Garrido Paulo Branco 2012

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Construction and Building Materials 29 (2012) 394–402

Contents lists available at SciVerse ScienceDirect

Construction and Building Materials


journal homepage: www.elsevier.com/locate/conbuildmat

Service life prediction of façade paint coatings in old buildings


M.A. Garrido ⇑, P.V. Paulo, F.A. Branco
Department of Civil Engineering and Architecture, Instituto Superior Técnico/ICIST, Technical University of Lisbon. Av. Rovisco Pais 1, 1049-001 Lisboa, Portugal

a r t i c l e i n f o a b s t r a c t

Article history: The financial resources available for infrastructure maintenance and rehabilitation are always limited,
Received 2 March 2011 creating a need for efficient resource management and for the ability to predict maintenance actions
Received in revised form 22 September throughout the infrastructure service period. In the context of bridge management, management systems
2011
have been developed and are already implemented around the world. Due to a much larger variety of
Accepted 2 October 2011
Available online 29 November 2011
maintenance problems and to a reduced number of buildings per owner, building management systems
are still quite rare. In fact efficient methodologies for service life prediction of building materials still need
to be developed.
Keywords:
Service life
This paper presents and tests a service life prediction methodology, based on the statistical analysis of
Façade paint coatings anomalies obtained from the inspection of in-use buildings and the corresponding degradation curves
Degradation factors obtained from deterministic models. The method was applied to the service life prediction of paint coat-
ings in old buildings’ façades and considers the quantification of the defect associated to paint peeling.
The influence of five construction/environment degradation factors on the behavior of paint coatings over
time is also analyzed.
Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction around the world [3]. Nevertheless, due to a much larger variety
of maintenance problems and to a reduced number of buildings
Today there is a strong motion towards a more efficient use of associated to each owner, building management systems are still
resources in every sector of human activity, with the goal of quite rare.
achieving and maintaining a human ‘‘sustainable development’’ The research here presented [4] aims to contribute to such
[1]. The construction industry is one of the most important indus- development, by exploring and testing a service life prediction
tries in this context, given the influence that the built environment methodology based on the statistical analysis of data obtained
has on the life of populations and on the economical development from the observation of long term degradation of buildings. The
of nations. In fact, the economic activity housed in built environ- method is achieved through the inspection of in-service buildings,
ments is significantly influenced by inputs given during the design, and it is applied to the specific case of service life prediction of
construction and maintenance phases of these environments [2]. paint coatings in old building façades. In a first step the inspections
Within the construction industry, the rehabilitation and main- quantified the degradation level associated to the defect of the
tenance strategies must also optimize the social and economic paint peeling in several buildings. The influence of five construc-
benefits of the built environment. In the present economical situa- tion/environment degradation factors on the paint coating perfor-
tion where the funds available for such actions are always very mance, over time, was also assessed during the inspections.
limited, there is a growing need to plan and prioritize the neces- Afterwards, using a statistical data analysis, degradation curves
sary maintenance works. The planning of such works is achieved were obtained simulating the paint performance over time, and
by predicting the moments when the critical elements of the built allowing for service life prediction considering a pre-defined max-
assets reach degradation levels that exceed acceptable values. To imum level of degradation.
be able to make such predictions, developments must be made
concerning the methodologies for service life prediction of building 2. Service life prediction methodologies
materials and components.
To solve this situation systems for bridge management were The development of service life prediction methodologies was
developed during the last 15 years and are already implemented greatly influenced by the work of several technical committees,
belonging to entities such as CIB (CIB W80), RILEM (RILEM 71-
⇑ Corresponding author. PSL, 100-TSL, 175-SLM) or ISO (ISO TC 59 SC 14) [5]. From the joint
E-mail address: majgarrido@gmail.com (M.A. Garrido). work of the CIB W80 and RILEM 71-PSL technical committees, an

