Seasonal effects of input parameters in urban-scale building energy simulation

Available online at www.sciencedirect.com Available online at www.sciencedirect.com

Available online at www.sciencedirect.com Available online www.sciencedirect.com Available online atatwww.sciencedirect.com

ScienceDirect Energy Procedia 00 (2016) 000–000 ScienceDirect Energy Procedia (2016) 000–000 Energy Procedia 122 (2017) 433–438 Energy Procedia 00 00 (2017) 000–000 Energy Procedia 00 (2016) 000–000

www.elsevier.com/locate/procedia www.elsevier.com/locate/procedia www.elsevier.com/locate/procedia www.elsevier.com/locate/procedia

CISBAT 2017 International Conference Future Buildings & Districts Energy Efficiency from Nano to International Urban Scale, CISBAT 2017 6-8 September & 2017, Lausanne, Switzerland from CISBAT CISBAT 2017 2017 International Conference Conference Future Future Buildings Buildings & Districts Districts Energy Energy Efficiency Efficiency from Nano to Urban Scale, CISBAT 2017 6-8 September 2017, Lausanne, Switzerland Nano to Urban Scale, CISBAT 2017 6-8 September 2017, Lausanne, Switzerland Seasonal effects of input parameters in urban-scale building energy

Seasonal input in urban-scale building simulation Seasonal effects effects ofInternational input parameters parameters inDistrict urban-scale building energy The 15thof Symposium on Heating and Cooling energy simulation Mart´ın Mosteiro-Romeroa,∗,simulation Jimeno A. Fonsecaa,b , Arno Schluetera,b Assessing the feasibility of using the heat demand-outdoor a,b Architecture and Building Systems, a,∗ ETH Zurich, Stefano-Franscini-Platz 1, ,Zurich 8093, Switzerland a,b Mart´ ı n Mosteiro-Romero , Jimeno A. Fonseca Arno Schlueter a,∗ a,b a,b Mart´ ı n Mosteiro-Romero , Jimeno A. Fonseca , Arno Schlueter Future Cities Laboratory, Singapore-ETH Centre, 1 Create Way, Singapore 13892, Singapore temperature function for a long-term district heat demand forecast Architecture and Building Systems, ETH Zurich, Stefano-Franscini-Platz 1, Zurich 8093, Switzerland a

b a a Architecture and Building Systems, ETH Zurich, Stefano-Franscini-Platz 1, Zurich 8093, Switzerland b Future Cities Laboratory, Singapore-ETH Centre, 1 Create Way, Singapore 13892, Singapore b Future Cities Laboratory, Singapore-ETH Centre, 1 Create Way, Singapore 13892, Singapore

Abstract a

I. Andrića,b,c*, A. Pinaa, P. Ferrãoa, J. Fournierb., B. Lacarrièrec, O. Le Correc

IN+ Center for Innovation, Technology and Policy Research - Instituto Superior Técnico, Av. Rovisco Pais 1, 1049-001 Lisbon, Portugal

