Design tool to make network tariffs more cost-reflective

This article follows on from our earlier articles here, here and here, and firstly describes a simple tool that can be used to visualise a tariff’s level of cost-reflectivity. It then uses the tool to illustrate the steps that can be used to make a tariff significantly more cost-reflective.

This work, as well as the views of other experts, will be presented at the APVI’s workshop ‘Cost-Reflective Pricing – some different perspectives’ on Wed 1 June.

Figure 1 shows the visual tool, which compares the unitised demand charge (y axis) with the customer demand at the time of the network peak (x axis). Where the demand charges have been unitised, they have been converted to an equivalent kW value.[1] We do this so that there is a more direct visual correlation between what a customer pays and the costs they impose on the network, and because it makes different demand tariffs easier to compare.

All the households above the red line would receive a unitised demand charge that is greater than their demand during the network’s annual peak. Thus, the correlation coefficient between payments under the demand charge and responsibility for the network peak is very low.

[1] In this case, the net annual demand charge was taken to be $1/kW but the months that have a higher demand charge are given a proportionally higher weighting. The addition of all the monthly demand charge rates equals 1. The unitised demand charge does not include any usage (c/kWh) charges or any separate fixed ($/day) charges.

Figure 2 shows the aggregated half hourly demand for the 3,876 households from the Smart Grid Smart City database for 2013. Although the annual peak is in summer, of the top 50 days, only six days are in summer 1^st, 2^nd, 5^th, 11^th, 24^th, and 27^th). All the rest are in winter. This means that if only the summer days are targeted in the demand charge, and customers respond, the winter days will quickly become the peak days.

Of course, if the network has already been sized to meet the summer peaks, the winter peaks will not be a problem – unless demand in general increases, or, if there is a genuine desire to reduce customer costs and so reduce the size of the network over time.

Figure 2. Annual aggregated SGSC load profile

In this case, the unitised demand charge should not be compared to a single network peak. In other work we have calculated that the correlation coefficient is greatest when the unitised demand charge is compared to each customer’s averaged demand during the first five network peaks (for this SGSC dataset). Figure 3 is equivalent to Figure 1 but for this five network peak comparison, and it can be seen the correlation is indeed better.

Figure 3. Unitised Standard Demand Charge vs Average Demand at Time of Five Highest Network Peaks

The correlation is further improved when the demand charge is applied only during the summer and winter months instead of every month of the year – Figure 4.

Figure 4. Unitised Demand Charge (applied only in summer and winter months) vs Average Demand at Time of Five Highest Network Peaks

Then if the demand charge is applied to the household demand at the time of the network peak (as recommended here) in each of the 12 months, the correlation improves further – Figure 5. Note that although residential network peaks are almost exclusively between 5pm and 7pm, a customer’s demand peaks can be at any time of the day (with two thirds of the SGSC households being outside the 5-7pm window). This means that a customer can be told when the network peak is likely to be, but will have little idea when their own peak is likely to be, and so it is easier for them to take action if the demand charge is based on the time of the network peak.

 

Then if the demand charge is applied to the household demand at the time of the network peak but only during the summer and winter months, the correlation improves further – Figure 6.

 

The tariff being assessed here included a minimum 1kW demand charge in each month, which acted as a proxy fixed charge, and if this is removed we get Figure 7.

Thus, for the load data used here, it would appear that the cost-reflectivity of the demand charge component can be improved significantly by simply applying it to the customer demand at the time of each summer and winter month’s network peaks.

Some points:

• From the customers point of view this would look like a standard demand charge tariff applied only during summer and winter.
• This analysis assumes that customers will respond to a demand charge and so decrease the highest network peaks.
• This design would work equally well with a ‘rebate-based’ tariff
• This process can be applied to any types of load profiles, no matter which season they peak in, it is simply a matter of applying the tariff periods to the appropriate seasons.

Rob Passey is a Senior Research Associate at the Centre for Energy and Environmental Markets (CEEM) at the University of NSW, Policy Analyst at the Australian PV Institute and Senior Consultant at IT Power (Australia).

Navid Haghdadi is a PhD Candidate at CEEM UNSW.

This work was part-funded by the Energy Consumers Australia.