2

Estimating construction waste generation in residential buildings: A fuzzy set
theory approach in the Brazilian Amazon

3

Luiz Maurício Furtado Maués, Brisa do Mar Oliveira do Nascimento, Weisheng Lu, Fan Xue

1

4

This is the peer-reviewed post-print version of the paper:
Maués, L., Nascimento, B., Lu, W., & Xue, F. (2020). Estimating construction waste
generation in residential buildings: A fuzzy set theory approach in the Brazilian
Amazon. Journal of Cleaner Production, Article ID 121779, In press. Doi:
10.1016/j.jclepro.2020.121779
The final version of this paper is available at: https://doi.org/10.1016/j.jclepro.2020.121779.
The use of this file must follow the Creative Commons Attribution Non-Commercial No
Derivatives License, as required by Elsevier’s policy.
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24

Abstract
The estimate of construction waste generation is the key decision-making information for policymakers, construction managers, and the like to devise informed waste management strategies.
However, estimating construction waste generated from projects is particularly onerous, as
numerous factors related to design, site, and construction are largely in a fuzzy nature when the
estimating job is conducted. Built upon previous studies, this paper seeks to develop a model
that can be used to estimate construction waste generation based on fuzzy set theory. It follows
a trilogy of methodology, including model development, sensitivity analysis, and model
validation. A set of IF-THEN rules are developed based on two independent variables, built area
and number of floors. A sensitive analysis was conducted to evaluate the influence of the
independent variables on waste generation. The model is further calibrated and verified through
a case study of 23 residential buildings constructed in the Brazilian Amazon. The model obtained
an accuracy of 64.29% in the development phase and 66.67% in the validation phase, showing
that the results are largely acceptable. By using this model, it is possible for a waste manager to
draw up a baseline graph to indicate the volume of construction waste generation as his/her
building project as it progresses. The research is also of novelty by using fuzzy set theory to deal
with the fuzzy nature of waste generation in construction projects. Further studies are
recommended to enhance the accuracy level of the model by engaging more factors and more
quality data.

25
26

Keywords: Construction waste; Building; Waste quantification; Fuzzy set theory; Brazil.

27
28
29
30
31
32
33

1. Introduction
The exponential growth of construction activities and their associated waste have given rise to
construction waste management (CWM) around the world. Construction and demolition (C&D)
waste, using interchangeably with the term construction waste, refers to the solid waste that
arises from construction activities, such as new building, site formation, renovation, or
demolition (HKEPD, 2019; Roche and Hegarty, 2006). It normally forms a signification portion

34
35
36
37

of total solid waste. Approximately 35% of the solid waste generated in the world comes from
construction. It is usually disposed of in landfills or uncontrolled/inadequately maintained
locations. Without proper management, C&D waste will cause severe damage to the natural
environment and quality of life.

38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55

Estimating construction waste generation lies at the kernel of any effort to properly manage
construction waste (Lu et al., 2016a). Such management strategies may include developing a
waste management plan comprising of onsite and offsite storage, transportation, reduction, reuse,
and recycling, which is also known as the “3R”. Numerous studies have been previously
conducted to estimate and forecast construction waste generation; these studies are normally put
under the nomenclature name of “quantifying waste generation’. Wu et al. (2014) conducted one
of the most comprehensive reviews on C&D waste quantification studies. They classified the
studies of this steam into six types: site visit, waste generation rate (WGR), lifecycle analysis,
classification accumulation, variables modeling, and others. The methods could be used
standalone or combined in the real-life estimate. Yu et al. (2019) provided a detailed elaboration
of WGR and its related data collection methods such as on-site investigation, analyzing waste
disposal records (Kleemann et al., 2016), or material flow onsite or offsite (Cochran and
Townsend, 2010). Researchers (e.g., Li et al., 2013; Lu et al. 2016b; Lu et al., 2015b) have
reported that unit of analysis of estimating C&D waste generation could be a project, a region,
or a nation. The unit of analysis of this research is a project, as it has a clear system boundary
and its result can be used to estimate the waste generation at a regional or national level, e.g., by
summing up the project level estimations.

56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71

Estimating construction waste generation is particularly onerous, as projects are different from
each other in terms of design (e.g., modular design, or parametric design), site and surrounding
(e.g., excavation, land terrace), and construction (e.g., technologies, material handling, and
craftsmanship). All these factors, particularly after combinations, will impact construction waste
generation to a varying extent. There exists vagueness between the factor descriptions and their
causal waste generation. For example, modular design is adopted as one of the curtailing
strategies to reduce construction waste. However, how to measure the level of modular design is
quite controversial. It is even far from clear what level of modular design will lead to what level
of construction waste generation. Dealing with the fuzziness is particularly important in
developing countries where exponentially growing construction activities generate massive
waste while statistics about it are largely in absence. Notably, fuzzy set logic derived from fuzzy
set theory is particularly useful to deal with some practical issues with uncertainties and
vagueness. Previous studies have seen its application across fields, including the estimate of
waste generation. However, rather limited research has applied fuzzy set theory to estimate
specifically construction waste generation, which is the research gap this study intends to fill.

72
73
74

This paper aims at developing a model to quantify the potential construction waste generation
from a project by taking the fuzziness of the project and its waste generation nature into
2

75
76
77
78
79
80
81
82
83
84
85
86

consideration. It does so by mainly following a trilogy of methodology, including model
development, sensitivity analysis, and model validation. The model is developed from a case
study of a series of high-rise residential buildings in the Amazon region, Brazil, where
construction activities are thriving, but the estimation of waste generation is still lacking. The
remainder of the paper is structured as follows. Following this introductory section is a literature
review of two issues: waste estimation studies, and fuzzy set theory. Section 3 is a detailed
description of the fuzzy logic-based model. Section 4 is to elaborate on the background of the
case study and data sample. Section 5 is data analyses and findings, followed by a discussion in
Section 6, and conclusions in Section 7. This research provides useful references to estimating
construction waste generation by considering the fuzziness of the subject under investigation. It
can be used in other projects, particularly when data is sparse, to estimate waste generation and
devise proper construction waste management plans.

