in sales

Overall, the complexity-related challenges with distributed (small scale) demand response can be summarized as outlined below. Despite its relatively low costs, these challenges will strongly limit the deployment rate of distributed demand response.

DSI challenges

The utility-scale alternative

In general, storage and demand response carried out on a large scale is simply much more versatile than it can ever be on a small scale. Whereas distributed storage options are essentially limited to batteries, utility-scale storage can be deployed in multiple forms covering all timescales applicable to energy storage. In addition, utility-scale storage options are generally more economical and easier to integrate into existing power networks. More information can be found in some previous posts on pumped hydro and CAESbatteries and synfuels.

For demand response, large industrial and commercial point loads are much simpler and more practical to control than millions of tiny loads. In addition, the long-term potential of demand-side management of these large loads is substantially larger and generally independent of the season.

An especially interesting longer-term prospect for demand response is synfuel production using advanced electrolysis processes and captured CO2. These processes can generate large loads and, with projected low capital expenditures and high efficiencies of processes like PEM electrolysis (see this review for example), could be used very efficiently to integrate large penetrations of intermittent renewables and/or baseload nuclear. This demand-side management can also occur outside of costly distribution networks, thereby avoiding unnecessary distribution capacity upgrades and line losses.

Finally, utility-scale solutions can add value in all locations around the globe, whereas distributed solar (and associated integration mechanisms) are suited to only a limited number of countries as discussed in the previous article. Utility-scale storage and demand response are suitable not only to intermittent utility-scale wind/solar, but also to baseload nuclear and CCS and, since it is much less complex than the distributed alternative, can potentially be rolled out rapidly and sustainably in developing nations.


This analysis on the potential of distributed storage and demand response has strengthened the conclusions about distributed generation drawn in the previous article. In comparison to the utility-scale alternative, distributed generation (primarily solar PV) has a fairly low potential and, in the vast majority of cases, will be unnecessarily expensive and complex.

This does not mean that distributed generation will not be deployed. Niche markets exist and the ideological attractiveness of this energy option remains very high and extremely marketable. What it does mean, however, is that distributed generation will most probably not make more than a minor contribution to the clean energy revolution that will have to take place this century. Headlines claiming that distributed solar will soon overthrow utilities everywhere should be patiently ignored until reality sinks in. As an example, the highly optimistic hi-Ren scenario in the PV Technology Roadmap from the IEA which has received broad PV-positive press lately forecasts about 8% of electricity from distributed PV by 2050. The vast majority of the remaining 92% will remain utility scale. It should also be mentioned that electricity accounts for only about 40% of primary energy consumption. 

The final conclusion from these two articles is twofold: 1) distributed generation is affordable, but far from economic and 2) distributed generation can contribute, but only to a minor degree. For these reasons, the ideological attractiveness of distributed generation presents a particularly difficult problem: we simply cannot afford to aggressively pursue uneconomic solutions with very limited potential when it comes to the energy and climate issues we face today. The time has come to leave ideology at the door and get pragmatic about the challenge before us.

Schalk Cloete's picture

Thank Schalk for the Post!

Energy Central contributors share their experience and insights for the benefit of other Members (like you). Please show them your appreciation by leaving a comment, 'liking' this post, or following this Member.


Jeff Watts's picture
Jeff Watts on October 10, 2014

Good article. Thanks for the information.

Clayton Handleman's picture
Clayton Handleman on November 2, 2014


You make some interesting points, however there appears to be considerable bias in the position overall.

Here are a few I picked out.

– Density:  For distributed generation density is generally is not an issue.  In fact, one of the selling points of distributed generation is that it is at the point of use and reduces rather than increasing the power line losses and stress on T&D resouces.  For distributed generation, density is a red herring.  It is also overblown for utility scale but I will focus on DG for now.

– EV curve flattening:  EVs are in an early adopter phase of an industry that is changing rapidly.  All agree that the limitations are mostly related to range and cost.  Your curve is for the current addressable market of affluent early adoptors.  Interestingly, even during this time of dropping gasoline prices, it shows an inflection with the slope beginning to increase.  It will be interesting to see how the slope changes in the next two years as, for example, Nissan’s leaf gets its new range doubling battery and the price of Li-ion batteries drops faster than most of the projections that were made prior to the announcement of the Tesla Giga factory.  Progress on batteries is moving at an extraordinary pace such that McKinsey is suggesting that we will hit a pricing threshold that will dramatically expand the EV addressable market before 2020, not the ‘decades’ that you suggest.

– Electricity is only ‘40% of primary energy consumption’:  This may be true today but that number will shift dramatically as EVs go mainstream over the next 10 – 15 years.  They will shift much of the transportation load to from FF to electric dramatially increasing the percentage of energy from electricity.

– Smart grid required for load shifting EVs is fraught with peril due to cyber security issues:  An enormous load shift benefit can accrue from EVs with nothing more than Time Of Use (TOU) metering which can be done with no increase in cyber vulnerability.  Tesla already has a timer built in to the car that allows the user to specify when to charge.  In most areas this can be shifted to known off-peak times such as after midnight.  Of course this does not take advantage of a noon solar peak but it does allow EVs to charge at off peak times.  Alternatively, with large solar penetration price drops would occur during late morning and early afternoon.  EV users could manually set supplemental daytime charging to occur during those peaks.  This has the added advantage that it supports a larger build-out of solar.  The charging load would switch off as the AC load moved toward peak.  But the larger build out would offer added capacity shifting the solar usefulness further into the evening hours.  This post looks at the potential for EV load shifting using real world wind and solar data.  It is also interesting to compare the real world August solar data in this image with that in your graph of peak AC loads.  It would appear that the mismatch you claim is exaggerated.   It shows a 6:00pm peak.  And while the PV is dropping off by that point it is still substantial and would contribute considerably to addressing demand.  Of course using trackers would extend PV so that it peak matches very well as is shown here


Clayton Handleman's picture
Clayton Handleman on November 2, 2014

Great source material.  I like the plot of ISO NE shaving the peak with load shifting.

It is interesting to note that the article you cite, about high AC loads, indicates that 20% of electricity costs arise from peaker operation for 100 hours per year.  That is about 1% of the time.  So peakers cost about 20x the cost of regular power.  As such, the economics of storage should be compared to that cost not the day to day cost of electricity.  In other words, in many cases the storage needed to offset renewable intermittency can be dual use, also offsetting peakers.  This dual use for storage dramatically improves the economics of backing up renewables.

Schalk Cloete's picture
Schalk Cloete on November 2, 2014

What is the link between power density and line losses / T&D?

I concur in the article that EVs have a good theoretical potential for distributed demand response, so the difference between our views is related only to differences in EV outlook which we have discussed a few times before. 

About TOU charging, this is fine for conventional baseload-dominated power systems where lowest prices occur consistently in the early morning. However, this article is about the potential for distributed PV integration which makes things more complicated. In a future system with a high penetration of intermittent renewables where the optimal charging time will shift from one day to the next, the mass consumer (and the grid operator) would want charging to be fully automated through smart systems. 

Get Published - Build a Following

The Energy Central Power Industry Network is based on one core idea - power industry professionals helping each other and advancing the industry by sharing and learning from each other.

If you have an experience or insight to share or have learned something from a conference or seminar, your peers and colleagues on Energy Central want to hear about it. It's also easy to share a link to an article you've liked or an industry resource that you think would be helpful.

                 Learn more about posting on Energy Central »