Our Future Energy Scenarios model

We have a companion site at ed.c4cs.org.uk, which contains our Future Energy Scenarios model.

The model is designed to help you test how various characteristics of our energy systems (e.g. different capacities of technologies, and different levels of demand in various uses) work together to produce a competitive or expensive and reliable or unreliable system.

The model uses various types of data from 3 years (2016-18). These years are not aggregated, as averaging tends to smooth out the real-world variations. They are available for multiple runs of the same scenario, based on different years’ data.

Hourly figures for:

the aggregated output of each generating technology provide a supply profile for that technology,
transmission-system demand provide a demand profile for conventional electricity uses,
interconnector flows provide profiles for the flows over each interconnector (as was in 2018), and
average temperatures are combined with seasonal heat-demand profiles to generate synthetic hourly heat demand figures, along the lines developed by Rob Sansom (of Imperial College London) and enhanced by Stephen Watson (of Loughborough University).

Road transport demand varies more consistently, and is less time-sensitive (because vehicles carry their own fuel-store). We construct a rough profile from monthly demand levels and a typical hourly profile for a day, modified by an assumption that re-fuelling will tend to adjust itself to flatten the curve (e.g. charging disproportionately at night if other demand is lowest then), rather than matching hourly consumption.

The model is also seeded with annual data or estimates for the period 2016-18, as the Base scenario. For most of these, it invites the user to input alternative figures for a future Proposed scenario, so the user and results are not bound by my estimates. These data items include:

Capacity, availability factor, capital cost per MW, fixed cost per MW/year, variable cost per MWh and carbon intensity of operation (tCO₂e/MWh) and construction (tCO₂e/MW) for 10 types of electricity generation (solar, biogas, nuclear, onshore wind, offshore wind, biomass, hydro, gas, coal and oil). I refer to availability factor rather than load or capacity factor because these figures are intended to be the load factor in the absence of any curtailment, i.e. the technical potential not real-world performance. The model works out the load factor after any curtailment. It does not differentiate between technologies per fuel (e.g. CCGT or OCGT), but it does differentiate between current and new capacity, allowing two sets of performance/cost parameters to be used per fuel.
Distribution losses, capacity, capital cost per MW, fixed cost per MW/year, variable cost per MWh and construction carbon intensity (tCO₂e/MW) for the electricity and gas transmission/distribution networks. It does not address regional issues (e.g. distribution bottlenecks). The gas part is currently un-utilised. The annual average distribution loss is converted into hourly figures on the assumption that losses are higher proportionately when flows are lower.
Overheads for electricity supply: capital cost, fixed cost p.a. and variable cost per MWh (based on the Big 6s Consolidated Segmental Statements).
Capacity (MWh), power (MW), round-trip efficiency (%), capital cost of capacity (/MW), capital cost of power (/MWh), fixed cost (/MW p.a.), variable cost (/MWh), charge at start of period (MWh, typically 50% of capacity) and construction carbon intensity (tCO₂e/MW) of three storage technologies (labelled pumped storage, compressed air and batteries, but can be used for any technology, e.g. hydrogen, although the base data will be inappropriate). It allows for separate current and new parameter-sets, as per generation.
Capacities of 5 interconnectors (France, Ireland, Netherlands, Northern Ireland and Belgium). They can be used simply as 5 flows, but the profile will take the historic profile of those interconnectors. This could do with updating, because several new interconnectors are now connected. Also, a single figure for the construction carbon intensity (tCO₂e/MW) of interconnectors, to be applied to each according to capacity increases in the Proposed scenario.
Downstream electricity demand (TWh) for “lighting” and air conditioning. Lighting really means all mainstream electricity demand in 2016-18, except air conditioning. Air-con is separated because its trajectory is likely to be opposite to other established demand. Conventional demand is falling with the deployment of efficient electrical equipment. Air conditioning should increase with rising temperatures, living standards and deployment of heat pumps. In many developed countries, it is a significant factor in energy consumption.
For existing homes:
- Number of homes (million) that have (a) cavity walls, (b) solid walls, and (c) lofts, split between insulated and uninsulated. Uninsulated cavities are differentiated by whether they are easy or hard to treat. Lofts are differentiated by whether they are insulated adequately or inadequately, and whether they are easy or hard to treat. Average cost of improvement as well as numbers for each sub-category, and the average saving (%) that can be expected as a result.
- Proportion of homes (%) with pull, partial or no double glazing, and the average costs and savings to be expected from improvements.
- The average heat demand (MWh) of a home before any improvements. This is estimated as the actual heat consumed, rather than the fuel used to produce that heat, to allow for losses in the boiler, because we need to be able to vary those assumed losses according to the mix of heating systems that we specify (particularly for the deployment of heat pumps). This figure should be and is slightly lower than the figures typically specified for home heat demand, which are usually estimated in terms of gas consumption, not heat output.
For new houses and flats (separately) by the date of the Proposed energy system:
- The number that will be built,
- The average FEE Level,
- The average floor area (m²)
- The base cost per m²
- Average space heating, hot water and cooking demand (MWh) per property.
For all (existing and new) houses and flats (separately), expected hot water and cooking demand (split out because building efficiency levels will not affect them much)
Total space heating, hot water and cooking demand (TWh) in the service (commercial and third) sector
Total space heating, low- and high-temperature process heating, drying and other heating demand (TWh) in the industrial sector
The level of heat demand per m² expected for new buildings depending on their FEE Level and type (house or flat)
Market share and efficiency of 13 different types of heating system (direct electric, air-source heat pumps, ground-source heat pumps, solar thermal, biomass boilers, wood fires/stoves, biomass CHP, biogas [AD], biogas [gasification], bioliquids, coal, oil and gas)
For four different types of transport (road, rail, air and water), total energy consumption, electricity consumption, fuel efficiency of non-electric, electrical efficiency of electric, fuel cost (£/MWh) for non-electric, network maintenance cost (£/MWh) and capital cost (£/MW) and vehicle O&M cost exc fuel (£/vehicle for road, £/MWh for others). Electricity consumption modified total energy consumption on the assumption that the total amount of work remains constant, because of much higher efficiency of EVs, e.g. increasing EVs’ share lowers total energy consumption. The relationship can be disabled (inadvisably)
Seasonal Performance Factors (now sometimes referred to as seasonal Coefficient of Performance, sCOP) for air-source and ground-source heat pumps.
Social cost of carbon (£/tCO₂e).

