Modelling of demolition waste generation: Application to Greek residential buildings

Abstract

The construction sector in Europe is among the biggest waste generators, producing 370 million tonnes of construction and demolition waste (CDW) every year, which contain important secondary materials. Quantification of CDW is important from their circular management and environmental impact point of view. Thus, the overall objective of this study was to develop a modelling methodology for estimating demolition waste (DW) generation. The volumes (m³) of individual construction materials contained in 45 residential buildings in Greece were accurately estimated using computer-aided design (CAD) software and the materials were classified according to European List of Waste. These materials will become waste upon demolition, with a total estimated generation rate of 1590 kg m⁻² of top view area and with concrete and bricks representing 74.5% of total. Linear regression models were developed to predict the total and individual amounts of 12 different building materials based on structural building characteristics. To test the accuracy of the models, the materials of two residential buildings were quantified and classified and the results were compared with the model predictions. Depending on the model used, the % differences between models’ predictions and CAD estimates for total DW averaged 11.1% ± 7.4% for the first case study and 2.5% ± 1.5% for the second. The models can be used for accurate quantification of total and individual DW and their management within the framework of circular economy.

Keywords

Demolition waste waste generation rate regression analysis modelling Greek residential buildings European list of waste

Introduction

Quantification of construction and demolition waste

Over the last years, many countries around the world (e.g. European Union (EU), China, Japan, Canada, UK) have promoted the circular economy (CE) as the only sustainable economic system (Korhonen et al., 2018). The three basic pillars of CE are the economy (e.g. less use of raw materials, energy saving), society (e.g. new employments) and environment (e.g. less waste, reduced carbon dioxide emissions, preservation of the natural resources) (Saidani et. al., 2017). For this purpose, EU adopted plans to transform into a CE society and identified construction and demolition waste (CDW) as a priority waste stream (European Commission, 2015). Over the last years, EU implemented many actions and set targets to improve the management of CDW. The Waste Framework Directive 2008/98/EC (Vardopoulos et al., 2021; Zorpas and Lasaridi, 2013) introduced the waste hierarchy and was amended by Directive (EU) 2018/851 strengthening the transition from a linear to the CE model, which would close loops by applying reduce–reuse–recycle activities, thus preventing the generation of wastes and converting them into resources (Zhang et al., 2022). However, the transition to the CE model does not prove to be easy, as there are still several obstacles until it can be fully adopted. Oluleye et al. (2022) classified those obstacles into seven categories and considered governments, industry, media and institutions responsible to implement the CE. Also, actions have been proposed on how to improve efficiency of CDW management and reduce disposal (Shooshtarian et al., 2022) or ways to improve the knowledge of those involved in CDW management (Pappas et al., 2022).

Construction, renovation and demolition activities are the main sources for production of the CDW, which are classified in the European List of Waste (LoW) in Chapter 17. The construction industry in EU-28 produced 372 million tonnes of CDW in 2018, representing 35.9% of total waste generated (Eurostat, 2022). CDW contain important secondary raw materials, including concrete, steel, glass, wood and bricks (Akhtar and Sarmah, 2018) that could be used to replace part of the primary raw materials in new activities, for example, in the pavement and road construction or for light reinforced sections (Puthussery et al., 2017; Tuladhar et al., 2020). They also contain materials with high market value, for example, concrete, metals, wood and ceramics (Zheng et al., 2017). Sometimes, hazardous materials (e.g. asbestos and lead-based paints) are also included in CDW that can cause health problems or release harmful substances into the environment (Kim and Yu, 2014). All hazardous materials must be recognized and removed from the waste stream, as they need special handling or treatment before CDW management (Bonifazi et al., 2019; Wu et al., 2022).

