The difficulties of meeting cyclic demand for electricity with intermittent renewable generation have been addressed in a number of previous posts, but with wind, solar etc. usually considered separately. Here we will examine a hypothetical scenario involving a diversified mix of renewable energy sources that supplies 100% of electricity consumption in unspecified future year 20XX in Atlantis, an imaginary island country that is very much like, but not exactly the same as, the UK.
Details of the 20XX generation mix in Atlantis are summarized in the following Table.
This mix generates the same amount of electricity as the UK, and we will assume Atlantis too, generated from dominantly thermal sources in 2013 (380 vs. 374 TWh). Dispatchable capacity (biomass and hydro) provides 35% of total generation and non-dispatchable capacity (wind, solar, tidal) the remaining 65%.
And because Atlantis is very much like UK the scenario assumes that demand in Atlantis in 20XX will be the same as it was in the UK in 2013, so the 2013 Gridwatch data for the UK are used to define 20XX Atlantis demand. Atlantis generation in 20XX was estimated using the following simplifying assumptions:
Wind generation is the same as UK generation in 2013 but scaled up in proportion to the increase in output.
Biomass is assumed to provide constant baseload generation, although biomass plants would probably be able to operate in semi-load-following mode in the same way as UK coal plants do at present.
Hydro generation is used in either baseload or load-following mode with a maximum output of 13GW and no ramping restrictions. (Hydro supplying 15% of the UK’s energy is of course a pipe dream, but it’s not a problem in Atlantis.)
Solar generation is estimated using total solar radiation values for latitude 53 north on the 15th day of the month and is kept constant through the month.
Tidal power generation assumes tidal generators spaced around the coast of Atlantis in such a fashion as to cancel out diurnal and semidiurnal tidal fluctuations. (Tide times around the coast of UK vary enough to allow this to be done). Generation is estimated by straight-lining between a spring tide maximum of 5GW and a neap tide minimum of 1GW in accordance with the 28-day lunar cycle.
Imports and exports are not taken into account.
Cost is no object.
Generation from all sources was estimated at ~5 minute intervals to match the Gridwatch reading interval.
To keep the post to a manageable length only the power balances in the months of July and January 20XX, which are assumed to be “typical” summer and winter months both in the UK and Atlantis, are considered.
Figure 1 shows total renewables generation by source for July 20XX, with wind generation factored from the July 2013 Gridwatch data, tidal generation estimated assuming peak spring tides on July 1 and generation from the other sources estimated using the assumptions listed earlier. Wind generation often shows diurnal variations, but these are only occasionally in antiphase with solar generation. At other times they are in phase with solar or phase-shifted relative to solar. The overall impact is no net smoothing effect (R squared wind vs. solar = 0.00):
Figure 1: Generation by source, July 20XX, 100% renewables
Figure 2 shows the Figure 1 data on a stacked bar chart. Hydro appears as a flat line because no attempt has been made to match generation to load:
Figure 2: Total generation by source, July 20XX, 100% renewables
Figure 3 compares total generation with the July 20XX demand (load) curve, which is assumed to be the same as the July 2013 demand curve:
Figure 3: Supply & demand, July 20XX, 100% renewables
Demand and total generation tend to be synchronized because July solar output is closely correlated with the daily load curve. Total generation from the 100% renewables system is also higher than demand (~24,000 vs. ~22,000 GWh). As would be expected, however, there are numerous occasions when the system generates more or less power than is needed, resulting in power surpluses or deficits. We can easily eliminate the surpluses by curtailing excess wind generation, but how do we handle the deficits?
The first approach is to ramp up hydro to its assumed maximum output of 13GW during deficits and ramp it down during surpluses. This gives the result shown in Figure 4. The worst of the deficits are gone but we are still left with a number exceeding 5MW, all occurring between 6 pm and midnight:
Figure 4: Supply & demand, July 20XX, 100% renewables, hydro follows load, surplus wind curtailed
The Figure 5 bar chart shows the Figure 4 generation mix. Note that hydro ramping rates would be lower if biomass also contributed to load-following:
Figure 5: Total generation, July 20XX, 100% renewables, hydro follows load, surplus wind curtailed
How to get rid of the residual deficits in Figure 4? One solution would be demand management, which in this case would be facilitated by the fact that the deficits all occur at the same time of day. However, it would probably be easier simply to increase installed capacity until the deficits disappear. But which generation source should get the added capacity? Here’s a summary of how much extra installed capacity would be needed for each (numbers approximate):
Solar: No amount of added capacity will eliminate the deficits because the sun has set.
Wind: 140 GW
Tidal: 60 GW
Biomass: 10 GW
Hydro: 7 GW
Hydro wins, but there are always practical limits on how much additional hydro capacity can be installed, and we will assume that 13GW is pushing them on Atlantis. Assuming no fuel supply constraints there are no such limits on biomass, which therefore emerges as the logical choice. So with 10GW more biomass, which would represent only a comparatively minor cost increment, demand in July 20XX could be met 100% of the time.
Now let’s turn to January 20XX.