0950-0618/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.conbuildmat.2011.10.057
M.A. Garrido et al. / Construction and Building Materials 29 (2012) 394–402 395

outline of a general methodology for service life prediction of Table 2


building materials and components was developed, and some re- Characterization and distribution of the study sample
according to coating ages.
search needs in that area were identified [6,7].
From the work developed in the CIB W80 and RILEM 100-TSL Coating age (years) Percent of sample (%)
technical committees, another contribution was presented to the Under 5 2
service life prediction problem, in which in-service exposure data Between 5 and 10 15
was explored [8]. Between 10 and 15 16
Between 15 and 20 23
For preliminary life estimation, the Architectural Institute of Ja- Between 20 and 30 13
pan proposed a methodology, based on a factorial method (consid- Between 30 and 40 9
ering construction and environment factors), which was later Between 40 and 50 15
adopted by the International Organization for Standardization in Above 50 7
the ISO 15686 standards [9,10].
A service life prediction methodology applied specifically to
paint coating systems was also developed by Martin et al., where Table 3
the use of reliability analysis was suggested in order to improve Characterization and distribution of the study sample according to the extension of
the quality of the service life estimates [11,12]. the paint peeling defect.
Recently, a systematic approach to service life prediction was Paint peeling (% of total coating area) Percent of sample (%)
developed by Paulo, where the materials characterization and the
Under 1% 47
life degradation conditions data is obtained from widespread Between 1% and 5% 14
building inspections and then analyzed using a web-based co- Between 5% and 20% 14
operative platform – BuildingsLife, considering several models of Between 20% and 50% 10
statistical analysis [13]. This platform, a building management sys- Above 50% 15
tem, was used in the development of the present study.

the inspected buildings, there was the concern to gather a sample


3. Inspection procedures
of coatings with a full range of paint peeling extents, i.e. from coat-
ings that presented no paint peeling, to coatings that were almost
3.1. Problem definition
completely peeled from their substrate. A brief characterization of
the study sample is presented in Tables 2–4.
In this study, the methodology of service life prediction was ap-
The inspections were performed with the following objectives:
plied to the façades of old buildings and to the particular defect
(i) to make a photographic registration of the façade, (ii) to collect
associated to paint peeling, considering its performance over time.
paint samples from the coatings, (iii) to make measurements of the
In this type of degradation, the effect of the following five degrada-
facades dimensions, and (iv) to perform an in situ interpretation of
tion factors (DF) was considered: (i) DF1: coating thickness; (ii) DF2:
the paint surface texture and substrate surface preparation. Addi-
paint binder; (iii) DF3: paint surface texture; (iv) DF4: substrate sur-
tionally, the last painting date was obtained from the municipal re-
face preparation; and (v) DF5: façade solar orientation. The data for
cords of the building maintenance works.
the study was obtained from the inspection of 100 in-service build-
The inspections allowed the conclusion that paint peeling fre-
ing façade coatings, and the associated defect quantification. The
quently begins near façade details (singularities) and in façade
inspected buildings are all located in the city centre of Lisbon, Por-
areas up to 1.5 m from the ground level, but these particular effects
tugal. Afterwards, the statistical data analysis was performed using
were not considered relevant for the global life estimation of the
degradation paths and deterministic models, considering three
coatings.
types of degradation curves: (i) Gompertz curves, (ii) Potential
The working photos of the building façades were obtained from
curves and (iii) Weibull curves. The analysis of this curve fitting al-
partial façade images that were later stitched together and cor-
lowed obtaining conclusions about the behavior of the façade
rected for perspective distortion in order to produce a single image,
painting along its service life.
similar to an elevation of the building façade. This process was
Given that all of the building inspections were performed in the
done using two image editing software applications (Hugin and
city centre of Lisbon, it is assumed that climatic conditions are
PTgui [13]). The method is illustrated in Fig. 1.
common to all the inspected buildings. A summary of monthly
The measurement of the façade dimensions was done using a
averaged temperatures and precipitation values in the city of Lis-
Stanley FatMax laser measuring device. The measured width and
bon is given in Table 1 [14].
height of the façades allowed for a scaling of the façade image.
Fig. 2 presents an example of the measurements done for one of
3.2. Building inspections
the inspected facades.

