Conference Agenda

Overview and details of the sessions of this conference. Please select a date or location to show only sessions at that day or location. Please select a single session for detailed view (with abstracts and downloads if available).

Please note that all times are shown in the time zone of the conference. The current conference time is: 6th Dec 2021, 21:54:17 CET

 
 
Session Overview
Session
Session W2.7 (Online Track): Ensuring high quality building simulations
Time:
Wednesday, 01/Sept/2021:
13:00 - 14:30

Session Chair: Charles S Barnaby, CSB Consulting
Location: Virtual Meeting Room 1

External Resource: Click here to join the Zoom Meeting
Show help for 'Increase or decrease the abstract text size'
Presentations
13:00 - 13:06

Identifying grey-box models from archetypes of apartment block buildings

Marius Eide Bagle1, Philip Maree2, Harald Taxt Walnum1, Igor Sartori1

1SINTEF Community, Norway; 2SINTEF Digital, Norway

Aim and Approach

(max 200 words)

A potentially large amount of flexibility resides in the space heating of residential buildings. To realize this potential, it is necessary to model heat demand with models that are accurate enough and suitable for real time control.

Well-suited for this purpose are grey-box models, which combine a relatively simple physical descriptions of the building with data-driven inference of key parameters. However, identification of grey-box models poses a challenge: alongside energy use and weather data, the indoor temperature must also be known. Such data are scarcely available. Furthermore, it is not given that measurements from normal operation of buildings provide datasets that are 'rich' enough to successfully drive the identification process, or if special test periods should be carried out. Such a test would require the manipulation of indoor temperatures, possibly in periods of non-occupancy, for several days; a challenging task on real buildings.

This paper presents a method that aims at overcoming this bottleneck by combining features of both white-box and grey-box modelling. A set of white-box models (specifically, IDA-ICE models) representing the Norwegian stock of apartment blocks is available, based on ca. 20 archetypes previously developed in the Tabula/Episcope project [1].

Scientific Innovation and Relevance

(max 200 words)

To enable a large-scale rollout of predictive control systems in the building stock, developing robust methods of model identification is of key importance. Provided that load profiles in the IDA-ICE models are validated (a parallel research activity), it is legitimate to assume that the indoor temperature profiles from the IDA-ICE archetypes are also representative for the real building stock, thus bypassing the need for measurement data.

Under this assumption, the grey-box models are identified from datasets generated by the IDA-ICE archetypes in two modalities: under normal operation with a seasonal (three-month) duration, and under special test periods with a duration of one to two weeks. During the test periods the IDA-ICE archetypes are excited with trains of heating events, Pseudo Random Binary Sequence (PRBS), aiming at exploring a wide and rapidly changing set of indoor temperatures around the comfort zone, ca. between 20 and 24 °C. It is also of interest to investigate the effect of temporal resolution in the identification process.

Preliminary Results and Conclusions

(max 200 words)

Preliminary results have been obtained using the software package CTSM-R [2], which uses an extended Kalman filter (EKF) to formulate a maximum likelihood estimation problem. This is combined with a quasi-Newton method to find the optimal set of parameters. With this setup, there are issues regarding the initial parameterization of the states, which essentially have to be found via a qualified guess. One weakness using the EKF is that it does not incorporate state constraints, nor does it utilize potential underlying non-linear dynamics. To such an extent, Moving Horizon Estimation (MHE) is presented as an alternative optimization-based approach to solving the maximum a-posteriori estimate. Recent advances in computational power makes MHE a viable solution as an asymptotically stable observer which has shown in other cases to provide improved state estimation and greater robustness to poor guesses of the initial state [3].

Results will be evaluated considering both the physical meaningfulness of the parameters, e.g. by ensuring that the energy balance of the model is unbiased [4], and that the parameter values are accompanied by reasonable confidence intervals. To enable a fair comparison between the algorithms, a fixed structure for the grey-box model will be decided a priori.

Main References

(max 200 words)

[1] “Tabula/Episcope - Norwegian archetypes.” [Online]. Available: https://episcope.eu/building-typology/country/no.html. [Accessed: 30-Jun-2020].

[2] H. Madsen et. al, “Continuous Time Stochastic Modeling in R User’s Guide and Reference Manual,” 2018.

[3] Haseltine, Eric L., and James B. Rawlings. "Critical evaluation of extended Kalman filtering and moving-horizon estimation." Industrial & engineering chemistry research 44.8 (2005): 2451-2460.

[4] R. De Coninck, F. Magnusson, J. Åkesson, and L. Helsen, “Toolbox for development and validation of grey-box building models for forecasting and control,” J. Build. Perform. Simul., vol. 9, no. 3, pp. 288–303, 2016.



13:06 - 13:12

Integrating thermal, energy, lighting, and acoustics in building design approach: Lesson learned from students assignments

Rizki A. Mangkuto, Anugrah Sabdono Sudarsono, R.S. Joko Sarwono

Institut Teknologi Bandung, Indonesia

Aim and Approach

(max 200 words)

This paper aims to report results and lesson learned from students assignments in integrating thermal, energy, lighting, and acoustics performance in a building design simulation project. The objectives are to observe the students approach in defining and obtaining the baseline, target, and optimum performance in all mentioned aspects, and to analyse the differences between the submitted design proposals.

The following hypothetical case was assigned to the students: A classroom with two glazed windows is situated on the second floor of a third-story building; the north side is exposed to the exterior, whereas the other three sides are adjacent to an unconditioned corridor. The students were asked to simulate and predict the baseline and target values of thermal, energy, lighting (all using Sefaira) and acoustics (using CATT v8) aspects of the classroom, and optimise them only by modifying the interior material properties. The submitted project reports were analysed statistically to observe the trend.

Scientific Innovation and Relevance

(max 200 words)

An integrated building performance simulation (BPS) approach considering thermal, energy, lighting, and acoustics performance altogether in building design is not so common. Integrating the first three mentioned aspects is indeed popular and is supported by current simulation tools, but the acoustics aspect is often not considered at the same time, i.e. it is not integrated in the design approach. This paper offers an insight on how graduate students (in Building Physics), some of which were professional architect and engineers, integrated and optimised all of those four aspects altogether, in an attempt to modify a classroom using BPS. There are also observations on how the proposed designs are compared to each other.