When the user is happy with the values they have chosen for the Proposed scenario, they press a button to Recalculate the hourly values for each component (it was originally regenerated live as values were changed, but the model is so heavy that frequent recalculation is a heavy burden on the user’s computer). The model generates a set of charts (some hourly, some annual totals) that can be displayed or hidden (to save CPU and memory) at the user’s choice. These include:

Retail electricity demand, pie charts of Base vs Proposed scenarios, showing shares and totals for the various electricity uses (used above)
Hourly retail electricity demand, stacked area chart, by same categories
Hourly wholesale electricity demand, line chart, based on the retail demand values plus the specified transmission losses
Hourly inflexible electricity generation (stacked area) vs wholesale demand (line chart). Inflexible electricity is solar, onshore wind, offshore wind, biogas and nuclear. We start from these on the assumption that they will operate at their potential for each hourly period (based on the historic data) so long as their combined output is less than wholesale demand. The aim is to estimate for each hour (a) how much additional demand needs to be met from dispatchable sources, whether flexible generation, storage or interconnectors, or (b) if their combined output would be more than wholesale demand, allocate the curtailment to bring them inline with demand according to the following merit order: solar, biogas, nuclear, onshore wind, offshore wind.
Electricity demand net of inflexibles, line chart, i.e. the hourly net value subtracting the stacked area (inflexible output) from the line chart (wholesale demand) in the previous chart. Shows how much dispatchable generation or inflexible curtailment is required in each hour.
Electricity supply (stacked area) and wholesale demand (line) by type. Types are inflexibles (i.e. the total of the stacked area chart two above), flexibles (all other generating technologies), storage and interconnectors. The merit order is inflexibles, storage, interconnectors, flexibles. Storage is charged if inflexibles output is materially more than demand if the storage capacity still has space, and discharged if inflexibles output is materially less than demand if the storage capacity still has some charge. Interconnector flow is estimated by balancing the residual need to import or export after inflexibles and storage, with the historic flows in that hourly period (which give us some idea of how much our neighbours needed to import or export at that time under those weather and operational conditions). It is a rough approximation, but better than the common approach of just assuming we can push or pull on the interconnectors as we see fit. It’s a two-way street, and the flows are influenced as much by our neighbours’ needs as our own. Unfilled gaps between the area chart and the demand line represent periods when we are unable to meet demand (i.e. demand shedding required). Where the area chart goes above the demand line, there are net outflows to storage and interconnectors.
Annual electricity supply by source, horizontal bar chart. The sources are each type of generation, storage and interconnector. Negative values mean net outflows (for the interconnectors and storage) over the year. Storage is always negative, because the inflows will always outweigh the outflows due to round-trip inefficiency.
Hourly electricity supply by source (stacked area) vs wholesale demand (line). Sources are each type of generating technology, plus hourly net values for storage and interconnectors. Gaps and overlaps between the stacked area and line charts represent supply shortfalls (demand shedding) and outflows respectively, as per two above.
Hourly electricity storage, with both charge level (stacked area chart, left vertical axis) and flows (line chart, right vertical axis) for each storage technology.
Annual marginal storage utilisation. Total storage capacity is divided into tranches (size depending on the total capacity specified). The first tranche shows the annual flows in (charge) and out (discharge) of the storage technologies assuming that tranche received priority for both charging and discharging. The second tranche shows the flows assuming that tranche received priority for whatever flows remain after the first tranche, and so on. This allows one to visualise the declining marginal utilisation of the storage as we add capacity, which is a critical factor in the economics of that storage. Where the annual amount charged for a tranche is close to 365 times the size of the tranche, the storage is probably reasonably economic, i.e. its charge/discharge cycle frequency is close to once a day. Where the ratio is much lower, the storage is probably not viable, unless a storage technology is invented with very low costs per energy capacity (MWh) and reasonable round-trip efficiency. There is currently no such technology at a mature stage of commercial development.
Hourly electricity supply margin (bar chart) vs wholesale demand (line). The supply margin is the amount of unused generating capacity (green where there’s plenty, orange where we’re in danger of a shortfall) or supply shortfall (red) in each hour. This is one of the charts used above. The amount of red and green relative to the wholesale demand line is a good indicator of whether our system has excess or insufficient capacity, or just the right amount. We want mostly green, not orange, because a system that often has just enough capacity will frequently fall over when there is an unexpected interruption to one of the generating units.
Annual margin-level distribution, bar chart. Each bar represents a level of supply margin, in 5% bands. The height of each bar represents the number of hours in the year when the supply margin was at that level. This is a good way to visualise whether we have got our electricity system about right. If our bars from 5% downwards are too high, we have an unstable system with high risk of frequent load shedding or blackouts. If our bars from 50% upwards are too high, we have a lot of excess capacity that has to be paid for without contributing much. We want the vast majority to lie between these values, and very few below 0%.
