Uncertainty analysis in building performance simulation for design support
TL;DR: In this article, a case study is performed based on an office building with respect to various building performance parameters and implications for the results considering energy consumption and thermal comfort are demonstrated and elaborated.
About: This article is published in Energy and Buildings.The article was published on 2011-10-01 and is currently open access. It has received 417 citations till now. The article focuses on the topics: Uncertainty analysis & Sensitivity analysis.
Summary (3 min read)
Jump to: [1. Introduction: Overview UA in BPS] – [2. Prototype description of applying UA] – [3. Case study of applying UA] – [3.1 Crude uncertainty analysis] – [3.1.1 Results crude uncertainty analysis] – [[[FIGURE 1]]] – [[[FIGURE 2]]] – [[[FIGURE 3]]] – [[[FIGURE 4]]] – [3.1.2 Discussion] – [3.2 Uncertainty in physical parameters] – [3.2.2 Robustness analysis] – [[[FIGURE 8]]] – [3.2.3 Stepwise regression and standardized rank regression coefficient] – [[[TABLE 5]]] – [3.3 Uncertainty in design parameters] – [3.4 Uncertainty in scenario parameters] and [4 Discussion and Conclusion]
1. Introduction: Overview UA in BPS
- The effective integration UA in BPS for design information and quality assurance is of high importance and will be discussed further on.
- Marques et al. [2005] for instance evaluate the reliability of passive systems by first identifying the sources of uncertainties and the determination of the important variables.
- Secondly, the uncertainties are propagated through a response surface.
- He quantifies the effects of uncertainty in building simulation by considering the internal temperature, annual energy consumption and peak loads.
- The total and relative impacts, the different groups have, will be demonstrated.
2. Prototype description of applying UA
- The intent of the study, the methodology, and the procedure in detail will be described in the following section.
- The generated files are passed to the BPS tool and the simulation is run 200 times.
- As mentioned, the focus of attention in the presented results is on energy consumption and thermal comfort.
- The same is valid for the weighted underheating hours.
3. Case study of applying UA
- On the one hand, with the help of UA it was aimed to show the effect of one group on the outcome in the uncertainty (normal distribution and range) and the sensitivity (order of most influential parameters).
- The simulation with VA114 was started with Matlab and conducted 200 times with different input files.
- In these files, the sampled parameters for material properties, building geometry, internal heat gains, infiltration rate and exchanged single/ double glazing are saved.
- The histogram and the normality plots are chosen for demonstrating the results of the UA.
- The values achieved in the end, are the indicator for the sensitivity of the parameter.
3.1 Crude uncertainty analysis
- In literature it is distinguished between two types of uncertainty: aleatory and epistemic uncertainties.
- The main focus of interest in the current research is the epistemic uncertainty that is reducible or even resolvable with the help of building performance simulation.
- To cover uncertainties in physical parameters in the presented case study, all material properties have been varied.
- For the uncertainties in design parameters adjustments in the geometry as well as glass surface and glass properties have been made.
- The uncertainties in boundary conditions are covered by internal parameters such as infiltration rate and internal gains (loads people, equipment and lighting).
3.1.1 Results crude uncertainty analysis
- The results will be shown in the beginning for all categories combined including those that address physical, design and scenario uncertainties at the same time.
- The figures on the right demonstrate in how far the distribution matches the assumptions by means of a normality plot.
- Its purpose as described earlier is to graphically assess whether the data follows a normal distribution.
[[FIGURE 1]]
- The results for the annual cooling vary between 1 and 33 kWh/m².
- The normality plot on the right hand side follows a normal distribution.
[[FIGURE 2]]
- The results for the weighted underheating hours vary between 20 and 140h.
- The normality plot on the right hand follows a normal distribution.
[[FIGURE 3]]
- The results for the annual heating vary between 30 and 117 kWh/m².
- The normality plot on the right hand follows a normal distribution.
[[FIGURE 4]]
- The results for the weighted overheating hours vary between 40 and 210h.
- The normality plot on the right hand does not follow a normal distribution.
3.1.2 Discussion
- The observed results are based on a normal distribution on assessed 95% confidence interval for all the parameters.
- The parameters ranked highest, such as infiltration rate, size of the room, etc., need deeper consideration.
- Furthermore the uncertainties addressed will be separated as they deserve focus also considering their difference in assessment.
- The data and knowledge on the various uncertainty types is limited.
- The three different categories of uncertainties differ in their nature, and therefore in the significance they have on simulation, performance, and the building design.
3.2 Uncertainty in physical parameters
- In this section only uncertainties in physical properties will be considered.
