This is the fourth (and final) blog post in a series about high-resolution carbon accounting. In the first post we introduced the concept and its background. The second post focused on the data processing and modelling required for high-resolution carbon accounting. The third post focused on the superior accuracy of high-resolution carbon accounting and why it matters. This last blog post focuses on the concept of ‘carbon efficiency’.
In a nutshell, high-resolution carbon accounting calculates the emissions intensity of the grid over small time intervals (30 minutes in our case). This is in contrast to the current practice of using an annual average value for the emissions intensity of the grid. High-resolution carbon accounting promises to be a more accurate method to estimate carbon emissions.
What is carbon efficiency?
Most people are familiar with energy efficiency; using as little energy as possible to accomplish a task. Energy efficiency plays an important role in reducing carbon emissions, as less energy used means less carbon emitted (when using the same energy source).
Carbon efficiency is similar but focuses on carbon emissions instead of energy use. The goal is to minimise the carbon emissions generated when accomplishing a task. Improving energy efficiency is one way to reduce carbon emissions. But it’s not the only option. Another option is to use energy with lower associated carbon emissions.
Reducing emissions through time-shifting
When it comes to electricity from the grid, sometimes solar and wind energy is abundant. The emissions associated with electricity imported from the grid at those times are low. At other times—at night, or on cloudy days with little wind*—solar and wind are less prevalent. Consequently, emissions associated with grid imports are higher at those times (because electricity generation is dominated by coal and gas).
To improve carbon efficiency, we want to use more energy from the grid when emissions from electricity generation are low. And less energy when emissions from electricity generation are high. In other words we need to time-shift our energy consumption to reduce carbon emissions.
Not all household appliances lend themselves to time-shifting. Your fridge for example needs to be on all the time. Other appliances are very well suited (e.g. pool pumps, EV chargers, washing machines and dishwashers). Still other loads can be time-shifted to a smaller degree (e.g. air conditioning for pre-cooling or pre-warming).
To be able to benefit from improved carbon efficiency through time-shifting in a carbon accounting context, high-resolution carbon accounting is a requirement. Annual average accounting ins’t suitable as it treats all energy as equal from an emissions point of view.
Simulation
To investigate the potential impact of time-shifting loads for residential sites, we’ll use the same dataset as in the previous blog posts. The dataset consists of grid imports of 488 residential sites over a year (1 July 2021–30 June 2022).
We will simulate all sites time-shifting some of their energy imports from a time the grid tends to have a high emissions factor to a time when the grid tends to have a lower emissions factor**. To do so, we use a heuristic, looking at average emissions intensity per state by time of day. Figure 1 through 4 below show the emissions factor profiles per state.
In these graphs we have highlighted the contiguous 2 hour period with (on average) the lowest emissions intensity. We also highlighted the contiguous 2 hour period with (on average) the highest emissions intensity. When determining these periods we only take “waking hours” (7am – 11pm) into account. Outside those hours most sites in our data set don’t have significant loads to shift ***.
We then simulate shifting a fraction of a site’s grid imports from the period with high emissions intensity (red) to the period with low emissions intensity (green).****
Aggregate carbon efficiency gains
Based on the above emissions intensity analysis, we simulate shifting 50% of the imports in the high-emissions intensity interval to the low-emissions intensity interval. We simply assume that this much load can be shifted. Analysing whether this is feasible based on the specific loads consuming electricity in the high emissions intensity interval is out of scope for this blog post.
Table 1 below shows both the average emissions reduction achieved and the percentage of overall imports that was shifted per state. Figure 5 shows the distribution of emissions reduction across the households in each state.
| Average % of imports shifted per site | Average emissions reduction per site | |
| NSW | 5.4% | 1.1% |
| QLD | 5.6% | 1.0% |
| SA | 7.2% | 3.8% |
| VIC | 5.9% | 0.4% |
We can see the biggest impact in South Australia, where we achieve an average emissions reduction of 3.8% by shifting 7.2% of imports. The largest emissions reduction for a single household was 8.3%. This is in stark contrast to Victoria, where the average emissions reduction is 0.4%. No household achieves more than an 0.8% reduction. New South Wales and Victoria both achieve around a 1% emissions reduction by shifting 5.5% of imports.
