Conference Agenda

Session
Session T3.2: Ensuring high quality building simulations
Time:
Thursday, 02/Sept/2021:
13:30 - 15:00

Session Chair: Dorota Brzezińska, Lodz University of Technology
Session Chair: Ralph Evins, University of Victoria
Location: Cityhall (Belfry) - Room 2


External Resource: Click here to join the livestream. Only registered participants have received the access code for the livestream.
Presentations
13:30 - 13:48

FMI Co-Simulation between 2D/3D component models and HVAC/control models

Andreas Nicolai, Andreas Söhnchen

TU Dresden, Germany

Aim and Approach

(max 200 words)

Detailed construction and building component models, including hygrothermal porous material transport models, can be used to model a large variatiy of modern energy transfer and storage systems. These include heated concrete slabs, shallow soil heat collectors, heated wall layers, combined photovoltaic and construction panels etc. The interaction with connected HVAC systems/energy distribution systems and/or control models is, however, often limited in such tools. Using three distinct application cases, the article describes the extension of the hygrothermal transport model DELPHIN with an FMI co-simulation interface and the setup of a coupled simulation with external models. The article covers the some details of the implemenation, but also derived best-practice approaches on FMI co-simulation algorithms and parameters, suitable for the tested application scenarios.

Scientific Innovation and Relevance

(max 200 words)

Description of tested, practice oriented co-simulation cases, using a detailed sub-model (building component model/hygrothermal transport model) and external control/energy distribution model. Description and demonstration of consequenced from Co-Simulation algorithm and parameter selection. Derivation of best-practice recommendations for setting up such coupled simulations, and defined interfaces and coupling parameters.

Preliminary Results and Conclusions

(max 200 words)

Co-Simulation support and use successfully demontrated in quite different use cases. Illustration, that supporting even the most basic variant of the FMI co-simulation standart enhances functionality of isolated models significantly. Discussion of co-simulation parameters show, that for application cases in building energy simulation/energy supply system simulation, a simple Gauss-Seidel co-simulation type with moderate time step selection is usually sufficient.

Main References

(max 200 words)

FMI co-simulation standard

Nicolai, A. and Paepcke, A.Entwicklung der Kopplungstechnologie von Komplexmodellen für Bauteil-, Raum- und Gebäudesimulation mit Modelica-basierten Anlagen-, Regelungs- und Nutzermodellen, 2018, Technischer Report

Nicolai, A. and Paepcke, A.; Co-Simulation between detailed building energy performance simulation and Modelica HVAC component models, 2017, 12th International Modelica Conference, Prague



13:48 - 14:06

Disaggregation of digital meter data for synthetic load profile generation

Toon Bogaerts2,3, Stef Jacobs1, Sara Ghane2,3, Freek Van Riet1, Wim Casteels2,3, Siegfried Mercelis2,3, Ivan Verhaert1, Peter Hellinckx2,3

1Energy and Materials in Infrastructure and Buildings, University of Antwerp, Belgium; 2IDLab, University of Antwerp, Belgium; 3Imec, Belgium

Aim and Approach

(max 200 words)

Building simulations require accurate Synthetic Load Profiles (SLP) of electricity consumption to research interaction with the grid or coping strategies for appliance-induced overheating. This is only possible by means of empirically validated user behavior profiles, i.e. based on in situ measurements. Logging for each single appliance separately is, however, expensive and labor-intensive. This means that centrally measured data should be disaggregated into the data of individual appliances.

The aim of this research is therefore to evaluate two event-based Non-Intrusive Load Monitoring (NILM) techniques for data disaggregation for appliance recognition: classification trees and timeseries analysis using deep learning . Moreover, the compatibility of these both techniques with low temporal resolution of the measurements is verified. Finally, based on the results, the applicability of the NILM-techniques on the digital energy meters in Belgium is discussed.

A public labelled dataset is considered as case study. The dataset contains one-week measurements with state transitions of the individual appliances as labels. The dataset is separated in order to train, validate and test the NILM-techniques.

Scientific Innovation and Relevance

(max 200 words)

According to the European Commission, digitalization of the energy system is a necessity for the transition towards a sustainable future. In this context, the roll out of digital energy meters in Europe has started. The next step is to translate the digital meters into “smart” meters. While great efforts have been made in previous research -including to the level of commercially available tools-, existing sources lack objective evaluation of disaggregation techniques. Therefore, no reliable tools exist for central measurements for generating Synthetic Load Profiles.

Therefore, this paper discusses the development and evaluation of different NILM-techniques for user behavior profile applications. Moreover, it compares different temporal resolutions to take into account different read-out frequencies. Indeed, e.g. only the more recent digital energy meters in Belgium are equipped with the high-frequency read-out S1 gate, while previous versions are limited to a frequency of 0.1 Hz.