The accurate quantification and classification of CDW are of great importance, as they will provide necessary information for reuse, recycling and disposal, leading to effective circular management and utilization in individual project cases or at the regional and country levels (Wu et al., 2014). It can also be used to assess environmental impact under field conditions and emergency situations, such as after earthquakes and war. Wu et al. (2014) classified the methods that had been used for CDW quantification into six main categories: (1) site visit (Hoang et al., 2020; Lafayette et al., 2018; Poon et al., 2004), (2) generation rate calculation (Banias et al., 2011; Cha et al., 2017; Fatta et al., 2003; Lu et al., 2015; Ram and Kalidindi, 2017), (3) lifetime analysis (Huang et al., 2013; Kalcher et al., 2017; Marinova et al., 2020; Wu et al., 2016), (4) classification system accumulation (Coelho and de Brito, 2011; Kim et al., 2017; Llatas, 2011; Solís-Guzmán et al., 2009), (5) variables modelling (VM) (Kern et al., 2015, 2018; Lu et al., 2021; Teixeira et al., 2020) and (6) others (Shi and Xu, 2006), but many researchers have used a combination of two or more of the above. Another approach for the estimation of CDW is building information modelling which provides building designs and data from a project in digital form, yielding geometric and semantic information for waste quantification (Cheng and Ma, 2013; Guerra et al., 2019; Hu et al., 2022; Su et al., 2021).

The VM method can estimate CDW generation using numerous relevant variables (Wu et al., 2014). For this purpose, machine learning (ML) models, such as linear regression (LR), artificial neural networks (ANNs), decision tree and grey models (GM), have been used to quantify generation of various solid waste streams (Dai et al., 2011; Duman et al., 2019; Golbaz et al., 2019; Wei et al., 2013). Several studies also used ML models to estimate individual materials or total CDW at the project, region or country level. Supplemental Table S1 lists some of these studies and includes information about the method, the input data needed, the output and the performance of each model. These studies are summarized as follows:

Three studies (Kern et al., 2015, 2018; Teixeira et al., 2020) used LR for estimating total construction waste (CW) at the building level in Brazil.

One study (Islam et al., 2019) developed LR models for estimating total CW, total demolition waste (DW) and individual components at the country level in Bangladesh.

One study (Coskuner et al., 2021) used ANN for estimating total CDW at the country level in Bahrain.

One study (Lu et al., 2021) used different ML models for estimating total CDW at the region level in China.

One study (Song et al., 2017) created a hybrid GM-support vector regression model and estimated total CDW and seven individual construction materials at the country level in China.

One study (Cha et al., 2022) developed four different ML and hybrid ML models to quantify total DW at the building level in South Korea and compared their performance.

Two studies (Akanbi et al., 2020; Cha et al., 2020) developed ANN and Random Forest models for estimating total DW and individual components at the building level in UK and South Korea, respectively.

Based on the above and Supplemental Table S1, the LR method has not been used specifically for DW at the building level. The latter is important and useful to professional engineers and contractors dealing with management of demolition projects.

In this study, to quantify DW generation at the building level in Greece, we chose LR, which is a widely used technique for model development, based on multiple parameters relevant to the modelling objective (Abdallah et al., 2020). LR has the advantage of simple mathematics, the well-developed statistical theory (Abbasi and el Hanandeh, 2016) and does not require very large amount of data, as other methods do (e.g. ANN) (Xu et al., 2021).

Along these lines, the objectives of this study were as follows: (1) to develop LR models to estimate the individual and total quantities of 12 materials contained in Greek urban residential buildings, (2) to estimate waste generation rates upon building demolition and (3) to provide guidance using case study examples on the use of the developed models, not only for waste generation, but also for estimating recycling cost. The implicit assumption is made that the construction material contained in a building will become waste upon demolition. Actual field measurement of these materials is difficult, because of the large volumes and bulky materials being produced upon demolition. To overcome this obstacle, the volume of construction materials contained in a known geometry residential building was accurately calculated using computer-aided design (CAD) software.

