Conference Agenda

Session
Session T5.2: Ensuring high quality building simulations
Time:
Thursday, 02/Sept/2021:
17:00 - 18:30

Session Chair: Andreas Nicolai, TU Dresden
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
17:00 - 17:18

A parametric combined design and simulation tool for the optimisation of domestic hot water systems in residential care centres

Martijn Holvoet, Elisa Van Kenhove, Klaas De Jonge, Lien De Backer, Wim Boydens, Jelle Laverge

UGent / Archipelago

Aim and Approach

(max 200 words)

In this master dissertation research is conducted into the proper design and dimensioning of Domestic Hot Water (DHW) systems in public building

typologies. A basis for the optimisation of the design is created through the development of a flexible and parametric simulation tool in Modelica that can be used in the design phase. An extensive design study and research

into parameters and characteristics concerning the building, system and demand leads to assumptions of average values and data to fill in the information that is still unknown at this stage. Together with the integration of the general methods of dimensioning, the simulation tool can function with only a limited number

of input data and can therefore be used in the design

phase of DHW systems in public buildings.

Scientific Innovation and Relevance

(max 200 words)

The tightening of energy performance requirements effectuates the ever-improving energy efficiency in buildings. Among others, the insulation level and the airtightness of the building envelope have evolved considerably. Only in the field of DHW, there is comparatively little innovation. The energy demand for DHW remained nearly unaltered over the years and consequently starts to represent an important share of the total energy demand of buildings. It is a domain with little research and innovation, certainly concerning public building typologies. Regarding the building design, more attention should be paid to the optimisation of DHW systems. As there are many degrees of freedom concerning the configuration of these systems, tools that allow to make an informed choice for the specific configuration of the DHW system at an early stage can be of great importance. In addition, nowadays, also indicators such as energy use play a crucial role, where previously only comfort related aspects were considered as important. Concretizing performance factors of certain design proposals in the early phase can be useful. Normally, dimensioning is only done at a later stage of the design, at a time when modifications to the design already entail major implications and are therefore no longer possible.

Preliminary Results and Conclusions

(max 200 words)

A parametric simulation tool for DHW systems that can be effortlessly used in the design phase and requires only a few input parameters, is developed and it can have the potential to be of great importance in the design and optimisation of DHW systems in public building typologies.

However, this requires a complex and extensive integration of algorithms and a research into parameters and characteristics concerning the building, the DHW system and the hot water demand. On the basis of a virtual model, the operation of the DHW system can be assessed. The tool can allow concretizing of the performance factors of different proposals for the configuration of DHW systems. DHW system designers will be able to estimate the impact of design decisions and to reduce energy demand for DHW, while keeping an equilibrium between healthy, comfortable and energy efficient buildings.

In addition, the tool can also be used in the optimisation of various parameters, from thickness of insulation to the care and washing procedure. It also has opportunities in the field of Legionella decontamination.

It can have the potential to be of great importance in the optimisation of DHW systems and in the definition of energy-saving methodologies.

Main References

(max 200 words)

*Van den Abeele L., Dinne K., De Cuyper K. and Bleys B., ‘Best Beschikbare Technieken (BBT) voor Legionella-beheersing in Nieuw Sanitaire Systemen’. VITO & WTCB, 2017.

*Van Kenhove E., ‘Coupled thermohydraulic and biologic modelling of legionella Pneumophila proliferation in domestic hot water systems’, 2018.

*Deutsches Institut für Normung, ‘DIN 1988-300: Technische Regeln für Trinkwasser-Installationen - Teil 300: Ermittlung der Rohrdurchmesser’. Berlin, 2012.

*A. Bertrand, A. Mastrucci, N. Schüler, R. Aggoune, en F. Maréchal, “Characterisation of domestic hot water end-uses for

integrated urban thermal energy assessment and optimisation”, Applied Energy, vol. 186, pp. 152–166, jan. 2017.