The results would be of relevance for building practitioners and educators, since the findings suggest that such integrated design can be quite challenging and is not trivial, particularly for those who are not familiar with the integrated approach, despite already having a background in the building industry.

Preliminary Results and Conclusions

(max 200 words)

Based on the submitted reports, it is found that only 50% of the students modified the interior materials by considering the four aspects at the same time, so that the results were optimum for all aspects. The remaining 50% chose to optimise the materials considering the first three aspects, as they could be done in Sefaira, and then modified the material with respect to acoustics aspect in CATT v8; without realising that the last step might also influence or change the thermal, energy, and lighting performance. While this seems like a typical beginner's mistake in BPS, it can also happen for instance to specialists, i.e. those who are experienced BPS users but do not master the four aspects altogether.

Among the aspects, lighting performance (sDA300/50%) is the most consistently predicted, with the coefficient of variance (CV) of 0.16 (baseline) and 0.07 (optimum). In opposite, thermal performance (annual overheating hours) is the least consistent, with CV of 1.68 (baseline) and 1.82 (optimum).

In conclusion, the findings can be applied to improve the content of BPS education by employing integrated building performance design approach.

Main References

(max 200 words)

I. Beausoleil-Morrison (2019) Learning the fundamentals of building

performance simulation through an experiential teaching approach, Journal of Building Performance Simulation, 12:3, 308-325.

P.P. Charles, C.R. Thomas (2009) Four approaches to teaching with building performance simulation tools in undergraduate architecture and engineering education, Journal of Building Performance Simulation, 2:2, 95-114.

E. Mendes, N. Mendes (2019) An instructional design for building

energy simulation e-learning: an interdisciplinary approach, Journal of Building Performance Simulation, 12:3, 326-342.



13:12 - 13:18

Statistical methodologies for verification of building energy performance simulation

Amin Nouri, Jérôme Frisch, Christoph van Treeck

Institute of Energy Efficiency and Sustainable Building (E3D), RWTH Aachen University, Germany

Aim and Approach

(max 200 words)

Building performance simulation tools are being increasingly deployed by researchers and professionals to predict the thermal behavior of buildings. In general, modeling and simulation techniques are employed to determine the thermal characteristics of the building. Validation methods are used to ensure the accuracy of simulation results. Several standardized procedures exist to assess building performance simulation tools, such as ASHRAE Standard 140 as well as the German VDI 6007 and VDI 6020 guidelines. This paper is conducted within a research project funded by the German Federal Ministry for Economic Affairs and Energy (BMWi), which addresses the development of quality standards for building and systems energy performance simulation. The objective of this project is to develop a validation methodology, to define standards for simulation applications and to transfer them into planning practice.

Scientific Innovation and Relevance

(max 200 words)

Buildings account for approximately 40% of the final energy consumption and 36% of the greenhouse gas emissions in Europe (European Commission, 2019). The Paris climate agreement aims to keep the global temperature rise to well below 2 °C above pre-industrial levels and to limit the temperature increase to a maximum of 1.5 °C. In the last few years, building performance simulation tools have been taken into consideration for the scientific community as well as industrial society. However, studies have revealed that there are substantial discrepancies between predictions of building performance simulation tools and the measured energy consumption in practice, which is referred to as the “performance gap”. Validation and verification procedures are essential elements within the development of building simulation tools to assess reliability of the simulation results as well as to find and eliminate eventual bugs in the simulation software.

Preliminary Results and Conclusions

(max 200 words)

The first part of this paper presents the comparison of the different standards and guidelines in the context of building and HVAC system performance simulation. The second part describes a systematic set of test cases for building and HVAC systems based on the ASHRAE Standard 140 and discusses the simulation approach in Modelica/Dymola. The third part presents the development and implementation of a validation methodology, which verifies the plausibility of simulation results and excludes simulation errors to the greatest extent. In this paper, first results as well as comparative analyses will be discussed. Another aspect of the research project is the development of a Platform. The main objective of the Platform is to provide a facility for defining individual test cases, to create individual simulation code, to perform a comparative validation, and to evaluate the accuracy of the simulation tools.

Main References

(max 200 words)

Wetter, M., van Treeck, C. (2017). „IEA EBC Annex 60: New Generation Computing Tools for Building and Community Energy Systems“.

Strachan, P., Svehla, K., Heusler, I., Kersken, M. (2016). „Whole model empirical validation on a full-scale building“. Journal of Building Performance Simulation.

Judkoff, R., Wortman, D., O’Doherty, B., Burch, J. (2008). „A methodology for validating building energy analysis simulations“. NREL, Golden.



13:18 - 13:24

Easy-to-implement simulation strategies for annual glare risk assessment based on the European Daylighting Standard

Bruno Bueno1, Abel Sepúlveda2, Christoph Maurer3, Simon Wacker4, Taoning Wang5, Tilmann E. Kuhn6, Helen Rose Wilson7

1Fraunhofer ISE, Germany; 2Taltech, Estonia; 3Fraunhofer ISE, Germany; 4Fraunhofer ISE, Germany; 5Lawrence Berkeley National Laboratory LBNL, Berkeley CA, USA; 6Fraunhofer ISE, Germany; 7Fraunhofer ISE, Germany

Aim and Approach

(max 200 words)

One of the most important functions of fenestration systems, which triggers user manipulation of the facade, is the protection from daylight glare. The new European daylight standard EN 17037 contains assessment procedures, criteria and information on glare evaluation. Still, mainstream building simulation programs include limited capabilities for glare risk assessment [1], which sometimes results in suboptimal façade design. The problem resides in the complexity and high requirements of state-of-the-art glare evaluation techniques, including those proposed in EN 17037 based on the DGP index [2,3]. In this study, we benchmark available methods for DGP calculation [4,5] and propose simulation strategies that could be easily implemented in simulation programs. This will foster the integration of glare risk assessment in the design process of buildings.

