Advanced search
1 file | 7.81 MB Add to list

Renovation of historical facades : the rescue or the kiss of death?

Klaas Calle (UGent)
(2020)
Author
Promoter
(UGent) and (UGent)
Organization
Abstract
“Belgians are born with a brick in their stomach.” This old saying is supported by our cultural heritage which displays a rich tradition of brick constructions. Historically valuable facades are often massive masonries. These constructions are subjected to the exterior climate, namely to mostly moisture and temperature fluctuations. These conditions in many cases lead to damage, which induces a need for restoration. At the same time legislation has the ambition to increase the energy efficiency of the building stock, in which heritage buildings are actually performing substandard. So next to restoration also a lot of renovations, which incorporate thermal measures, are performed to increase the thermal comfort of the buildings. Because of the high heritage value of most exterior facades and their massive composition, interior insulation is often the only option. This insulation intervention increases the moisture contents of the construction and decreases the temperatures in the construction, which inadvertently leads to increased damage related risks. Frost damage, decay of wooden beam heads and mould growth at the interior surface are the main issues investigated in this research but the developed methodologies are also applicable to investigate other moisture related damage phenomena such as corrosion, algae growth, … To predict which impact specific measures on the construction might have, a representative simulation model is necessary. To reach this goal, multiple Heat, Air and Moisture models on the level of the building component have been developed. The complexity in the use of these models, and with that often the stumbling block in practical application, lies in the correct translation of the historic masonry construction and its environmental conditions into specific properties, boundary conditions and compositions in the model. Here we find various challenges that are explained and investigated in this dissertation. Historic masonry constructions consist of various materials such as pointing mortars, bed mortars and bricks. The necessary 2- and 3D effects render simulations computationally expensive. To prevent an endless calculation time it has now been investigated in which situations one is allowed to homogenize masonries, that consist of brick and mortars, to one representative material, being the (in volume) dominant material: the brick. In this perspective especially the moisture transport properties of the represented materials must be considered. For mortars with low transport properties, like cement based mortars, this homogenisation appears to be mostly a conservative approach. For lime mortars, which have higher transport and storage properties, it has been found one should be aware that moisture contents deep in the construction could be underestimated, which could be important for damage phenomena such as decay of wooden beam heads and mould growth at the interior surface. Next to the geometry and composition of the wall, also the properties of the historic materials themselves vary significantly, as a consequence of their production process and the natural origin of the used aggregates. These characteristics can in some cases have a severe impact on the hygrothermal behaviour of the construction. The historical context in which these heterogeneities in masonries have developed is discussed based on the available historic literature namely tendencies in dating, typology, geography and geology. On the more technical side, cluster analysis on detailed material databases seems to allow for the derivation of generic materials, which in turn allow us to generate a representative material for a construction by fitting the generic material towards simple experimental measurements on the construction. This way, within specific conditions, intensive time consuming laboratory measurements for practical applications can be avoided. But not only on the brick and mortars themselves there is a large spread; also the hygric interaction between the materials is uncertain. In literature hygric interface resistances (IR) between bricks and cement mortars are found, but as historic masonries mostly comprise lime based mortars, essential knowledge was missing to be able to make well-founded statements. By X-ray scanning during an absorption test of a combined brick lime mortar sample the moisture transport was visualized and quantified in this study. The measured moisture contents were precisely calibrated by using a staircase shaped water trough. This method yields the attenuation coefficient of the polychromatic X-rays in water in function of the amount of penetrated water. This way hygric interface resistances of 10^9 and 10^8 m/s were found for respectively dry and wet cured lime mortars. Dry/wet refers to the dry or capillary saturated state in which the brick was brought before the fresh mortar was applied. Hence, it was shown that IR for lime-based mortars are an order of magnitude lower than the IR for cement based mortars. Next to the variations discussed above, the impact of water infiltrations in masonries through cracks and hollow spaces is often a big question mark. In practice, water infiltrations occur often, but these are generally not included in HAM simulations as there is no methodology available to deal with their highly uncertain character. Two methods for the implementation of water infiltrations in HAM models are discussed in this research: specific modifications in the material properties and the implementation of moisture sources into the construction in various ways. - In a first step, the feasibility of implementing a single specific crack in the simulation by introducing a local increase in the liquid transport properties when the capillary moisture content is reached, is investigated. Though this is technically