87
88
89
90
91
92
93
94

2. Quantification of construction waste generation
Wu et al. (2014) highlight the importance of quantifying construction waste generation, seeing
it “a prerequisite for the implementation of successful waste management”. They conducted one
of the most comprehensive literature reviews by focusing on 57 journal papers with a niche of
quantifying construction waste generation. There is no need for this paper to “reinvent the wheel”
to repeat such review. Rather, here it is to highlight some relevant findings and update the latest
references since then, with a view to providing a solid literature foundation for this study.

95
96
97
98
99
100
101
102
103

Wu et al. (2014) identified three primary waste generation activities: (i) new construction, (ii)
demolition of old facilities, and (iii) civil and infrastructure works. This echoes with the
definition of construction waste as the solid waste that arises from the construction activities as
rehearsed above. The 57 papers reviewed by them did not allow the renovation surface as a major
construction activity leading to waste generation. This reflects the fact that renovation waste has
not been well documented (Lu et al., 2016). Renovation actually generates a considerable amount
of waste, but such activities are often small-scale and dispersed around, making it difficult to be
traced and documented. As will be elaborated later, this study focuses on new buildings.

104
105
106
107
108
109
110
111
112
113
114
115

Their review also identified two “estimation levels” (i.e., units of analysis): project level and
regional level. The focus of this paper is at a project level. What is most relevant to the review
is the quantification methodology. The methods adopted in previous studies were categorized
into six major categories, which consist of site visit (e.g., direct or indirect measurement), WGR
calculation (e.g., per-capita, per-floor area), classification accumulation, lifetime analysis (e.g.,
lifecycle analysis, and material lifetime analysis), modelling based on multiple parameters, and
other particular methods. Particularly, WGR has been broadly adopted to measure the waste
generation, and also the effectiveness of waste management. Table 1 is a non-exhaustive list of
the WGR indicators adopted in previous studies. There is little uniformity in the literature in
relation to the indicators of either the volume or the mass of construction waste generated. For
example, in Table 1, the largest indicator of the volume of construction waste generated is 0.4193
3

116
117
118
119
120
121

m³/m², more than three times as the smallest volume of 0.118 m³/m². Lu et al. (2011) provided
a comprehensive review of WGRs before using them in their empirical studies. The bright side
is that construction waste is tangible and able to be measured. The indicators can be converted
from one another based on some experiential formula. Whether to use one indicator or another
is also dependent on the availability of data or construction quantity convention in a region (Lu
et al., 2011).

122
123

Table 1 Waste generation indicators

124

Author
Dias
Villoria Sáez et al.

Year
(2013)
(2012)

Country
Brazil
Spain

Llatas

(2011)

Spain

Katz and Baum
Ortiz et al.
Kofoworola and Gheewala
Kharrufa
Maña I Reixach et al.
Pinto

(2011)
(2010)
(2009)
(2007)
(2000)
(1999)

Israel
Spain
Thailand
Iraq
Spain
Brazil

Indicator
0.128 m³/m²
0.2 m³/m²
0.1388 m³/m² without soil
0.4193 m³/m² with soil
0.200 m³/m²
205.9 kg/m²
21.38 kg/m2
215 kg/m²
0.118 m³/m²
150 kg/m²

125
126
127
128
129
130
131
132
133
134
135
136
137
138

Previous studies have also well explored the factors that can lead to different waste generation
level. From a project perspective, Wimalasena et al. (2010) established a quantification
framework, which identified five generic factors to estimate construction waste generation: “(1)
activity specific factors; (2) labour and equipment-related factors; (3) material and storagerelated factors; (4) site condition and weather-related factors; (5) company policies”. Using big
data analytics, Chen and Lu (2017) found that location, building types, and the clients of a
building project have an impact on demolition waste generation to a varying extent in the Hong
Kong context. Although construction waste is generated during the construction stage, it is also
determined by design stage (Baldwin et al., 2009; Lu et al., 2017); With these factors, researchers
have developed different models to quantify construction waste generation (See Table 2).
Generally, as shown in Table 2, the most commonly adopted model to estimate construction
waste generation in projects is statistical quantification factors, while statistical multiple
regression and web-based construction waste estimation system are less frequently adopted.

139
140

Table 2 Models to estimate construction waste generation in projects
Author
Country
Model
Mah et al. (2016)
Malaysia
Statistical quantification factors
Parise Kern et al. (2015)
Brazil
Statistical multiple regression
Li and Zhang (2013)
Hong Kong
Web-based construction waste
estimation system
4

Villoria Sáez et al. (2012)
Katz and Baum (2011)
Llatas (2011)
Kofoworola and Gheewala (2009)
Cochran et al. (2007)

Spain
Israel
Spain
Thailand
USA

Statistical quantification factors
Statistical multiple regression
Statistical quantification factors
Statistical quantification factors
Statistical quantification factors

141
142
143
144
145
146
147
148
149
150
151

A major shortcoming of previous models is that they are yet to consider the complexity of the
factors, their interactions, and in particular their fuzziness. The majority of the factors mentioned
above cannot be clear-cut defined and measured in construction projects. For example, in
addition to the modular design as a strategy to curtail construction waste, there are many 3R
factors, which cannot be simply put in classical bivalent sets. Rather, they can be straddling in
different grades, i.e., they presenting fuzziness. Foundation design and construction is different
from one project to another, which will lead to different waste generation. Demolition methods,
such as blaster, deconstruction, or their mixture will be different and in turn, determine waste
generation levels. It is not surprising that previous models yield quite divergent estimates (e.g.,
Lu et al., 2011).

152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173

3. Research methodology
3.1 Fuzzy set theory
Unlike the classic set theory wherein membership of an element in a set is in dichotomous terms
(i.e., either 0 or 1), the fuzzy set theory allows partial membership, which means it is containing
elements that have varying degrees of membership in the set; it follows a membership function
valued interval [0,1]. The theory was introduced by Zadeh (1965) to reflect the fact that most
modes of human reasoning in uncertain and imprecise environments are approximate in nature.
Fuzzy set theory is a paradigm shift from traditional set theory. The theory can be applied in a
wide range of domains, particularly useful under the condition where information is incomplete
or imprecise. For example, in the domain of bioinformatics, Liang et al. (2006) identified
disease-associated genes from microarray gene expression profiles based on the fuzzy set theory.
Fuzzy logic is derived from fuzzy set theory. It is capable of handling inherently imprecise (i.e.,
vague, inexact, rough, or inaccurate) concepts. Both fuzzy set theory and fuzzy logic thus
received widespread applications. In the scientific literature, for example, fuzzy models are used
to make diffuse predictions in several areas of knowledge (Carvalho and Costa, 2017; Ghadi et
al., 2016; Islam et al., 2017). Some of these possible applications in the area of geotechnics
include the hybrid model using fuzzy logic and Bayesian networks to evaluate the risk, as to the
quality and safety in the execution of deep tubular foundations of buildings in urban areas (Zhou
et al., 2011). Fuzzy logic was also used in the evaluation of the experiments where ceramic
material residues were added as material to be added to asphalt composition (Kara and Karacasu,
2017).