Scientific Innovation and Relevance

(max 200 words)

For a reliable glare risk assessment, it is crucial to account for the direct-direct optical transmittance and cut-off angle of fenestration systems. In this study, we analyse different options to represent the optical behaviour of fenestration components in simulation environments. Bi-directional scattering distribution functions (BSDF) with different resolutions [6] are compared with other strategies such as peak extraction. These representations are based on datasets that have to be experimentally obtained [7] and digitally transferred to simulation environments. The simulation approach, the timestep sampling and the optical representation of the shading technologies determine their applicability in simulation environments and the possibility to be integrated in the design process of buildings.

Preliminary Results and Conclusions

(max 200 words)

In this study, state-of-the-art glare evaluation methods are compared. We propose new strategies based on optical data such as low-resolution BSDF, which is becoming available in international databases, combined with recent approaches such as peak extraction. Efficient sampling strategies reduce the computational cost of dynamic raytracing simulations without loss of accuracy. DGP calculations based on these strategies could be implemented in building simulation environments.

Main References

(max 200 words)

[1] Bueno B, Sepúlveda A. A specific building simulation tool for the design and evaluation of innovative fenestration systems and their control. Building Simulation Conference, Rome, 2-4 September; 2019.

[2] J. Wienold, J. Christoffersen, Evaluation methods and development of a new glare prediction model for daylight environments with the use of CCD cameras, Energy and Buildings (2006).

[3] J. Wienold, T. Iwata, M. Sarey Khanie, E. Erell, E. Kaftan, R.G. Rodriguez, J.A. Yamin Garreton, T. Tzempelikos, I. Konstantzos, J. Christoffersen, T.E. Kuhn, C. Pierson, M. Andersen, Cross-validation and robustness of daylight glare metrics, Lighting Research and Technology (2019).

[4] E.S. Lee, D. Geisler-Moroder, G. Ward, Modeling the direct sun component in buildings using matrix algebraic approaches: Methods and validation, Solar Energy (2018).

[5] M. Abravesh, B. Bueno, S. Heidari, T.E. Kuhn, A method to evaluate glare risk from operable fenestration systems throughout a year, Building and Environment (2019).

[6] G. Ward, M. Kurt, N. Bonneel. Reducing anisotropic BSDF measurement to common practice. In: Proceedings of the Eurographics 2014 Workshop on Material Appearance Modeling: Issues and Acquisition. Lyon, France: Eurographics Association; 2014, p. 5–8.

[7] P. Apian-Bennewitz. New scanning gonio-photometer for extended BRTF measurements. In: Proceedings of SPIE; 2010.



13:24 - 13:30

A machine-learning framework for daylight and visual comfort assessment in early design stages

Hanieh Nourkojouri1, Zahra Sadat Zomorodian1, Mohammad Tahsildoost1, Zohreh Shaghaghian2

1Shahid Beheshti University, Iran, Islamic Republic of; 2Texas A&M University, College Station, United States

Aim and Approach

(max 200 words)

This research is mainly focused on daylight assessment in early stages of design through a new framework which is based on machine learning algorithms. A dataset was primarily developed from 2880 simulations derived from Honeybee for Grasshopper. The simulations were done for a shoebox space with a one side window. The alternatives emerged from different physical features including room dimensions, interior surfaces reflectance, window dimensions and orientations, number of windows and shading states. The metrics used for daylight evaluations included UDI, sDA, mDA, ASE and sVD. Quality Views were analyzed for the same shoebox spaces via a grasshopper-based algorithm which was developed from the LEED v4 evaluation framework for Quality Views. The dataset which had finally daylight, visual comfort and quality views outputs indices was further analyzed with Artificial Neural Network algorithm written in Python. The developed model could be used in early design stages analyzes without the need for time consuming simulations in previously used platforms and programs.

Scientific Innovation and Relevance

(max 200 words)

This work presents a concept for development of a new method to evaluate daylight assessment and views at the same time without the need for time conserving simulations. The mentioned method includes 5 indices for daylight and visual comfort evaluations and 2 indices for evaluation of quality views. Having all these 7 metrics together could give the designers a comprehensive vision for the designed space’s performance in daylight and visual comfort. Furthermore, with the application of machine learning algorithms like ANN which is used in this research, these results could be delivered to the designers and stakeholders quickly with determination of some simple physical features of the designed spaces and there will be no need for vast knowledge in building physics modeling and simulations for architects and designers using them.

Preliminary Results and Conclusions

(max 200 words)

In this research a model was developed for prediction of daylight, visual comfort and quality view assessment at the same time in a space. The model takes some simple physical features of the space as inputs and the estimated outputs include 7 metrics indicating the daylight and visual comfort performance of the defined space. The model is based on an optimized ANN algorithm which achieved a prediction accuracy of about 97% and MSE index of 0.002.

Main References

(max 200 words)

Ayoub, M., 2020. A review on machine learning algorithms to predict daylighting inside buildings, Solar Energy, Vol 202, p249-275.

Mardaljevic, J., 2015. Climate-based daylight modelling and its discontents. Presented at the Simple Buildings Better Buildings? Delivering performance through engineered solutions, CIBSE Technical Symposium,London, April 16-17th

Ngarambe, J., Irakoze, A., Young Yun, G., Kim, G., 2020, Comparative performance of machine learning algorithms in prediction of indoor daylight illuminances, Sustainability 2020,12(11), 4471

Tregenza, P., Mardaljevic, P., 2017, Daylighting Buildings: Standards and the needs of designer, Lighting Research and Technology, Vol 50, p63-79.

LEED® green building program, v4 for Building design and construction.



13:30 - 13:36

Reducing heat island effect: a mathematical model of green roof design

Jing Hong1, Dennis Michael Utzinger2

1Rivion; 2University of Wisconsin- Milwaukee

Aim and Approach

(max 200 words)

Green roofs ease the heat island effect. Optimizing green roof design helps achieve this goal more efficiently. This paper proposes an energy model of green roofs to estimate surface temperature and validates them with experimental evidence that measured on the green roof at the University of Wisconsin - Milwaukee. Two energy balance equations separately indicate the heat flux through the bare soil and vegetation-covered surfaces. The energy flux density and surface temperature of bare soil and vegetation-covered surface were modeled by the proposed mathematical models. In the meantime, the effects of color, soil depth, and plant type on the surface temperature were analyzed to conclude the optimal green roof design.