possible, practical implementation is not evident because multiple cracks occur, with varying geometries and frequencies. A probabilistic implementation would then result in infinitesimal increases in the transport properties or a weighted scaling according to the distribution. As it is today not feasible to derive such distribution, a general scaling of the liquid transport properties is selected as the most reliable method to include the effect of cracks in material properties. - For the second method, the implementation of moisture sources in HAM simulations, the quantification of the amount of water that should be applied is off course a key factor. Next to this it is important to define where the moisture source should be implemented in the simulation. For the quantification, extensive series of watertightness tests were performed on test walls. It was found -in contrast to what is often assumed- that not the pressure difference during the test but the amount of runoff water at the surface and the quality of the bricklaying are key factors. Subsequently a method was provided to make the translation from the infiltration rates measured during the test towards the moisture source in the simulation. Based on these findings this was partly based on the saturation degree of the facade surface during the simulation. The incorporation of the material properties, the indoor and outdoor climate conditions, the thickness of the wall and other impact parameters as variable inputs in simulations is computationally very expensive. This is because the number of required simulations to generate a representative reflection of the impact of the variability of one impact parameter easily rises, compared to the variability of a damage indicator. This prompted the need for alternative methods to answer the sensitivity related research questions. The solution was found in Kriging based surrogate models. Hereby ‘new’ models are trained to represent the correlation between the input and the damage indicator, based on a limited number of simulations of the original HAM model. These correlations were specified based on the surrogate model with specific coefficients such as Sobol indices to indicate the impact of each of the input variables on the output. The most important impact factors for each of the investigated damage phenomena are the orientation and rain exposure coefficient. These input variables determine the rain intensity and were found to be very dominant for moisture related damage of the masonry construction. Next to this, the brick type also is determinative: this categorical parameter was defined as the combined trend of the moisture retention and moisture transport function. For frost damage at the exterior surface of the masonry and decay of wooden beam heads embedded in the construction the absorption coefficient tended to be of main importance as well. Next to the spread on the material properties, the indoor and outdoor climate conditions and the thickness of the wall, the impact of hygric interface resistances, interior insulation and water infiltrations was evaluated in perspective of the damage indicators and with the aid of the same surrogate modelling methodology. The hygric interface resistances with the same order of magnitude of the ones measured for lime mortars during absorption, tended to have a very limited impact on the damage indicators. For historical masonries based on lime mortars, the IR’s can thus be abstracted. Resistances in the order of magnitude of those for cement mortars might have a severe impact on the hygric behaviour. Beware: due to poor workmanship it could be that hollow spaces are present between materials, and these capillary interruptions could be regarded as large hygric resistances. These can for example provoke high moisture contents behind pointing mortars with an increased chance on frost damage as a consequence. Interior insulation has both a thermal and hygric effect on the construction. The purely thermal effect of insulation proved negligible in perspective of the investigated damage indicators. The hygric effect was found to be important. The reduction in the drying of the construction towards the indoor climate was limited due to the vapour resistance of the insulation and the blocking of moisture transport towards the interior plaster. The last one is by far the most important parameter. But it is typically not discussed since most insulation materials are assumed to stop capillary moisture transport. The implementation of variable water infiltrations –quantified by lab experiments– seemed to clearly impact each of the damage indicators. A thorough analysis of the building envelope was thus found to be a prerequisite to be able to estimate the chance of potential water infiltrations occurring and with this the potential risk of damage. From these numerous simulations and their respective input and output, classification trees can be derived. These indicate which parameter in the input provokes the highest spread in the output or damage indicator. All simulations are then split in two subgroups based on a specific value of this dominant input parameter, which maximally limits the spread on the outputs in the newly generated subgroups. This Poisson based process can be repeated until a classification tree is derived. For each of the final branches the average of the evaluated damage indicator is mentioned. This tree can also be used the other way around in practice, as a decision tree. For a specific situation with specific inputs the damage indicator can be predicted without performing a case specific HAM simulation. Of course these trees do not render all HAM simulations superfluous. But for interior insulation for example, for which now all historic buildings are in the “grey” risk zone based on the actual guidelines, this method distinguishes more specifically whether small, moderate or large risks are to be expected.
Keywords
HAM, Historical masonries, Sensitivity analysis, Damage assesment

