Skip to main content

Energy Storage and Flexible Demand - Two Sides of the Same Coin

 For this previous post, we have built a very simplistic LP model to determine the optimal resource allocation for a power grid that is based on wind, solar and storage.

This model has shown an obvious, significant need for storage to balance variable and non-dispatchable wind and solar generation to a fixed demand on timescales ranging from hourly to seasonal.

The key to decarbonizing high emission sectors like transport, buildings and industry is going to be aggressive electrification of these sectors, which is going to increase electricity demand significantly and put a strain on aging power distribution infrastructure.

The good news is that the vast majority of this new demand is going to be somewhat elastic and flexible in time - potentially at no extract cost  and without loss in comfort or quality of service.

Instead of using storage to match variable supply to a rigid load, we can exploit demand flexibility to match  adaptive loads to existing supply and hence reduce the need to invest in expensive storage.

Most of the flexibility that comes at little or no additional cost is in fact taking advantage of already existing equivalent storage in another domain:

  • For transportation, the additional electricity will be used to charge batteries in electric vehicles, which are typically quite oversized for the average daily use and may only need to charged once a week - resulting in flexibility to shift the demand around in the order of days.
  • For building and industrial heat production, the additional electricity will be used to power heat pumps. Buildings which are heated or cooled can have significant inertia - thermal storage in the form of the buildings material and structure. Or there might still be a hot water storage tanks in the boiler room from previous systems or such thermal storage tanks could be added at much less cost than comparable electricity storage. 
  • For industrial processes which need heat at higher temperatures, specialised thermal storage systems based on sand, volcanic rock or molten salt can also provide cost efficient heat storage at temperatures ranging  from ambient to several hundred degree Celsius with a capacity to store heat for days or months.
  • For some energy intensive production processes, it might be possible to shift production to when electricity is most cheaply available and effectively store the energy in the warehouse as produced goods.

In order to evaluate the potential impact of flexible demand, we extend the previous model to assume that a fraction of the existing load could be somewhat flexible in time. We are again using the actual load and wind/solar production data from the ENTSO-E transparency platform for the Netherlands between Spring 2022 to 2023 with a target availability of 100%.

We are assuming that some fraction of this load (0 to 60% respectively) which could be shifted in time anywhere from 0 to 12 hours. This would be a reasonably conservative assumption for both everyday electric car charging and many building heat applications:

0% 20% 40% 60%
Load (total / avg / peak) 99.6TWh / 11GW / 17.7GW 99.6TWh / 11GW / 17.7GW 99.6TWh / 11GW / 17.7GW 99.6TWh / 11GW / 17.7GW
Generation (total / avg / peak) 116.4TWh / 13.3GW / 48.9GW 115.6TWh / 13.2GW / 48.5GW 114.2TWh / 13.1GW / 47.8GW 111.9TWh / 12.8GW / 46.8GW
Generation PV / ONW / OFFW 39.4% / 60.6% / 0.0% 39.2% / 60.8% / 0.0% 38.3% / 61.7% / 0.0% 37.4% / 62.6% / 0.0%
Annual Cost / Cost per MWh 8.5B€ / 85.3 €/MWh 8.2B€ / 82.3 €/MWh 7.9B€ / 79.8 €/MWh 7.8B€ / 78.1 €/MWh
System Efficiency 85.5% 86.2% 87.1% 89.0%
Surplus 1.5% 1.5% 1.3% 0.8%
Storage contribution 28.2TWh (28.4%) 21.9TWh (22.0%) 17.6TWh (17.7%) 15.2TWh (15.3%)
SDS Power 7.6GW 2.9GW 0GW 0GW
SDS Capacity / Duration 39.9GWh / 5.2h 16.4GWh / 5.7h 0GWh / 0.0h 0GWh / 0h
SDS contribution 9.5TWh (9.5%) 3.4TWh (3.4%) 0TWh (0%) 0TWh (0%)
LDS Power (charge/ discharge) 14.1GW / 11.0GW 13.9GW / 10.8GW 13.5GW / 10.6GW 12.8GW / 10.1GW
LDS Capacity / Duration 5489.6GWh / 500h 5456.0GWh / 505h 5431.4GWh / 513h 5453.0GWh / 537h
LDS contribution 18.8TWh (18.9%) 18.6TWh (18.7%) 17.6TWh (17.7%) 15.2TWh (15.3%)

As one might expect, with a growing share of flexible load the need for short duration storage declines and goes away completely at about 50% of flexible load. The demand for long-duration or seasonal storage capacity does not decline much from this short-term demand flexibility, but the need for its expensive charge and discharge capacity is reduced as the short-term demand flexibility can help to reduce the maximum instantaneous production surplus and shortfall peaks that might occur.

Increasing the flexibility window to 24 hours only moderately improves the efficiency and reduces the cost of the system:

0% 20% 40% 60%
Load (total / avg / peak) 99.6TWh / 11GW / 17.7GW 99.6TWh / 11GW / 17.7GW 99.6TWh / 11GW / 17.7GW 99.6TWh / 11GW / 17.7GW
Generation (total / avg / peak) 116.4TWh / 13.3GW / 48.9GW 114.3TWh / 13.1GW / 47.9GW 111.6TWh / 12.8GW / 46.7GW 109.8TWh / 12.6GW / 45.8GW
Generation PV / ONW / OFFW 39.4% / 60.6% / 0.0% 38.9% / 61.1% / 0.0% 37.8% / 62.2% / 0.0% 36.3% / 63.7% / 0.0%
Annual Cost / Cost per MWh 8.5B€ / 85.3 €/MWh 8.0B€ / 80.8 €/MWh 7.7B€ / 77.5 €/MWh 7.5B€ / 75.7 €/MWh
System Efficiency 85.5% 87.1% 89.2% 90.6%
Surplus 1.5% 1.0% 0.7% 0.7%
Storage contribution 28.2TWh (28.4%) 20.1TWh (20.2%) 15.1TWh (15.2%) 12.6TWh (12.7%)
SDS Power 7.6GW 3.3GW 0.5GW 0GW
SDS Capacity / Duration 39.9GWh / 5.2h 10.8GWh / 3.3h 0.6GWh / 1.1h 0GWh / 0h
SDS contribution 9.5TWh (9.5%) 2.4TWh (2.4%) 0.1TWh (0.1%) 0TWh (0%)
LDS Power (charge/ discharge) 14.1GW / 11.0GW 13.4GW / 10.3GW 12.2GW / 9.6GW 11.0GW / 8.8GW
LDS Capacity / Duration 5489.6GWh / 500h 5448.4GWh / 526h 5412.6GWh / 564h 5290.0GWh / 603h
LDS contribution 18.8TWh (18.9%) 17.8TWh (17.9%) 15.0TWh (15.1%) 12.6TWh (12.7%)

Even if we assume that the system were to accomodate for 60% of the load to be flexible anywhere within a 24h range, only a small part of that potential is actually used. Most of the load can still be satisfied close to the time when it is nominally requested:

This simulation assumes a lossless distribution network without capacity constraints. In reality most power grids face severe limitation and will require significant capacity increase to keep up with the expected growth in load. We should take advantage of the inherent flexibility for most this new load. This would be significantly cheaper than building new transmission capacity. Or in other words: silicon and software are generally much cheaper than copper!