Scientific Innovation and Relevance

(max 200 words)

In the proposed energy model, the calculation of radiation, convection, and evaporation on surface temperature is simplified to use fewer unknown variables. For that reason, this model only requires basic weather data and an initial soil temperature reading to estimate the green roof surface temperature in any equation-solving program.

Preliminary Results and Conclusions

(max 200 words)

The energy balance models explain how surface color, soil depth, and plant types affect the surface temperature of a green roof. In conclusion, the green roof surface temperature can be reduced by lighter surface color, shallower soil depth, and plants with lower internal leaf resistance and larger leaf size.

Main References

(max 200 words)

[1] Susca, Tiziana, Stuart R. Gaffin, and G. R. Dell’Osso. "Positive effects of vegetation: Urban heat island and green roofs." Environmental pollution 159, no. 8-9 (2011): 2119-2126.

[2] Sailor, David J. "A green roof model for building energy simulation programs." Energy and buildings 40, no. 8 (2008): 1466-1478.

[3] Duffie, John A., and William A. Beckman. Solar engineering of thermal processes. John Wiley & Sons, 2013.

[4] Berdahl, Paul, and Marlo Martin. "Emissivity of clear skies." Solar Energy 32, no. 5 (1984): 663-664.

[5] Watmuff, J. H., W. W. S. Charters, and D. Proctor. "Solar and wind induced external coefficients-solar collectors." Cooperation Mediterraneenne pour l'Energie Solaire (1977): 56.

[6] Farouki, Omar T. Thermal properties of soils. No. CRREL-MONO-81-1. Cold Regions Research and Engineering Lab, Hanover NH, 1981.

[7] Oke, Timothy R. Boundary layer climates. Routledge, 2002.

[8] Gates, David M. Biophysical ecology. Courier Corporation, 2012.

[9] Monteith, John, and Mike Unsworth. Principles of environmental physics: plants, animals, and the atmosphere. Academic Press, 2013.

[10] American Society of Heating, Refrigerating and Air-Conditioning Engineers. 2013 Ashrae Handbook: Fundamentals. Inch-pound ed. Atlanta, Ga.: Ashrae, 2013

[11] Oyj, Vaisala. "Humidity Conversion Formulas—Calculation formulas for humidity." Vaisala: Helsinki, Finland (2013).



13:36 - 13:42

Semantic web ontologies for buildings objects and their performance data

Eleanna Panagoulia, Zachary Lancaster, Tarek Rakha

Georgia Institute of Technology, United States of America

Aim and Approach

(max 200 words)

The normative ontologies of tools, federated and tailored by the Architecture, Engineering, Construction and Operations industry (AECO), constrain collaboration and interoperability due to the creation of specific and bounded representational spaces, outside of which, software cannot operate. Although Building Information Modeling (BIM) provides a foundation for collaboration and data exchange within a common platform, professional practice still relies heavily on document sharing to circumvent the limits of proprietary software ontologies, especially when integrating performance analytics.

Current approaches of using monolithic data schemas to communicate information (i.e. IFC) constrain the domain of possible content, hence reducing the data expressivity and failing to capture the diversity of data in the built environment. To overcome this constraint we must provide access to information across platforms without presupposing information loss. This requires a modular, consistent, specific enough vocabulary, to describe any element, without at the same time being ambiguous.

This paper proposes one such vocabulary in the form of a semantic web-based ontology for representing building envelopes and their linked data. We demonstrate the use of this ontology in a platform agnostic exchange between envelope performance analysis and industry standard design software; enabling decision-making for building envelopes in the context of high-performance retrofitting design.

Scientific Innovation and Relevance

(max 200 words)

The ability to represent building elements as semantically complete objects would contribute to higher performance architectures. However, the identification and classification of building objects depends largely on user interpretation and appears to operate without an explicitly documented or comprehensive ontology. This impedes the consistent linking of analytical outcomes and auxiliary data to envelope systems and prevents the creation of a common information repository. This lack of consistent understanding of a project, across all disciplines, results in miscommunication and thus, deviation between the actual and anticipated performance.

Previous research indicated that a single interface would be insufficient to handle the multifarious datasets associated with the building objects and their performance. This paper’s innovation is in the employment of a semantic web approach that proposes the use of consistent and practical ontologies describing both building elements and building performance data, while also describing the relationship between building elements and their analytical data. This ontological definition focuses on a bi-directional exchange between platforms, and makes performance and simulation data available to practitioners.

Preliminary Results and Conclusions

(max 200 words)

This research specifies two novel ontological definitions and a software interface that makes use of these definitions. The first one is defined as the Ontology for Building Objects (OBO), describes an explicit graph-based, rule-driven description of building envelopes as discrete elements and the relationships between them. While OBE focuses primarily on a description of geometric objects, the second, the Ontology for Building Performance Data (OBPD), is a minimal definition for building performance analytics, its data requirements, outputs and the relationship between an analysis and a building element or set of elements. While OBO is strongly aligned with OBPD, it is also aligned with other known building ontologies.

The paper showcases an application that operates as middleware between different data stores, the user and an interface for data serialization towards a common data environment. We demonstrate the above using Revit’s API and a dataset common to energy audits of existing buildings. The result is an automated method of identifying components enabled by the OBO’s building objects definition and linking them to performance analytics based on OBPD’s relationship specification. We present a lossless exchange and mapping of data that provides the ability to represent non-native, analytical data directly in the BIM environment.

Main References

(max 200 words)

[1] R. Volk, J. Stengel, and F. Schultmann, “Building Information Modeling (BIM) for existing buildings - Literature review and future needs,” Automation in Construction. 2014.

[2] E. Kamel and A. M. Memari, “Review of BIM’s application in energy simulation: Tools, issues, and solutions,” Autom. Constr., vol. 97, no. October 2018, pp. 164–180, 2019.

[3] P. Pauwels, M. Poveda-Villalón, Á. Sicilia, and J. Euzenat, “Semantic technologies and interoperability in the built environment,” Semant. Web, 2018.

[4] U. Knaack, “Potential for innovative massive building envelope systems – Scenario development towards integrated active systems,” J. Facade Des. Eng., 2015.