To obtain the data, the façade coatings of 100 buildings were in-
spected in the city of Lisbon. The buildings were chosen as built 3.3. Coating thickness measurements (DF1)
prior to 1940 (prior to concrete structures), corresponding to the
pre-pombaline type (up to XVIII century), pombaline type (XVIII The film thickness measurements were performed with the
century) and gailoeiro type (XIX and beginning of XX centuries) paint samples collected during the building inspections. The sam-
[13]. These buildings are stone/masonry based. In the choice of ples were first cleaned, to remove any pieces of mortar that might

Table 1
Average monthly temperature and precipitation values in the city of Lisbon, Portugal [14].

January February March April May June July August September October November December
Average temperature (°C) 11.3 12.6 14.3 15.3 17.3 20.3 22.7 23.0 21.7 18.4 14.8 12.4
Average precipitation (mm) 96.8 90.2 51.2 64.7 55.6 17.2 6.1 6.8 28.5 79.8 107.1 121.8
396 M.A. Garrido et al. / Construction and Building Materials 29 (2012) 394–402

Table 4
Characterization and distribution of the study sample according to the degradation
factors considered.

Degradation factor Class Percent of


sample (%)
DF1: coating thickness (quantification performed
on 100% of the sample)
DF2: paint binder PVA 26
Acrylic resins 6
Oil-based 13
Unidentified 55
DF3: paint surface texture Textured 35
Smooth – plastic 52
Smooth – oil 13
DF4: substrate surface preparation Repaint removing paint 59
Repaint over paint 41
DF5: solar orientation North 9
South 11
East 48
Fig. 2. Example of measured façade dimensions (width and height, meters).
West 32

be attached, and then their thickness was measured using an


Elcometer 355 Coating Thickness Gauge [4]. Ten measurements were
done for each paint sample, from which an average film thickness
was calculated.
In cases where more than one layer of paint existed (i.e. layers
from previous coatings that were not removed when the latest
coating was applied), the coating thickness considered was the
one associated with the most recent paint.

3.4. Paint binder identification (DF2)

To perform the paint binder identification, FTIR analyses were


performed on the paint samples. For this, a Perkin Elmer 1600 Series
FTIR was used (Fig. 3). This equipment allows the characterization Fig. 3. Perkin Elmer 1600 Series FTIR.
of substances through a solid potassium bromide (KBr) window
[4].
In a first attempt to identify the paint binders, the FTIR analyses 3.5. Paint surface texture determination (DF3)
were performed on samples of raw paint film. This led to transmit-
tance spectra that showed peaks associated to the usual paint The identification of the paint surface texture was performed
binders, but that showed also the peaks associated with inorganic through visual inspection. Three types of surface textures were
compounds that exist in paints, associated to pigments and fillers. considered: textured, smooth – plastic and smooth – oil. Examples
As a result, the obtained spectra were not clear enough for binder of these texture types can be found in Figs. 5–7, respectively,
identification. allowing for an in situ visual characterization.
To overcome this difficulty, a new method was adopted which
involved the extraction of the paint binder from the paint samples.
This was done through the use of a solvent, methyl ethyl ketone 3.6. Identification of the substrate surface preparation (DF4)
(MEK), and allowed for the paint binders to be isolated and identi-
fied, and clear FTIR spectra were obtained (Fig. 4) [4]. The identification of the exact conditions in which the paint
With this method, four different types of binders were identi- coatings were applied, it is not an easy task, as the corresponding
fied in the painted façades: polyvinyl acetate (PVA), acrylic resin, records are not usually kept. This makes it difficult to assess the
styrenated acrylic resin and oil-based. The acrylic resin and styrenat- real conditions existing during the application of the coatings.
ed acrylic resin binders were joined for the data analysis as acrylic For this reason, the substrate preparation factor was ap-
resins. proached in a simplified way, with two basic types of substrate

Fig. 1. Stitching and perspective correction processed to obtain the façade image.
M.A. Garrido et al. / Construction and Building Materials 29 (2012) 394–402 397

Fig. 4. Transmittance spectrum obtained from FTIR analysis of a paint binder, polyvinyl acetate (PVA).