Looking at the emissions intensity profiles (Figures 1–4) the results are not surprising. South Australia, with its abundant solar and wind generation, has by far the highest difference between the lowest and highest emissions intensities throughout the day.
However, the emissions reduction associated with time-shifting doesn’t just depend on the difference in emissions intensity between the time-shifting intervals. It’s also a function of the amount of energy that’s available to shift away from the high-emissions intensity interval.
In the next section we’ll dig down into a single site to explore this further.
Single site analysis
Figure 6 shows the average imports profile of a single site in South Australia.
Figure 7 shows the same site’s imports profile after we apply the time-shifting simulation outlined in the previous section. The interval we shifted loads away from is depicted in red and the interval we shifted loads to are depicted in green. This time shift resulted in a 1.6% overall emissions reduction.
It’s easy to see that the interval we shifted load away from has a pretty low volume of imports. This limits the absolute emissions reduction gain that we can achieve.
We might be able to achieve higher emissions reduction by selecting a different interval to shift load from. This will not be optimal from an emissions intensity perspective, but this may be outweighed by the available volume that can be shifted.
Figure 8 shows the imports profile resulting from simulating time shifting with an alternative interval. Again, the interval we shifted loads away from is red and the interval we shifted the loads to are green. This time shift resulted in a 3.9% overall emissions reduction, more than a doubling.
Forecasting
The above simulation showed that we can reduce carbon emissions by time-shifting energy consumption. We note, however, that the simulation was far from optimal. We used the average emissions factor by time of day over the whole year to determine our time-shifting intervals. Effectively we forecasted every day to have the same emissions-factor profile.
In reality the emissions-factor profile changes day-to-day, depending on many factors, most importantly the weather. A more accurate model to forecast the emissions factor of the grid over time would allow for more optimised time-shifting and a bigger reduction in carbon emissions associated with grid imports.
Conclusions
High-resolution carbon accounting opens up another pathway to reduce carbon emissions beyond energy efficiency. It’s also an incentive to use energy when it’s clean and abundant.
It’s possible to improve the carbon efficiency of imports from the grid by time-shifting from intervals where the grid has a higher emissions intensity to intervals where the grid has a lower emissions intensity.
The simulated time-shifting presented in this article led to modest emissions reductions, based on shifting a modest percentage of imports. Emissions reductions were most significant for South Australia (3.8% emissions reduction by time-shifting 7.2% of imports), as its large penetration of renewables in the grid results in significant periods of low emissions intensity. Opportunities for emissions reductions from time-shifting will improve in other states as more renewables enter the grid in coming years.
This article focused on time-shifting loads from a single time interval to another single interval. In real world applications there is no need for this limit. Load-shifting from any period of higher emissions intensity to a period with lower emissions intensity is beneficial. Moreover, the simulation was based on a very crude heuristic. Forecasting the emissions intensity of the grid more accurately to determine intervals to shift from and to will lead to bigger emissions reductions.
In short: in settings where a reasonable percentage of loads is shiftable, time-shifting combined with high-resolution carbon accounting can be a worthwhile way to reduce emissions today (e.g. as part of a net zero or real zero strategy). And with more renewables entering the grid in coming years, it will only become a more exciting proposition.
Adriaan Stellingwerff is a Senior Software and Data Engineer at Wattwatchers.
* There’s (of course) a German word for this: Dunkelflaute
** In this simulation we are purely time-shifting grid imports. That is, imports shifted from their original time interval will result in imports in the destination interval. We don’t analyse whether there is excess generation from rooftop solar available at the destination time interval.
*** Typically, electric hot water systems are one of the only significant overnight loads that are shiftable.
**** This is a theoretical analysis. We assume that we can shift a percentage of electricity imports, without checking the specific loads that consume the energy. In practice not all loads can be shifted. E.g. energy consumption from a fridge can’t be shifted, there’s usually only limited scope to shift air conditioning, etc.
ACKNOWLEDGEMENT
The My Energy Marketplace project received funding from ARENA as part of ARENA’s Advancing Renewables Program. The views expressed herein are not necessarily the views of the Australian Government, and the Australian Government does not accept responsibility for any information or advice contained herein.