17:18 - 17:36

Simulation of Legionella risks when restarting a sanitary sport facility installation after a period of inactivity during COVID-19

Elisa Van Kenhove, Lien De Backer, Jelle Laverge

Ghent University, Belgium

Aim and Approach

(max 200 words)

During the COVID-19 lockdown it was repeatedly reported in the media that the prolonged closure of buildings (e.g. hotels, sport facilities, student homes) is not without danger as stagnant water is an important risk factor for Legionella growth. LoWatter, a service of Ghent University, carries out scientific research and consultancy into Legionella in sanitary hot water systems [1]. Based on an in-house developed simulation tool, the Legionella concentration can be predicted dynamically in the entire hot water circuit [2]. The aim of the presented case study is to demonstrate what the consequences are of a temporary shutting down of a system with regard to the risk of a Legionella pneumophila contamination of the system. To do so, the simulation tool is applied on a fictive, but representative case of a sport facility. Parameters, such as number of changing rooms and showers were derived from a comparative study. The distribution system consists of a circulation loop, located in the basement, with vertical distribution to each changing room. Sizing of pipe diameters, insulation thickness, circulation pump and nominal power of the production unit was done within the simulation tool according to DIN1988-300 [3], [4].

Scientific Innovation and Relevance

(max 200 words)

The simulation tool applied for the described case study, is written in the Modelica language. This in-house developed tool is a parametric tool that allows to build up a simulation model of the hydraulic scheme quickly by defining building characteristics: in case of a sport facility such parameters are the number of changing rooms and number of showers. The sizing of all subcomponents present in the hot water system (production system, pump, pipes, valves,) is done automatically. Although this tool was used in this paper to study the impact of temporary shutting down, the tool can also be used in other situations that may cause Legionella contamination of in the water system. Examples are an improper balance of the sanitary hot water system and an undersized production system. This case-specific simulation model can then be used to first “virtually” test the effectiveness of possible solutions on a contaminated system (modifications on the system design, decontamination techniques), before applying this in practice. Based on the results, it is possible to provide advice on which case-specific measure(s) will ensure that the contamination disappears and does not reoccur. Furthermore, it allows a building owner to get a better understanding of the system.

Preliminary Results and Conclusions

(max 200 words)

The case study of the sports facility shows the effect of a prolonged stagnation of the water in a domestic hot water system. As expected, the combination of stagnation and the relatively high ambient temperatures that were present during April till August resulted in legionella growth in the system. The model also showed the importance of the correct application of proposed measures at the start-up of the system: flushing each tap point at a high temperature for a sufficiently long time.

Furthermore, the case study also shows the potential of the developed tool. The sizing that was calculated in the background of the tool was compared to a fully manual calculation for the same system and results in the same sizing of the system. There are still a few work points in the parametric tool such as the integration of the iterative calculation process in case the pressure drop in the pipes requires to an adjustment of the pipe diameter. This will be further improved.

Main References

(max 200 words)

[1] “Lowatter - Controlling Legionella in tapwater,” 2020. [Online]. Available: www.lowatter.com.

[2] E. Van Kenhove, L. De Backer, M. Delghust, and J. Laverge, “Coupling of modelica domestic hot water simulation model with controller,” in Proceedings of Building Simulation 2019 : 16th Conference of IBPSA, 2019, pp. 924–931.

[3] S. Kreps, K. De Cuyper, S. Vanassche, and K. Vrancken, “Best Beschikbare Technieken (BBT) voor Legionella-beheersing in Nieuwe Sanitaire Systemen,” 2017.

[4] DIN Deutsches Institut für Normung e.V., “DIN 1988-300 Technische Regeln für Trinkwasser-Installationen - Ermittlung der Rohrdurchmesser; Technische Regel des DVGW,” 2012.