174
175
176

In construction management, fuzzy logic empowered by other tools was used to perform
modeling for risk analysis, e.g., Khazaeni et al. (2012) supporting decision making in the
5

177
178
179
180
181
182
183
184
185
186
187
188
189

allocation of resources in contract management; Lam et al. (2007) in relation to the risk between
the contracting parties of construction projects. Other applications can be found in Afshar et al.
(2017), (Mirahadi and Zayed, 2016). In the environmental field, Huang and Hsu (2011) created
a model using fuzzy to evaluate the performance of environmental indicators of sustainable
construction. Ighravwe and Oke (2019) developed a multi-criteria method with the use of the
diffuse technique to evaluate the adaptation of sustainable requirements in public buildings.
Zarghami et al. (2018) elaborated a method to prioritize the aspects that should be used in Iranian
buildings with respect to sustainability requirements. Particularly, Li and Chen (2011) developed
a stochastic linear hybrid model to aid the estimate of municipal waste generation under
uncertainties. Cai et al. (2007) also used fuzzy logic to create a hybrid model to assist Municipal
Solid Waste (MSW) management under uncertainty. It can be summarized from the literature
review that fuzzy set logic, properly developed, can handle the uncertainties and vagueness of
waste generation from construction projects.

190
191
192
193
194
195
196
197
198
199
200
201
202

A classification model of the potential of waste generation was created based on diffuse logic,
the choice of the fuzzy methodology occurred in function of the same one to provide the creation
of models for the decision making in uncertain and imprecise environments, using two
independent variables (built area and number of floors of buildings), and the triangular and
trapezoidal pertinence functions were used in the modeling. Basically, the research methodology
can be known as trilogy, from fuzzy model development stage where some simple fuzzy logic
can be derived from experienced practitioners, then the sensitivity analysis stage to identify the
two independent variables, the built area and number of floors, whichever is more sensitive to
the waste generation, to the final model validation stage where the percentage of assertiveness
of the model can be verified with another batch of data independent from the data used in the
model development stage. The detailed description for each stage of the methodology is
elaborated separately as follows.

203
204
205
206
207
208
209
210
211
212
213
214

3.2 Fuzzy modelling
With the elaboration of this method in the modeling, the number of rules used was 25. Therefore,
the IF-THEN rules that were developed in this study were generated from the combination of
the two functions of relevance for each of the variables. IF-THEN rules are normally used to
formulate the conditional statements comprising fuzzy logic. For this, 18 construction engineers
were interviewed by means of a questionnaire, aiming to know the opinions of specialists to
elaborate the fuzzy rules, the mode values of the interviews were used to define the functions of
pertinence, as shown in the following Table 3. Here, in Table 3, two conditional statements,
which are number of floors and built area are required to elicit the result, construction waste
generation. For example, if number of floors is high while built area is low, then the estimate of
construction waste generation is medium.

215
216

Table 3 List of “IF-THEN” rules for estimating waste (m3) based on area and number of floors
6

Area
Very low
Low
Medium
High
Very high
217

Very low
VL
VL
L
L
L

Low
VL
VL
L
L
M

Number of floors
Medium
High
L
M
L
M
M
H
M
H
H
VH

Very high
H
H
VH
VH
VH

Note: VL= very low, L= low, M= medium, H= high, and VH= very high

218
219
220
221
222
223
224

In order to perform the modeling, the following fuzzy characteristics were used: Let X be the
universe of discourse of the linguistic variable Zi: i = 1;2;...; n (n ≤ T) and ϕ (Zi) a fuzzy number
represented by using a triangular and trapezoidal fuzzy set Aj (j = 1;2;. . .;m) associated with Zi.
Then, ϕ (Zi): X → [0;1] and Aj will be considered a triangular and trapezoidal fuzzy set if it is
defined on an interval xj ⊂ X (j = 1;2;...;m) according to the function with respect to the
following:

225
226

-Triangular fuzzy set

227

µ A(x)

228
229
230

0, if Zi ≤ aj
(Zi - aj) / (bj- aj), if Zi ϵ ] aj, bj [
(cj - Zi) / (cj - bj), if Zi ϵ [bj, cj [
0, se Zi ≥ cj

231
232

and

233
234

- Trapezoidal fuzzy set

235
236
237
238
239

µ A(x)

0, if Zi ≤ dj
(Zi - dj) / (ej-dj), if Zi ϵ ] dj, ej]
1, if Zi ϵ ] ej,fj]
(gj- Zi) / (gj - fj), if Zi ϵ [fj, gj [
0, if Zi ≥ gj

240
241
242
243
244

where aj, cj, and bj are scalar parameters that show the position of the vertices and the midpoint
of an isosceles triangle, respectively; dj and gj are the scalar parameters that indicate the position
of the vertices of the bottom end of a trapezoid; ej and fj are the vertices of the upper base of the
ends of a trapezoid; and uj is the support of the triangular fuzzy sets m associated with Zi.

245
246
247
248
249
250

The application of the mathematical concepts of fuzzy logic to the construction of the model
was, therefore, used from the two functions described above because the combination of the two
types of pertinence function allowed the model to have good precision, as opposed to the use of
only one type of function, which resulted in low assertiveness. The applied model is represented
in Fig. 1. For example, when a building has a fuzzy number of floors at 80% “Low” and 20%
7

251
252
253
254
255

“Medium” and a fuzzy area at 40% “Low” and 60% “Medium,” the chance of “Very Low” waste
in the model will be 32% according to Table 3; while the chances of “Low” and “Medium”
wastes will be 56% and 12%, respectively. The task of the fuzzy model calibration in this paper
is thus to find the optimal parameters to maximize the predictions made by the model shown in
Table 3 and Fig. 1.