The novelties of this paper are the following: (1) It uses volumes of individual construction materials contained in residential buildings in Greece, estimated by CAD software. Field measurements of these materials are difficult, time-consuming and expensive. (2) Using regression analysis, equations have been developed to predict the quantities of 12 individual materials, some of which (plaster, insulation and marble) have not been reported by previous LR studies (Islam et al., 2019). In the past, Fatta et al. (2003) estimated total CDW generation at the country level in Greece, using data from construction and demolition licenses and, later, Banias et al. (2011) developed a web-based application to estimate DW for different types of Greek buildings.

The LR models developed in this study can be used by professionals and researchers for accurate quantification of total and individual DW, which is important information in management and decision-making. These models cannot be applied for DW prediction in countries with different construction codes but the methodology developed here can be used for creating similar models.

Building construction in Greece

Estimation of DW in Greece is an interesting case study, with DW generation affected by environmental factors. Specifically, Greece is a highly seismic country and throughout history earthquakes have caused heavy damage to buildings (Riga et al., 2021). Inland areas experience low temperatures in winter and high temperatures in the summer, which affects building integrity (Papamanolis, 2005). Most urban residential buildings are multi-storey apartment buildings, containing three to five floors with basements and balconies, and the height of each floor is a little less than 3 m (Papamanolis, 2005). The regulations of building construction were established initially in 1959 with the first ‘Greek Seismic Code’, which was updated in 1984, and the most recent one in 1995. Reinforced concrete was the main construction material for the frame while bricks and plaster were used for the walls. Another important regulation in building construction was the ‘Thermal Insulation Code’ in 1979, which brought the use of thermal insulation materials to reduce energy consumption. Based on the Greek construction codes, the time period in which a building was constructed has important implications on the total quantities of materials contained into it.

Many urban areas were mostly developed between 1970 and 1990 (Theodoridou et al., 2011). In the following years, the expected gradual replacement of these buildings will generate an increased amount of DW. In 2020, Greece generated 3.4 million tonnes of non-hazardous and 0.69 million tonnes of hazardous CDW, respectively, representing 14.1% of total waste (Hellenic Statistical Authority, 2022). Regarding CDW management, Greece adopted Directive 2008/98/EC with Greek Law 4042/2012 (Vardopoulos et al., 2021; Zorpas and Lasaridi, 2013) and most recently Directive 2018/851/EU with Law 4819/2021.

Materials and methods

Data collection

The initial raw data were obtained in 2019 from the departments of urban planning of three different areas (Thessaloniki, Kavala and Xanthi) in Greece and contained licenses of buildings constructed during the period 1955–1997. This period was chosen because of the increase in urbanization after the war, which also resulted in the construction of many buildings (Zambon and Salvati, 2019).

Building licenses contained construction drawings and technical reports. A total of 45 urban residential buildings were used, and their main characteristics or parameters are presented in Supplemental Table S2. Using CAD software (AutoCAD, 2016), the three-dimensional (3D) building models were created (Supplemental Figure S1) over the period 2019–2021. Then, the volume (m³) of each component material (concrete, iron, bricks, plaster, wood, glass, ceramic tiles, insulation, roof tiles, mortar, marble and aluminium) contained in the building was calculated. It was decided to include only those building characteristics that could be more easily available in a demolition project. When a building license did not contain sufficient data (in some old building licenses, some details were missing), reasonable assumptions were made after discussion with experts in the field. For example, ceramic tiles thickness was set to 2 cm, plaster thickness on walls to 2 cm and on ceilings 1.5 cm, and for concrete reinforcement, stirrups were set with a diameter of 8 mm and spacing of 400 mm in columns and 300 mm in beams. The component materials were shown in the drawings or the technical report that accompanied the building license. Visual inspection of the buildings’ exteriors was performed before creation of the 3D models.

Regression analysis

The depended variable of the problem is the material volume and the independent variables are the building characteristics or parameters, listed in Supplemental Table S2. To examine whether correlations existed between variables, the Pearson Correlation coefficients were calculated, using the Minitab v.17 software. If the absolute value of these coefficients is close to 1, then the relationship between the two variables is strong (Edwards, 1976: 33–46). The same software was used to conduct the regression analysis. Regression models then were created to predict the amount of each construction material contained in a building (equations (1)–(12), Table 1). To select the best regression model, all the parameters were included initially in the equations. Βy sequentially dropping the non-significant terms, the simplest equations were chosen, for which the p value for each term was statistically significant (p < 0.05).