17:36 - 17:54

Towards the integration of energy performance certificates (EPC) and simplified building performance simulations using machine learning: initial findings

Roberto Boghetti1, Roberto Rugani1, Marco Picco2, Giacomo Salvadori1, Marco Marengo2, Fabio Fantozzi1

1University of Pisa, Pisa, Italy; 2University of Brighton, Brighton, United Kingdom

Aim and Approach

(max 200 words)

A broad range of policies and supportive measures aimed at improving the performance of existing and new buildings has been introduced in the recent years. Among these, energy performance certificates provide a rating scheme to assess the energy efficiency of buildings based on quasi-steady state methods. The adoption of these certificates, however, is currently facing some critical issues, such as the lack of a quality control, a limited access to data for practitioners, and regional differences in the way the certificates are managed. Furthermore, the quasi-steady state calculation method used, while less onerous than a dynamic simulation, has a lower precision and still carries numerous overheads [1]. The most common risk is to use energy certification methods to predict the real consumption of buildings. Data-driven approaches may solve some of these problems and provide a faster yet powerful alternative to quasi-steady state methods which could also be easily integrated with existing building cadasters. The goal of this study is to examine the reliability of existing energy performance certificates and to investigate the possibility of using a data-driven approach, namely an artificial neural network (ANN), as an alternative to the current quasi-steady state method.

Scientific Innovation and Relevance

(max 200 words)

This study is a preliminary step towards a broader research intent: developing a simplified simulation tool based on machine learning to predict both the energy consumption of buildings and their associated energy performance score. Machine learning has been successfully used on a research level to predict energy consumption and carbon emissions of buildings[2-3-4], using both measured and simulated data. Given the complexity of the problem and the difficulty of gathering the needed data, however, the resulting models were always limited to the respective case studies and lacked the generality to be applied in practical situations. Focusing on energy performance certificates partially solves this issue as several public repositories of these certificates were made available by different european countries and institutions. To cope with the fact that these databases are not standardized, in particular for what concerns the descriptive parameters of the buildings, an intermediate step will be made to generate a surrogate model of each considered building using the available information, so that it will be possible to create a single predictive model despite having heterogeneous input data. In the view of the authors, this approach could represent a valuable and novel addition to the ongoing scientific discussion.

Preliminary Results and Conclusions

(max 200 words)

Preliminary results were obtained using both the old and new version of the Lombardy’s energy performance certificates database (CENED). The two databases, created according to two subsequent versions of the Italian norm, differ for the provided input data and for the calculation method. In both cases, the available certificates were vastly incorrect [5], with many entries incorrectly reported. The proposed data-driven model, which calculates the energy performance index (EPH) in the heating season, demonstrated high precision on the old database, with a mean error of around 10%. On the newer database, however, where some characteristics of the buildings were not available anymore, the resulting error was unacceptable, ranging around 40%. To demonstrate that the problem was due to the lack of important information, a third attempt to calculate the updated EPH value using inputs from the old database was made. Even if the number of available buildings in this case was smaller than in the previous attempts, the error, concentrated in highly efficient buildings, was reduced by 35%. The positive results on this initial step indicate that a data-driven approach to EPC calculation could yield reliable estimations and therefore represents a promising support to the current methodology.

Main References

(max 200 words)

[1] Ballarini, I., Primo, E. and Corrado, V., 2018. On the limits of the quasi-steady-state method to predict the energy performance of low-energy buildings. Thermal Science, 22(Suppl. 4), pp.1117-1127.

[2] Ahmad, M.W., Mourshed, M. and Rezgui, Y., 2017. Trees vs Neurons: Comparison between random forest and ANN for high-resolution prediction of building energy consumption. Energy and Buildings, 147, pp.77-89.

[3] Paterakis, N.G., Mocanu, E., Gibescu, M., Stappers, B. and van Alst, W., 2017, September. Deep learning versus traditional machine learning methods for aggregated energy demand prediction. In 2017 IEEE PES Innovative Smart Grid Technologies Conference Europe (ISGT-Europe) (pp. 1-6). IEEE.

[4] Boghetti, R., Fantozzi, F., Kämpf, J.H., Salvadori, G., 2019, Understanding the performance gap: a machine learning approach on residential buildings in Turin, Italy. Journal of Physics: Conference Series (1343), CISBAT 2019

[5] Khayatian, F. and Sarto, L., 2016. Application of neural networks for evaluating energy performance certificates of residential buildings. Energy and Buildings, 125, pp.45-54.