[5] S. B. Sadineni, S. Madala, and R. F. Boehm, “Passive building energy savings: A review of building envelope components,” Renew. Sustain. Energy Rev., 2011.

[6] A. Mahdavi and D. Wolosiuk, “A Building Performance Indicator Ontology : Structure and Applications”, 2019.

[7] M. H. Rasmussen, M. Lefrançois, P. Pauwels, C. A. Hviid, and J. Karlshøj, “Managing interrelated project information in AEC Knowledge Graphs,” Autom. Constr., 2019.

[8] M. Bonduel, M. Vergauwen, R. Klein, M. H. Rasmussen, and P. Pauwels, “A novel workflow to combine bim and linked data for existing buildings,” eWork Ebus. Archit. Eng. Constr. - Proc. 12th Eur. Conf. Prod. Process Model. ECPPM 2018.



13:42 - 13:48

A machine-learning framework for acoustic design assessment in early design stages

Reyhane Abarghooie1, ZahraSadat Zomorodian1, Mohammad Tahsildoost1, Zohreh Shaghaghian2

1Shahid Beheshti University, Tehran, Iran; 2Texas A&M University, College Station, United States

Aim and Approach

(max 200 words)

Predicting acoustic performance by utilizing simulation tools is a widely used approach being favored over time-costly scale model studies. In this field, building acoustic simulation tools are complicated by several challenges, including the high cost of acoustic tools, the need of acoustic expertise, and the time-consuming process of acoustic simulation. This research presents a methodology for measuring acoustic comfort using a soft-sensing approach in the early design stages of the building. This work presents a proof of concept for a novel machine learning method to estimate a set of typical room acoustics parameters by using only geometrical information as input features.

The proposed model is trained and evaluated by using a novel dataset composed of acoustical simulation of a single room with 2916 different configurations. In the stimulation process features that include room dimensions, window size, material absorption coefficient, furniture, and shading type has been analyzed by using Pachyderm acoustic software. The mentioned dataset is used as the input of a machine-learning model based on Deep Neural Network (DNN). The machine learning model is fully-connected DNN with 5 hidden layers.

Scientific Innovation and Relevance

(max 200 words)

Acoustic comfort is one of the most important design parameters that affect people's satisfaction and productivity. Therefore predicting the sound conditions in the building is noticeable in both the research and the industrial environments. Traditional methods to study the sound propagation inside the rooms can be divided into three approaches: the geometrical models, the wave-based models, and the statistical models. The parameter values that render the acoustical comfort of a room should be estimated based on the geometry of the virtual room. While the complete physical models are typically computationally too demanding, an approximation can be made by using simplified mathematical methods such as Sabin or Eyring. Formulas that can only calculate RT, and do not often hold in typical environments such as offices, schools, accommodations, commercial, and healthcare environments. In this way, we aim to estimate some of the acoustical indexes which, include Reverberation Time (RT), Early Decay Time (EDT), Speech Transmission Index (STI), Clarity (C80), and Definition (D50), where the estimated values should be accurate enough for the plausible rendering of acoustic virtual reality. The goal of this project is to introduce a model with short calculation time to estimate the mentioned indexes in the nominated spaces.

Preliminary Results and Conclusions

(max 200 words)

This work presents a machine-learning-based method to estimate the room acoustic indexes. The model takes geometric and physical features of a room as input and output the estimated RT, EDT, STI, C80, and D50 indexes values as a function of frequency. All models were trained and evaluated in a dataset that contains parametric acoustic simulation.

The estimation was performed by using a machine-learning model based on a DNN. The baseline model achieves appropriate results by using geometrical features extracted from the 3D model of rooms. The stimulation model achieves a prediction accuracy of approximately more than 96% during the mentioned indexes.

Main References

(max 200 words)

Rossing, T.D. (2014). Springer Handbook of Acoustics. New York, Ny Springer.

Falcón Pérez (2018). Machine-learning-based estimation of room acoustic parameters. Master’s thesis. Aalto University, Espoo

J. Bianco, M., Gerstoft, P., Traer, J., Ozanich, E., A. Roch, M., Gannot, S. and Alban Deledalle, C. (2019). Machine learning in acoustics: Theory and applications. The journal of the acoustical society of America.

H. and Zhai, Z. (John) (2016). Advances in building simulation and computational techniques: A review between 1987 and 2014. Energy and Buildings, 128, pp.319–335.

Huang, B., Pan, Z., Liu, Z., Hou, G. and Yang, H. (2017). Acoustic amenity analysis for high-rise building along urban expressway: Modeling traffic noise vertical propagation using neural networks. Transportation Research Part D: Transport and Environment, 53, pp.63–77.

Lovedee-Turner, M. and Murphy, D. (2018). Application of Machine Learning for the Spatial Analysis of Binaural Room Impulse Responses. Applied Sciences, 8(1), p.105.



13:48 - 13:54

A Python library for Radiance matrix-based simulation control and EnergyPlus integration

Taoning Wang1, Greg Ward2, Eleanor S. Lee1

1Lawrence Berkeley National Lab, United States of America; 2Anyhere Software

Aim and Approach

(max 200 words)

In the realm of building simulation, there are many reasons why accurate, efficient ray-tracing-based solar radiation and daylighting simulations are needed: 1) accurate thermal and visual comfort predictions rely on detailed maps of solar radiation and luminance on an occupant’s body or field of view. 2) broader adoption of energy efficiency, indoor environmental quality, and healthy building standards require accurate modeling of innovative fenestration solutions, and 3) increased trends toward integrated, advanced building design and control solutions require efficient models to work seamlessly within co-simulation environments, such as Spawn-of-EnergyPlus. Radiance matrix-based simulation methods provide efficient and accurate ways to simulate annual/dynamic daylighting and solar radiation, particularly optically-complex operable shades, and dynamic glazings. However, real-world adoption could be more pervasive if the model setup and simulation workflow were not so difficult to learn and prone to user errors.

The goal of this work is to enable users and software developers to adopt advanced Radiance matrix-based simulation methods without extensive knowledge and experience. A Python-based open-source library frads was developed, along with a series of command-line programs, to automate and speed up the use of these advanced simulation methods.