Annual capacity (or load) factors, bar chart, by generating technology, comparing Base (green) and Proposed (blue) scenarios. Where a blue bar is above the green bar for a technology, this represents new installations of that technology having higher availability factors than the old installations, and not much curtailment. Where the blue is much below the green bar, the technology is running less frequently than before, i.e. fewer hours when it is called in the balancing market. These technologies will be more expensive per MWh as a result. That does not mean they are intrinsically expensive, but rather that they have to cover their costs over fewer MWh.
Annual marginal generation utilisation. Divided into tranches, as per the marginal storage chart above. For each technology, the first tranche shows how many MWh it would produce annually if that capacity was the highest priority for that technology. The second tranche is how much the next chunk of capacity would produce if it had the next highest priority, etc. The values decline as lower-priority capacity is not required to meet demand in some periods. The solid lines show the values for the Proposed scenario. The dashed lines show the values for the Base scenario. Where they drop off on the last tranche, that is just a statistical artefact, not a realistic projection. Declining marginal utilisation (e.g. commonly for gas) indicates the rising marginal cost per unit of the balancing services provided by that technology. As the lines head towards the horizontal axis, we head for a point where the cost of the balancing services they provide is not justified by their level of utilisation. At some level, it will be more efficient to constrain demand than to ensure it can be met every hour.
Annual carbon emissions from fuel combustion, bar chart, by energy-use (electricity, heat, transport), comparing Base and Proposed scenarios. This is only a partial carbon footprint, not including construction and decommissioning emissions.
Annual carbon emissions from electricity, bar chart, by carbon-emitting generating fuel (fossil fuels plus biomass). Again, excluding construction and decommissioning emissions.
Emissions from electricity infrastructure construction, bar chart, by technology (generation, transmission, storage and interconnectors). For the difference between the Base and the Proposed scenario (i.e. the Base bar is zero).
Annual carbon emissions from non-electric heat production, bar chart, by heating technology (fossil fuels and biomass technologies). We assume the carbon emissions for producing the electricity for electric heating technologies are captured in the electricity chart, and exclude them from the heat totals to avoid double counting. Again, no construction emissions in this chart.
Annual carbon emissions from non-electric transport, bar chart, by transport mode. Electricity for transport and construction emissions excluded as per previous chart.
Annual average cost per MWh per technology (generation, transmission and storage), bar chart, Base vs Proposed, plus lines to show the average cost when you aggregate the bars for each scenario. These costs do not count the costs of demand shedding, unlike the following charts.
Annual cost of energy, bar chart, by major component (electricity, heat, transport, and social cost of carbon). Base vs Proposed. Numbers above each bar are the social cost (private cost plus carbon). Numbers labelled “Sys” are total private costs excluding the social cost of carbon.
Annual cost of transport, bar chart, by transport mode plus social cost of carbon (combined for all modes). Base vs Proposed. Numbers at top as per previous chart. Includes electric transport, which is why the figures will not match the figures for transport in the previous chart (all energy) if you have a material amount of electric transport.
Annual cost of heat, bar chart, by type of heating system plus social cost of carbon. All as per previous chart.
Electricity system annual cost, bar chart, by technology (generation, storage and transmission), supply overheads, external cost of demand shedding, and social cost of carbon. All as per previous chart.
Annual electricity system cost per MWh (supplied), bar chart, Base vs Proposed. As per previous chart, with values divided by annual MWh supplied (i.e. net of downstream losses) in each scenario.
Annual electricity system cost per MWh (generated), bar chart, Base vs Proposed. As per previous chart, but divided by annual MWh generated (i.e. gross of downstream losses) in each scenario.
Capital, writedown and carbon costs of infrastructure construction or decommissioning, bar chart, by technology (generating, transmission and storage) to get from the Base to the Proposed scenario.
Hourly cost of electricity generation, line chart, showing average cost (cost of all generation that hour divided by total MWh generated), marginal operating cost (operating cost that hour of the technology required to meet the marginal MWh in that hour; sometimes lower than the average cost because it does not include any amortised capital or fixed-cost components), and marginal net cost (as per marginal operating cost, but allocating capital and fixed costs in that hour’s proportion to total hours operated by that technology).
Annual marginal generation cost per MWh, line chart, by technology, in tranches as per marginal storage utilisation and marginal generation utilisation. Base (dashed) vs Proposed (solid). Shows how costs rise as you add capacity of a technology, according to the extent that utilisation falls (curtailment / load factor based on frequency of being called in the balancing market). As the line heads for the top of the chart, we can assume that it is not worth installing this capacity of this technology to balance the market, and more efficient to shed some demand (balancing) or rely more on dispatchable options rather than inflexible generation (curtailment).
Hourly storage flows (left vertical axis) and electricity cost/income (right vertical axis), line chart, by storage technology, in the Proposed scenario. Each line is a combination of the estimated flows in or out of that technology in each hour, and the marginal cost/value of electricity in that hour. The requirement of viable storage is for the values to be materially higher above the horizontal axis than below it, which would represent the difference between the buying and selling price being greater than the losses due to the round-trip efficiency. If the values are roughly equal above or below, or worse still heavier below the axis, then the storage is uneconomic – it cannot earn enough when discharging to cover its electricity purchases when charging and the operating and capital costs.