17:54 - 18:12

Interactions of building- / urban-, and multi-energy systems design variables

Adam Bufacchi1, Georgios Mavromatidis2, Arno Schlüter1, Christoph Waibel1

1Architecture and Building Systems, ETH Zurich, Switzerland; 2Sustainability and Technology, ETH Zurich, Switzerland

Aim and Approach

(max 200 words)

Significant synergies between the energy demand and supply side have been shown to exist which should be exploited in the design of low-carbon buildings and neighbourhoods by considering building design parameters and energy systems design parameters simultaneously, rather than addressing them sequentially (Waibel et al., 2019) (Ferrara et al., 2019). However, existing literature lacks an explicit quantification of such synergies, particularly at the building geometry scale. Therefore, this work aims to investigate the degree of parameter interdependence between parameters influencing energy demand (i.e. building construction, use, and geometry) and parameters defining the energy supply (i.e. the design of a neighbourhood’s decentralised multi-energy system).

We apply a Morris Screening on a wide range of parameters, which informs us through a qualitative ranking of these parameters’ importance. Subsequently, we compute first, second and total order Sobol indices: Global Sensitivity Analysis indicators, quantifying the parameters’ influence on the design output individually, when interacting with one-, or all other parameters respectively. Hereby we seek to quantify the degree of parameter interdependence. The workflow includes sampling real building geometries (LoD1 data of Zürich as case study), energy demand simulations with Energyplus/Honeybee, and multi-energy systems design optimisation using a Mixed Integer Linear Programming model.

Scientific Innovation and Relevance

(max 200 words)

Related work has investigated the significance of coupling the demand side with the supply side on larger district or urban scales, and with simplified, or archetypal geometrical parameters (Shi et al., 2020)(Mavromatidis et al., 2018). Our study, however, uses more detailed geometrical properties of individual buildings derived from existing building geometry data in its sensitivity analysis. The footprint and building height of randomly selected buildings are used to generate geometries of multi-story buildings with varying roof types and windows. Derived geometrical characteristics (such as compactness and significant faces) are calculated to understand the importance of different architectural features.

Furthermore, the sensitivities to the building- and energy parameters of not only the overall costs and emissions are analysed, as is typically the case, but also the sensitivities of the selection and sizing of the technologies in the energy hub are investigated.

Finally, due to the large number of dependent and independent variables considered in this study, we cluster several parameters into topical groups in the analysis (e.g. by overall performance of individual technologies, or by individual attributes such as size, efficiency or lifetime of the energy systems), which helps in understanding possible interactions between the energy demand and supply.

Preliminary Results and Conclusions

(max 200 words)

The preliminary results of the Morris screening are consistent with the existing literature: parameters directly affecting the energy demand, such as the overall geometry, ventilation, the climate, temperature setpoints and wall insulation appear to have a larger impact on the overall cost and emissions than technical energy system parameters such as investment costs or efficiencies of various technologies. In the subsequent Sobol Analysis, we will obtain quantitative measures on (first and second order) interaction effects between architectural design parameters and energy system design parameters.

Main References

(max 200 words)

Ferrara, M., Prunotto, F., Rolfo, A., & Fabrizio, E. (2019). Energy Demand and Supply Simultaneous Optimization to Design a Nearly Zero-Energy House. Applied Sciences. https://doi.org/10.3390/app9112261

Waibel, C., Evins, R., & Carmeliet, J. (2019). Co-simulation and optimization of building geometry and multi-energy systems: Interdependencies in energy supply, energy demand and solar potentials. Applied Energy, 1661–1682. https://doi.org/10.1016/j.apenergy.2019.03.177

Mavromatidis, G., Orehounig, K., & Carmeliet, J. (2018a). Uncertainty and global sensitivity analysis for the optimal design of distributed energy systems. Applied Energy, 214, 219–238. https://doi.org/10.1016/j.apenergy.2018.01.062

Shi, Z. ;, Hsieh, S. ;, Fonseca, J. A. ;, Schlueter, A., Shi, Z., Hsieh, S., & Fonseca, J. A. (2020). Interdependencies between the design of street grids and the cost-effectiveness of district cooling systems ETH Library 1 Interdependencies between the design of street grids and the cost-effectiveness of district cooling systems. https://doi.org/10.3929/ethz-b-000391818



18:12 - 18:30

Using a surrogate model to analyze the impact of geometry on energy efficiency of buildings.