Scientific Innovation and Relevance

(max 200 words)

Ray-tracing algorithms have typically been too time-consuming for practical use in annual simulations. With matrix-based methods, the time needed to conduct annual simulations using ray-tracing tools has been reduced by several orders of magnitude. With the advent of version 9.3, EnergyPlus has exposed its core simulation engine to any external software environment, which enables runtime Radiance and EnergyPlus integration. The entire workflow is implemented as a command-line program, but software developers can use frads to incorporate the workflow into any third-party software.

frads consist of many standard Radiance operations allowing for automated model translation from EnergyPlus, matrices generation, and runtime data-exchange with EnergyPlus. frads also comes with several command-line programs for routine operations, and these programs also serve as examples of using the frads library. The chief among these programs is mrad, an executive program that automates the Radiance two- to six-phase simulation workflow. Users need to provide the model description (e.g., geometry, material, and location) and parts of the model that are parametrized. mrad will then choose the suitable matrix phase method based on user-specified speed and accuracy requirements and carry out matrix generation and multiplication workflow, outputting irradiance, illuminance or luminance results per time step.

Preliminary Results and Conclusions

(max 200 words)

The framework for frads has been developed with further refinements currently underway. Early tests indicate that Radiance and EnergyPlus run-time integration requires minimal to no user intervention, enabling rapid widespread adoption. One of the key advantages of run-time data-exchange between Radiance and EnergyPlus is that it enables the evaluation of advanced model predictive control of building dynamic facade in a multizone building. During runtime, at each timestep, EnergyPlus simulation pauses; given a dynamic facade control signal (e.g., via Spawn/ Modelica or manufacturer component model), Radiance computes resulting facade energy transfer data, which are then used in EnergyPlus to complete the relevant heat balance calculations.

With these accurate simulation methods as part of the toolset, engineers and designers are able to differentiate facade product performance through simulation, which in turn spurs innovations from inventors and manufacturers. Integration of Radiance and EnergyPlus in a co-simulation environment further boosts the confidence among engineers and designers in the whole-building energy simulation results.

Main References

(max 200 words)

Gehbauer, C., Blum, D. H., Wang, T., & Lee, E. S. (2020). An assessment of the load modifying potential of model predictive controlled dynamic facades within the California context. Energy and Buildings, 210, 109762.

Wang, T., Ward, G., & Lee, E. S. (2018). Efficient modeling of optically-complex, non-coplanar exterior shading: Validation of matrix algebraic methods. Energy and Buildings, 174, 464-483.

Lee, E. S., Geisler-Moroder, D., & Ward, G. (2018). Validation of the Five-Phase Method for Simulating Complex Fenestration Systems with Radiance against Field Measurements.

McNeil, A., & Lee, E. S. (2013). A validation of the Radiance three-phase simulation method for modelling annual daylight performance of optically complex fenestration systems. Journal of Building Performance Simulation, 6(1), 24-37.



13:54 - 14:00

Functional mock-up unit based generic threat injection framework for energy efficient and resilient buildings

Yangyang Fu, Xing Lu, Zheng O'Neill

Texas A$M University, United States of America

Aim and Approach

(max 200 words)

Fault detection and diagnosis in building energy and control systems relies much on accessible faulty data, which can be obtained from experiments or dynamic simulation. While experiments that inject faults to actual building energy and control systems are risky to building operators, high-fidelity dynamic modeling and simulation can provide flexibility of injecting all types of faults to the system in order to investigate their impact.

EnergyPlus as a popular building energy simulator has integrated plenty of faulty models. However, the development exposed several limitations. First, it is difficult to model control-related faults, such as inappropriate PID parameters and communication delays etc. Second, the injection of pressure-involved faults is not straightforward. For example, to model duct fouling, users need to adjust the pressure head, minimum and maximum air flowrate in the fan model. Third, faults cannot be injected in the middle of simulation. Forth, fault evolution patterns are not considered.

This paper presents a flexible FMU-based fault injection framework that can address the above-mentioned issues. This framework will consider a comprehensive list of typical faults in the HVAC system. The injection method will be mathematically presented. Component-level and system-level faulty modeling and simulation will be demonstrated in the case study.

Scientific Innovation and Relevance

(max 200 words)

This paper presents a flexible and generic functional mock-up unit (FMU) based threat injection frame- work in Python. FMU is a simulation model that is compliant to the functional mock-up interface (FMI), which defines a standardized interface for a model so that such a model can be simulated in a different en- vironment for model exchange or co-simulation. The utilization of FMU in this paper is to provide a standardized model interface for different building energy simulators so that a generic threat library can be connected to it and perform threat evaluation. The proposed framework can address most of the identified research gaps in terms of threat modeling and simulation. Currently, EnergyPlus, TRNSYS and Modelica models can all be compiled to FMU models. Therefore, this proposed framework can be reused for all three simulators with minimum modifications when the different software structures of these three simulators should be taken into considerations.

Preliminary Results and Conclusions

(max 200 words)

This paper presents a generic and flexible threat in- jection framework based on Functional Mock-up Unit (FMU). Several Python libraries are developed to address automating the generation of FMU wrap- per models, define threats and threat injection, and perform step-wise simulation for both baseline and threat-injected systems. The standalone definition of threats and the use of FMU as modeling base enables this framework to be easily expanded to other FMU- supported building energy simulators. The framework is demonstrated in a Modelica/FMU environment by modeling and simulating single-/multiple- order of threats. The simulation results show that the framework can provide flexibility of injecting different types of threats at different starting time for different duration.

Main References

(max 200 words)

Kim, Janghyun, Stephen Frank, James E. Braun, and David Goldwasser. "Representing small commercial building faults in EnergyPlus, Part I: Model development." Buildings 9, no. 11 (2019): 233.

Kim, Janghyun, Stephen Frank, Piljae Im, James E. Braun, David Goldwasser, and Matt Leach. "Representing Small Commercial Building Faults in EnergyPlus, Part II: Model Validation." Buildings 9, no. 12 (2019): 239.

Granderson, Jessica, Guanjing Lin, Ari Harding, Piljae Im, and Yan Chen. "Building fault detection data to aid diagnostic algorithm creation and performance testing." Scientific data 7, no. 1 (2020): 1-14.