256

257
258

Fig. 1. Input and output of the fuzzy subset

259

Note: VL= very low, L= low, M= medium, H= high, and VH= very high

260
261
262
263
264
265

3.3 Developing a fuzzy classifier: A classification of the constructions was carried out from the
perspective of the potential volume of residues to be generated during the works. A fuzzy
classifier was created based on the volume of waste generated. As shown in Table 4, this
classification occurs through the results of the inference from the model described in the previous
step.

266
267

Table 4 The fuzzy set classifier in relation to waste generation potentials
Classification regarding the
Very low
Low
Medium High
Very
potential to produce waste
(class 1)
(class 2) (class 3) (class 4) high
(class 5)
Classification intervals regarding [0, 1800]
(1800, (3800,
(5800, (7300,
the volume of waste to be
3800]
5800]
7300]
+∞)
generated

268
269
270
271
272

This classification becomes an important indicator in the phase of identification of the
construction waste volume. Therefore, a classification with five classes was created (see Table
4). Each class created has an interval regarding the volume of waste to be generated from the
lowest [0, 1800] to the highest (7300, +∞) and also a descriptive classification regarding the

8

273
274

potential to produce waste from the lowest ‘Very low (class 1)’ to the highest ‘Very high (class
5)’.

275
276
277
278
279
280
281
282
283
284
285
286

3.4 Sensitivity analysis: In this stage, a sensitivity study of the results generated through the
proposed model was conducted to evaluate the influence of the independent variables (built area
and number of floors of the building) in the generation of waste. For this, the values of built-up
areas were stated in the fuzzy model, starting at 8,000 m2 on a growing scale every 2,000 m2, up
to the limit of 40,000 m2, leaving another fixed independent variable (number of floors) constant.
For each value released, the value inferred by the fuzzy system regarding the volume of waste
generated was recorded. Afterward, the same procedure was performed by inverting the
independent variables, varying the number of floors (starting with 14 going up to 36 type
pavements) and leaving the variable built area fixed. In this way, a graph can be constructed to
obtain the potential volume of waste to be generated as a function of the two independent
variables.

287
288
289
290
291
292
293
294
295

3.5 Model verification: After performing fuzzy inference using the proposed model, a
comparison of the obtained results was performed, through the classifier of the potential of waste
generation of the work, aiming at verifying the percentage of assertiveness of the model in
relation to the data collected in the works. This assertiveness was performed by means of the
percentage of correct classifications found among the five classes defined in item “d”. After the
validation of the model, the classification (among the five classes, see item “d”) in terms of the
number of works that fit each classification in relation to the potential of waste generation
volume was performed.

296
297
298
299
300
301
302
303
304
305
306
307

4. Case description
The case area in this study is located in the city of Belém, the Brazilian Amazon (see Fig. 2).
Data were collected from 23 residential buildings that have completed their works between the
years 2016 and 2018 (see Fig. 3). All these projects located in the center of the city. It is important
to note that construction companies in Brazil do not have a tradition of collecting and
disseminating indices, whether they are performance, consumption, productivity and mainly
about the generation of waste in their works. This makes it difficult to access information and
conduct research. The information used in this work was obtained after a large number of
dialogue with the companies, so that they would allow access to their construction sites and
allowing to collect the data used in this research; a fact that highlights the importance of this
information.

308
309
310
311

9

312
313

Fig. 2. Location of the research targeted buildings

314

315
316

Fig. 3. A glance of some of the construction projects participated in this research

317
318
319
320
321
322
323
324
325
326

The construction projects in this samples have the following characteristics: deeply excavated
substructures, reinforced-concrete superstructures using cast in-situ technology, and ceramic
blocks with interior and exterior vertical envelopment systems. The exterior walls and floors are
made of concrete, and with sand mortar and ceramic tiles as the finishing. The interior floors are
fitted with ceramic or porcelain tiles. Kitchens and bathrooms are fitted with wall ceramic tiles.
This construction system is widespread in the region for vertical constructions (Maués et al.,
2017). In this way, knowing that all the research works have the same characteristics and
constructive methods, it was possible to make a comparison according to the homogeneity of the
samples.

327
328
329
330
331
332

Data collection was carried out by a master’s degree student in civil engineering under the
supervision of a professor at Universidade Federal do Pará. Monthly visits were conducted at all
working sites where information on the volume of waste generated in the month was collected.
This information came from the records of the number of containers that were used by the
company responsible for transport and disposal, as this service was performed by outsourced
10

333
334
335
336
337
338
339

companies. The quantification of the residues was conducted based on the construction waste
routinely removed from the construction site in special containers by a specialized company. The
brief information of the 23 sample buildings is listed in Table 5 where each labelled building has
three characteristics, including total construction area, number of floors and respective generated
construction waste by volume. Of the 23 construction sites in the sample, 14 projects (B1-B14)
were randomly selected to train the fuzzy logic model, and the other 9 projects (B15-B23) were
used to validate the model.

340
341

Table 5 Brief information of the 23 sample projects

Model Validation

Model Construction

Buildings

Total
Number Construction
constructed area of floors Wast (m³)

B1
B2
B3
B4
B5
B6
B7
B8
B9
B10
B11
B12
B13
B14
B15
B16
B17
B18
B19
B20
B21
B22
B23

20.542,02
8.375,49
12.347,91
18.877,59
40.691,34
30.718,78
18.079,21
15.303,61
15.023,38
22.667,32
8.605,74
10.919,69
14.790,55
16.732,00
14.831,00
42.000,00
25.320,00
9.870,39
10.919,69
9.858,00
12.840,50
28.297,96
41.559,52

35
25
26
14
35
35
30
32
28
26
33
35
28
26
29
12
14
21
34
28
25
29
17

1.588,22
4.092,00
1.925,00
5.320,00
12.375,00
11.450,00
6.020,00
4.145,00
3.305,00
3.520,00
2.880,00
2.635,00
3.279,00
2.045,00
5.360,00
4.660,00
4.510,00
2.130,00
3.140,00
3.355,00
3.435,00
4.780,00
5.030,00