Fig. 5. Example of a ‘‘textured’’ paint surface texture. Fig. 8. Example of a ‘‘repaint over paint’’.

Fig. 6. Example of a ‘‘smooth – plastic’’ surface texture.


Fig. 9. Example of a ‘‘repaint removing paint’’.

surface preparation being considered: repaint over paint and re-


paint removing paint (or without paint). These types of substrate
preparation are illustrated in Figs. 8 and 9, respectively.
The repaint over paint surface preparation comprises coatings
that were applied as a repaint, without the removal of the previous
paint coating. The repaint removing paint surface preparation in-
cludes both coatings applied after the full removal of any previous
paint coatings, and coatings applied for the first time on the façade
render that serves as substrate.

3.7. Solar orientation (DF5)

The solar orientation of the façades was determined using a


compass, with four main solar orientations being considered:
Fig. 7. Example of a ‘‘smooth – oil’’ surface texture. North, South, East and West. The grouping of façades into these
398 M.A. Garrido et al. / Construction and Building Materials 29 (2012) 394–402

In this study, the best fitting of the degradation curves was eval-
uated through a process of mean squared error (MSE) minimiza-
tion. The calculation of MSE values can take into account the
errors determined either in the vertical or horizontal axis, i.e. given
a coating age one can estimate the extent of degradation and com-
pare that with the real degradation (vertical MSE, or MSED, as
shown in Eq. (4)), or given a degradation extent one can determine
the age at which the model will predict that degradation will occur
and compare that with the real coating age (horizontal MSE, or
Fig. 10. Solar orientation system adopted.
MSEt, as shown in Eq. (5)). To achieve curves that will provide
the best overall fit, an indicator was created, the Combined Mean
main orientations was performed according the angles illustrated Squared Error (CMSE), given by the expression presented in Eq.
in Fig. 10. (6) [15].
X
MSED ¼ ½ðDi;model  Di Þ2 =n ð4Þ
3.8. Paint peeling defect quantifications
X
The defect quantifications were made using the Photo Measure MSEt ¼ ½ðti;model  t i Þ2 =n ð5Þ
application of the BuildingsLife platform [13]. This application al-
lows areas and distances to be directly measured over images of
CMSE ¼ MSEt þ MSED  103 ð6Þ
building façades. The façade images obtained with the previously
presented photo correction processes were uploaded onto Build-
ingsLife, and scales were defined using the façade dimensions mea-
5. Results and discussion
sured during the inspections.
This allowed for the quantification of the coating areas affected
5.1. Global degradation curves
by paint peeling, which were then converted into percentages of
peeled coating after the additional measurement of the total coat-
In Fig. 11 the total data is presented, along with the associated
ing areas.
degradation curves (the associated equations are shown in Table
By using high resolution façade images, it was possible to quan-
5). A significant dispersion in the data is observed as it considers
tify even very small areas of paint peeling. In fact, the quantifica-
data from different types of coatings which present different per-
tion method allowed accurate measurement of areas as small as
formances over time. In fact, the dispersion in this global data re-
10 cm2. Additionally, the measurement of peeled areas with irreg-
flects the different influences that the degradation factors have
ular borders presented no difficulty, as this measurement was
on the performance over time of paint coatings.
made by defining the exact border of the area to be quantified.
The CMSE values presented in Table 4 indicate that the Gom-
As such, the quantifications obtained by this method were very sat-
pertz curve, from the three curves considered, provides the best fit-
isfyingly reliable and accurate, given the scale of the analysis con-
ting. This result was consistently found in the majority of the
ducted in this research.
degradation graphs subsequently analyzed with the application
of the degradation factors. For this reason, the Gompertz curves
4. Data analysis will be the degradation curves presented in the current study.
From Fig. 11, it can be seen that in a global analysis, paints usu-
The inspection data is then a sample of points (each building), ally support around 20–25 years in a reduced/no peeling state,
each representing the relation between the time elapsed since reaching 20% of the façade after 30 years.
the last maintenance action (which in the present case is the appli-
cation of the last paint coating) and the value of the painting deg-
5.2. Influence of the degradation factors
radation. The degradation was the extension of the peeled coating
area, expressed as a percentage of the total painted area in the
The application of the degradation factors, which act as data fil-
façade.
ters grouping together coatings with similar characteristics, re-
Based on the previous data, the degradation modeling was ana-
duces the data dispersion and makes the different trends in
lyzed considering deterministic models associated to curve fitting
performance, over time, become more evident.
of the data, with the following curves Gompertz (Eq. (1)), Potential
This was done for each one of the factors listed previously,
(Eq. (2)) and Weibull (Eq. (3)) types.
either individually or in combination of more than one factor. Deg-
DG ¼ exp½a  expðb  tÞ ð1Þ radation curves were then fitted to the corresponding data.