Bhumika Bhatta1, Ralph Evins2, Paul Westermann3

1University of Victoria, Canada; 2University of Victoria, Canada; 3University of Victoria, Canada

Aim and Approach

(max 200 words)

Parametric exploration and optimization of building geometry is a powerful tool for designing energy efficient buildings. However, in practice this process is computationally expensive and time-consuming. In this research, we explore the use of surrogate models, i.e. efficient statistical approximations of ex-pensive physics-based building simulation models, to lower the computational burden of large-scale build-ng geometry analysis. For this purpose, we developed a novel dataset of 38,000 residential building models derived from real world floor plans from (Wu et al. (2019))[5] and train a surrogate model to emulate their simulated annual energy performance. We extract up to 20 parameters as surrogate model inputs to represent the building geometry and show that the trained surrogate model reaches a high accuracy (R2score = 0.999, MSE =0.007 and RMSE = 0.022) on test data. The current setup forms the basis for further research where the complexity of the building models will be increased.

Scientific Innovation and Relevance

(max 200 words)

Surrogates are the statistical models which can be used to provide rapid approximations of more expensive models. Surrogate models can be taken as an alternative approach to replace the detailed simulations with simplified approximate simulations, thereby sacrificing accuracy for reduced computational time. In this research work we are going to develop a surrogate model with two different approaches (physics-based and machine learning-based) which will approximate the impact of geometry on energy demand. Previous research work has been done on studying the impact of building geometry on its energy demand. For example, AlAnzi et al. [1] take the geometry of building into account for the energy performance analysis but consider the shape of façade only rather than the overall footprint. Most research studies sought to identify the relationship between the relative compactness or building volume and building energy loads [2],[3],[4] On the other hand, many research studies have considered hypothetical building shapes with constant floor area and height or constant floor area with varying heights among different shapes. The work presented in this paper will use a model, based on Building Technology Assessment Platform (BTAP) small and medium office building for Victoria, BC, Canada.

Preliminary Results and Conclusions

(max 200 words)

In this paper we provide a framework to develop neural network based surrogate models that predicts the EUI over different geometries. The frame work consists of conversion of 2D raster images of floorplans into 3D building simulation models, extraction of 20-building features for geometry, surrogate model training, and performance analysis of the impact of each feature. We developed the dataset consisting of 38,000 building simulation IDF files derived from real-world floor plans (Wu et al. (2019)). Our results for the training of the surrogate models show that off-the shelve neural network surrogate models pro-vided with manually engineered features are capable of emulating the simulation outcomes really well (R2score = 0.999, MSE = 0.007 and RMSE= 0.022) on test data. Overall, this paper shows that a simple machine learning model can perform very well in predicting energy use for various geometries, which provides architects with a great tool to get building performance estimates in real time while they are exploring various designs without the burden of simulating building designs individually.

Main References

(max 200 words)

1. Adnan AlAnzi, DonghyunSeo, and MoncefKrarti. Impact of building shape on thermalperformance of office buildings in kuwait. Energy Conversion and Management, 50(3):822–828, 2009

2. JosifasParasonis, Andrius Keizikas, Audron ̇eEndriukaityt ̇e, and Diana Kalibatien ̇e. Architec-tural solutions to increase the energy efficiency of buildings. Journal of civil engineering and management, 18(1):71–80, 2012

3. Werner Pessenlehner and Ardeshir Mahdavi. Building morphology, transparence, and energy performance. na, 2003

4. Carlo Ratti, Dana Raydan, and Koen Steemers. Building form and environmental performance: archetypes, analysis and an arid climate. Energy and buildings, 35(1):49–59, 2003.

5. Wu, W., X.-M. Fu, R. Tang, Y. Wang, Y.-H. Qi, and

L. Liu (2019). Data-driven interior plan generation for residential buildings. ACM Transactions

on Graphics (SIGGRAPH Asia) 38 (6)