To establish disaggregation of appliances, we look to use beyond the state of the art time series classification techniques such as appliance fingerprinting, timeseries classification and feature extraction combined with classification trees. More specifically, the use of Long Short Term Memory cells will be used to analyze time series.

Preliminary Results and Conclusions

(max 200 words)

Preliminary research shows that an optimized decision tree classifier is able to identify appliances in a similar fashion as statistical methods. Feature engineering improved the performance of the tree. We look to further improve these result by using recurrent neural networks such as LSTM’s to extract complex features from the timeseries data. These features can be passed to a fully connected multilayer perceptron classifier to distinguish the different appliances. Furthermore, the influence of the sample rate did not affect the decision tree. This relation will be further analyzed to estimate the importance of the s1 gate. Finally, we will look into the possibility of continues labeling of timeseries with a relation between the hidden state of the recurrent neural network and the states of appliances. With these techniques we look to achieve 5%-10% better performance in means of accuracy beyond statistical methods.

Before the discussed algorithms can be applied for the generation of Synthetic Load Profiles, future work should focus on their applicability on buildings with other user behaviour and appliances than represented by the used dataset.

Main References

(max 200 words)

• Anderson, K. et al., BLUED: A Fully Labeled Public Dataset for Event-Based Non-Intrusive Load Monitoring Research, in proceedings of ACM SustKDD'12, 2012

• NGUYEN, M., et al. A novel feature extraction and classification algorithm based on power components using single-point monitoring for NILM. In: 2015 IEEE 28th Canadian Conference on Electrical and Computer Engineering (CCECE). IEEE, 2015. p. 37-40.

• KELLY, Jack; KNOTTENBELT, William. Neural nilm: Deep neural networks applied to energy disaggregation. In: Proceedings of the 2nd ACM International Conference on Embedded Systems for Energy-Efficient Built Environments. 2015. p. 55-64.



14:06 - 14:24

Measurement of the building envelope thermal performance in collective housings

Lorena de Carvalho Araujo1,2, Simon Thébault1, Laurent Mora2, Thomas Recht2

1CSTB, France; 2Univérsité de Bordeaux, France

Aim and Approach

(max 200 words)

Building energy efficiency is a key factor in reducing CO2 emissions and assuring the comfort level for inhabitants. Governments have been valorizing the energy performance standards through thermal regulations and economic incentives. These are often based on results from building simulation softwares, achieved during building design stage. However, the real thermal performance can significantly deviate from the predicted one [1]. It is important to have reliable performance indicators to assure new building quality and to estimate the gains accruing after renovation works. The application of an in-situ method after construction or retrofitting phases enables the measurement of such indicators, as the whole heat loss coefficient (HLC) [2] and the heat loss coefficient by transmission (Htr) [3]. Collective housing counts for an important part of building stock, for this reason, mature technologies to measure its thermal performance are necessary. The current paper studies the applicability of a short duration test for identifying the HLC and Htr in collective housings and how to optimize the test protocol.

Scientific Innovation and Relevance

(max 200 words)

There are different available methods in the literature for measuring the building envelope thermal performance (for instance: average method, energy signature, PSTAR, EBBE, co-heating, ISABELE, QUB, and others) [2],[3],[4],[5],[6]. They present variations concerning the mathematical approach, the duration, the protocol modalities and the applicability [7],[8]. Among those, methods like energy signature and EBBE can be applied to collective housing. However, they use static models, presenting long measurement periods that usually lasts for more than one season extending upto a period of few years. Presently, there are not short duration tests that have been validated regarding the HLC and Htr estimation quality for this building typology. The propose of the current paper is to study a dynamic approach using grey box models, that allows the reduction of test protocol for identifying the thermal performance of collective housing’s envelope. In addition, variations of the test protocol, allows the study of optimal test conditions applied virtually to a medium collective housing. Furthermore, the relevancy of test protocol is verified by its application in-situ in a real building. This research suggests an alternative to evaluate the whole building heat loss coefficient of collective housings, with measurements duration shorter than one week.

Preliminary Results and Conclusions

(max 200 words)

We modelled in a thermal dynamic simulation software (Pléiades + COMFIE) a collective housing from the Residence Figuières Vignettes in Feyzin, France. It presents 1300 m² divided in four floors and sixteen apartments, 21 thermal zones and the thermal properties of components are in a level of a retrofitted building. 336 variations from a protocol inspired by ISABELE method were applied to this model in order to study the impact of several key parameters of the protocol (duration, heating power, set point temperature, preheating) in the quality of Htr estimation.

The test duration and the temperature difference from the beginning and the end of the test were the most influent parameters in the quality of the Htr indicator. For durations equal or superior to four days of measurement, the tests presented a Htr bias inferior to 15% for moderate internal temperature variation.

Besides the virtual experiments, we applied a test into a real collective housing composed of three apartments located in Sallanches, France. The overall heat loss coefficients level of this building was measured using the SEREINE method for a period of one week. After two days of test, the results are stable and present an uncertainty inferior to 15%.