- Physical uncertainties refer to physical properties of materials such as thickness, density, thermal conductivity, etc., of wall, roof and floor layers.
- As a matter of fact, they are always there, and thus, inevitable.
- Despite the designers best attempt of quality assurance there will always remain a degree of uncertainty that he has no influence on.
- That means that the thermal insulation is according to article 5.2 and 5.3 Bb and the thermal resistance Rc of the envelope, floor and roof construction should be equal or higher than 2.5m²K/W.
3.2.2 Robustness analysis
- If the distribution has outliers, the assumption and therefore also the parameters estimates, confidence intervals, etc., become unreliable.
- To provide the decision maker with the guarantee of reliable results, a robustness analysis is conducted.
- This is done by iteratively re-weighting least squares.
- The resulting figure shows a scatter plot with two fitted lines.
- Both lines match each other for the performance aspect annual cooling.
[[FIGURE 8]]
- For the performance aspect weighted underheating hours as shown in Figure 9 a mismatch between both regressions is noticeable.
- This mismatch results in less robustness of the model.
- Bringing the right-most data point closer to the least squares line makes the two fitted lines nearly identical.
- The adjusted right-most data point has significant weight in the robust fit.
- For the infiltration rate considered in this study it leads to the conclusion that a variation above 0.05ACH should be assumed.
3.2.3 Stepwise regression and standardized rank regression coefficient
- Additional methods such as linear or non-linear regression models or regression in a stepwise manner also exist.
- Two of them will be exemplified showing sensitivity analysis: the non-linear regression model SRRC and a stepwise regression analysis.
- The coefficient R² is the square of the correlation coefficient between the output of the model and the values used for prediction.
- The regression model is shown for the weighted underheating hours.
- This new parameter is determined based on R² containing the infiltration rate and the remaining variables.
[[TABLE 5]]
- The above steps signify the movements taken in the stepwise regression.
- The steps determine all parameters in a stepwise manner that have the most dominant affect.
- This procedure continues until the consideration of an additional parameter does not lead to an increase of R².
- It can be noticed that infiltration rate already causes a regression coefficient of more than 0.91.
- The further consideration of further parameters only increases the value slightly.
3.3 Uncertainty in design parameters
- Uncertainties in design parameters can be described as design variations that occur during the planning process.
- They are fully determined by the decision maker/ designer himself.
- In the conceptual design, aspects such as building mass (heavy/lightweight) or orientation might be unknown.
- Opposed to this, in the detailed design the designer is more indecisive regarding the type of glazing or the type of system and so on.
- The consideration of design uncertainties could therefore improve and enable design decision support, in particular if it would be augmented by sensitivity analysis.
3.4 Uncertainty in scenario parameters
- Uncertainties in scenario conditions are very different compared to physical and design uncertainties in the sense that they can change during the building's life time.
- Scenario uncertainties or uncertainty in boundary conditions can be sub-divided into internal or external scenario uncertainties.
- Scenario uncertainties are based on a random process.
- In [ de Wit, 2001] for instance, a distinction between energy-friendly user and less energy friendly was conducted.
- At present, models are created dealing with the user behaviour in buildings.
4 Discussion and Conclusion
- A realistic case study has been simulated adapting UA.
- Other advantages are that different sensitivity analysis techniques such as standardized rank regression and stepwise regression are available when using these methods.
- Three different sets of parameters were considered: uncertainty in physical, design, and scenario parameters.
- Their existence is inevitable, however, they can be identified and quantified with measurements and tests.
- The integration of uncertainties in BPS provides evidence based decision support in design team meetings and dialogues with building partners.
Did you find this useful? Give us your feedback
Figures (15)
Table 4 List of the properties for uncertainties in scenario conditions and in design variations. 
Figure 1 Frequency distribution and normality plot of annual cooling when considering uncertainty in all parameters. 
Figure 2 Frequency distribution and normality plot of weighted underheating hours when considering uncertainty in all parameters. 
Figure 11 Frequency distribution of weighted over- and underheating hours when considering uncertainty in design variations as shown in Table. 
Figure 12 Sensitivity plot showing the sensitive parameters based on annual cooling and compared to annual heating. 
Figure 10 Frequency distribution of annual cooling and annual heating when considering uncertainty in design variations as shown in Table. 
Table 3 Description of the material properties and deviations for the roof construction of the case study. 
Table 6 List of the properties for uncertainties in s design variations. 
Table 5 Comparison of stepwise regression analysis and the standardized rank regression coefficient for the 8 most affecting parameters on the weighted underheating hours. 
Figure 9 Robustness analysis comparing robust fit to least square when considering infiltration rate shown in relation to weighted underheating hours. 