Downloads

  • (...).pdf
    • full text (Published version)
    • |
    • UGent only
    • |
    • PDF
    • |
    • 7.81 MB

Citation

Please use this url to cite or link to this publication:

MLA
Calle, Klaas. Renovation of Historical Facades : The Rescue or the Kiss of Death? Universiteit Gent. Faculteit Ingenieurswetenschappen en Architectuur, 2020.
APA
Calle, K. (2020). Renovation of historical facades : the rescue or the kiss of death? Universiteit Gent. Faculteit Ingenieurswetenschappen en Architectuur.
Chicago author-date
Calle, Klaas. 2020. “Renovation of Historical Facades : The Rescue or the Kiss of Death?” Universiteit Gent. Faculteit Ingenieurswetenschappen en Architectuur.
Chicago author-date (all authors)
Calle, Klaas. 2020. “Renovation of Historical Facades : The Rescue or the Kiss of Death?” Universiteit Gent. Faculteit Ingenieurswetenschappen en Architectuur.
Vancouver
1.
Calle K. Renovation of historical facades : the rescue or the kiss of death? Universiteit Gent. Faculteit Ingenieurswetenschappen en Architectuur; 2020.
IEEE
[1]
K. Calle, “Renovation of historical facades : the rescue or the kiss of death?,” Universiteit Gent. Faculteit Ingenieurswetenschappen en Architectuur, 2020.
@phdthesis{8660833,
  abstract     = {“Belgians are born with a brick in their stomach.” This old saying is supported by our cultural heritage which displays a rich tradition of brick constructions. Historically valuable facades are often massive masonries. These constructions are subjected to the exterior climate, namely to mostly moisture and temperature fluctuations. These conditions in many cases lead to damage, which induces a need for restoration. At the same time legislation has the ambition to increase the energy efficiency of the building stock, in which heritage buildings are actually performing substandard. So next to restoration also a lot of renovations, which incorporate thermal measures, are performed to increase the thermal comfort of the buildings. Because of the high heritage value of most exterior facades and their massive composition, interior insulation is often the only option. This insulation intervention increases the moisture contents of the construction and decreases the temperatures in the construction, which inadvertently leads to increased damage related risks. Frost damage, decay of wooden beam heads and mould growth at the interior surface are the main issues investigated in this research but the developed methodologies are also applicable to investigate other moisture related damage phenomena such as corrosion, algae growth, … 
To predict which impact specific measures on the construction might have, a representative simulation model is necessary. To reach this goal, multiple Heat, Air and Moisture models on the level of the building component have been developed. The complexity in the use of these models, and with that often the stumbling block in practical application, lies in the correct translation of the historic masonry construction and its environmental conditions into specific properties, boundary conditions and compositions in the model. Here we find various challenges that are explained and investigated in this dissertation.
Historic masonry constructions consist of various materials such as pointing mortars, bed mortars and bricks. The necessary 2- and 3D effects render simulations computationally expensive. To prevent an endless calculation time it has now been investigated in which situations one is allowed to homogenize masonries, that consist of brick and mortars, to one representative material, being the (in volume) dominant material: the brick. In this perspective especially the moisture transport properties of the represented materials must be considered. For mortars with low transport properties, like cement based mortars, this homogenisation appears to be mostly a conservative approach. For lime mortars, which have higher transport and storage properties, it has been found one should be aware that moisture contents deep in the construction could be underestimated, which could be important for damage phenomena such as decay of wooden beam heads and mould growth at the interior surface.
Next to the geometry and composition of the wall, also the properties of the historic materials themselves vary significantly, as a consequence of their production process and the natural origin of the used aggregates. These characteristics can in some cases have a severe impact on the hygrothermal behaviour of the construction. The historical context in which these heterogeneities in masonries have developed is discussed based on the available historic literature namely tendencies in dating, typology, geography and geology. On the more technical side, cluster analysis on detailed material databases seems to allow for the derivation of generic materials, which in turn allow us to generate a representative material for a construction by fitting the generic material towards simple experimental measurements on the construction. This way, within specific conditions, intensive time consuming laboratory measurements for practical applications can be avoided.
But not only on the brick and mortars themselves there is a large spread; also the hygric interaction between the materials is uncertain. In literature hygric interface resistances (IR) between bricks and cement mortars are found, but as historic masonries mostly comprise lime based mortars, essential knowledge was missing to be able to make well-founded statements. By X-ray scanning during an absorption test of a combined brick lime mortar sample the moisture transport was visualized and quantified in this study. The measured moisture contents were precisely calibrated by using a staircase shaped water trough. This method yields the attenuation coefficient of the polychromatic X-rays in water in function of the amount of penetrated water. This way hygric interface resistances of 10^9 and 10^8 m/s were found for respectively dry and wet cured lime mortars. Dry/wet refers to the dry or capillary saturated state in which the brick was brought before the fresh mortar was applied. Hence, it was shown that IR for lime-based mortars are an order of magnitude lower than the IR for cement based mortars.
Next to the variations discussed above, the impact of water infiltrations in masonries through cracks and hollow spaces is often a big question mark. In practice, water infiltrations occur often, but these are generally not included in HAM simulations as there is no methodology available to deal with their highly uncertain character. Two methods for the implementation of water infiltrations in HAM models are discussed in this research: specific modifications in the material properties and the implementation of moisture sources into the construction in various ways.