Urban Building Energy Models areRecherche powerful&tools for estimating future states of energy b Abstract Veolia Innovation, 291 Avenue Dreyfous Daniel, 78520consumption Limay, Franceand energy generation in Abstract c to the complexity of these systems, large amounts of data are required, which are often incomplete or unavailable. buildings. Due Département Systèmes Énergétiques et Environnement - IMT Atlantique, 4 rue Alfred Kastler, 44300 Nantes, France Urban Building Energy Models are powerful tools for estimating future states of energy consumption and amount energy generation in Through the implementation of building archetypes, models such as the City Energy Analyst minimize the input data. Urban Building Energy Models are powerful tools for estimating future states of energy consumption and energy of generation in buildings. Due to the complexity of these systems, large amounts of data are required, which are often incomplete or unavailable. However, these simplifications inherently uncertainty the expected results. buildings. Due to the complexity of these increase systems,the large amounts of data are required, which are often incomplete or unavailable. Through the implementation of building archetypes, modelsproperties such as the City Energy Analyst minimizedensity the amount of input leakdata. This paper presents a sensitivity analysis of architectural (window-to-wall ratio, occupant and envelope Through the implementation of building archetypes, models such as the City Energy Analyst minimize the amount of input data. However, these simplifications inherently increase the uncertainty of the expected results. iness), thermal (U-values, G-values, thermal mass and emissivity of building Abstract However, these properties simplifications inherently increase the uncertainty of the expected results.surfaces), operating parameters (set point This paper and presents a sensitivity analysis of architectural properties ratio, occupant density and leaktemperatures ventilation rates) and internal loads (heat gains due to(window-to-wall occupancy, appliance use and lighting). Forenvelope this, the study This paper presents a sensitivity analysis of architectural properties (window-to-wall ratio, occupant density and envelope leakiness), thermal properties (U-values, G-values, thermal mass and emissivity ofofbuilding surfaces), operating parameters (set point combines a two-step process of sensitivity analysis with Saltelli’s extension the Sobol method and the City Energy Analyst. iness), thermal properties G-values, thermalinmass emissivity of building surfaces), operating parameters (set point District heating networks(U-values, are commonly addressed the and literature as one of the most effective solutions for decreasing the temperatures and ventilation rates) and internal loads gains due to occupancy, comprising appliance use and lighting).with Forpredominantly this, the study The methodology is appliedfrom to athe case study area in (heat central Zurich, Switzerland, 284 buildings temperatures and emissions ventilation rates) and internal loads (heat due to occupancy, appliance use and lighting). Forthrough this, thethe study greenhouse gas building sector. Thesegains systems require high investments which are returned heat combines a two-step process of sensitivity analysis with Saltelli’s extension of the Sobol method and the City Energy Analyst. educational, hospital and residential uses. combines a two-step process of sensitivity analysis Saltelli’s extensionpolicies, of the Sobol method and the future City Energy sales. Due to the changed climate conditions andwith building renovation heat demand in the could Analyst. decrease, The methodology is applied tocooling a case demand study area in central Zurich, Switzerland, comprising with predominantly The results showed that the in the area was very strongly influenced by the284 set buildings point temperature, with other The methodology is appliedreturn to a case study area in central Zurich, Switzerland, comprising 284 buildings with predominantly prolonging the investment period. educational, hospital and residential uses. variables having aofrelatively minor influence. For the heating case larger number– outdoor of variables were needed in order to explain educational, hospital andpaper residential uses. the feasibility The main scope this is to assess of was using thea strongly heat demand temperature for heat demand The results showed that the cooling demand in the area very influencedrates by the set buildings. point function temperature, with other variations in demand, primarily the thermal properties of the envelope and air exchange of the This was generally The results showed that the cooling demand in the area was very strongly influenced by the set point temperature, with other forecast. The district of Alvalade, located in Lisbon (Portugal), was used as a case study. The district is consisted of 665 variables having a relatively minor influence. For the heating case a larger number of variables were needed in order to explain true for allhaving occupancy types, shapes, sizes and locations, showing the ofscenarios accurate estimates ofneeded these parameters urban variables a relatively minor influence. For and the heating case aimportance larger number of variables were in order toinexplain buildings that vary in both construction period typology. Three weather (low, medium, high) and three district variations in demand, primarily the thermal properties of the envelope and air exchange rates of the buildings. This was generally building energy modeling. a the broader sense, theintermediate, results contribute the development of urban energy simulations that are both variations in scenarios demand, primarily thermal properties of the envelopetoand exchange ofobtained the buildings. This was generally renovation wereOn developed To air estimate the rates error, heat demand values were true for alland occupancy types, shapes, sizes(shallow, and locations, showingdeep). the importance of accurate estimates of these parameters in urban practical accurate. true for all occupancy types, shapes, sizes and locations, showing the importance of accurate estimates of these parameters in urban compared with results from a dynamic heat demand model, previously developed and validated by the authors. building energy modeling. On a broader sense, the results contribute to the development of urban energy simulations that are both ©The 2016 The Authors. Published by Elsevier Ltd. building energy modeling. On a only broader sense, the results contribute the to the development of urban simulations are both results showed that when weather change is considered, margin of error could be energy acceptable for somethat applications practical andunder accurate. Peer-review responsibility of the scientific committee of the CISBAT 2017 International Conference Future Buildings & practical andinaccurate. (the error annual demand was lower than 20% for all weather scenarios considered). However, after introducing renovation © 2016 The Authors. Published by Elsevier Ltd. Scale. Districts Energy Efficiency from Nano to Urban © 2016 The Authors. Published by Elsevier Ltd. © 2017 Thethe Authors. Published byofElsevier Ltd. committee scenarios, error value increased upscientific to 59.5% (dependingofon weather2017 and International renovation scenarios combination considered). Peer-review under responsibility the thethe CISBAT Conference Future Buildings & of the scientific scientific committee theCISBAT CISBAT 2017 International Conference Future Buildings Peer-review under responsibility of the committee ofofthe Conference Future Buildings The value of slope coefficient increased on average within range of2017 3.8%International up to 8% per decade, –that corresponds to&& the Districts Energy Efficiency frommodels, Nano to Urban Scale. Urban building energy Sensitivity analysis, Sobol the method, City Energy Analyst Keywords: Efficiency totoUrban Scale. Districts Energy Efficiency fromNano Nano Urban Scaleduring the heating season (depending on the combination of weather and decrease–Energy in the number of from heating hours of 22-139h building energy models, Sensitivity analysis, Sobol method, Cityincreased Energy Analyst Keywords: renovation Urban scenarios considered). On the other hand, function intercept for 7.8-12.7% per decade (depending on the Keywords: Urban building energy models, Sensitivity analysis, Sobol method, City Energy Analyst coupled scenarios). The values suggested could be used to modify the function parameters for the scenarios considered, and improve the accuracy of heat demand estimations.