Li, Yanfei, and Zheng O’Neill. "A critical review of fault modeling of HVAC systems in buildings." In Building Simulation, vol. 11, no. 5, pp. 953-975. Tsinghua University Press, 2018.

Zhang, Rongpeng, and Tianzhen Hong. "Modeling of HVAC operational faults in building performance simulation." Applied Energy 202 (2017): 178-188.



14:00 - 14:06

Real-time light simulation methodology for expedited comparison and optimization studies through agent-based photon modeling

Vishal Vaidhyanathan, Jichen Wang

Carnegie Mellon University, United States of America

Aim and Approach

(max 200 words)

Daylight Simulations for illumination analyses for tasks involving comparative evaluation and optimization are time taking due to their dependency on external simulation engines. Conventional daylight analysis workflows for quantitative analysis in early phase design decisions for multiple design iterations can be cumbersome and require running thousands of test cases to optimize a workspace for point in time illuminance. This research proposes a methodology for utilizing light simulation as a metric for optimization that uses agent based modelling of photon particles, mimicking their properties using certain policies to generate instantaneous daylight illumination results for comparative and qualitative analysis of architectural spaces. This algorithm removes external dependencies of simulation engines and runs on the CPU. Since it works with direct geometry collision and particle based methods, it works with non-orthogonal and non-linear geometries as well at real-time. A large analysis test mesh can also be executed in real time using multi-threading of particle groups. This enables quick performance optimization for real time design decision making. It also consists of a front end interface for non-intuitive users to perform comparative studies with ease where fitness graphs are visualized in real time.

Scientific Innovation and Relevance

(max 200 words)

Daylight simulations for tasks involving comparative evaluation, optimization, etc. - are time taking. They require external simulation-engine dependencies and take a while to return results. While the required tasks are comparative in nature, and multiple design options are to be rapidly evaluated and ranked, the performance evaluation process must be expedited (almost real-time), easy to setup (non-intuitive) and computationally non-intensive. In such scenarios, it is also sufficient if results are not absolute but rather comparative and “rankable”. Agent Based Modelling (ABMs) is an approach to model systems composed of autonomous and interacting agents that have a set of policies or behavioral rules and are constrained in an environment. Parallels can be drawn between ABM and the properties of light particles (Photons). This research project explores if ABM can be used to emulate light behavior to an extent that it is possible to compare multiple design options and make daylight performance directed design decisions, in an expedited way and in real-time. The amount of illuminance and its uniformity are the two most critical criteria to validate good lighting conditions and are thus evaluated in this tool. The main inputs for this tool are the sun angle and geometry surfaces.

Preliminary Results and Conclusions

(max 200 words)

This methodology with light simulation and optimization is flexible and extremely user friendly where input parameters can be changed during the design process that gives quantifiable standards for comparing design options. The tool gives the designer the power to come up with design choices by themselves, where the evaluation tool is just a helper for decision making. By alternating the input parameters and the definition of geometries, it can also be applied in other kinds of design processes. With this agent based algorithm, one can simulate the light behavior with a rough accuracy and fast calculation. Given that there are already plenty of well established light simulation tools in the field, a comparison was made between this tool and ladybug for grasshopper. The results though not as accurate as ladybug, resulted in the run time being very short which was almost comparable to a real time calculation. Moreover the input settings are much simpler than other tools given that it can take any surface type objects as the geometry input. These benefits make this simulator a very good tool in daylight evaluation and optimization by exploring a large amount of possibilities with multiple geometrical inputs to find the optimized solution.

Main References

(max 200 words)

1. Illuminating Engineering Society. IES Approved Method: Spatial Daylight Autonomy (sDA) and Annual Sunlight Exposure (ASE). New York: IES; 2012. (IES LM-83-12).

2. Chamilothori, Kynthia & Chinazzo, Giorgia & Rodrigues, João & Dan-Glauser, Elise & Wienold, Jan

& Andersen, Marilyne. (2019). Subjective and physiological responses to façade and sunlight pattern

3. United States Green Building Council (USGBC), retrieved 2020.



14:06 - 14:12

Radiant spectral energy for simulation in the built environment

Joseph Del Rocco, Joseph T. Kider Jr.

University of Central Florida, School of Modeling, Simulation, and Training

Aim and Approach

(max 200 words)

Fast and accurate daylighting and energy performance simulations are crucial for real-time control systems (RCS) in the built environment. RCS include responsive facades, adaptive e-glass, and smart HVAC control systems. State-of-the-art building monitoring systems driving the RCS require spectral energy inputs to take full advantage of both light and heat of solar and sky radiation. Additionally, building designers also require spectral energy for building and fenestration design and material decisions relating to circadian rhythm daylighting. Unfortunately, modern daylighting and energy simulations often neglect spectral energy in efforts to reduce computation time, and therefore do not provide the full global-illumination and energy solution needed by fine-grained RCS. We present an accurate, interactive, physically-based approach that utilizes a Transition Portal radiosity engine (Kider et al., 2019) to compute a full global-illumination solution with spectral energy for building simulations in real-time. This approach allows building simulations to leverage the visible spectrum for more accurate daylighting and the full spectrum for energy analysis. This novel method is intended for real-time control systems in the built environment and building adaptation for circadian daylighting.

Scientific Innovation and Relevance

(max 200 words)

Modern building performance simulation often ignores real-time spectral energy input for various reasons. The calculations can be time-consuming and the data needs to be transformed and marshaled through a diverse array of legacy and modern daylighting and energy software systems (modeling packages, fenestration software, OpenStudio, EnergyPlus, Radiance, etc.). However, daylighting and energy performance software must eventually use full-spectrum solar and sky radiation to provide the most accurate results. Doing so may even help to simplify the building performance pipeline. We have anticipated this and progressively worked towards a solution for fine-grained real-time control systems manipulating both light and heat from the same spectral inputs. We have developed a novel system that interfaces with our radiosity engine which processes the spectral inputs and produces spectral outputs and can interface with modern tools. Spectral energy is generated from sky conditions through a machine-learned model (Del Rocco, 2020) from both real-world skydome photographs and synthetic sky renders to demonstrate our system under varied times of day and sky conditions. We also propose and simulate a novel adaptive e-glass solution that filters heat during warm climates but allows it during cool climates, to maximize energy performance with natural heating.