Buildings

Total constructed
area

Number
of floors

Construction
Wast (m³)

B1
B2
B3
B4
B5

20.542,02
8.375,49
12.347,91
18.877,59
40.691,34

35
25
26
14
35

1.588,22
4.092,00
1.925,00
5.320,00
12.375,00

Model
Construction

342

11

Model Validation
343
344
345
346
347
348
349
350
351
352

B6
B7
B8
B9
B10
B11
B12
B13
B14
B15
B16
B17
B18
B19
B20
B21
B22
B23

30.718,78
18.079,21
15.303,61
15.023,38
22.667,32
8.605,74
10.919,69
14.790,55
16.732,00
14.831,00
42.000,00
25.320,00
9.870,39
10.919,69
9.858,00
12.840,50
28.297,96
41.559,52

35
30
32
28
26
33
35
28
26
29
12
14
21
34
28
25
29
17

11.450,00
6.020,00
4.145,00
3.305,00
3.520,00
2.880,00
2.635,00
3.279,00
2.045,00
5.360,00
4.660,00
4.510,00
2.130,00
3.140,00
3.355,00
3.435,00
4.780,00
5.030,00

5. Results and findings
The general structure of the model developed is a fuzzy logic-based decision-making model with
Mamdani inference (see Table 6). The accuracy of the model developed is dependent on whether
the actual volume of waste generated (left half of Table 6) and the predicted volume of waste
with the model (right half of Table 6) fall within the same classification interval. During the
construction of the fuzzy model, 9 of the 14 sample values were correctly classified in relation
to the volumes of waste generated during the construction of the projects. This represents an
accuracy of 64.29%, and a 27.20% MAPE (mean abs percentage error) for waste volume (m3)
prediction. Of these, four works are small waste generators (28.57%), three (21.42%) are
medium waste generators, and two (14.28%) are very high waste generators.

353
354
355
356
357
358
359
360
361
362
363
364
365
366

After analyzing the result generated by the model, the margin of accuracy was considered
acceptable and therefore met its expectations. A validation phase was then performed with the
purpose of validating the model, this time with the other nine works of the initial sample that
were not used in the construction of the model (being chosen randomly among the 23 works of
the sample) (See Table 7). As shown in Table 7, there are totally 6 buildings (B16, B19, B20,
B21, B22, and B23) of which their actual waste generated and predicted waste fall in the same
classification interval, which indicates a 66.67% accuracy in the second modelling. Among
these, two works (60.86%) were classified with a small risk and seven (30.43%) with a medium
risk in the generation of residues. In terms of construction waste volume (m3) prediction, the
MAPE was 33.62%; in comparison, the linear regression (Waste = -1,595.309 + 0.312* area +
19.086* No. of floors) returned an MAPE at 86.25%. The performance of positive classifications
in the modelling (among the 23 works of the sample in the target region of the research) shows
that there is a tendency of the works to present a potential in the generation of residues such as
12

367
368

small (34.78%), medium (21.74%) and very large (08.69%); the other 34.78% of the samples
were not correctly classified by the model.

13

Table 6 The fuzzy classifier of construction waste generation – Model development
Fuzzy Classifier Risk - Construction waste
Construction
Buildings
Wast (m³)
B1
B2
B3
B4
B5
B6
B7
B8
B9
B10
B11
B12
B13
B14

1588
4092
1925
5320
12375
11450
6020
4145
3305
3520
2880
2635
3279
2045

very
low
[0,
1800]
1

B1
B2
B3
B4
B5

high

very high

(1800,
3800]

(3800,
5800]

(5800,
7300]

(7300,
+∞)

1
1
1
1
1
1
1
1
1
1
1
1

Construction
Wast (m³) [0, 1800]

1588
4092
1925
5320
12375

medium

1

very low

Buildings

low

Wast
Class

Fuzzy
Modeling
Construction
Wast (m³)

1
3
2
3
5
5
4
3
2
2
2
2
2
2

2080
2400
3260
4780
7010
7110
2260
3940
3850
4370
2430
3000
3780
3780

very
low
[0,
1800]

low

medium

high

very high

(1800,
3800]
1
1
1

(3800,
5800]

(5800,
7300]

(7300,
+∞)

1
1
1
1
1
1
1
1
1
1
1

low

medium

Fuzzy Classifier Risk - Construction waste
Fuzzy
high
very high
very low

low

medium high

(1800,
3800]

(3800,
5800]

(5800,
7300]

(1800,
3800]

(3800,
5800]

(7300, +∞)

Wast
Class

1

1
3
2
3
5

1
1
1
1

Modeling
Construction [0, 1800]
Waste (m³)

2080
2400
3260
4780
7010

Fuzzy
Modeling
Wast
Class
2
2
2
3
5
5
2
3
3
3
2
2
2
2

Fuzzy
Modeling
(5800,
Wast
(7300, +∞)
7300]
Class

very high

1
1
1
1
1

2
2
2
3
5

B6
B7
B8
B9
B10
B11
B12
B13
B14

11450
6020
4145
3305
3520
2880
2635
3279
2045

1
1
1
1
1
1
1
1
1

5
4
3
2
2
2
2
2
2

7110
2260
3940
3850
4370
2430
3000
3780
3780

1
1
1
1
1
1
1
1
1

5
2
3
3
3
2
2
2
2

Table 7 The fuzzy classifier of construction waste generation - Model validation
Fuzzy Classifier Risk - Construction waste -Model Validation
very
low
Construction
Buildings
Wast (m³)

[0,
1800]

low
(1800,
3800]

medium
(3800,
5800]

high
(5800,
7300]

very high
(7300,
+∞)

Wast
Class

Fuzzy
Modeling
Construction
Waste (m³)

B15

5360

1

3

2400

B16

4660

1

3

5600

B17

4510

1

3

3070

B18

2130

1

2

1530

B19

3140

1

2

2200

very
low

low

medium

high

very high

[0,
1800]

(1800,
3800]

(3800,
5800]

(5800,
7300]

(7300,
+∞)

1

2
1

1
1

Fuzzy
Modeling
Wast
Class

3
2
1

1

2
15

B20

3355

1

2

2400

1

2

B21

3435

1

2

2400

1

2

B22

4780

1

3

5230

1

3

B23

5030

1

3

5600

1

3

Buildings

Construction
Wast (m³)