1. Introduction

© Introduction 2017 The Authors. Published by Elsevier Ltd. 1. 1.Peer-review Introduction Urban Building Energy Models (UBEM) areCommittee expectedof toThe become a key planning tool for public utilities, municipalunder responsibility of the Scientific 15th International Symposium on District Heating and ities, urban planners and architects. Currently, the two major obstacles UBEM must tackle are input data availability Cooling. Urban Urban Building Building Energy Energy Models Models (UBEM) (UBEM) are are expected expected to to become become aa key key planning planning tool tool for for public public utilities, utilities, municipalmunicipalities, urban planners and architects. Currently, the two major obstacles UBEM must tackle are input Keywords: demand;and Forecast; Climate Currently, change ities, urbanHeat planners architects. the two major obstacles UBEM must tackle are input data data availability availability ∗

Corresponding author. E-mail address: [email protected] ∗ Corresponding author. ∗ Corresponding author. E-mail address: [email protected] 1876-6102 © 2016 The Authors. Published by Elsevier Ltd. E-mail address: [email protected] 1876-6102 under © 2017 The Authors.ofPublished by Elsevier Ltd.of the CISBAT 2017 International Conference Future Buildings & Districts Energy Peer-review responsibility the scientific committee 1876-6102 © under 2016 The Published by Elsevier Ltd. of The 15th International Symposium on District Heating and Cooling. Peer-review responsibility of the Scientific Committee Efficiency from Nano to Authors. Urban Scale. © 2016 The Authors. Published by Elsevier Ltd. Ltd. 1876-6102 2017 The Authors. Published by Elsevier Peer-review under responsibility of the scientific committee of the CISBAT 2017 International Conference Future Buildings & Districts Energy Peer-review underresponsibility responsibility of the scientific committee the CISBAT 2017 International Conference – Future&Buildings Peer-review under of the scientific committee of theofCISBAT 2017 International Conference Future Buildings Districts & Energy Efficiency–from NanoEfficiency to Urban Scale. Districts Energy from Nano to Urban Scale Efficiency from Nano to Urban Scale. 10.1016/j.egypro.2017.07.459