Preliminary Results and Conclusions

(max 200 words)

We performed daylighting simulations and energy analysis on three separate building spaces with the following real-time control systems: automated blinds, a kinetic facade, and various e-glass configurations. Our preliminary results show that our spectral energy system is definitely fast enough to be used by building monitoring systems to drive real-time control systems, as well as by building designers using parametric design to plan shading devices and materials. We also demonstrate that our proposed adaptive e-glass solution can properly filter heat energy from the upper spectra during warm climates (allowing it during cool climates) based on configured limits, contributing to an overall more energy-efficient building performance solution. A cyber-physical prototype building monitoring system with e-glass control is proposed, which monitors the skydome with an all-sky camera as the root input of spectral energy, and then simulates the propagation of the spectral energy input throughout the building efficiently. These experiments demonstrate the feasibility of a system that can handle spectral energy calculations for building performance simulations and control systems in real-time.

Main References

(max 200 words)

Kider, J. T., Walter, B., Fang, S., Sekkin, E., & Greenberg, D. P. (2019). Transition Portal for daylighting calculations in early phase design. Energy and Buildings.

Del Rocco, J., Bourke, P. D., Patterson, C. B., & Kider, J. T. (2020). Real-time spectral radiance estimation of hemispherical clear skies with machine learned regression models. Solar Energy.

S. F. Rockcastle, M. Danell, L. Petterson, and M. L. Ámundadóttir (2020). The Impact of Behavior on Healthy Circadian Light Exposure Under Daylight and Electric Lighting Scenarios. ACEEE Summer Study on Energy Efficiency in Buildings 2020.

Balakrishnan, P., & Jakubiec, J. A. (2019). Spectral Rendering with Daylight: A Comparison of Two Spectral Daylight Simulation Platforms. IBPSA Conference.

Rockcastle, S., Ámundadóttir, M. L., & Andersen, M. (2019). The Case for Occupant-Centric Daylight Analytics: a Comparison of Horizontal Illumination and Immersive View.International Conference of the International Building Performance Simulation Association

Adaptive Lighting for Alertness (ALFA) Software. https://solemma.com/Alfa.html

Inanici, M., and ZGF Architects LLP (2015). LARK Spectral Lighting plugin to Grasshopper. https://faculty.washington.edu/inanici/Lark/Lark_home_page.html

Ward, G. J. (1994, July). The RADIANCE lighting simulation and rendering system. In Proceedings of the 21st annual conference on Computer graphics and interactive techniques.



14:12 - 14:18

Short-term forecasting of building energy consumption with deep generative learning

Yichuan X. Ma

The University of Hong Kong, Hong Kong S.A.R. (China)

Aim and Approach

(max 200 words)

Short-term building energy consumption forecasting is highly valuable from both a technical and an economic point of view. In this paper, a deep generative learning model (GAN-Plus) taking account of short-term future meteorological data is proposed to accurately forecast building energy consumption in the near future. A conventional machine learning model (multilayer perceptron, MLP) and a non-meteorology-based version of the proposed model (GAN-Zero) were developed and comparatively tested as baseline models. Multi-year hourly meteorological data and actual energy consumption measurements from two office buildings in Shanghai were used for modelling and testing.

Scientific Innovation and Relevance

(max 200 words)

(1) A novel generative short-term forecasting framework was delineated.

(2) Customised conditional GAN-based model with a 1D-UNet generator under the generative short-term forecasting framework were developed and quantitatively evaluated under multiple granularity settings.

(3) Historical and future meteorological information was taken into consideration to improve forecasting accuracy.

(4) Cross-case generalisability, which is an underappreciated issue in the field of building energy forecasting, was discussed.

Preliminary Results and Conclusions

(max 200 words)

State-of-the-art accuracy and decent cross-case generalisability of the proposed GAN-based models were demonstrated. The proposed model outperformed the chance level and all the baseline models, achieving accuracies of 85.48% with hourly granularity and 94.77% with daily granularity. MLP failed in the cross-case forecasting task with an hourly CV-RMSE notably larger than 30%. Though similar in terms of performance in hourly forecasting, GAN-Plus showed superior accuracy (with a difference > 1%) than GAN-Zero in daily forecasting. In the noise robustness analysis, GAN-Plus demonstrated strong capability in dealing with the uncertainties in the future meteorological clues, with less than 1.5% deterioration of accuracy with a noise level of 20%. The advantage and potential risks of using the proposed GAN-Plus model were further discussed.

Main References

(max 200 words)

[1] Deb, C., F. Zhang, J. Yang, S. E. Lee and K. W. Shah (2017). "A review on time series forecasting techniques for building energy consumption." Renewable and Sustainable Energy Reviews 74: 902-924.

[2] Fan, C., Y. Sun, Y. Zhao, M. Song and J. Wang (2019). "Deep learning-based feature engineering methods for improved building energy prediction." Applied energy 240: 35-45.

[3] Goodfellow, I. J., J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville and Y. Bengio (2014). "Generative adversarial networks." arXiv preprint arXiv:1406.2661.

[4] Ma, Y. X. and C. Yu (2020). "Impact of meteorological factors on high-rise office building energy consumption in Hong Kong: From a spatiotemporal perspective." Energy and Buildings 228: 110468.

[5] Mirza, M. and S. Osindero (2014). "Conditional generative adversarial nets." arXiv preprint arXiv:1411.1784.

[6] Monfet, D. and P. Radu Zmeureanu PhD (2009). "Calibration of a building energy model using measured data." ASHRAE Transactions 115: 348.

Ronneberger, O., P. Fischer and T. Brox (2015). U-net: Convolutional networks for biomedical image segmentation. International Conference on Medical image computing and computer-assisted intervention, Springer.



 
Contact and Legal Notice · Contact Address:
Privacy Statement · Conference: Building Simulation 2021
Conference Software - ConfTool Pro 2.6.142+TC
© 2001 - 2021 by Dr. H. Weinreich, Hamburg, Germany