Feedback on a data analysis script that enabled energy savings during Covid-19 lockdown
Published on 26 Jun 2020
The health crisis caused by Covid-19 has affected the whole world, practically taking everybody by surprise. Following the announcement of the lockdown, non-essential stores along with numerous corporate offices had to close. Giulia Caputo, Data Scientist at Deepki, explains to us how she and her team were able to detect potential net savings of 123 000 € per week at more than 3 500 buildings courtesy of data analysis during this exceptional period.
Identifying consumption anomalies during lockdown
Backed by its experience in data, Deepki’s R&D team decided from the very beginning of lockdown to design an analysis pipeline in order to automatically detect sites that were presenting abnormally high consumption despite their closure.
“For the R&D team, lockdown presented itself as an interesting situation from the intellectual and scientific point of view. We immediately saw this as an opportunity to allow our clients to avoid wasting energy while proving to them the importance of regulating their equipment.” – Giulia Caputo, Data Scientist at Deepki
Adopting a pragmatic and efficient data methodology
Accustomed to collecting and analyzing its customers’ data in order to reduce their environmental impact, Deepki’s R&D team has been mobilized from the very beginning of lockdown in order to alert its clients as quickly as possible in the event of any consumption anomalies.
“Since our core business is the collection and analysis of large data volumes, everything was already in place to enable us to detect poorly regulated sites in their stock. We were therefore very swiftly able to put forward a response to our clients’ emerging needs as they faced lockdown.” – Giulia Caputo, Data Scientist at Deepki
In order to achieve this, it was first necessary to characterize sites’ consumption by asking the following questions: What is normal consumption for this site? And what is to be considered abnormal?
To do this, Giulia and her team chose to use load curves – that is, the power demand every 10 minutes – which are recovered automatically thanks to the API provided by Enedis.
Glossary API (Application Programming Interface): privileged communication channel allowing connections to be established between several software programs for the purpose of exchanging data.
The team then worked to determine the pre-lockdown and post-lockdown consumption levels for each building in order to calculate the difference in consumption between the two situations.
With the aim of determining a target consumption level for each site during the lockdown period, the R&D team used the consumption recorded at night between 2 am and 5 am in the month prior to lockdown – since the buildings were not expected to be consuming much energy at those times.
The level of residual consumption thus determined then corresponds to the level of consumption achievable at any time of day during lockdown for sites that are not accessed by any visitors.
“The use of data analysis algorithms allows us to detect anomalies at thousands of sites in the blink of an eye.” – Giulia Caputo, Data Scientist at Deepki
Finally, based on the principle that any consumption above the pre-defined “minimum acceptable level of consumption” can be reduced, the R&D team calculated a wattage ratio before and after lockdown in order to identify the sites with poor regulation, and thus the energy and financial gains to be made. At more than 3 500 sites, 40% of them had not adapted their consumption level during the lockdown period.
Glossary Minimum acceptable level of consumption: power level corresponding to residual consumption outside periods of activity, for example at night when a store is closed. Remember, some equipment sometimes needs to be kept on, such as freezers or data centers.
Load curve presenting a normal “minimum level of consumption”, then an abnormality on 20/11/2019
Realizing energy and financial gains during lockdown
The conducted survey in figures:
Number of buildings monitored: 3 500
Number of projects concerned: 17
Gains detected: 123 000 euros excl. tax/week of lockdown
“The exceptional lockdown situation that we have gone through has allowed us to perform a study that clearly demonstrates to our clients the importance of correct regulation.” – Giulia Caputo, Data Scientist at Deepki
The health crisis due to Covid-19 is an opportunity to question your energy management and rethink your transition to sustainable real estate. Indeed, it is now – more than ever – necessary to avoid waste by making use of data and artificial intelligence.
Zero carbon, a critical ESG challenge for the Real Estate sector
Faced with ever-increasing pollution and greenhouse gas (GHG) emissions, the world is currently in a phase of global warming. The question is no longer how to avoid this trend, but how to curb it and adapt as well as possible. We’re also seeing more pressure from the market, where there’s a growing awareness of issues around ESG and an increasing focus on assets’ extra-financial value.
Resilience: Confronting Climate Change and Adapting Your Real Estate Asset Management Accordingly
The real estate sector accounts for more than a third of CO2 emissions in Europe, yet only 1% of building inventory is renovated each year. This reality has led to growing market awareness, with the emergence of new European and national environmental regulations.
The rise of ESG: the evolution of environmental regulations in Europe
The real estate sector in Europe accounts for nearly half of the region’s energy consumption and more than a third of CO2 emissions. Despite this reality, only 1% of real estate inventory undergoes renewal each year. Faced with the climate crisis, European countries have set themselves a common goal: achieving carbon neutrality by 2050.