Table 1.

Best reduced regression models with respective code numbers and entry type according the European LoW.

Code	Equation no.	Material (m³)	Regression model	R², %	p Value^a
17 01 01 MNH	1	Concrete	$\begin{array}{l} = 37.3 (5.11) \times L e v e l s 1 + 1.67 (0.152) \times A v e r a g e a r e a 1 - 0.111 \\ (0.013) \times C o n s t r u c t i o n y e a r \end{array}$	94.97	0.17
17 04 05 MNH	2	Iron	$\begin{array}{l} = 0.375 (0.074) \times L e v e l s 1 + 0.019 (0.002) \times A v e r a g e a r e a 1 - \\ 0.001 (0.0002) \times C o n s t r u c t i o n y e a r \end{array}$	90.70	0.26
17 01 02 MNH	3	Bricks	$= 0.076 (0.009) \times T o t a l a r e a 1 + 1.19 (0.139) \times D o o r s$	97.19	0.07
17 09 04 MNH	4	Plaster	$\begin{array}{l} = 0.0281 (0.00319) \times T o t a l a r e a 1 + 0.724 (0.143) \times T o t a l \\ h e i g h t 1 + 1.88 (0.199) \times B a l c o n i e s \end{array}$	98.96	0.45
17 02 01 MNH	5	Wood	$= 0.00771 (0.00114) \times T o t a l a r e a 2 + 0.0528 (0.0210) \times R o o m s$	88.65	0.15
17 02 02 MNH	6	Glass	$= 0.000702 (0.000034) \times T o t a l a r e a 1$	91.05	0.30
17 01 03 MNH	7	Ceramic tiles	$\begin{array}{l} = - 0.000023 (0.00001) \times A v e r a g e a r e a 2^{2} + 0.00454 \\ (0.000383) \times L e v e l s 2 \times A v e r a g e a r e a 2 \end{array}$	93.90	0.21
17 06 04 MNH	8	Insulation	$\begin{array}{l} = - 6.76 (2) \times L e v e l s 3 + 0.178 (0.0308) \times A v e r a g e a r e a 2 \\ + 1.19 (0.254) \times L e v e l s 3^{2} \end{array}$	90.17	0.59
17 01 03 MNH	9	Roof tiles	$\begin{array}{l} = 0.0352 (0.00378) \times T o t a l a r e a 4 - 0.000080 (0.000023) \times \\ T o t a l a r e a 4^{2} \end{array}$	98.99	0.81
17 01 01 MNH	10	Mortar	$\begin{array}{l} = - 0.00899 (0.00181) \times C o n s t r u c t i o n y e a r + 0.14 (0.024) \times \\ A v e r a g e a r e a 2 + 2.16 (0.26) \times T o t a l h e i g h t 1 \end{array}$	93.15	0.06
17 05 04 MNH	11	Marble	$= 0.211 (0.025) \times R o o m s - 0.00566 (0.0011) \times W i n d o w s^{2}$	69.51	0.06
17 04 02 MNH	12	Aluminium	$= 0.0017 (0.000091) \times T o t a l a r e a 1$	88.48	0.81
–	13	Total1	$= 0.514 (0.00803) \times T o t a l a r e a 3$	98.91	0.59
–	14	Total2	$\begin{array}{l} = - 0.13 (0.0267) \times C o n s t r u c t i o n y e a r + 19.1 (3.96) \times T o t a l \\ h e i g h t 2 + 2.63 (0.304) \times A v e r a g e a r e a 3 + 10.5 (3.33) \times B a l c o n i e s \end{array}$	96.98	0.64
–	15	Total3	=Equations (1) + (2) + (3) + (4) + (5) + (6) + (7) + (8) + (9) + (10) + (11) + (12)	–	–

The standard errors of the corresponding coefficients are in parentheses.

