Unlocking flexibility with dynamic price contracts for residential consumers

tl;dr I wrote an interactive application to calculate the impact of switching to a dynamic price contract, and how much you should aim to shift your consumption to reap tangible rewards.

An introduction to dynamic prices

It stands without doubt that flexibility in (residential) electricity demand is one of the key enablers of the energy transition. When electricity is increasingly supplied via renewable (intermittent) sources, the supply side becomes more volatile (yet not necessarily unpredicatable). The necessity to ensure the constant equilibrium between electricity supply and demand will, in such an environment, emphasize the important role of demand response.

Dynamic price contracts are contracts for the supply (or injection) of electricity where a consumer is exposed to the hourly (day-ahead) market price. This requires consumers to have smart, or digital, meters installed. As the unit price paid for electricity (in €/MWh) corresponds to the day-ahead market price, there is an incentive to shift consumption from higher-priced hours to lower-priced hours.

This requires an active approach for consumers:

• Consumers need to timely monitor day-ahead prices. These are known at least 11h (@13:00 D-1) before consumption starts.
• Consumers need to adapt their consumption to this market price signal: shifting from higher-priced to lower-priced hours within the day or across days.
• Consumers need to have a sizeable "volume" of flexible consumption: not all appliances can programmed to run at certain moments, and some electricity use cannot be postponed or anticipated. Electric vehicles, heat pumps and residential storage systems typically can provide this required flexibility.

These elements are important, for a very simple reason. Given that day-ahead prices are established at the intersection of supply and demand, they tend to be higher when electricity demand is higher. Electricity demand is higher when residential consumers require more electricity. As simple as this may sound, this is actually a significant implicit barrier to the integration of dynamic contracts: a consumer with a typical consumption profile will tend to use more electricity during hours where electricity is more expensive (as the consumers' behaviour is actually one of the drivers of electricity prices).

This, in particular, is one of the reasons that the CREG - Commission for Electricity and Gas Regulation - Belgium , as well as me personally, always recommend a cautious approach regarding the adoption of dynamic prices. The average electricity consumer does not necessarily tick off the three elements listed above, and without significant behavioural changes, the impact of a switch to dynamic price contracts may not yield (substantive) benefits.

Such a warning begs some very important questions: how active does a consumer need to be? As of which point can a consumer overcome the problem that his own demand is correlated with aggregated demand, which tends to increase the cost of dynamic contracts? In other words: how flexible does one need to be?

How much flexibility is required in a typical residential consumption profile, in order for dynamic price contracts to provide a significant savings potential?

The answer to this question is as simple as it is useless: "it depends". Actually, it essentially depends on two elements:

• the market prices signal and its evolution (this is known: the hourly day-ahead prices);
• the consumption profile (this is unknown: can be approximated or obtained individually).

To explain the intricacies of these elements, I created a tool to assess rather precisely how shifting more consumption from higher-priced hours to lower-priced hours contributes to a lower (total and unit) cost of energy, and vice versa. This tool is available via https://nicoschoutteet.shinyapps.io/DynamicPriceContracts/

What does the tool do?

Based on the input provided by the user, the tool calculates the total energy cost and the savings' potential of dynamic contracts, by varying the flexibility potential of a residential consumer. The tool can, in its current form, only be used to simulate contracts based on the Belgian day-ahead prices for 2023. In other words, the price of different contracts are assessed from 1 Jan. 2023 until 31 Dec. 2023.

Users can define their desired consumption profile and the share of hourly consumption which can be postponed or anticipated ("the flexibility potential"), as well as the number of hours during which to shift consumption.

By default, a standard consumption profile is uploaded (based on the Belgian real load profiles, provided by Synergrid). Choosing this option allows you to define the total annual consumption (in kWh).

What's most interesting, thoug (at least in my view) is that obviously (and fortunately) not all consumers follow the same load profile. That's why I added the option to upload your own consumption profile. Right now, this is only available for Flemish consumers who have digital meters with quarter-hourly registration activated with Fluvius. You can have a look on Fluvius' web page on how to access and download these data.

This is just a practical choice, though, and a necessary one for me to be able to test the tool with the data from my own digital meter. As time goes by, I intend to provide a template so that other user-defined profiles can be uploaded as well.

What can we learn from this?

The example

In order to demonstrate how the tool calculates the cost of dynamic contracts, a visual example is included. Users can select any of the days in 2023 (below is 28 May 2023), and the first figure automatically shows the hourly day-ahead prices, indicating the highest-priced hours ("H") and the lowest-priced hours ("L"). This is done to mimic the price signal to which consumers are supposed to react.

Secondly, the consumption profile is shown, based on the default profiles from Synergrid (with an annual consumption of 3.500 kWh). The green line reflects the chosen flexibility percentage: during the highest-priced hours, consumption was reduced by 25%. This consumption is then spread evenly across the six lowest-priced (actually negatively priced in this example) hours. The result of this shift is that the daily total actual consumption equals the shifted consumption: there is no shift of consumption to preceding or following days. In order to demonstrate the impact of a wrong shift (from low-priced to high-priced) hours, the red line is added: reflecting a judgement error of consumers, either actively or through the result of a passive approach.