434 2

Martín Mosteiro-Romero et al. / Energy Procedia 122 (2017) 433–438 M. Mosteiro-Romero et al. / Energy Procedia 00 (2016) 000–000

and input uncertainty [1]. Input data uncertainty is a key aspect of UBEM and as such, UBEM inputs hold a fundamental role for accurate model predictions. Computational models such as the City Energy Analyst (CEA) [2,3] are powerful tools to estimate future states of energy consumption and energy generation in districts accounting for hourly exchanges of energy among buildings, users and the environment. Such a level of detail requires a vast number of user inputs, including information about 3D geometry, materials, occupancy and HVAC components of buildings. However, these input data are often incomplete or unavailable. Through a wide database of building properties or archetypes, UBEM such as the CEA aim to minimize these inputs. However, these simplifications in input parameters inherently increase the uncertainty of the expected results. Sensitivity analysis studies how the uncertainty of input parameters is assigned to different output parameters. Sensitivity methods are classified into local and global methods. While local methods evaluate the effect of one input on one output, global sensitivity methods sample the complete input space and are therefore able to calculate overarching sensitivities [4]. Global sensitivity analysis is a generic term referring to various sensitivity methods. The Morris method is a one-at-a-time method widely used to rank a set of input variables according to their qualitative influence on the output of a computational model [5,6]. In contrast to the Morris method, the Sobol method is a variance-based method that furthermore estimates the percentage of variance caused by the variability of a certain input [7,8]. It is useful to determine, in a quantitative way, the effects of an input variable on the output of a computational model. The objective of this paper is to analyze the seasonal effects (heating and cooling season) of architectural properties (window-to-wall ratio, occupant density, envelope leakiness), thermal properties (U-values, G-values, thermal mass and emissivity of building surfaces), operating parameters (set point temperatures and ventilation rates) and internal loads (heat gains due to occupancy, appliance use and lighting) on the demand for heating and cooling of urban areas. For this, the study combines a two-step process of sensitivity analysis using the Sobol method and the UBEM City Energy Analyst [9]. The method is applied to an area in central Zurich, known as the Hochschulquartier. The area hosts 284 buildings in the residential, educational, services and healthcare sectors.

2. Method 2.1. Data collection The CEA demand is based on a resistance-capacitance model of the buildings in a district and on the application of construction archetypes to minimize the amount of input data required. The necessary information about 3D geometry, materials, occupancy and mechanical components was obtained from GIS data, owner information and the archetype database. Data on energy-relevant retrofits for the main building components was scarce and thus estimated. Table 1 presents the probability density functions of input variables and their references. Twenty-three different variables and their corresponding probability density functions were selected to cover different sources in uncertainty, including the effects of buildings’ architectural and thermal properties, operating parameters and internal loads. The means and standard deviations shown in the table were calculated assuming a triangular distribution from published minimum and maximum values. 2.2. Sensitivity analysis The Saltelli series [7] was used to create stratified samples out of the probability distributions in Table 1 with a sample size N of 1000. CEA was executed on every sample to determine yearly heating and cooling needs for each building. The Sobol method was applied on the data with the computational implementation of SALib [16]. The number of simulations needed for this methodology depends strongly on the number of variables sampled, however, with N · (2k + 2) simulations required for k variables and a sample size N. Thus, in order to reduce the computational time required, pre-screening was used to select the most sensitive variables to the yearly consumption of heating and cooling in buildings. First, a reduced case study was created that consisted of ten representative buildings covering the main usage types in the area, construction years and building sizes. CEA was then executed on 48’000 samples to determine yearly heating and cooling needs of these ten representative buildings. The parameters that caused 90% of the observed effects on the demand were selected. Through pre-screening, the number of variables



Martín Mosteiro-Romero et al. / Energy Procedia 122 (2017) 433–438

435

M. Mosteiro-Romero et al. / Energy Procedia 00 (2016) 000–000

3

Table 1: Probability density function parameters for key input variables. Variable Name Window-to-wall ratio Occupant density Air change rate at 50 Pa Overall thermal transmittance coefficient of basement ceiling a Overall thermal transmittance coefficient of exterior walls a Overall thermal transmittance coefficient of roof a Overall thermal transmittance coefficient of windows a Solar energy transmittance of window glazing Internal heat capacity of building Emissivity of external walls b Emissivity of windows b Emissivity of roofs b Solar absorption coefficient of external walls b Solar absorption coefficient of roof b Ratio of gross floor area that is heated or cooled Set-point temperature for space cooling Set-point temperature for space heating Set-back temperature for space cooling Set-back temperature for space heating Minimum air ventilation rate per person Sensible heat gain due to occupancy Maximum electrical consumption due to appliances Maximum electrical consumption due to lighting a b