LoW: list of waste; MNH: mirror non-hazardous.

p > 0.05 indicates normal distribution of residuals.

Two regression equations were developed that could calculate the total amount of construction materials in a building, in m³. The first equation (equation (13), Table 1) that includes all the surfaces (total area) contained in a building (apartments, basements, common areas, balconies, terraces and foundations) and a second one (equation (14), Table 1) that is more simplified and includes only the closed interior surfaces (apartments, basements and common areas).

Results

Materials distribution in buildings

The quantification of the building materials showed that concrete and bricks contribute 74.5% by volume, with the contribution of plaster, mortar, insulation, wood, marble, ceramic tiles, iron, stone, aluminium, glass and roof tiles being less than 10% per material by volume (Supplemental Figure S2).

Supplemental Figure S3 shows the boxplots of the measured materials, except the stone. The materials originate mainly from three- to five-storey buildings and from one nine-storey building. Stone material appeared only in two buildings with stone walls. These buildings were rebuilt into pre-existing ones and kept their old walls. Supplemental Figure S3 shows that the building materials do not follow normal distribution and are strongly skewed towards higher values, as indicated by the outliers. This distribution resulted in the high standard deviations of Supplemental Table S3. The visual lack of normality was also verified by the Anderson–Darling test (Ramachandran and Tsokos, 2021).

Pearson correlation coefficients

Supplemental Table S4 presents only the statistically significant Pearson correlation coefficients. Concrete and iron coefficients were statistically significant for all measured variables. Bricks, plaster, wood, glass, ceramic tiles, insulation, mortar, marble and aluminium were not significantly associated only with the construction year, while roof tiles were statistically significantly correlated with the total area and the average area.

Regression analysis

The construction materials were classified as mirror non-hazardous according to the LoW and are presented in Table 1 with their respective LoW six-digit code numbers and the different regression equations that were developed for each material (equations (1)–(12)). For the selection of the most proper equation, the method described by Berthouex and Brown (2002) was followed for developing the best reduced models. On each equation, the confidence interval of the estimated regression coefficients was examined and if zero was included in the interval, then that term was removed from the model. Afterward, the residuals were examined to verify if they were normally distributed (p > 0.05), using the Anderson–Darling test. The p values for the normal distribution of residuals are listed in Table 1.

Number of levels, average area and total area are the main variables that were included in most of the regression equations (equations (1)–(10) and (12)) In contrast, construction year, total height, doors, balconies, rooms and windows were included in a smaller number of regression equations (Table 1). Plaster (equation (4)) and roof tiles (equation (9)) had the highest R², 98.99% and 98.96%, respectively, while marble (equation (11)) and aluminium (equation (12)) had the lowest, 88.48% and 69.51%, respectively. Lastly, three different equations which can estimate the total amount of materials in a building were developed: In equation (13), all the surfaces of the buildings were used and in equation (14) only the closed interior surfaces were used for the analysis. In equation (15), amounts of individual construction materials, estimated by respective regression models, were added (Table 1). The definition and explanation of input parameters of the regression models of Table 1 are presented in Table 2.

Table 2.

Definition of input parameters in the regression equations of Table 1.

Parameter	Description
Levels1	Number of all levels of a building – without counting tile roofs (N)
Levels2	Number of all levels of a building – without counting foundations and tile roofs
Levels3	Number of all levels of a building – without counting foundations
Average area1 (m²)	Average area of the levels – without counting tile roofs (Σarea_i/N)
Average area2 (m²)	Average area of the levels – without counting foundations and tile roofs
Average area3 (m²)	Average area of the levels – without counting foundations, tile roofs, terrace roofs and balconies
Total area1 (m²)	Area of all levels – without counting foundations and tile roofs
Total area2 (m²)	Area of all levels – without counting foundations and terrace roofs
Total area3 (m²)	Area of all levels
Total area4 (m²)	Area of tile roof
Total area5 (m²)	Area of all levels – without counting foundations, tile roofs, terrace roofs and balconies
Total height1 (m)	Height of all levels – without counting foundations and tile roofs
Total height2 (m)	Height of all levels
Windows	Total number of windows of the building
Rooms	Total number of rooms of the building
Doors	Total number of doors of the building
Balconies	Total number of balconies of the building