Figure 6 shows total generation by source for the month, with wind generation again factored from the January 2013 Gridwatch data, tidal generation estimated with peak spring tides on January 16 (counting lunar cycles back from July 1) and generation from other sources estimated as before. Wind generation more than doubles over July (from 6,600 to 15,300 GWh; the wind in Atlantis blows harder in winter, as it does in UK) but solar generation is down by a factor of more than four (from 4,800 to 1,100 GWh; the factor would be larger if increased winter cloudiness were allowed for):
Figure 6: Generation by source, January 20XX, 100% renewables
Figure 7 shows the Figure 6 data on a stacked bar chart. Solar generation has faded into insignificance:
Figure 7: Total generation by source, January 20XX, 100% renewables
And although monthly generation is equal to monthly demand (at ~29,000 GWh) the 65% non-dispatchable capacity in the generation mix again produces too much electricity when the wind is blowing and too little when it isn’t (Figure 8). Because of the minimal solar output there is also no “natural” synchronization of supply and demand, such as there was in July:
Figure 8: Supply & demand, January 20XX, 100% renewables
And the load-following hydro capacity can’t handle the resulting deficits. They are just too large:
Figure 9: Supply & demand, January 20XX, 100% renewables, hydro follows load, surplus wind curtailed
Figure 10 shows the adjusted generation mix in bar chart form. (Note that the demand shortfalls occur even when hydro runs at 13 GW maximum capacity for days on end. Whether hydro could sustain this level of output depends on how much of the 13GW is “conventional” hydro and how much pumped, but if it were all pumped more than a terawatt-hour of storage would be needed):
Figure 10: Total generation, January 20XX, 100% renewables, hydro follows load, surplus wind curtailed
Deficits of this size will not easily succumb to demand management, so we are back to considering how much extra capacity is needed to make them disappear. Here’s a summary of requirements by generation source (numbers again approximate):
Solar: No amount of added capacity will eliminate the deficits because the sun has set.
Wind: ~1,000 GW
Tidal: 90 GW
Biomass: 30 GW
Hydro: 25 GW
Adding another 25 GW of hydro is out, so the choice would be to add another 30 GW of biomass, thereby increasing total installed biomass capacity from 11 to 41 GW. This would allow the system to meet January 20XX demand, and since no other month is as problematic as January to meet demand through the rest of the year as well.
There is, however, no safety margin in this number. An additional ~10 GW would be needed to cover the worst-case scenario, which occurs when there is a coincidence of cold winter weather, high peak demand, little or no wind, darkness and neap tides. But after adding this extra 10 GW we have a generation mix containing 51 GW of biomass and 13GW of hydro. Assuming equivalence with UK this would be about equal to the capacity of the coal, nuclear and CCGT plants currently operating in Atlantis, and it performs the same function – supplying dispatchable baseload and load-following generation.
These results are of course subject to uncertainty, but what they basically tell us is this. Atlantis can install a 100% renewables generating system, but it will be capable of meeting demand only if the presently-existing thermal dispatchable generation is replaced kilowatt-for-kilowatt by renewable dispatchable generation.
As for non-dispatchable capacity, with adequate dispatchable generation it can be added at will. The problem is that the wind has a habit of blowing when we don’t need it to and not blowing when we do. Because of this up to 50% of the wind energy generated in July 20XX and up to 80% of the wind energy generated in January 20XX gets “spilt”, raising the question of why Atlantis installed so much surplus wind capacity in the first place.
That concludes the analysis of 100% renewables generation in the mythical country of Atlantis, where 100% renewables generation is actually less problematic than it would be in the UK because of the 15% hydro contribution, something the UK is unlikely ever to achieve. For those interested in a UK-specific analysis a study from the University of Glasgow identifies the same problems without coming up with any solutions:
Across the whole year (2025), there are large periods of electricity deficit during winter and large periods of electricity surplus during summer. The maximum supply deficit (electricity demand minus electricity production) present across the year in the model is 53.8GW, which is almost equal to the demand at that time (54.5GW). This occurs on January 21st at 17:00, which is also the time of maximum demand that day. By 17:00 on this day, the Sun had gone below the horizon and so solar production was nil. Additionally, wind production also happened to be very low at this time. This is an example of maximum demand coinciding with minimum production, which is one of the major problems that will challenge future grid designs.
While another from the Journal of Power Sources: indicates how 100% renewables generation might be made to work in the Eastern US, although after a detailed reading I’m not sure it would. (H/T Dennis Coyne for the link):
Our model evaluated over 28 billion combinations of renewables and storage, each tested over 35,040h (four years) of load and weather data. We find that the least cost solutions yield seemingly-excessive generation capacity—at times, almost three times the electricity needed to meet electrical load. This is because diverse renewable generation and the excess capacity together meet electric load with less storage, lowering total system cost. At 2030 technology costs and with excess electricity displacing natural gas, we find that the electric system can be powered 90%–99.9% of hours entirely on renewable electricity, at costs comparable to today’s—but only if we optimize the mix of generation and storage technologies.