Distribution Symbol win − wall Occ n50 Ubase Uwall Uroo f Uwin Gwin Cm ewall ewin eroo f awall aroo f Hs T cs,set T hs,set T cs,setback T hs,setback Ve Qs Ea El

Unit [-] [m2 /p] [h−1 ] [W/m2 -K] [W/m2 -K] [W/m2 -K] [W/m2 -K] [-] [Wh/m2 -K] [-] [-] [-] [-] [-] [-] [°C] [°C] [°C] [°C] [L/s] [W/p] [W/m2 ] [W/m2 ]

µ 0.36 14 3.17 0.62 0.35 0.29 1.78 0.70 55 0.91 0.60 0.78 0.62 0.52 0.80 26 21 28 12 10 70 7 15.9

min 0.20 8 1 0.15 0.11 0.09 0.90 0.50 22 0.84 0.02 0.09 0.3 0.3 0.75 22 20 22 12 8 60 3 11.6

Ref. max 0.90 14 6 2 1.5 1 3.1 0.85 103 0.95 0.89 0.95 0.85 0.85 0.95 28 24 28 24 12 90 10 16.9

[10] [11] [10] [10] [10] [10] [10] [11] [12] [13] [13] [13] [13] [13] [14] [10,15] [10,15] [15] [15] [10,15] [10] [10] [10]

Includes linear transmittance losses (10% more of standard value). Calculated for different typical building materials.

to be sampled was reduced from 23 parameters to 11. The Saltelli series was then used again to create 24,000 samples out of the probability distributions of these variables and CEA was again executed for the 284 buildings in the area. For each variable i, the Sobol method generates an associated sensitivity measure S i (first order sensitivity coefficient) and a total effect index S T i , which measures the total (i.e., first and higher order) effects. In order to analyze not only the direct effects of a variable on the model results but also indirect interactions, in the following results section the total effect index S T i is used to quantify the primary and secondary effects of each of these variables on the demand of each building in the area.

3. Results 3.1. Effect of occupancy Building archetypes such as the ones used by CEA assume different thermal and architectural properties, system set points and controls, and internal gains for different occupancy types. Thus, the sensitivity of each of these parameters was analyzed for the various occupancy types found in the case study. For this, the following generalized building occupancy types were defined: educational (including classrooms, libraries and laboratories), hospital (including hospital laboratories), office (including both private and university office spaces) and residential (including multidwelling units and hotels). Other usages found in the area, such as exhibition and retail spaces, were scarce and thus not included in this analysis. Figures 1 and 2 present a comparison of seasonal effects (winter and summer) on heating and cooling consumption for the area of study. The results are categorized by the type of occupancy. For the heating case, the building envelope’s thermal properties are responsible for more than half of the variation, while the air exchange rate is responsible for about a quarter and the rest of the parameters have a much smaller influence. Hospital buildings were found to be particularly sensitive to the air exchange rate, whereas window U-

Martín Mosteiro-Romero et al. / Energy Procedia 122 (2017) 433–438 M. Mosteiro-Romero et al. / Energy Procedia 00 (2016) 000–000

436 4

Total

Education

Ubase

Uwall

Hospital

Uwin

Gwin

win − wall

Office

n50

T hs,set

Residential

T hs,setback

Fig. 1: Sensitivity effects on yearly space heating demand (MWh/yr) for the total area and per occupancy type.

Total

Education

Ubase

Uwall

Uwin

Gwin

Hospital

win − wall

n50

Office

T cs,set

T cs,setback

Fig. 2: Sensitivity effects on yearly space cooling demand (MWh/yr) for the total area and per occupancy type. (Note: residential buildings typically do not have cooling in Zurich.)