Figure 6 Frequency distribution of weighted over- and underheating hours when considering uncertainty in physical parameters as shown in Appendix B. 
Figure 4 Frequency distribution and normality plot of weighted overheating hours when considering uncertainty in all parameters. 
Figure 5 Frequency distribution of annual cooling and annual heating when considering uncertainty in physical parameters as shown in Appendix B. 
Figure 3 Frequency distribution and normality plot of annual heating when considering uncertainty in all parameters. 
Figure 13 Sensitivity plot showing the sensitive parameters based on weighted underheating hours and compared to weighted overheating hours.
Citations
More filters
TL;DR: The review indicates that future researches should be oriented towards improving the efficiency of search techniques and approximation methods for large-scale building optimization problems; and reducing time and effort for such activities.
1,009 citations
TL;DR: In this article, the typical steps of implementation of sensitivity analysis in building analysis are described, and a number of practical issues in applying sensitivity analysis are also discussed, such as the determination of input variations, the choice of building energy programs, how to reduce computational time for energy models.
Abstract: Sensitivity analysis plays an important role in building energy analysis. It can be used to identify the key variables affecting building thermal performance from both energy simulation models and observational study. This paper is focused on the application of sensitivity analysis in the field of building performance analysis. First, the typical steps of implementation of sensitivity analysis in building analysis are described. A number of practical issues in applying sensitivity analysis are also discussed, such as the determination of input variations, the choice of building energy programs, how to reduce computational time for energy models. Second, the sensitivity analysis methods used in building performance analysis are reviewed. These methods can be categorized into local and global sensitivity analysis. The global methods can be further divided into four approaches: regression, screening-based, variance-based, and meta-model sensitivity analysis. Recent research has been concentrated on global methods because they can explore the whole input space and most of them allow the self-verification, i.e., how much variance of the model output (building energy consumption) has been explained by the method used in the analysis. Third, we discuss several important topics, which are often overlooked in the domain of building performance analysis. These topics include the application of sensitivity analysis in observational study, how to deal with correlated inputs, the computation of the variations of sensitivity index, and the software issues. Lastly, the practical guidance is given based on the advantages and disadvantaged of different sensitivity analysis methods in assessing building thermal performance. The recommendations for further research in the future are made to provide more robust analysis in assessing building energy performance.
671 citations
TL;DR: In this paper, the importance of urban physics related to the grand societal challenges is described, after which the spatial and temporal scales in urban physics and the associated model categories are outlined.
627 citations
Cites methods from "Uncertainty analysis in building pe..."
...BES is applied for evaluating the energy performance of buildings, including energy consumption and thermal comfort [149-155]....
[...]
TL;DR: The findings indicate a breakthrough in using evolutionary algorithms in solving highly constrained envelope, HVAC and renewable optimization problems and some future directions anticipated or needed for improvement of current tools are presented.
360 citations
TL;DR: The data sources of uncertainty in building performance analysis are described to provide a firm foundation for specifying variations of uncertainty factors affecting building energy, and several applications of uncertainty analysis in building energy assessment are discussed.
Abstract: Uncertainty analysis in building energy assessment has become an active research field because a number of factors influencing energy use in buildings are inherently uncertain. This paper provides a systematic review on the latest research progress of uncertainty analysis in building energy assessment from four perspectives: uncertainty data sources, forward and inverse methods, application of uncertainty analysis, and available software. First, this paper describes the data sources of uncertainty in building performance analysis to provide a firm foundation for specifying variations of uncertainty factors affecting building energy. The next two sections focus on the forward and inverse methods. Forward uncertainty analysis propagates input uncertainty through building energy models to obtain variations of energy use, whereas inverse uncertainty analysis infers unknown input factors through building energy models based on energy data and prior information. For forward analysis, three types of approaches (Monte Carlo, non-sampling, and non-probabilistic) are discussed to provide sufficient choices of uncertainty methods depending on the purpose and specific application of a building project. For inverse analysis, recent research has concentrated more on Bayesian computation because Bayesian inverse methods can make full use of prior information on unknown variables. Fourth, several applications of uncertainty analysis in building energy assessment are discussed, including building stock analysis, HVAC system sizing, variations of sensitivity indicators, and optimization under uncertainty. Moreover, the software for uncertainty analysis is described to provide flexible computational environments for implementing uncertainty methods described in this review. This paper concludes with the trends and recommendations for further research to provide more convenient and robust uncertainty analysis of building energy. Uncertainty analysis has been ready to become the mainstream approach in building energy assessment although a number of issues still need to be addressed.