B15
B16
B17
B18
B19
B20
B21
B22
B23

5360
4660
4510
2130
3140
3355
3435
4780
5030

very low

low

Fuzzy Classifier Risk - Construction waste -Model Validation
Fuzzy
very low
medium high
very high

low

medium high

[0, 1800]

(1800,
3800]

(3800,
5800]

(1800,
3800]

(3800,
5800]

1
1
1
1
1
1
1
1
1

(5800,
7300]

(7300, +∞)

Wast
Class

Modeling
Construction
Waste (m³)

3
3
3
2
2
2
2
3
3

2400
5600
3070
1530
2200
2400
2400
5230
5600

[0, 1800]

1
1
1
1
1
1
1
1
1

(5800,
7300]

very high
(7300,
+∞)

Fuzzy
Modeling
Wast
Class

2
3
2
1
2
2
2
3
3

16

Based on the results obtained, a sensitivity analysis was performed to test the reliability of the
model. This analysis was carried out based on the variation of the independent variables (area
and floors), by generating a graph that easily allows the prediction of the possible volume of
construction waste to be generated in future construction projects in the target city. Results in
Fig. 4 show that the volume of construction waste is strongly related to the construction area of
the building, representing the growth of waste generation in a more significant way, as there was
a 248% increase in waste between the range of 8,000 m2 and 40,000 m2. The increase in relation
to the higher number of floors, for the interval between 14 and 36 floors, was 66.67%.

Fig. 4. A sensitivity analysis of potential waste generation volume in new construction
This result also shows no variation in the increase of construction waste between buildings that
have between 14 and 18 floors, as compared to those between 26 and 30 floors. This analysis
and the plotting of the graph allow managers to easily estimate the construction waste generated
in new construction simply by defining the area and number of floors of the residential building.
Thus, the information will be available without the need to use the model created through fuzzy
logic. This could restrict its applicability.
With this study, it was still possible to identify the average construction waste volume generated
by the 23 works surveyed, being 0.21 m3/m2 of construction floor area, which serves as the initial
indicator for works in the region, given that there is no reference value in it. This value is very
close to that found by Villoria Sáez et al. of 0.20 m³/m² (2012) in Spain and much higher than
those found by Kern et al. (2015) in a city in the southern region of Brazil, which was 0.128
m³/m², and by Maña I Reixach et al. (2000), 0.118 m³/m², also in Spain, thus demonstrating that

there is still considerable potential for action to be taken to reduce the impact of waste from
building construction.
6. Discussion
Estimating construction waste generation can provide critical decision-making information for
the many efforts to manage waste in construction projects. Researchers have developed a popular
measurement of WGR, e.g., volume (m3) or quantity (tons) of waste generated per m2 of Gross
Floor Area (GFA) (Lu and Yuan, 2011), assuming that by timing the WGR and the GFA an
overall volume of construction waste generation can be derived. However, there are two primary
hurdles preventing this WGR-based approach from practical applications. First, the WGRs are
too disparate to be accepted as the estimate. Therefore, researchers have used big data to refine
the resulted WGR, as suggested by the Law of Large Numbers, the mean of the results obtained
from numerous tries should tend to become convergent as more data is available (Sen and Singer,
1993). Lu et al. (2015a; 2015b) proved this law, but in many developing economies such as the
Brazilian Amazon, big data is not available. Second, construction waste generation from a
project is rarely linear. For example, Lu et al. (2016b) modeled it as an S-curve in high-rise
concrete buildings. Wu et al. (2019) used an off-site snapshot methodology to estimate building
waste. Construction waste generation is impacted by a large number of factors which are largely
in a fuzzy nature.
The fuzzy logic-based approach developed in this study overcame the above hurdles to a large
extent. It uses some simple fuzzy logic derived from experienced practitioners. It captures the
tacit knowledge residing in these practitioners' mind about the fuzzy nature of waste generation.
It requires no heavy data, but the estimated result of 66.67% is still acceptable for developing a
waste management plan or the like, particularly when the project is at a very early stage. Future
studies can be conducted to refine the estimate result further. Actually, in real-life construction
time and cost management, no one will naively stick to an initial estimate without any change as
the project progresses. Rather, experienced managers will rely on emerging information to finetune their baseline estimate and devise interventions accordingly. To this end, other factors can
be incorporated into the fuzzy logic model to refine the estimate for practical use continuously.
7. Conclusions and future works
Construction waste generation is the key information for devising reduce, reuse, and recycle (i.e.,
3R) measures. However, estimating waste generation from a new project is rather difficult as it
is essentially an attempt to extrapolate current understanding to a future situation that is often
full of uncertainties and fuzziness. This research tried to capture the fuzzy nature of construction
waste generation estimate by developing a fuzzy logic model from a case study in the Brazilian
Amazon. We used the sample data from 14 construction projects and further validated the model
using the sample data from nine other projects. The construction projects were classified on a
five-class scale for the potential of waste generation, and a graph that allows us to estimate the
volume of waste to be generated by new building projects was plotted. An accuracy level of
18

64.29% in the classification of waste generation was achieved by the model. In the validation
stage, the level of accuracy was improved and reached to 66.67%. It is discovered that the floor
area is much more significant than the number of floors in predicting the volume of waste
generation. It also enabled the creation of a graph that allows the agents involved in the
construction to estimate the volume of waste to be generated by new enterprises quickly and with
an adequate degree of reliability.
One of the limitations of the model is its generalizability. It cannot be applied to other localities
or other types of construction projects, given the fact that in each region, the volume of waste
residues is well-differentiated according to the cultural characteristics and the construction
techniques used. However, the research provides a good reference for comparative studies in the
future. The overall methodology is also inspiring for quantifying waste generation from
construction projects, particularly by incorporating in their fuzziness. Future studies are
recommended to refine further the fuzzy logic model, e.g., by introducing bigger data, to enhance
its level of accuracy in estimating. It may also be possible to seek to use other methods such as
Artificial Neural Network (ANN) or Long Short-Term Memory (LSTM) when data is allowed.
Studies are recommended to redevelop the model for other sectors of the construction industry,
with respect to their project typologies and regional particularities. It is expected to extend the
validation to other locations and other project sets or small urban areas to calibrate the method.
Glossary table
Symbol
CWM
C&D
GFA
MAPE
MSW
WGR

