International Industrial Ecology Day 2021
Building material flow characterization model : quantification and propagation of uncertainties of numerical and categorical variables.
Knowledge of the stocks and flows of materials and the spatial and temporal dynamics of the building stock is necessary to improve the circularity of resources and, consequently, reduce the inflow of raw materials and the outflow of demolition waste. The material stocks and flows analysis in the built environment has received increased attention in urban metabolism in recent years. Although the general objective of stock and flow characterization models is to capture the complexity of the built stock by being as simple as possible to provide a useful abstraction, it is desirable to arrive at parsimonious models with a balance between the complexity of the model and the quality of the results obtained. However, many of these models that estimate building material stocks and flows are based on assumptions where the reliability is not always assessed or partially assessed; therefore, the results can include preponderant uncertainties that are not visible and are not communicated to the users.
This study makes an analysis, coupling, and quantification of the uncertainties coming from numerical and categorical variables for the stock and flow characterization model of construction materials (BTP-Flux) developed by CSTB, which uses a macrocomponent based bottom-up approach. A Monte Carlo method is used to propagate uncertainties and calculate the total uncertainty of the model; nevertheless, the innovative contribution of this study is the use of confusion matrices for the quantification of uncertainties coming of categorical variables. The propagation of uncertainties was tested on two main stages of BTP-Flux: geometry of the evaluated buildings and the coupling stage between the generic categorical data of buildings and their macrocomponents. The study results allow identifying the main and secondary stages or variables of the model regarding their contribution to the total uncertainty of the model.
TIRADO FABIAN Deanira Rafaela1,2, HABERT Guillaume2, LAURENCEAU Sylvain1, MAILHAC Adélaïde1
1 : Université Paris-Est, Scientific and Technical Center for Building (CSTB), France
2 : Sustainable construction department, IBI, ETH Zurich, Zurich, Switzerland
This study makes an analysis, coupling, and quantification of the uncertainties coming from numerical and categorical variables for the stock and flow characterization model of construction materials (BTP-Flux) developed by CSTB, which uses a macrocomponent based bottom-up approach. A Monte Carlo method is used to propagate uncertainties and calculate the total uncertainty of the model; nevertheless, the innovative contribution of this study is the use of confusion matrices for the quantification of uncertainties coming of categorical variables. The propagation of uncertainties was tested on two main stages of BTP-Flux: geometry of the evaluated buildings and the coupling stage between the generic categorical data of buildings and their macrocomponents. The study results allow identifying the main and secondary stages or variables of the model regarding their contribution to the total uncertainty of the model.
TIRADO FABIAN Deanira Rafaela1,2, HABERT Guillaume2, LAURENCEAU Sylvain1, MAILHAC Adélaïde1
1 : Université Paris-Est, Scientific and Technical Center for Building (CSTB), France
2 : Sustainable construction department, IBI, ETH Zurich, Zurich, Switzerland
Author(s)
| Name | Affiliation |
|---|---|
| Rafaela TIRADO | ETH-Zurich & CSTB |
« Back to Modelling uncertainty and variability in Industrial Ecology