266 citations
Cites background or methods from "Uncertainty analysis in building pe..."
...The same uncertainty range is also used for material thickness parameters in the research by Hopfe and Hensen [3]....
[...]
...automatically create many building energy models, including Matlab [3], Python [121], Excel VBA [2, 133], and R environment [134, 135]....
[...]
...Researchers in the field of building energy simulation have proposed several classifications of uncertainty to represent the different characteristics of uncertainty in building energy analysis [3, 22, 29, 30]....
[...]
...Latin hyper-cube sampling (LHS) is the most widely used sampling method in the field of building energy analysis as it can produce converged results with a considerably reduced number of samples [3, 51, 52, 58, 59, 107, 111, 116, 123]....
[...]
...provides a number of sampling methods, including the random sample, Latin hyper-cube sampling, and Sobol method [3, 52, 59, 123, 128]....
[...]
References
More filters
Book•
01 Apr 2004
TL;DR: In this paper, the authors present a method for sensitivity analysis of a fish population model using Monte Carlo filtering and variance-based methods, which is based on the Bayesian uncertainty estimation.
Abstract: PREFACE. 1. A WORKED EXAMPLE. 1.1 A simple model. 1.2 Modulus version of the simple model. 1.3 Six--factor version of the simple model. 1.4 The simple model 'by groups'. 1.5 The (less) simple correlated--input model. 1.6 Conclusions. 2. GLOBAL SENSITIVITY ANALYSIS FOR IMPORTANCE ASSESSMENT. 2.1 Examples at a glance. 2.2 What is sensitivity analysis? 2.3 Properties of an ideal sensitivity analysis method. 2.4 Defensible settings for sensitivity analysis. 2.5 Caveats. 3. TEST CASES. 3.1 The jumping man. Applying variance--based methods. 3.2 Handling the risk of a financial portfolio: the problem of hedging. Applying Monte Carlo filtering and variance--based methods. 3.3 A model of fish population dynamics. Applying the method of Morris. 3.4 The Level E model. Radionuclide migration in the geosphere. Applying variance--based methods and Monte Carlo filtering. 3.5 Two spheres. Applying variance based methods in estimation/calibration problems. 3.6 A chemical experiment. Applying variance based methods in estimation/calibration problems. 3.7 An analytical example. Applying the method of Morris. 4. THE SCREENING EXERCISE. 4.1 Introduction. 4.2 The method of Morris. 4.3 Implementing the method. 4.4 Putting the method to work: an analytical example. 4.5 Putting the method to work: sensitivity analysis of a fish population model. 4.6 Conclusions. 5. METHODS BASED ON DECOMPOSING THE VARIANCE OF THE OUTPUT. 5.1 The settings. 5.2 Factors Prioritisation Setting. 5.3 First--order effects and interactions. 5.4 Application of Si to Setting 'Factors Prioritisation'. 5.5 More on variance decompositions. 5.6 Factors Fixing (FF) Setting. 5.7 Variance Cutting (VC) Setting. 5.8 Properties of the variance based methods. 5.9 How to compute the sensitivity indices: the case of orthogonal input. 5.9.1 A digression on the Fourier Amplitude Sensitivity Test (FAST). 5.10 How to compute the sensitivity indices: the case of non--orthogonal input. 5.11 Putting the method to work: the Level E model. 5.11.1 Case of orthogonal input factors. 5.11.2 Case of correlated input factors. 5.12 Putting the method to work: the bungee jumping model. 5.13 Caveats. 6. SENSITIVITY ANALYSIS IN DIAGNOSTIC MODELLING: MONTE CARLO FILTERING AND REGIONALISED SENSITIVITY ANALYSIS, BAYESIAN UNCERTAINTY ESTIMATION AND GLOBAL SENSITIVITY ANALYSIS. 6.1 Model calibration and Factors Mapping Setting. 6.2 Monte Carlo filtering and regionalised sensitivity analysis. 6.2.1 Caveats. 6.3 Putting MC filtering and RSA to work: the problem of hedging a financial portfolio. 6.4 Putting MC filtering and RSA to work: the Level E test case. 6.5 Bayesian uncertainty estimation and global sensitivity analysis. 6.5.1 Bayesian uncertainty estimation. 6.5.2 The GLUE case. 6.5.3 Using global sensitivity analysis in the Bayesian uncertainty estimation. 6.5.4 Implementation of the method. 6.6 Putting Bayesian analysis and global SA to work: two spheres. 6.7 Putting Bayesian analysis and global SA to work: a chemical experiment. 6.7.1 Bayesian uncertainty analysis (GLUE case). 6.7.2 Global sensitivity analysis. 6.7.3 Correlation analysis. 6.7.4 Further analysis by varying temperature in the data set: fewer interactions in the model. 6.8 Caveats. 7. HOW TO USE SIMLAB. 7.1 Introduction. 7.2 How to obtain and install SIMLAB. 7.3 SIMLAB main panel. 7.4 Sample generation. 7.4.1 FAST. 7.4.2 Fixed sampling. 7.4.3 Latin hypercube sampling (LHS). 7.4.4 The method of Morris. 7.4.5 Quasi--Random LpTau. 7.4.6 Random. 7.4.7 Replicated Latin Hypercube (r--LHS). 7.4.8 The method of Sobol'. 7.4.9 How to induce dependencies in the input factors. 7.5 How to execute models. 7.6 Sensitivity analysis. 8. FAMOUS QUOTES: SENSITIVITY ANALYSIS IN THE SCIENTIFIC DISCOURSE. REFERENCES. INDEX.