values had a much greater influence on residential buildings. Set point temperatures were generally less influential, while setback temperatures only had an effect in educational and office buildings. For the cooling case, the set point temperature was responsible for the vast majority of the variation, especially in hospital buildings. The window-to-wall ratio and window U-values showed a relatively high effect, in particular in educational and office buildings. All other variables had a rather small impact. 3.2. Effect of building shape The case study area also presented a variety of building scales and typologies. Thus, the correlations to building shape were also analyzed. The buildings were then categorized by their envelope factor [17] as compact (envelope factor less than 0.8), medium (0.8 – 1.4) and non-compact (greater than 1.4). Of all the heated buildings in the area, 15% are compact, 48% medium, and 36% are non-compact. The results in Figures 3 and 4 show a strong correlation of seasonal effects to building compactness, with less compact buildings showing a greater effect of building envelope parameters such as U-values and window-to-wall ratios. Compact buildings, on the other hand, showed a greater influence of the set point temperatures and air exchange rates. For the cooling season, however, the set point temperature was again most influential in all cases. Compact

Ubase

Uwall

Medium

Uwin

Gwin

win − wall

Non-compact

n50

T hs,set

T hs,setback

Fig. 3: Sensitivity effects on yearly space heating demand (MWh/yr) per level of compactness.



Martín Mosteiro-Romero et al. / Energy Procedia 122 (2017) 433–438 M. Mosteiro-Romero et al. / Energy Procedia 00 (2016) 000–000

Compact

Ubase

Uwall

Medium

Uwin

Gwin

win − wall

437 5

Non-compact

n50

T cs,set

T cs,setback

Fig. 4: Sensitivity effects on yearly space cooling demand (MWh/yr) per level of compactness.

A similar effect was seen when analyzing the correlation of building heating and cooling energy demand to buildings’ heated floor area. In particular, building envelope properties had a stronger effect on smaller buildings than larger ones due to their higher surface to volume ratio. 3.3. Spatial effects The spatial distribution on the sensitivity results are represented in Figure 5. For the heating case, the top variable for each building is represented. Large buildings tend to have a stronger influence from air exchange rate in the heating case, while buildings with a higher exposure to sunshine such as those on the top left and in the center of the area show a stronger influence from the window U-values. For the cooling case, since the set point temperature had the greatest effect on all buildings, the second most impactful variable is represented, with the majority of the remaining effects relating to the window properties. Cooling (MWh/yr)

Heating (MWh/yr)

Ubase

Uwin

n50

T cs,setback

Ubase

Uwin

n50

win − wall

Fig. 5: Spatial distribution of the variables with the highest effect on each building for the heating case (left). Since the set point temperature had the highest effect for every building in the cooling season, the map on the right shows the variables with the second highest effect.

4. Discussion and conclusion Sensitivity analysis using Saltelli’s extension of the Sobol method was carried out on an urban area comprising 284 buildings in central Zurich, Switzerland, using the City Energy Analyst urban building energy model. The analysis considered variations in seasons, occupancy type, building shape and size, and spatial distribution. The results showed that the cooling demand in the area was very strongly influenced by the set point temperature, with other variables having a relatively minor influence. For the heating case a larger number of variables were needed

438 6

Martín Mosteiro-Romero et al. / Energy Procedia 122 (2017) 433–438 M. Mosteiro-Romero et al. / Energy Procedia 00 (2016) 000–000