DP ¼ a  t b ð2Þ

DW ¼ 1  exp½ðt=gÞb  ð3Þ
In the above expressions, DG DP and DW represent the extension
of peeled coating, t represents the time elapsed since the applica-
tion of the paint coating, and a, b, g and b are the parameters to
be adjusted for the best fitting of the curves to the data.
These curves are frequently used in life data analysis, with the
Weibull distribution being particularly relevant in this field [15].
In another way, the Gompertz curves usually show a good fitting
to degradation data of construction materials [13], as the data ana- Fig. 11. General degradation data and degradation curves for all the studied
lyzed in this study. Potential curves frequently are less effective [13]. coatings.
M.A. Garrido et al. / Construction and Building Materials 29 (2012) 394–402 399

Table 5
Parameters and CMSE values obtained for the global degradation curves.

Gompertz curve Potential curve Weibull curve


Parameters CMSE Parameters CMSE Parameters CMSE
a = 67.98536 164.83 a = 1.54581E6 178.65 g = 45.54301 170.55
b = 0.12081 b = 3.30210 b=5

5.2.1. Coating thickness (DF1)


For the analysis of the coating thickness degradation factor, the
sample was divided according to thickness intervals. The results
shown correspond to a division of the data in the following two
thickness levels:

 Level 1 – film thickness under 400 lm;


 Level 2 – film thickness greater than or equal to 400 lm.

