EnergyNumbers-Balancing: Modelling wind and PV on the grid

Design your own system for


Name Max power
into store
Max power
out of store
Select what the lower chart displays:

Click ? to see the tutorial again.

What am I looking at?

This model is designed to give you some idea of the trade-offs, of what could be possible, and of the scale of what's involved, in integrating lots of wind and PV into a national grid; and of some of the contributions required from storage.

This is a simulation model. You give it a set of inputs - a scenario - and the model simulates one set of possible outcomes.

Hover over the lines and bars in the charts to get detailed information.

OK, so what do I do?

You can set the amount of generation as a percentage of demand equivalent: for example, generation of 100% means that over the whole data period, total generation would be the same amount of energy as total demand, assuming that generation was never curtailed because there was no use for it.

You can specify up to five stores, with different characteristics. The top store absorbs as much surplus energy as it can (subject to its max power in constraint, and its storage capacity); the second store then tries to absorb any remaining surplus after that, and so on down the table. Similarly, when the grid needs energy from storage, it first tries to take it from the top store, and having done that, if it still needs more energy, it then tries the second store, and so on down the table.

Drag the up/down arrows to change the relative priority of the different stores.

Assigning a value of 0 to the storage capacity (GWh) means that the store is only constrained by its maximum power limits, not by its energy storage capacity: this is a reasonable assumption for imports and exports via "interconnectors" - the connections to neighbouring countries' electricity grids. You can model differing quantities of guaranteed import and export capacity (even though they'd physically be carried across the same wires), by entering them as two different stores: an import store with zero max power into store, and zero (i.e. unlimited) storage capacity; and an export store with zero max power out of store and zero (i.e. unlimited) storage capacity.

If you want to share a store between two (or more) rows, just include the word 'shared' in the name of each store you want to share. Only the capacity of the highest store in the list will be used.

Storage is taken to be half-full at the start of the simulation.

And what do the results mean?

Because generation of wind and solar doesn't follow demand, but rather follow the wind and the sun respectively, then only part of the generation will meet the demand in real time. That means that some generation could be wasted ("curtailed"); and some demand would not be met by real-time generation. On-grid storage is one of several ways to handle this.

Changes to storage levels are reported. If any store shows a net decrease over time, that could be a problem. If all stores together show a net decrease over time, that could be a problem.

What's in the 100%-renewables example scenarios?

An example scenario has been created for each country. Each provides 100% renewables with no shortfalls: demand is always met. Each illustrates several features of the way that stores are modelled.

Long-term storage is provided by power-to-gas: electricity is used to synthesise hydrogen (or methane), from water (and carbon dioxide). The gas is stored until it is needed. That means that storage can be provided over long periods of time; the amount of energy stored can easily be scaled up - we have a lot of experience of storing large quantities of gas, and of burning it to generate electricity. The amount of energy that can be stored can be adjusted independently of the rate at which energy goes into the store; and both of these can be adjusted independently of the rate at which energy can be taken out of the store. We have two long-proven techniques for generating electricity from gas: the cheaper, but less efficient Open-Cycle Gas Turbines (OCGT); and the more expensive and more efficient Combined-Cycle Gas Turbines (CCGT). These turbines all share a common gas store in the model: this is triggered in the model by addding the word "shared" into the stores' names.

The order of the stores is significant: for this reason, interconnectors are specified as two separate stores - one for imports, one for exports. Imports are given higher priority than the gas store. Exports, though they'd use the same wires, are given lower priority than the gas store. So when extra power is needed, it's taken from imports first; and if there's still more needed, then it's taken from the gas store. The more efficient CCGTs are drawn on, and if even more power is needed, then the OCGTs are used. Whereas when surplus power is available, it first goes into the store, and then any surplus that can't be stored is available for export.

How the model works

The model uses historic demand data, and historic (half-)hourly capacity factors for PV and wind, to simulate the extent to which demand could be met by some combination of wind, PV and storage. For each (half-)hour in any model run, the capacity factor of each generation source is exactly what was observed historically; it's just scaled up so that the total energy produced over the whole dataset matches what you've set in the controls screen.

This simulation assumes that sufficient resource would be available and can be harnessed. It extrapolates based on observed generation and demand data.

No model can model everything, so some simplifications have been made:

  1. Demand is just as it was in the historic dataset - there's no demand-side response to time-of-use pricing or other incentives.
  2. There's not yet enough data to separate out onshore and offshore wind (this will change soon), or to include hydro.
  3. Increasing the amount of wind or PV does not change the historic shape of the generation curve of either: only the height of the curve.
  4. There are no significant within-country transmission constraints - the model has a lot of detail about time, but no geographic detail. The assumption is made that sufficient capacity will be built.
  5. Although the model matches demand and supply at all timescales from (half-)hourly to years, it does not do so at timescales of milliseconds to minutes. A half-hour is the smallest timescale in this model. Balancing at the scale of a single cycle to one hour is just as important as balancing at the scale of hours to years. This model draws the line at half an hour (one hour for Germany, as that's the data interval of the NetzTransparenz data).
  6. There's no demand-side storage in here yet: in particular, there is no hot-water storage, and no heating-demand profile, so it can't (yet) model the advantages of future electrification of heating. That's a shame, as that's one of the cheapest, most scalable forms of energy storage we've got, and as heat is such a large proportion of annual energy demand, the use of hot-water storage to time-shift heating demand represents a significant extra resource to help balance energy supply and demand every second of the day.

The model version number at the bottom of the page has two separate elements: the first part, between the "v" and the ".", shows the version of the input time-series for supply and demand: this is increased by one each time the input data is updated. The second part, after the ".", is the version number of the software, which includes both the front-end and the back-end.

There's not much data for offshore wind for Germany yet, and it's not yet possible to split out offshore and onshore wind data in Britain; but I'll implement that when there's enough to make some sensible extrapolations. Biomass isn't in the model yet, but it may not be too much work to include it. I'd also like to add more countries, as long as that didn't increase the maintenance costs of keeping the data up to date. Ping me on Twitter if you want to help.

My thanks to:

Calculating ...

Supply from generation and storage was sufficient to meet demand in every interval, though this may not last, as stores were depleted over the period.

% of demand could not be met by generation or storage, across a total of events.