Meaning
Construction waste management
Construction and demolition
Gross floor area
Mean abs percentage error
Municipal solid waste
Waste generation rate

Acknowledgment
This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível
Superior - Brasil (CAPES) - Finance Code 001 and The author thanks all the construction firms
that contributed information.
References
Afshar, M.R., Alipouri, Y., Sebt, M.H., Chan, W.T., 2017. A type-2 fuzzy set model for
contractor
prequalification.
Autom.
Constr.
84,
356–366.
https://doi.org/10.1016/j.autcon.2017.10.003
Baldwin A, Poon CS, Shen LY, Austin S, Wong I. Designing out waste in high-rise residential
buildings: analysis of precasting methods and traditional construction. Renew Energy,
2009;34:2067–73.

19

Cai, Y., Huang, G.H., Nie, X.H., Li, Y.P., Tan, Q., 2007. Municipal Solid Waste Management
Under Uncertainty: Environ. Eng. Sci. 24, 338–352. https://doi.org/10.1089/ees.2005.0140
Carvalho, J.G., Costa, C.T., 2017. Identification method for fuzzy forecasting models of time
series. Appl. Soft Comput. 50, 166–182. https://doi.org/10.1016/j.asoc.2016.11.003
Chen, X. and Lu, W., 2017. Identifying factors influencing demolition waste generation in Hong
Kong. Journal of Cleaner Production, 141 (2017) 799-811.
Cochran, K., Townsend, T., Reinhart, D., Heck, H., 2007. Estimation of regional building-related
C&D debris generation and composition: Case study for Florida, US. Waste Manag. 27,
921–931. https://doi.org/10.1016/j.wasman.2006.03.023
Cochran, K.M., Townsend, T.G., 2010. Estimating construction and demolition debris
generation using a materials flow analysis approach. Waste Manag. 30, 2247–2254.
https://doi.org/10.1016/j.wasman.2010.04.008
Dias, M.F., 2013. MODELO PARA ESTIMAR A GERAÇÃO DE RESÍDUOS NA
PRODUÇÃO DE OBRAS RESIDENCIAIS VERTICAIS. Tese (Doutorado em
Engenharia de Recursos Naturais) - Programa de Pós-Graduação em Engenharia Civil,
Ghadi, Y.Y., Rasul, M.G., Khan, M.M.K., 2016. Design and development of advanced fuzzy
logic controllers in smart buildings for institutional buildings in subtropical Queensland.
Renew. Sustain. Energy Rev. 54, 738–744. https://doi.org/10.1016/j.rser.2015.10.105
HKEPD (Hong Kong Environmental Protection Department), 2019.Glossary: Construction and
demolition (C&D) materials. Available from: shorturl.at/dopK2 [Access March 9 2020]
Huang, R., Hsu, W., 2011. Framework development for state-level appraisal indicators of
sustainable construction. Civ. Eng. Environ. Syst. Vol. 28, 143–164.
Ighravwe, D.E., Oke, S.A., 2019. A multi-criteria decision-making framework for selecting a
suitable maintenance strategy for public buildings using sustainability criteria. J. Build.
Eng. 24, 100753. https://doi.org/10.1016/j.jobe.2019.100753
Islam, M.S., Nepal, M.P., Skitmore, M., Attarzadeh, M., 2017. Current research trends and
application areas of fuzzy and hybrid methods to the risk assessment of construction
projects. Adv. Eng. Informatics 33, 112–131. https://doi.org/10.1016/j.aei.2017.06.001
Kara, Ç., Karacasu, M., 2017. Investigation of waste ceramic tile additive in hot mix asphalt
using fuzzy logic approach. Constr. Build. Mater. J. 141, 598–607.
https://doi.org/10.1016/j.conbuildmat.2017.03.025
Katz, A., Baum, H., 2011. A novel methodology to estimate the evolution of construction waste
in
construction
sites.
Waste
Manag.
31,
353–358.
https://doi.org/10.1016/j.wasman.2010.01.008
Kharrufa, S., 2007. Reduction of building waste in Baghdad Iraq. Build. Environ. 42, 2053–
2061. https://doi.org/10.1016/j.buildenv.2006.03.011
Khazaeni, G., Khanzadi, M., Afshar, A., 2012. Optimum risk allocation model contracts: fuzzy
TOPSIS aproach.pdf. Can. J. Civ. Eng. 39, 789–800. https://doi.org/10.1139/L2012-038.
Kleemann, F., Lederer, J., Aschenbrenner, P., Rechberger, H., Fellner, J., 2016. A method for
determining buildings’ material composition prior to demolition. Build. Res. Inf. 44, 5162.
20