The Gompertz degradation curves obtained are shown in


Fig. 12, and the equations for all the types of degradation curves
are presented in Table 6.
Fig. 13. Degradation curves for factor ‘‘DF2: paint binder’’.
The coatings in Level 2 thickness show a better performance
over time when compared to Level 1 coatings. Level 2 leads to an
increase of around 5 years without significant peeling, and this
The results show that coatings with the acrylic resins binder
age difference is approximately kept for other percentages of peel-
have the best performance in the initial part of the degradation
ing. In fact, considering a reduced/no peeling state in which the
curves leading to an increase of around 13 years in relation to
peeled area is inferior to 5% of the total coating area, it is possible
PVA, for a reduced/no peeling situation. The oil based paints have
to observe that Level 1 coatings remain in this state for approxi-
an intermediate behavior. The reduced/no peeling periods are
mately 23 years whereas Level 2 coatings prolong this period to
25 years for PVA paints, 38 years for acrylic paints and 31 years
approximately 28 years.
for oil paints.
Nevertheless, the dispersion shown in the data is still signifi-
The good fitting of the curves indicates that the binder is very
cant, which indicates that even though film thickness is a very rel-
relevant in its influence to the performance over time of the paint
evant factor in the performance over time of paint coatings, there
coatings.
are still other factors that influence the rate at which coatings de-
grade over time.
5.2.3. Paint surface texture (DF3)
The analysis of the influence of the paint surface texture degra-
5.2.2. Paint binder (DF2) dation factor led to the curves presented in Fig. 14. The equations
In the paint binder degradation factor analysis, three categories for these degradation curves are shown in Table 8.
of binders were considered: PVA, acrylic resins and oil-based. The textured paint films seem to present the best performance
The Gompertz degradation curves obtained in this analysis are over time of the three texture types analyzed. This was an expect-
shown in Fig. 13, and the equations for all the types of degradation able result, since this type of coating is usually considered to have
curves are presented in Table 7. higher resistance to degradation agents and, as a result, improved
durability over other smooth – plastic paint films.
Additionally, as the film thickness in textured paints is usually
higher than the average film thickness of smooth – plastic or smooth
– oil paints, and given the results found related to film thickness,
the result obtained is expectable.
After the degradation initiation period, the smooth – oil paint
films show the quickest degradation of the three paint texture
types, which justifies the fact that this solution has been practically
abandoned in the construction industry.
In this case the time periods associated to reduced/no peeling
among the three types of texture are as follows: 29 years for tex-
tured paints, 25 years for smooth – plastic paints and 31 years
Fig. 12. Degradation curves for factor ‘‘DF1: coating thickness’’. for smooth – oil paints.

5.2.4. Substrate surface preparation (DF4)


Table 6 The degradation curves obtained through the application of this
Gompertz degradation curves for ‘‘DF1: coating thickness’’. degradation factor are presented in Fig. 15 with the corresponding
Level 1 Level 2
parameters shown in Table 9.
The curves do not show significant differences in performance
Parameters CMSE Parameters CMSE
over time between the ‘‘repaint removing paint’’ and the ‘‘repaint
Gompertz curves over paint’’ solutions. The results were affected by a level of disper-
a = 74.57801 134.22 a = 68.84021 169.61
sion that suggests that the individual application of this factor is
b = 0.14159 b = 0.11481
not enough to reveal different trends in performance over time
400 M.A. Garrido et al. / Construction and Building Materials 29 (2012) 394–402

Table 7
Gompertz degradation curves for factor ‘‘DF2: paint binder’’.

PVA Acrylic resins Oil-based


Parameters CMSE Parameters CMSE Parameters CMSE
Gompertz curves
a = 15.13874 111.34 a = 1086.07431 264.20 a = 742.13802 201.76
b = 0.06690 b = 0.15705 b = 0.18346

Fig. 14. Degradation curves for factor ‘‘DF3: paint surface texture’’.
Fig. 15. Degradation curves for factor ‘‘DF4: substrate surface preparation’’.

of paint coatings, meaning that this factor should be combined


with additional degradation factors. Table 9
Nevertheless, the analyses of the obtained degradation curves Gompertz degradation curves for factor ‘‘DF4: substrate surface preparation’’.
indicate reduced/no peeling periods of 25 years for repaints with
Repaint removing paint Repaint over paint
removal of previous paint coatings and 28 years for repaints over
previous paint coatings. However, this small difference may not Parameters CMSE Parameters CMSE

