Home Solar Batteries: 2.2d Demand - High Resolution Model

Home Solar Batteries - Modelling Performance

Introduction

This post specifically addresses the charging of domestic battery systems using solar PV and forms part of a series of entries looking to establish a reasonable view as to how the combination of solar PV and battery storage system should be modelled to ensure that information available to consumers should be considered accurate enough to establish a reasonable justification for investment in storage technologies.

If this is the first time you've accessed this information it is recommended that all of the relevant posts are considered in order through accessing the 'relevant posts' links

High Resolution Demand Model

In this series we've been looking at household energy consumption on progressively increasing data resolutions from basic annual through to hourly totals to assess their effectiveness in driving an accurate model of battery performance.

Previously we looked at HHM (Half Hourly Metering) provided by smart-meters as our demand data source, which to almost any standard household, would be the most accurate source of consumption available, however, just as has been the case in all other period we've considered, doing so would still smooth variability to a level which would often create significant levels of inaccuracy.

Let's elaborate on this through example:

When previously looking at the variability of solar PV generation it was apparent that any comparative model would need to include provision for the high level of variation which occurs within any particular day and between days within a given period.

Home Solar Battery - Model Intraday Generation Variability

The chart above was previously introduced as a representative of generation for a spring month (April), so we'll take this month as our source for looking at demand, but to see detail we'll zoom in on a couple of consecutive days with typically different demand patterns in order to see the data in higher resolution.

So as not to concentrate on the generation variability this exercise will look at the first Friday & second Saturday in the month, that's the 7th & 8th day shown in the chart, and analyse the generation against demand in 5 minute intervals over the entire 48 hours, which results in the following output.

Home SolarPV with Battery Performance Modelling - High Resolution Data

In the chart above we can clearly see that generation (green) on the Friday morning was impaired by clouds until mid morning when conditions obviously changed to intermittent sun, then in the afternoon the sky looks to have cleared apart from the odd small cloud. The bell curve representation of generation on the Saturday shows that the sky was clear of clouds almost all day.

When we analyse demand data (red) at this level of detail we can capture the impact of individual loads and switching and compare them to the available solar generated energy by time-slice, thus obtaining an accurate view of excess solar PV generation which is available to export or divert to battery storage as represented by the grey trace where positive data equates to the level of energy export with and negative showing import requirements.

We'll now take a little time to look at both days to see what the data could be saying.

Friday

The early spikes on Friday morning showing as between 225 & 250 Wh within a 5 minute period look to be a kettle boiling for morning hot drinks & maybe the use of a toaster or hairdrier.

Following a period of early consumption there seems to be a period of little electricity demand apart from a regular pattern of background load switching, which is likely to be the result of a refrigerator & freeze cycling. In comparison to other daytime periods there seems to be a lack of high regular daytime loads, which leads to the conclusion that the property was probably unoccupied for a few hours.

Around midday the energy consumption pattern changes, suggesting someone has returned home, including at least one lunchtime hot drink. Later in the afternoon we see a period with initial consumption of ~450Wh within 5minutes, so somewhere around a 5.4kW load, before settling to approximately 3kW for a longer period, certainly a pattern that looks like like meal preparation and cooking to me.

Through the evening it looks like the peaks & troughs of the regular background energy consumption pattern, which we've attributed to refrigeration cycling, have been raised, so it looks like evening lighting & a night in front of the TV was enjoyed by someone in the household, they even look to have had a final hot drink before bedtime!

Saturday

The energy usage pattern on Saturday is completely different as would by expected in many households at weekends. We can still see the short high spikes which likely result from boiling kettles for hot drinks, and the morning seems to also have additional spikes of around 1kW load, which likely result from small kitchen appliances of one form or another.

The period of raised energy consumption in mid morning suggests that a reasonably high but variable load is in use for a couple of hours, maybe suggesting a dishwasher, washing machine or something similar. It's also clear that in late morning the background level of energy consumption increases by a few hundred watts, before dropping back again in mid afternoon, so something different is happening here, possibly multiple devices, PCs, TVs etc for 3-4 hours?

Later in the afternoon we see a long period of raised demand. Would this be consistent with preparing & cooking a later evening meal or baking for later in the week, with the secondary consumption spike maybe the use of a dishwasher?, but whichever the case, it looks like a hot drink was enjoyed not long afterwards!

48 hour Household Energy Summary

Within the 48 hours used in example period described above the following totals applied

Solar PV Generation 42.8kWh

Total Energy Consumption 19.2kWh

Net Import/Export 23.6kWh

Export 32.2kWh

Import 8.6kWh

Summary

When looking at analysing solar PV generation against consumption using all previous data timescales, whether annual or half hourly, there has always been an issue with the effect of data averaging, however, as demonstrated above, the ability to perform energy analysis of in 5 minute buckets provides an interesting solution.

The ability to capture historical actual consumption data or simulate demand to a resolution which is capable to show individual high demand loads such as a single kettle being boiled or variations in baseline demand resulting from the cycling of refrigeration appliances should provide a level of data detail for a model to create a pretty realistic estimate of a combined battery storage and solar PV installation's performance and thereby form the basis for relatively accurate financial justification before committing to purchase a storage solution.