Figure 1 compares the estimates of the regression equations of Table 1 with the actual data produced by the CAD analysis of buildings. The respective errors sum of squares are presented in Supplemental Table S5.

Generation rates for total DW

The implicit assumption was made that the construction material contained in a building will become waste upon demolition. From the total amount of building materials measured by CAD analysis, two different generation rates for total DW were estimated (Table 3). The Generation rate1 was 1.00 tonnes/m² or 0.51 m³/m². The Generation rate2 was 1.59 tonnes/m² or 0.81 m³/m². The generation rate expressed in m³/m² refers to the volume of DW produced per m². Using the densities of the quantified material components (Supplemental Table S6), the total density calculated for DW was 1.95 tonnes/m³ (Table 3). The Generation rate3 was developed in the project ‘Development of best management systems for high priority waste streams in Cyprus’, LIFE03 TCY/CY/000018 and since then it has been used as the main generation rate for demolition projects and reports in Greece. It is listed in Table 3 for comparison.

Table 3.

Comparison of generation rates and densities for total DW in Greece with other countries.

Country	Tonnes/m²	m³/m²	Tonnes/m³	Reference
Kuwait	1.45	–	–	Kartam et al. (2004)
Norway	1.10	–	–	Bergsdal et al. (2007)
Spain	–	1.27	–	Solís-Guzmán et al. (2009)
Spain	1.35	–	–	Martínez Lage et al. (2010)
Portugal	1.37	–	–	Mália et al. (2013)
Malaysia	1.04	–	–	Mah et al. (2016)
Bangladesh	1.62	–	–	Islam et al. (2019)
Greece (Generation rate1)	1.00	0.51	1.95	This study
Greece (Generation rate2)	1.59	0.81	1.95	This study
Greece (Generation rate3)	1.28	0.80	1.60	LIFE03 TCY/CY/000018

DW: demolition waste.

Comparison of methods for total DW estimation

The predictions of all methods for total waste estimation alongside with prediction of the Generation rate3 method were visually compared in Figure 2, with respect to the actual data produced by the CAD analysis. To select the best estimation method, the models were ranked with increasing values of their error sum of squares (Table 4). Thus, the models were ranked with order of decreasing preference for estimation in m³ (Table 4 top) as follows: Total1 (equation (13)) > Total based on Generation rate1 (Table 3) > Total3 (equation (15)) > Total2 (equation (14)) > Total based on Generation rate3 (Table 3) > Total based on Generation rate2 (Table 3).

Table 4.

Ranking of six methods for total DW estimation based on increasing error sum of squares.

	Total1 (equation (13))	Generation rate1 (Table 3)	Total3 (equation (15))	Total2 (equation (14))	Generation rate3 (Table 3)	Generation rate2 (Table 3)
Error sum of squares for m³ prediction (units m⁶)	130,373	130,484	210,283	285,489	342,033	381,294
	Generation rate1 (Table 3)	Total3 (equation (15))*	Generation rate2 (Table 3)	Total1 (equation (13))*	Generation rate3 (Table 3)	Total2 (equation (14))*
Error sum of squares for tonnes prediction (units tonnes²)	498,408	803,215	1,456,428	2,024,223	2,279,806	2,752,401

DW: demolition waste.

Equations (13)–(15) with respective densities were used in calculations of generation in tonnes.

Using the respective material densities (Supplemental Table S6) and the tonnes/m³ values (Table 3), the estimation methods for total DW in tonnes were ranked in Table 4 (bottom): Total based on Generation rate1 (Table 3) > Total3 (equation (15)) > Total based on Generation rate2 (Table 3) > Total1 (equation (13)) > Total based on Generation rate3 (Table 3) > Total2 (equation (14)).

