Demo Site information:
Partner: Aalborg University
Location of the site: Aalborg District Heating
Function of the building: District Heating Network
Implementations: Data from 1664+houses will be included in the analysis.
The initial part of the project focuses on detecting and cleaning both the systematic and random errors, and on structuring and visualizing large datasets coming from smart meters. AAU will apply the tidying methods based on statistical modelling, such as automatic tidying, which includes outliners detection techniques, or macro-editing, in particular distribution method. Different visualisation methods, such as table-plots and carpet-plots will be tested in order to develop best tailor-made automatic editing technique for SM data. The open source statistical language R will be used. It is expected to create a novel and effective method that has wide application in all types of monitoring and modelling data coming from the built sector and related to temperature and fluid flows, e.g. indoor air temperature, air and water flow in HVAC systems.
In the second part, the objective is to decompose the heating usage into space heating (SH) and domestic hot water (DHW). This is a crucial step, since the SH demand can be used to identify primarily features on building characteristics, and the DHW demand features on household and domestic installations characteristics. Secondly, statistical and machine learning techniques, such like advanced regression models and clustering algorithms, e.g. K-Means, will be applied to extract variety of features e.g. building topologies, expected household size, technology used for producing DHW (instantaneous heat exchanger or water tank). The open source statistical language R in combination with Matlab will be used. It is expected that the developed method lays the groundwork for new exploration of the knowledge captured in hourly time series from SM.
In the last part we will combine the heat time series from previous task with weather, GIS and register data, which deliver information on boundary conditions, such like solar radiation, wind speed and direction, building local context, building envelope details, e.g. area of windows, actual number of occupants. By merging this datasets, estimates of the actual fluctuations of indoor temperature and the energy flexibility for each customer can be made. The heat balance equation in combination with recurrent artificial neural networks and hidden Markov models will be used.
Reasons for the intervention: The chosen demo site is a mix of detached and semi-detached houses equipped with the smart heat meters and no additional knowledge about availability of high-tech technologies, smart devices or BEMS systems in the houses themselves is public. Therefore, from the perspective of district heating network, these houses, represent the low-tech end of possible proactive users, which are the vast majority of residential building stock in Denmark and EU. Consequently, the main reason for implementing this demo is to develop new methods and tools that unlock the knowledge on end-users captured in smart heat meter data.
- increase the service level for end-customers by providing tailored-made messages for individual customers and thus facilitating them into being the proactive users
- generate energy and financial savings for both the individual buildings and the district heating utilities
- contribute to the bottleneck-free district operation
- create new value based on data
The District Heating represents excellent replication potential: detached and semi-detach houses represent the 53% of all residential buildings connected to district heating in Denmark. The distribution of the EPCs for these two topologies in Klarup has the same pattern as in Aalborg Municipality and in whole Denmark. Moreover, the installed monitoring technology, namely the smart meters use the standardized logging protocol and thus the developed methodology/tool can be easily applied to analyse the data from various smart meters, which by October 2020 should be installed in every Danish household connected to district heating.