Page 28 - InnoRenew CoE International Conference 2023
P. 28
Enhancing spatial resolution in durability assessment of
wood at different scales: a data-driven approach for
moisture content prediction
J. Niklewski1*, H. Hosseini2
1 Lund University, Department of Building and Environmental Technology, jonas.niklewski@kstr.lth.se
2 Lund University, Department of Building and Environmental Technology, hasan.hosseini@tvrl.lth.se
* Corresponding author
The durability of wood in exposed environments is influenced by a range of biotic and abiotic factors. Among
these factors, fungal decay poses a significant threat as it breaks down the wood structure. Wood moisture
content is a critical factor that affects the development of fungal decay. Therefore, accurate modeling of wood
moisture content is essential for the design of wooden structures and commodities. Existing mechanistic models
based on finite element modeling provide accuracy but demand substantial computational resources, limiting
their applicability.
In this work, we explore data-driven approaches to predict moisture content in durability applications requiring
high spatial resolution. Specifically, we focus on two different scales: decay risk of building envelopes and the
development of decay hazard maps. We used a time-lagged neural network trained on simulated data with inputs
of relative humidity, temperature, and precipitation to predict moisture content. The model was used with the
CERRA reanalysis dataset as input, enabling us to assess the decay hazard of wooden specimens under different
European climates at a horizontal resolution of 5.5 km. Moreover, for a specific location, we used the same model
with a simplified environmental analysis to predict decay risk at a resolution of approximately 10x10 mm over a
building envelope.
This work aims to contribute to user-friendly and open-source tools which facilitate performance-based
service life assessment of wood. In doing so, we aim to enhance the resilience and reliability of wood in diverse
applications, optimizing its potential for sustainable and enduring construction solutions. Future work will focus
on the development of similar models with data stemming from field measurements rather than simulation.
Keywords: wood, modeling, moisture, data-driven, durability
Acknowledgement: The authors acknowledge receiving funding from the Slovenian Research Agency for the
project Using questionnaires to measure attitudes and behaviours of building users [Z5-1879] and from the
European Cooperation in Science and Technology for the InnoRenew project [grant agreement #739574] under
the H2020 Spreading Excellence and Widening Participation Horizon2020 Widespread-Teaming program.
REFERENCES
Hedstrom, G.S. Sustainability: What It Is and How to Measure It; De|G Press: Berlin, Germany, 2018; ISBN
1-5474-1660-2
Renda, A., Pelkmans, J., Schrefler, L., Luchetta, G., Simonelli, F., Mustilli, F., Wieczorkiewicz, J., et al., 2014. The
EU Furniture Market Situation and a Possible Furniture Products Initiative: Final Report. European Commission
DG Enterprise and Industry.
28 1 3 – 1 4 S E PT E M B E R 2 0 2 3 I I Z O L A , S LOVE N I A
wood at different scales: a data-driven approach for
moisture content prediction
J. Niklewski1*, H. Hosseini2
1 Lund University, Department of Building and Environmental Technology, jonas.niklewski@kstr.lth.se
2 Lund University, Department of Building and Environmental Technology, hasan.hosseini@tvrl.lth.se
* Corresponding author
The durability of wood in exposed environments is influenced by a range of biotic and abiotic factors. Among
these factors, fungal decay poses a significant threat as it breaks down the wood structure. Wood moisture
content is a critical factor that affects the development of fungal decay. Therefore, accurate modeling of wood
moisture content is essential for the design of wooden structures and commodities. Existing mechanistic models
based on finite element modeling provide accuracy but demand substantial computational resources, limiting
their applicability.
In this work, we explore data-driven approaches to predict moisture content in durability applications requiring
high spatial resolution. Specifically, we focus on two different scales: decay risk of building envelopes and the
development of decay hazard maps. We used a time-lagged neural network trained on simulated data with inputs
of relative humidity, temperature, and precipitation to predict moisture content. The model was used with the
CERRA reanalysis dataset as input, enabling us to assess the decay hazard of wooden specimens under different
European climates at a horizontal resolution of 5.5 km. Moreover, for a specific location, we used the same model
with a simplified environmental analysis to predict decay risk at a resolution of approximately 10x10 mm over a
building envelope.
This work aims to contribute to user-friendly and open-source tools which facilitate performance-based
service life assessment of wood. In doing so, we aim to enhance the resilience and reliability of wood in diverse
applications, optimizing its potential for sustainable and enduring construction solutions. Future work will focus
on the development of similar models with data stemming from field measurements rather than simulation.
Keywords: wood, modeling, moisture, data-driven, durability
Acknowledgement: The authors acknowledge receiving funding from the Slovenian Research Agency for the
project Using questionnaires to measure attitudes and behaviours of building users [Z5-1879] and from the
European Cooperation in Science and Technology for the InnoRenew project [grant agreement #739574] under
the H2020 Spreading Excellence and Widening Participation Horizon2020 Widespread-Teaming program.
REFERENCES
Hedstrom, G.S. Sustainability: What It Is and How to Measure It; De|G Press: Berlin, Germany, 2018; ISBN
1-5474-1660-2
Renda, A., Pelkmans, J., Schrefler, L., Luchetta, G., Simonelli, F., Mustilli, F., Wieczorkiewicz, J., et al., 2014. The
EU Furniture Market Situation and a Possible Furniture Products Initiative: Final Report. European Commission
DG Enterprise and Industry.
28 1 3 – 1 4 S E PT E M B E R 2 0 2 3 I I Z O L A , S LOVE N I A
