Master's Thesis
A Framework for Systematizing and Understanding Façade Soiling Patterns
— 2026
Key information
Authors:
Supervisors:
Published in
June 16, 2026
Abstract
Soiling is one of the most prevalent pathologies affecting building façades. Although often treated as an aesthetic issue, soiling patterns record how surfaces interact with moisture, rainfall and environmental degradation agents, encoding information on runoff paths, architectural detailing and material behaviour that manual inspection cannot reliably capture. Systematic, reproducible classification of these patterns remains an open challenge, as traditional visual surveys are subjective and time-consuming. This dissertation presents an automated, image-based framework that converts digital photographs of ten façades in Lisbon, Portugal, into quantified soiling maps. Each pixel is assigned a Soiling Index (SI) based on its Euclidean distance to user-defined clean and dirty prototypes in RGB colour space, while zone-based dirty fractions, anisotropy ratios and CIELAB colour metrics describe spatial organisation and colour signatures of soiling. Ten façades across four case study buildings were analysed, with SI values ranging from 0.186 to 0.615 (0 = clean, 1 = fully soiled) and dirty-pixel fractions from 7.7% to 83.4%, spanning Light to Very Heavy severity classes. Detail-level CIELAB analysis distinguished co-existing soiling mechanisms colourimetrically, with ΔE values between 4.9 and 42.0, spanning visible to very strong contrasts relative to the just noticeable difference of about 2.3. The framework revealed a clear orientation-dependent soiling hierarchy, with north-facing façades exhibiting the most severe and biologically influenced staining, and confirmed the dominance of eave zones driven by roof-edge runoff and runoff termination. Together, these results show that the method provides objective, physically interpretable indicators to support structured façade inspection.
Publication details
Authors in the community:
Gabriela Lenkiewicz
ist1119380
Supervisors of this institution:
Inês Flores-Colen
ist24597
Fields of Science and Technology (FOS)
civil-engineering - Civil engineering
Publication language (ISO code)
por - Portuguese
Rights type:
Embargoed access
Date available:
May 1, 2027
Institution name
Instituto Superior Técnico