in order to explain variations in demand, with an especially strong influence of the thermal properties of the building envelope and air exchange rates. This was generally the case for all occupancy types, even though variations could be observed for different activities, stressing the importance of these input parameters on the models capability to provide more accurate predictions. It is worth noting that due to the very different of use types within hospital buildings (e.g. operating room, laboratory, bedroom), the sensitivity results presented here might not be representative of each of these uses. In terms of shape and size, small and non-compact buildings generally showed a greater influence of the thermal envelope due to their relatively larger exposed surfaces, while compact and large buildings showed a stronger influence of air exchange rates and set point temperatures. The effects due to spatial distribution were relatively minor. The methodology shown in this paper provides insight into the parameters most relevant for CEA models with different building typologies and occupancy types. The results show which parameters are most sensitive in this scale and thus provide key variables that need to be calibrated in order to accurately predict the demands of urban areas. Further work will be carried out on calibrating these parameters using measured energy demand data for the case study area. Acknowledgements We would like to thank to the CEA developer team for assistance. Special regards to Daren L. Thomas for enhancing connectivity to the High-Performance Computing cluster at ETH Zurich. This research is financed by the EU ERA-NET Cofund Smart Cities and Communities and the National Research Foundation of Singapore under the program Future Cities Laboratory II. References [1] C. F. Reinhart, C. Cerezo Davila, Urban building energy modeling A review of a nascent field, Building and Environment 97 (2016) 196–202. doi:10.1016/j.buildenv.2015.12.001. [2] J. A. Fonseca, T.-A. Nguyen, A. Schlueter, F. Marechal, City Energy Analyst (CEA): Integrated framework for analysis and optimization of building energy systems in neighborhoods and city districts, Energy and Buildings 113 (2016) 202–226. doi:10.1016/j.enbuild.2015.11.055. [3] J. Fonseca, D. Thomas, A. Willmann, A. Elesawy, A. Schlueter, The City Energy Analyst Toolbox V0.1, in: Expanding Boundaries: Systems Thinking in the Built Environment, vdf Hochschulverlag AG an der ETH Zrich, Zurich, 2016, pp. 584–588. doi:10.3218/3774-6. [4] W. Tian, A review of sensitivity analysis methods in building energy analysis, Renewable and Sustainable Energy Reviews 20 (2013) 411–419. doi:10.1016/j.rser.2012.12.014. [5] F. Campolongo, J. Cariboni, A. Saltelli, An effective screening design for sensitivity analysis of large models, Environmental Modelling & Software 22 (10) (2007) 1509–1518. doi:10.1016/j.envsoft.2006.10.004. [6] M. D. Morris, Factorial Sampling Plans for Preliminary Computational Experiments, Technometrics 33 (2) (1991) 161–174. doi:10.1080/00401706.1991.10484804. [7] A. Saltelli, P. Annoni, I. Azzini, F. Campolongo, M. Ratto, S. Tarantola, Variance based sensitivity analysis of model output. Design and estimator for the total sensitivity index, Computer Physics Communications 181 (2) (2010) 259–270. doi:10.1016/j.cpc.2009.09.018. [8] I. Sobol’, Global sensitivity indices for nonlinear mathematical models and their Monte Carlo estimates, Mathematics and Computers in Simulation 55 (1-3) (2001) 271–280. doi:10.1016/S0378-4754(00)00270-6. [9] D. Thomas, J. A. Fonseca, G. Happle, B. K. Sreepathi, M. Mosteiro-Romero, S. Hsieh, C. K. Wee, P. Neitzel, A. Elesawy, Z. Shi, architecturebuilding-systems/CEAforArcGIS: City Energy Analyst 2.2 (2017). doi:10.5281/zenodo.556165. [10] J. A. Fonseca, A. Schlueter, Integrated Model for Characterization of Spatiotemporal Building Energy Consumption Patterns in Neighborhoods and City Districts, Applied Energy 142 (2015) 247–265. [11] SIA 2024: Standard-Nutzungsbedingungen f¨ur Energie- und Geb¨audetechnik., Schweizerischer Ingenieur- und Architektenverein (SIA), Zurich, Switzerland, 2015. [12] ISO 13790: Energy performance of buildings - Calculation of energy use for space heating and cooling., International Organization for Standardization, Switzerland, 2008. [13] T. L. Bergman, A. S. Lavine, F. P. Incropera, D. P. Dewitt, Fundamentals of Heat and Mass Transfer, 7th Edition, John Wiley & Sons, Hoboken, NJ, USA, 2011. [14] L. Girardin, A GIS-based Methodology for the Evaluation of Integrated Energy Systems in Urban Areas, PhD diss., cole Polytechnique Federale de Lausanne (EPFL), Lausanne, Switzerland, 2012. [15] M. H. Kristensen, S. Petersen, Choosing the appropriate sensitivity analysis method for building energy model-based investigations, Energy and Buildings 130 (2016) 166–176. doi:10.1016/j.enbuild.2016.08.038. [16] Will Usher, Jon Herman, Calvin Whealton, David Hadka, xantares, Fernando Rios, bernardoct, Chris Mutel, Joeri van Engelen, Salib/Salib: Launch! (2016). doi:10.5281/zenodo.160164. [17] SIA 380/1: Thermische Energie im Hochbau, Schweizerischer Ingenieur- und Architektenverein (SIA), Zurich, Switzerland, 2009.