2,297 citations
TL;DR: This paper demonstrates practical approaches for determining relative parameter sensitivity with respect to a model's optimal objective function value, decision variables, and other analytic functions of a solution.
Abstract: In applications of operations research models, decision makers must assess the sensitivity of outputs to imprecise values for some of the model's parameters. Existing analytic approaches for classic optimization models rely heavily on duality properties for assessing the impact of local parameter variations, parametric programming for examining systematic variations in model coefficients, or stochastic programming for ascertaining a robust solution. This paper accommodates extensive simultaneous variations in any of an operations research model's parameters. For constrained optimization models, the paper demonstrates practical approaches for determining relative parameter sensitivity with respect to a model's optimal objective function value, decision variables, and other analytic functions of a solution. Relative sensitivity is assessed by assigning a portion of variation in an output value to each parameter that is imprecisely specified. The computing steps encompass optimization, Monte Carlo sampling, ...
958 citations
TL;DR: Three sensitivity analysis techniques, differential sensitivity analysis (DSA), Monte Carlo analysis (MCA), and stochasticensitivity analysis (SSA), are appraised using three detailed finite difference simulation programs, ESP, HTB2, and SERI-RES.
380 citations
TL;DR: Uncertainty analysis has been used in several research projects in order to estimate reliability of results, especially in empirical validation-based projects where both measured and predicted uncertainty bands need to be evaluated as discussed by the authors.
236 citations
DOI•
01 Jan 2009
TL;DR: In this paper, a prototype simulation based environment that provides add-ons like uncertainty and sensitivity analysis, multi-criteria and disciplinary decision making under uncertainty, and multi-objective optimization is presented.
Abstract: Building performance simulation (BPS) uses computer-based models that cover performance aspects such as energy consumption and thermal comfort in buildings. The uptake of BPS in current building design projects is limited. Although there is a large number of building simulation tools available, the actual application of these tools is mostly restricted to code compliance checking or thermal load calculations for sizing of heating, ventilation and air-conditions systems in detailed design. The aim of the presented work is to investigate opportunities in BPS during the later phases of the design process, and to research and enable innovative applications of BPS for design support. The research started from an existing and proven design stage specific simulation software tool. The research methods applied comprise of literature review, interviews, rapid iterative prototyping, and usability testing. The result of this research is a prototype simulation based environment that provides add-ons like uncertainty and sensitivity analysis, multi-criteria and disciplinary decision making under uncertainty, and multi-objective optimization. The first prototype addressing the uncertainties in physical, scenario, and design parameters provides additional information through figures and tables. This outcome helps the designer in understanding how parameters relate to each other and to comprehend how variations in the model input affect the output. It supports the design process by providing a basis to compare different design options and leads therefore to an improved guidance in the design process. The second approach addresses the integration of a decision making protocol with the extension of uncertainty and sensitivity analysis. This prototype supports the design team in the design process by providing a base for communication. Furthermore, it supports the decision process by providing the possibility to compare different design options by minimizing the risk that is related to different concepts. It reduces the influence of preoccupation in common decision making and avoids pitfalls due to a lack of planning and focus. The third and last approach shows the implementation of two multi-objective algorithms and the integration of uncertainty in optimization. The results show the optimization of parameters for the objectives energy consumption and weighted overand underheating hours. It shows further how uncertainties impact the Pareto frontier achieved. The applicability and necessity of the three implemented approaches has further been validated with the help of usability testing by conducting mock-up presentations and an online survey. The outcome has shown that the presented results enhance the capabilities of BPS and fulfil the requirements in detailed design by providing a better understanding of results, guidance through the design process, and supporting the decision process. All three approaches have been found important to be integrated in BPS.
206 citations