DEXMA Detect features

Follow

As described in article Discover New Detect,  Detect is considered as the first step in energy efficiency journey: a savings detection tool. It determines which buildings among your portfolio offer the greatest potential for savings and what technologies/energy savings measures  can be adopted to be more competitive on a recurrent basis.

The logic behind this product is founded in a combination of energetic, simulation, and artificial intelligence (AI) models that can be performed massively  and without the need of installing meters or any type of hardware. Let’s look into its insides:

DEXMA Detect architecture

Below you can see an overall view of the system infrastructure. Further details concerning the data calculations can be found in the following document points.

Note: Some of the modules described here are only available for selected sectors & countries. 

Screenshot 2023-08-08 at 12.15.58.png

Data processing

Once the account has been configured in DEXMA’s platform as explained in this article,  DEXMA Detect’s API gathers all the required inputs such as weather, electrical energy consumption and metadata (surface*, prices, tags, zip code, activity…) 

*If  the surface is not configured, DEXMA Detect counts on a module called Surface imputer that estimates the surface of a building  based on a model/ linear regression per activity that has been trained with real surfaces from our database. 

As part of this data processing, DEXMA Detect validates the quality of the data (gaps without data, extreme energy consumption,  surface and activity undefined) and discards those locations that do not match the required criteria. The following list shows the discarded reasons:

  • If the consumption of 12 months is lower than 1500 kWh the location is discarded if the threshold is not reached for any data period.
  • If the trust score (% of days with data per month) of 12 months is lower than 50% the location is discarded if the threshold is not reached for any data period.
  • The location is discarded if its coordinates are not configured.
  • The location is discarded because it doesn’t have any reference device and its sub-locations neither.
  • The location is discarded if it doesn’t have energy active readings.
  • The location is discarded because it doesn’t have surface after the surface imputation is being calculated.
  • The location is discarded because it has a wrong surface after the surface imputation module is being calculated:
    • Lower than 10 m².
    • Median of monthly consumption per square meter is greater than 200 kWh/m².
  • The location is discarded because the geolocation provider cannot get a valid address from the configured coordinates.

Clustering

The clustering module analyzes all buildings with the same electricity consumption pattern to form tightly defined groups of buildings with a similar consumption behavior. 

DEXMA_Benchmarking.gif

Three-dimensional plot of clustering results for a given consumption pattern. Red dot represents a building,orange color represents similar buildings/neighbors to that building and  blue represents other buildings. 

Using state-of-the-art clustering and dimensionality reduction algorithms, this module permits us to only take up to 100 similar buildings into account when comparing and benchmarking their energy efficiency. In order to strike the right balance between data quantity and quality, we utilize the aforementioned consumption pattern hierarchy to ensure a minimum number of buildings for each clustering analysis.

Global energy performance/Benchmarking

The benchmarking module estimates the energy efficiency and savings potential for a given building, based on a comparison within the cluster of similar buildings obtained in the clustering module.

Each building is given an energy efficiency score that ranges from 0 to 100, with 100 denoting a building that is already operating at maximum efficiency. Additionally, for each building we provide a monthly comparison chart of energy consumption values with the average and top-10% of buildings in each cluster. This comparison allows us to evaluate the savings potential for each building, expressed in three ways: energy (KWh), revenue, and as a percentage of the building’s current consumption.

With this information the user can quickly identify buildings with a high margin for improvement as well as the most energy-efficient buildings, and proceed to Dexma Analyse to implement and validate any further efficiency solutions.

mceclip2.png

Global energy performance/ Benchmarking outputs provided by DEXMA Detect

NILM /non intrusive load monitoring* for selected sectors and countries

The Non-Intrusive Load Monitoring (NILM) module breaks down the electricity consumption of a building into different sub-consumptions, effectively revealing what percentage of a building’s total consumption is a result of lighting, heating, cooling, and other devices.

This disaggregation is calculated without using any hardware, so no capital expenditure is required on the client’s part. DEXMA leverages its existing database with tens of thousands of buildings to build statistical models that are then applied to obtain an accurate virtual submetering of a building’s main supply.

mceclip3.png

Monthly HVAC and Lighting energy disaggregation results for a two-year period, computed for  a building from one of DEXMA’s analysis datasets. True consumption is shown in blue, while predicted consumption is shown in orange.

This not only brings more information to the end-customer about the electricity consumption of their buildings, but also allows Dexma Detect to calculate recommendations based on the amount of energy consumed for each disaggregation group (lighting, HVAC, others)* (for selected activities and countries)

Energy conservation measures/recommendations & energy behaviour

Using data from the last month of hourly energy consumption, you will get metrics that analyse how the building has been performing monthly, such as deviations from average consumption and climatic severity, the day and time of the peak consumption, and monthly comparisons. This information complements your electricity bill and helps out to understand your consumption habits and be able to optimise them. Besides the behaviour metrics, energy efficiency recommendations and tips are virtually simulated for your building at a technical and economic implementation level. Each recommendation has a detail of the simulation carried out and has been adapted per country or climatic zone. . 

For further  information please check the following  article about Detect energy conservation measures.

mceclip4.png

Virtual audit recommendations in DEXMA Detect user interface

Reports + User interface

All the outputs of DEXMA Detect are shown through an interface inside DEXMA’s platform and can be exported in 3 ways:

  • Through a branded report that could include your branded colors defined in the platform. Check article.
  • Through a csv file that shows the details of Detect calculations
  • Through a csv file that shows the details of Detect energy measures simulations

mceclip5.png

Example of branded  Detect report 

Other key modules

  • Building energy modeling* for selected sectors and countries
    With the help of building energy modeling, we can simulate how your building might theoretically behave, and use the model for disaggregating your energy consumption, recommendations simulations and compare scenarios.

mceclip6.png

Example of a BEM simulation for an office’s energy consumption vs a measured one in winter and summer. 

  • Heating & cooling type classification* for selected sectors and countries
    With the help of an algorithm Detect can estimate which is the energy source used for heating/cooling systems, that algorithms logic follows one of the following techniques:
    • Correlation between consumption and weather and check if heating or cooling type is ELECTRICAL based on a threshold.
    • Computing a linear regression using HDD as explanatory variable and HVAC consumption as target and check if heating type is ELECTRICAL comparing the regression slope with a threshold.
    • Using a  pre-trained classification models to decide heating and cooling type. This last method has been validated using data generated by building energy modeling. 
  • Exporter * for selected customers
    Under request, Detect is able to export all the reports from each execution to for example an external SFTP, Salesforce.

Was this article helpful?