Not everyone has the same consumption profile, though, as evidenced below. In the spirit of full transparency I'm sharing my personal consumption profile, based on my digital metering data obtained from Fluvius. It's been a while, so I don't recall what exactly our household did at 10:00 in the morning, but given that it was a Sunday, probably breakfast? In any case: comparing my personal profile with the default profile shows that 25% for one does not equal 25% for another, and this obviously impacts the profitability of dynamic contracts.

The results

The example above is repeated for every single day of the year 2023. For a default consumer, this leads to the numbers in the table below:

For a default consumption profile (RLP0 with 3.500 kWh), the energy component of a dynamic price contract (without subscription fees, indexation co?ffici?nts or markup) reaches 354,21 €/year, in case consumption remains identical (i.e. no shift). In case 25% of the total consumption is shifted from 6 high-priced to 6 low-priced hours, the total cost decreases to 338,05 €/year. Inversely, when consumption is shifted from low-priced to high-priced hours, the cost rises to 367,17 €/year.

When the selected flexibility percentage increases, the total savings increase. In a similar sense, when this flexibility is applied to more hours (8, or 10, or even 12), total savings increase.

The tool calculates the effect of a 100% shift during the selected hours as well, and shows the possible savings potential during each day of the year, as shown in the heatmap below. This savings' potential can be higher than 100%: when the total daily cost becomes negative by shifting all of the consumption to high-priced hours to negatively-priced hours (the dark green tiles).

Interestingly, the savings' potential is highest during weekend days, or holidays: probably simply because of the high spreads between high-priced and low-priced (or often negatively priced) hours.

The final figure shows the actual answer to the initial question, by calculating the total cost of the dynamic price contract in function of the flexibility potential, in both directions (good and bad), with 10% increments.

It is clear that, the more consumption is shifted in the right direction, the lower the cost becomes. At 100% from 6 high-priced to 6 low-priced hours, the total energy cost is decraesed with ~19% (from 354,2 € to 289,5 €).

Again, for me personally, this is a different story. Given that my consumption profile 2023 was apparently weighted more towards higher consumption during more expensive hours, the savings potential is larger, and the risk of a higher cost (by wrong behaviour) is, relatively speaking, lower, as shown below. Interestingly, the cost for my consumption profile without any shift would reach 355,1 €/year, despite my consumption being only 3.279 MWh. A linear extrapolation from 3.279 to 3.500 MWh would yield a cost of 378,9 €/year, or 6,6% more than a default consumer with the same total annual consumption.

Disclaimer

I have created this tool as a personal contribution to the ongoing debates on residential flexibility and dynamic price contracts. I have tried my best to explain the link between these two elements as neutrally as possible, yet any opinions expressed, coming from an interpretation of the results, is my responsibility only: it does not constitute an official position, nor guidance, of the CREG, nor should it be considered as a formal price comparison tool, such as the CREG Scan or V-Test.

What's next?

Even though this is merely a tool for educative purposes, I definitely intend to see if it can be further developed to reflect more closely the reality, making the simulations more realistic. A couple of things I have in mind:

• Extending the geographical coverage. This is a fairly easy one: uploading all European day-ahead prices and adding a dropdown list in the user interface.
• Linked to the above: providing a template for the user-defined profiles. Currently the tool uses the Fluvius data structure as input, meaning that users outside of the Flanders region cannot upload their data. I have no doubt that all these individual system operators will have different data formats, so I can just generate a template for users to copy-paste their data in.
• Focusing on the six highest-priced and lowest-priced hours does not guarantee consecutive hours: this makes adapting consumption behaviour a bit more difficult (increasing consumption, then decreasing, increasing again, etc.). It would perhaps be more user friendly to identify the six consecutive highest-priced hours as a single "consumption block", to be shifted to the six consecutive lowest-priced hours.
• Implementing more realistic pricing formulas. Obviously I am not developing a price comparison tool, but it is a fact that dynamic tariffs have specific subscription fees, indexation coefficients or markups. One idea would be to allow the users to upload these parameters, to be taken into account in the cost of the energy components. This would better reflect the risk associated to suppliers for purchasing and delivery electricity at certain prices.

That's it. Any other ideas which haven't crossed my mind are more than welcome, I will definitely consider any suggestions for improvements (subject to the reasonability and possibility to fit this into the coding).

As always: please share your comments and questions. This is a fairly new topic to me (and most of us), so there's definitely a lot of room for learning from each other's experiences!

Finally, for those interested: the tool is obviously freely accessible, and I'm happy to share the source code with all of you - just have a look at: https://github.com/nicoschoutteet/DynamicPriceContracts