- In a first step, the feasibility of implementing a single specific crack in the simulation by introducing a local increase in the liquid transport properties when the capillary moisture content is reached, is investigated. Though this is technically possible, practical implementation is not evident because multiple cracks occur, with varying geometries and frequencies. A probabilistic implementation would then result in infinitesimal increases in the transport properties or a weighted scaling according to the distribution. As it is today not feasible to derive such distribution, a general scaling of the liquid transport properties is selected as the most reliable method to include the effect of cracks in material properties.
- For the second method, the implementation of moisture sources in HAM simulations, the quantification of the amount of water that should be applied is off course a key factor. Next to this it is important to define where the moisture source should be implemented in the simulation. For the quantification, extensive series of watertightness tests were performed on test walls. It was found -in contrast to what is often assumed- that not the pressure difference during the test but the amount of runoff water at the surface and the quality of the bricklaying are key factors. Subsequently a method was provided to make the translation from the infiltration rates measured during the test towards the moisture source in the simulation. Based on these findings this was partly based on the saturation degree of the facade surface during the simulation.
The incorporation of the material properties, the indoor and outdoor climate conditions, the thickness of the wall and other impact parameters as variable inputs in simulations is computationally very expensive. This is because the number of required simulations to generate a representative reflection of the impact of the variability of one impact parameter easily rises, compared to the variability of a damage indicator. This prompted the need for alternative methods to answer the sensitivity related research questions. The solution was found in Kriging based surrogate models. Hereby ‘new’ models are trained to represent the correlation between the input and the damage indicator, based on a limited number of simulations of the original HAM model. These correlations were specified based on the surrogate model with specific coefficients such as Sobol indices to indicate the impact of each of the input variables on the output.
The most important impact factors for each of the investigated damage phenomena are the orientation and rain exposure coefficient. These input variables determine the rain intensity and were found to be very dominant for moisture related damage of the masonry construction. Next to this, the brick type also is determinative: this categorical parameter was defined as the combined trend of the moisture retention and moisture transport function. For frost damage at the exterior surface of the masonry and decay of wooden beam heads embedded in the construction the absorption coefficient tended to be of main importance as well.
Next to the spread on the material properties, the indoor and outdoor climate conditions and the thickness of the wall, the impact of hygric interface resistances, interior insulation and water infiltrations was evaluated in perspective of the damage indicators and with the aid of the same surrogate modelling methodology.
The hygric interface resistances with the same order of magnitude of the ones measured for lime mortars during absorption, tended to have a very limited impact on the damage indicators. For historical masonries based on lime mortars, the IR’s can thus be abstracted. Resistances in the order of magnitude of those for cement mortars might have a severe impact on the hygric behaviour. Beware: due to poor workmanship it could be that hollow spaces are present between materials, and these capillary interruptions could be regarded as large hygric resistances. These can for example provoke high moisture contents behind pointing mortars with an increased chance on frost damage as a consequence.
Interior insulation has both a thermal and hygric effect on the construction. The purely thermal effect of insulation proved negligible in perspective of the investigated damage indicators. The hygric effect was found to be important. The reduction in the drying of the construction towards the indoor climate was limited due to the vapour resistance of the insulation and the blocking of moisture transport towards the interior plaster. The last one is by far the most important parameter. But it is typically not discussed since most insulation materials are assumed to stop capillary moisture transport.
The implementation of variable water infiltrations –quantified by lab experiments– seemed to clearly impact each of the damage indicators. A thorough analysis of the building envelope was thus found to be a prerequisite to be able to estimate the chance of potential water infiltrations occurring and with this the potential risk of damage.
From these numerous simulations and their respective input and output, classification trees can be derived. These indicate which parameter in the input provokes the highest spread in the output or damage indicator. All simulations are then split in two subgroups based on a specific value of this dominant input parameter, which maximally limits the spread on the outputs in the newly generated subgroups. This Poisson based process can be repeated until a classification tree is derived. For each of the final branches the average of the evaluated damage indicator is mentioned. This tree can also be used the other way around in practice, as a decision tree. For a specific situation with specific inputs the damage indicator can be predicted without performing a case specific HAM simulation. Of course these trees do not render all HAM simulations superfluous. But for interior insulation for example, for which now all historic buildings are in the “grey” risk zone based on the actual guidelines, this method distinguishes more specifically whether small, moderate or large risks are to be expected.
},
  author       = {Calle, Klaas},
  isbn         = {9789463553339},
  keywords     = {HAM,Historical masonries,Sensitivity analysis,Damage assesment},
  language     = {eng},
  pages        = {292},
  publisher    = {Universiteit Gent. Faculteit Ingenieurswetenschappen en Architectuur},
  school       = {Ghent University},
  title        = {Renovation of historical facades : the rescue or the kiss of death?},
  year         = {2020},
}