Lam, K.C., Wang, D., Lee, P.T.K., Tsang, Y.T., 2007. Modelling risk allocation decision in
construction
contracts.
Int.
J.
Proj.
Manag.
25,
485–493.
https://doi.org/10.1016/j.ijproman.2006.11.005
Liang, L. R., Lu, S., Wang, X., Lu, Y., Mandal, V., Patacsil, D., & Kumar, D. (2006). FM-test:
a fuzzy-set-theory-based approach to differential gene expression data analysis. BMC
bioinformatics, 7(S4), S7.
Li, H.X., Al-Hussein, M., Lei, Z., Ajweh, Z., 2013. Risk identification and assessment of
modular construction utilizing fuzzy analytic hierarchy process (AHP) and simulation.
Can. J. Civ. Eng. 40, 1184–1195. https://doi.org/10.1139/cjce-2013-0013.
Li, J., Ding, Z., Mi, X., Wang, J., 2013. A model for estimating construction waste generation
index for building project in China. Resour. Conserv. Recycl. 74, 20–26.
Li, P., Chen, B., 2011. FSILP: Fuzzy-stochastic-interval linear programming for supporting
municipal solid waste management. J. Environ. Manage. 92, 1198–1209.
https://doi.org/10.1016/j.jenvman.2010.12.013
Li, Y., Zhang, X., 2013. Web-based construction waste estimation system for building
construction
projects.
Autom.
Constr.
35,
142–156.
https://doi.org/10.1016/j.autcon.2013.05.002
Llatas, C., 2011. A model for quantifying construction waste in projects according to the
European
waste
list.
Waste
Manag.
31,
1261–1276.
https://doi.org/10.1016/j.wasman.2011.01.023
Lu, W., Ye, M., Flanagan, R., Ye, K., 2015. Corporate Social Responsibility Disclosures in
International Construction Business: Trends and Prospects. J. Constr. Eng. Manag. 142,
04015053. https://doi.org/10.1061/(asce)co.1943-7862.0001034
Lu, W., Yuan, H., Li, J., Hao, J.J.L., Mi, X., Ding, Z., 2011. An empirical investigation of
construction and demolition waste generation rates in Shenzhen city, South China. Waste
Manag. 31, 680–687. https://doi.org/10.1016/j.wasman.2010.12.004
Lu, W. and Yuan, H.P. 2011. A Framework for Understanding Waste Management Studies in
Construction. Waste Management, 31, 1252-1260.
Lu, W., Chen, X., Ho, C.W.D., and Wang, H.D. 2015b. Analysis of the construction waste
management performance in Hong Kong: the public and private sectors compared using
big data. Journal of Cleaner Production, 112(1), 521-531.
Lu, W., Chen, X., Peng, Y., and Shen, L.Y. 2015a. Benchmarking construction waste
management performance using big data. Resources, Conservation & Recycling. 105(A),
49-58.
Lu, W., Peng, Y., Chen, X., Skitmore, M., and Zhang, X.L. 2016a. The S-curve for forecasting
waste generation in construction projects. Waste Management, 56, 23-34.
Lu, W., Webster, C., Peng Y., Chen, X., and Zhang, X.L. 2016b. Estimating and calibrating the
amount of building-related construction and demolition waste in urban China. International
Journal of Construction Management, 17(1), 13-24.

21

Mah, C.M., Fujiwara, T., Ho, C.S., 2016. Construction and demolition waste generation rates for
high-rise buildings in Malaysia. Waste Manag. Res. 34, 1224–1230.
https://doi.org/10.1177/0734242X16666944
Mañà i Reixach, F., Gonzàlez i Barroso, J.M., Sagrera i Cuscó, A., 2000. Plan de Gestión de
Residuos en las obras de construcción y demolición: Situación actual y perspectivas de
futuro de los residuos de la construcción. Institut de Tecnologia de la Construcció de
Catalunya - ITeC, 1a edición Catalunya.
Maués, L.M.F., Santana, W.B., Santos, P.C., Neves, R.M., Duarte, A.A.A., 2017. Construction
delays: a case study in the Brazilian Amazon. Ambient. Construido 17, 167–181.
https://doi.org/10.1590/s1678-86212017000300169
Mirahadi, F., Zayed, T., 2016. Simulation-based construction productivity forecast using NeuralNetwork-Driven
Fuzzy
Reasoning.
Autom.
Constr.
65,
102–115.
https://doi.org/10.1016/j.autcon.2015.12.021
Moghaddam, T.B., Soltani, M., Karim, M.R., Shamshirband, S., Petkovic, D., Baaj, H., 2015.
Estimation of the rutting performance of Polyethylene Terephthalate modified asphalt
mixtures by adaptive neuro-fuzzy methodology. Constr. Build. Mater. 96, 550–555.
https://doi.org/10.1016/j.conbuildmat.2015.08.043
Ortiz, O., Pasqualino, J.C., Castells, F., 2010. Environmental performance of construction waste:
Comparing three scenarios from a case study in Catalonia, Spain. Waste Manag. 30, 646–
654. https://doi.org/10.1016/j.wasman.2009.11.013
Parisi Kern, A., Ferreira Dias, M., Piva Kulakowski, M., Paulo Gomes, L., 2015. Waste
generated in high-rise buildings construction: A quantification model based on statistical
multiple regression. Waste Manag. 39, 35–44.
Pinto, T. de P., 1999. Metodologia para a gestão diferenciada de resíduos sólidos da construção
urbana. Universidade de São Paulo, São Paulo, 1999.
Roche, T.D., Hegarty, S., 2006. Best Practice Guidelines on the Preparation of Waste
Management Plans for Construction & Demolition Projects.
Sen, P.K. and Singer, J.M., 1993. Large Sample Methods in Statistics. Chapman & Hall, Inc.
Villoria Sáez, P., del Río Merino, M., Porras-Amores, C., 2012. Estimation of construction and
demolition waste volume generation in new residential buildings in Spain. Waste Manag.
Res. 30, 137–146. https://doi.org/10.1177/0734242X11423955
Wimalasena, B.A.D.S., Ruwanpura, J.Y., Hettiaratchi, J.P.A., 2010. Modeling Construction
Waste Generation towards Sustainability, in: Construction Research Congress 2010. pp.
1498–1507.
Wu, Z., Yu, T.W.A, Shen, L., Liu, G., 2014. Quantifying construction and demolition waste: an
analytical review. Waste Manage. 34 (9), 1683–1692.
Wu, Z., Yu, T.W.A., Poon, C.S., 2019. An off-site snapshot methodology for estimating building
construction waste composition-a case study of Hong Kong. Environmental Impact
Assessment Review 77, 128-135.
Yu, B., Wang, J., Li, J., Zhang, J., Lai, Y. and Xu, X., 2019. Prediction of large-scale demolition
waste generation during urban renewal: A hybrid trilogy method. Waste Manage. 89, 1-9.
22

Zadeh, L., 1965. Fuzzy Sets-Information and Control. University of California, Berkeley.
Zarghami, E., Azemati, H., Fatourehchi, D., Karamloo, M., 2018. Customizing well-known
sustainability assessment tools for Iranian residential buildings using Fuzzy Analytic
Hierarchy
Process.
Build.
Environ.
128,
107–128.
https://doi.org/10.1016/j.buildenv.2017.11.032
Zhou, H., Ph, D., Asce, M., Zhang, H., Ph, D., 2011. Risk Assessment Methodology for a Deep
Foundation Pit Construction Project in Shanghai, China. J. of Construction Eng. ing
Manag. 137, 1185–1194. https://doi.org/10.1061/(ASCE)CO.1943-7862.0000391.

23