be representative for the above state reasons. Gompertz curves


a = 59.84923 166.38 a = 81.68746 158.39
b = 0.12106 b = 0.12152
5.2.5. Solar orientation (DF5)
The application of this factor revealed that coatings in façades
facing North have the best performance over time, whereas coat-
ings in façades facing West have the worst performance over time,
followed by coatings in façades facing South and East. This is a re-
sult that is consistent with the levels of average solar exposure in
each orientation.
Yet, it is interesting to observe that the degradation initiation
periods are quite similar among the different solar orientations,
with the major differences in performance over time coming from
the degradation rates after this period. In fact, the reduced/no peel-
ing periods are as follows: 26 years for North-facing façades,
25 years for South-facing façades, 26 years for East-facing façades
and 24 years for West-facing façades. Fig. 16. Degradation curves for factor ‘‘DF5: solar orientation’’.
The degradation curves obtained are shown in Fig. 16 and the
respective parameters are presented in Table 10.
With the study sample of 100 buildings, the simultaneous
5.2.6. Combination of degradation factors application of more than two degradation factors usually leads to
Filtering the data through the simultaneous use of various deg- data sets not containing enough buildings for the results to be rep-
radation factors, allows a better data segmentation, grouping the resentative. For this reason, the combination of degradation factors
paint coatings according to their respective common characteris- was performed in this study with a maximum of two simultaneous
tics, thus providing degradation curves that are better adjusted factors.
to each specific coating, given its inherent characteristics, applica- As an example of the type of analysis performed, and of the results
tion conditions and degradation agents. it yields, the combination of the ‘‘DF3: paint surface texture – smooth

Table 8
Gompertz degradation curves for factor ‘‘DF3: paint surface texture’’.

Textured Smooth – plastic Smooth – oil


Parameters CMSE Parameters CMSE Parameters CMSE
Gompertz curves
a = 37.59611 125.52 a = 62.65949 158.78 a = 742.13802 201.76
b = 0.08874 b = 0.12391 b = 0.18346
M.A. Garrido et al. / Construction and Building Materials 29 (2012) 394–402 401

Table 10
Gompertz degradation curves for degradation factor ‘‘DF5: solar orientation’’.

North South East West


Parameters CMSE Parameters CMSE Parameters CMSE Parameters CMSE
Gompertz curves
a = 13.29133 25.57 a = 52.04740 257.85 a = 57.32268 117.64 a = 130.73159 197.34
b = 0.05828 b = 0.11698 b = 0.11399 b = 0.16008

degradation curves as they were the most consistent, presenting


lower CMSE error values.
The analysis of the degradation curves led to the following main
conclusions:

 DF1: coatings with film thickness above 400 lm show improved


performance over time when compared to coatings with infe-
rior film thicknesses;
 DF2: acrylic resin based coatings seem to present the best initial
performance, with oil-based coatings showing the worst perfor-
mance of the three binders;
 DF3: textured coatings present significantly better performance
Fig. 17. Degradation curves for simultaneous factors ‘‘FD3: paint surface texture: over time when compared to smooth coatings (plastic and oil);
smooth – plastic’’ and ‘‘FD4: substrate surface preparation’’.  DF4: the individual analysis of subtract preparation is not
conclusive, but in combination with ‘‘DF3: paint surface texture
– smooth – plastic’’ yields interesting results, with the smooth
– plastic coatings clearly presenting a better performance when
Table 11 applied directly over a clean substrate;
Gompertz degradation curves for simultaneous degradation factors ‘‘DF3: paint  DF5: coatings applied in North facing façades show the best
surface texture – smooth – plastic’’ and ‘‘DF4: substrate surface preparation’’. performance, whereas coatings in West facing façades show
Smooth – plastic + repaint removing Smooth – plastic + repaint over the quickest degradation, reflecting the influence of solar and
paint paint UV exposure.
Parameters CMSE Parameters CMSE
In conclusion, the methodology proved to be effective in provid-
Gompertz curves
a = 49.48653 153.05 a = 91.85606 147.76 ing analytical tools that can be used in the service life prediction of
b = 0.10702 b = 0.14933 façade paint coatings, and allowing for a cost estimation for the
best maintenance strategy if labor and material costs are available.