Main References

(max 200 words)

[1] Wang, Liping & Mathew, Paul & Pang, Xiufeng. Uncertainties in energy consumption introduced by building operations and weather for a medium-size office building. Energy and Buildings. (2012).

[2] Bauwens G. In Situ Testing of a Building’s Overall Heat Loss Coefficient – Embedding Quasi-stationary and Dynamic Tests in a Building Physical and Statistical Framework. Doctoral Thesis. (2015) .

[3] Thébault, Simon. Contribution à l’évaluation in situ des performances d'isolation thermique de l'enveloppe des bâtiments. Thèse de doctorat. (2017).

[4] Cohen M. et al., EPILOG Livrable n° 2 Rapport de synthèse sur la méthodologie employée avec tests sur cas d’étude théoriques par simulation. PACTE. (2017).

[5] Nordström, Gustav, Helena Johnsson, and Sofia Lidelöw. "Using the energy signature method to estimate the effective U-value of buildings." Sustainability in Energy and Buildings. Springer, Berlin, (2013).

[6] Wingfield, Jez, et al. "Whole house heat loss test method (Coheating)." Leeds Metropolitan University (2010).

[7] IEA Annex 58. Reliable Building Energy Performance Characterisation Based on Full Scale Dynamic Measurements. (2015)

[8] Roels, Staf, et al. "On site characterisation of the overall heat loss coefficient: Comparison of different assessment methods by a blind validation exercise on a round robin test box." Energy and Buildings 153 (2017).



14:24 - 14:42

Cooling demand reduction approaches for typical buildings in a future city district in mid-Sweden

Sana Sayadi, Abolfazl Hayati, Jan Akander, Mathias Cehlin

Universuty of Gävle, Sweden

Aim and Approach

(max 200 words)

The increase in population and living standards, as well as global warming and heatwaves due to climate change, have created a challenge to meet the cooling demand in buildings. Using currently available sources of energy endangers future energy security[1]. Therefore, implementing new approaches to reduce energy requirements in buildings to pave the path for energy transition is an area of interest. This study aims to analyze and minimize the cooling requirement for a multifamily building through simulations in a new city district in mid-Sweden. Buildings must meet the Near Zero Energy Building (NZEB) requirements based on the new Swedish National building regulations [2]. This study first explores the cooling demand of the building by means of simulations with IDA Indoor Climate and Energy (IDA-ICE) software, then investigates the effect of different mechanical ventilation strategies, window properties and orientations. The characteristics are aligned with Key Performance Indices (KPIs) which are based on the proposed list from IEA Annex 80: Resilient cooling of buildings. Climate files of normal and extreme conditions are considered for the simulations [3]. After implementing the changes in the building, results and their effect on cooling demand is investigated.

Scientific Innovation and Relevance

(max 200 words)

Fulfilling the latest building regulations and implementing the most energy-efficient characteristics in the building, aligned with Annex 80’s proposed KPIs help meeting the NZEB requirements. Performance of the multifamily building with focus on robustness and resilience for a future city district has to be considered. Implementing optimum building and window specification and using different climate files help fulfilling the future resilient NZEB buildings. Today’s residential buildings, mainly in Sweden, have been designed to fulfill heating requirements but are seldom designed and equipped with systems for space cooling. This study investigates the future cooling requirement in terms of building regulation requirement and future climate conditions containing heat-waves.

Preliminary Results and Conclusions

(max 200 words)

The cooling demand is expected to rise as the climate changes, therefore, buildings should be resilient to the future heat-waves. The chosen model meets NZEB requirements by implementing different optimized characteristics of a building aligned with Annex 80’s KPIs and Swedish National Building regulations. The results envision the minimum cooling demand through optimum combination of the building’s specifications in the new city district in mid-Sweden to provide the required comfort for the residents

Main References

(max 200 words)

[1] Z. X. Jing, X. S. Jiang, Q. H. Wu, W. H. Tang, and B. Hua, “Modelling and optimal operation of a small-scale integrated energy based district heating and cooling system,” Energy, 2014, doi: 10.1016/j.energy.2014.06.030.

[2] Boverket, ”Konsekvensutredning BFS 2020:4 Boverkets föreskrifter om ändring i verkets byggregler (2011:6) – föreskrifter och allmänna råd, BBR, avsnitt 5 och 9”, [In Swedish] Report on proposed NZEB building regulations version BBR 29, 2020.

[3] A. Machard, C. Inard, J.-M. Alessandrini, C. Pelé, and J. Ribéron, “A Methodology for Assembling Future Weather Files Including Heatwaves for Building Thermal Simulations from the European Coordinated Regional Downscaling Experiment (EURO-CORDEX) climate data,” Energies, 2020, 13 (13), p. 3424, doi: 10.3390/en13133424.