Behind the scenes: Our response

5 min readMar 4, 2023

Introduction

CarbonLaces collaborated with the Sunday Times to publish an article highlighting the potential pitfalls of Energy Performance Certificates (EPCs). Much debate and we welcome this with kindness and positivity. @JanRosenow, said, “I should have been more careful in sharing the Times piece without a deep dive into the methodology”.

So here goes.

CarbonLaces used energy readings from smart meters and compared them to the modelled EPC data. Specifically, we look at primary energy consumption guidance with the energy consumption current for 10.2 SAP with factors of 1.13 for gas and 1.501 for electricity resulting in an average of 170 kWh/m2. We further did an update to compare the metered primary energy consumption using primary energy factors from SAP 9.2 1.22 for gas and 3.07 for electricity, which results in an average consumption of 240 kWh/m2

The difference in the average primary energy consumption across the EPC ratings between metered and EPC models for SAP 9.2 was up to 3X, and SAP 10.2 SAP is up to 4.3X.

Compare this to metered consumption. See below

Comparing SAP 9.2 V SAP10.2 Median primary energy intensity with the factors (stating the obvious)

What data did we use?

Energy consumption of homes is complex; each is unique; some are all-electric, others use off-grid sources (LPG, community heating, etc.), and by far the majority uses the classical gas and electricity mix where gas is used for heating, warm water and cooking and electricity for anything else. We focused on classic gas and electricity sets to compare. The property sample was chosen given the following criteria:

· Each property needed to have a smart meter gas and electricity meter.

· At least 300+ days of meter data needed to be available for both meters.

· Each property needs to have a valid EPC.

· For properties with multiple EPCs, we chose the one that was issued last.

After applying these criteria, our sample size, initially > 60k, was reduced to 17k. We abided by all data governance to ensure data privacy while processing this.

Primary energy consumption

We took Primary energy consumption from the EPC guidance for the write-up. Using this subset of properties, we then went ahead to determine the annual primary energy consumption for each smart meter of a property based on the last n (300 ≤ n ≤ 365) days of half-hourly meter data available and the relevant primary energy factor for the given month as defined in the latest SAP 10.2. So, we have considered both the SAP 9.2 and 10.2 factors with grid carbon, and not much changes.

A note on the comparison of metered data to EPC

We don’t wish to replicate EPCs. Further, we know that people are under-heating from actual metered consumption, sadly even before the energy crisis, and it’s particularly acute now over the last 36 months. People have unique behaviours.

Still, fuel poverty has been a real thing before and has increased exponentially since the current energy crisis. There are no binary outcomes. Further, can we afford to spend £500 for home measurements when 3 million social housing at some level requires an upgrade?

Our analysis was to show the stark difference between primary energy consumption according to EPC and smart meters, even when comparing SAP v9.2 and 10.2.

When using the primary energy factors from SAP v9.2 to compute the primary energy intensity, we obtain a distribution similar in shape to what was obtained using SAP v10.2 primary energy factors, but which is, on average, 70 kWh/m2

CarbonLaces stands by those staggering differences.

So here is a further report from JLL highlighting the same issue, slightly dated from 2012, for Commercial Properties here

Maarten De Groote a Senior Expert at EnergyVille, states the same from the Flanders study.

“The gap between real energy performance and EPC-calculated performance can be significant and is a source of confusion for many. They show a clear correlation between calculated and measured, although with a prebound effect on the worse labels and rebound on the better ones (as such confirmation of many other studies) (Vlaams Energie- en Klimaatagentschap — Ghent University)

Further U.K. research done by SERL using SAP methodology reaches similar conclusions.

http://dx.doi.org/10.31219/osf.io/jn3v6

It worked when EPCs were used to make occasional decisions on fuel poverty and buying and selling. Yet, EPCs are being used for policy, lending, leasing, renting and even pension, as pointed out in this wonderful podcast. EPCs are now used to making decisions they weren’t designed for.

Jonathan Ducker’s post highlights these and further points where single, one-time, isolated measures are the norm.

Isn’t this obsession with the binary analysis with carbon fuelled growth as the only measure of success that got us here in the first place?

CarbonLaces stands by its analysis in the Times and calls on people to dialogue. We welcome talking to BRE, CCC and anyone who wishes to have a constructive dialogue.

It’s 2023, and we must stop fighting turf metric and come together to fight climate change. We see a win-win for all and are ready to help.

Advocating for accessible data

CarbonLaces has released its first dataset, which provides public and private institutions with insights into actual anonymised metered energy consumption by property type, EPC Energy Efficiency Rating, and MSOA. We believe this dataset, along with others that will be released in the coming weeks, will be a valuable tool for institutions to:

· More accurate estimations

· Assess the potential outcomes of retrofit actions.

· Identify potential fuel poverty in MSOAs

· Innovate through open collaboration.

There are numerous ways in which this data can prove helpful, and we encourage anyone with interest in this dataset to contact us at contact@carbonlaces.com.