Acknowledgments
– plastic’’ factor and the ‘‘DF4: substrate surface preparation’’ factor
is presented. The obtained degradation curves are shown in Fig. 17, The authors wish to acknowledge Isabel Eusébio, Maria Paula
with the respective parameters being presented in Table 11. Rodrigues and Helena Silva from LNEC, as well as Maria Amélia
In the degradation curves shown in Fig. 17, it is seen that the Dionísio from CEPGIST for their support in the development of
improved performance over time of coatings applied directly over some of the experimental tests.
the façade render (repaint removing paint), when compared to the
coatings applied over previous paint layers (repaint over paint).
The reduced/no peeling period is of 27 years for repaint removing References
paint and 23 years for repaint over paint.
[1] Brundtland GH. Our common future: the world commission on environment
This means that the surface substrate preparation has a real and development. Oxford University Press: Oxford; 1987.
influence on the performance over time of paint coatings present- [2] Pearce D. The social and economic value of construction. The construction
ing a smooth – plastic texture. In fact, this suggests that these coat- industry’s contribution to sustainable development, nCRISP, London, UK; 2003.
[3] Branco FA, Brito J. Handbook of concrete bridge management – Ed. Am Soc Civ
ings are significantly vulnerable to a less careful preparation of the Eng, Reston, Dez; 2004. p. 468.
substrate, prior to their application. [4] Garrido M. Service life prediction of facade paint coatings in traditional
buildings. Methodology based on the inspection of in-use buildings, in
Portuguese, MSc thesis. IST Tech Univ Lisbon; 2010.
[5] Lacasse MA, Sjöström C. Recent advances in methods for service life prediction
6. Conclusions of building materials and components – an overview. In: Proceedings of the
CIB world building congress, Toronto, Ontario, Canada; 2 May, 2004. p. 1–10.
[6] Masters LW, Brandt E. Prediction of service life of building materials and
Aiming to contribute to the development of building manage- components, RILEM Technical Committee, CIB W80/RILEM 71-PSL Final Report.
ment systems, a service life prediction methodology is presented, Mater Struct 1987;20:55–77.
for the service life prediction of façade paint coatings in old [7] Masters LW, Brandt E. Systematic methodology for service life prediction of
building materials and components, RILEM recommendations, CIB W80/RILEM
buildings. The method was tested through its application to a
71-PSL TC. Mater Struct 1989;22:385–92.
sample of 100 buildings and considering the peeling as the coat- [8] Sjöström CH, Brandt E. Collection of in-service performance data: state-of-the-
ing defect. art and approach by CIB W80/RILEM 100-TSL, RILEM technical committees,
With the use of degradation curves it was possible to observe prediction techniques of service life, CIB W80/RILEM 100-TSL. Mater Struct
1991;24:70–6.
the influence of the five degradation factors in the performance [9] Architectural Institute of Japan (AIJ). The english edition of principal guide for
over time of paint coatings. The research used the Gompertz service life planning of buildings, AIJ, Japan; 1993.
402 M.A. Garrido et al. / Construction and Building Materials 29 (2012) 394–402

[10] International Organization for Standardization (ISO). Buildings and [13] Paulo PV. A building management system (BuildingsLife): application of
constructed assets – service life planning. Part 1: General principles, ISO deterministic and stochastic models with genetic algorithms to building
15686-1; 2000. façades, PhD thesis in Civil Engineering. Instituto Superior Técnico; 2009.
[11] Martin JW, Saunders SC, Floyd FL, Wineburg JP. Methodologies for predicting [14] Instituto de Meteorologia de Portugal (IM). Normais climatológicas 71-00 –
the service lives of coating systems. Gaithersburg, MD, USA: NIST Building Lisboa/Geofísico. Statistical data collected from 1971 to 2000. <http://
Science Series 172; 1994. www.meteo.pt/pt/oclima/clima.normais/012> [viewed 13.09.2011].
[12] Martin JW. A systems approach to the service life prediction problem for [15] Meeker WQ, Escobar LA. Statistical methods for reliability data, Wiley series in
coating systems, service life prediction of organic coatings: a systemic probability and statistics. New York: Wiley-Interscience; 1998. 1998.
approach, ACS symposium series. Am Chem Soc. vol. 722; 1999. p. 1–20.

You might also like