To choose the values to set in your Proposed scenario, and/or the charts to display, click the arrow on the right to pop out the menu (and then again when you are ready to close it). When you have chosen your values, click the Recalculate button near the top on the right. You can run the scenario multiple times by selecting the base year and the type of weather that could have been experienced in that year, then clicking Recalculate again.

You can download or print each chart by clicking the “hamburger” button (three horizontal lines) at the top right of each chart.

At the bottom, you can download your scenario as a JSON file by clicking Save Scenario. You can upload it to carry on with those values by clicking the Load Scenario button and selecting the JSON file. You can download an Excel file with some of the data (quite incomplete) by clicking the Download Data button.

This is a private effort and not heavily resourced, so hopefully people won’t hit it too heavily. That said, most of the processing happens on your computer, not the server. You can use it on most systems, but you’d want a decent amount of processing power if you don’t want it to gunge up your browser. It’s javascript-based (React), so you would need javascript enabled.

There are lots of improvements and additions I would like to make. I am unlikely to make them in the near future. I took a large amount of work time to get it to this point, and owe it to my company (Summerleaze) and colleagues to set a limit on that. You can give me your comments if you like, but they are unlikely to turn into actions quickly. Enjoy it for what it is. It should be useful for indicative and testing purposes, but don’t assume that it’s right. “All models are wrong, but some are useful.”