Case study

To evaluate the models, the construction materials of two residential buildings were quantified by CAD and their results were compared with the predictions of the six methods presented above for total DW and the regression models of Table 1 for individual waste materials. The unit management cost for each material is included in Supplemental Table S7. ‘Building A’ was a seven-floor building, constructed in 1989, and ‘Building B’ was a three-floor building, constructed in 1991. All the input parameters are included in Supplemental Table S8.

Table 5 presents the estimated DW from the two buildings. Building A produced 1679 m³ of total materials as calculated by CAD. The comparison with the predictions of the six methods above showed that the absolute differences were less than 22.7%. Generation rate1 prediction had the highest accuracy (1758 m³) with a 4.7% difference from the CAD value and the Generation rate2 had the lowest accuracy (2059 m³) with a 22.7% difference. The average and standard deviation of the absolute value % differences for total materials was 11.1% ± 7.4%. The regression equations predictions were less than 10% from the CAD values (actual materials) for bricks, plaster, ceramic tiles, insulation and mortar. The exceptions were the predictions for wood (149.6%), glass (31%) and marble (−86.2%) from the CAD values. The total recycling cost to be charged to the producer for non-mixed CDW was 21,504€ and for mixed CDW was 38,071€, that is, an increase of 77% (without including the transport cost to the plants).

Table 5.

Material estimation (m³) and recycling cost estimates for the two-building case study.

	CAD value (m³)	Regression models (m³) (equations (1)–(12))	Total1 (m³) (equation (13))	Total2 (m³) (equation (14))	Total based on Generation rate1 (m³) (Table 3)	Total based on Generation rate2 (m³) (Table 3)	Total based on Generation rate3 (m³) (Table 3)	Total cost (non-mixed CDW) (€)	Total cost (mixed CDW) (€)
Building A
Concrete	861	751 (−12.8)	–	–	–	–	–	13023€	22263€
Bricks	393	411 (4.6)	–	–	–	–	–	4645€	7941€
Plaster	156.8	167.9 (7.1)	–	–	–	–	–	2279€	3895€
Iron	9.39	8.15 (−13.2)	–	–	–	–	–	25€	819€
Wood	10.4	25.99 (149.6)	–	–	–	–	–	98€	168€
Glass	1.60	2.10 (31)	–	–	–	–	–	0€	68€
Ceramic tiles	10.64	10.37 (−2.5)	–	–	–	–	–	180€	307€
Insulation	97.5	88.7 (−9)	–	–	–	–	–	0€	34€
Tile roof	0.0	0.00 (0)	–	–	–	–	–	0€	0€
Mortar	81.2	82.56 (1.7)	–	–	–	–	–	1245€	2128€
Marble	52.76	7.30 (−86.2)	–	–	–	–	–	4€	263€
Aluminium	4.49	5.09 (13.2)	–	–	–	–	–	6€	184€
Total	1679	1560	1762	1521	1758	2059	2023	21504€	38071€
% Difference with respect to actual (CAD) value for Total		−7.1	5	−9.4	4.7	22.7	20.5
Building B
Concrete	162	168 (3.9)	–	–	–	–	–	2921€	4994€
Bricks	60	64 (7.7)	–	–	–	–	–	729€	1246€
Plaster	25.1	21.9 (−12.9)	–	–	–	–	–	297€	507€
Iron	1.77	1.76 (−0.1)	–	–	–	–	–	6€	177€
Wood	3.4	4.34 (26.8)	–	–	–	–	–	16€	28€
Glass	0.28	0.30 (6.7)	–	–	–	–	–	0€	10€
Ceramic tiles	2.06	1.47 (−28.7)	–	–	–	–	–	25€	44€
Insulation	11.1	15.7 (40.9)	–	–	–	–	–	0€	6€
Tile roof	0.0	0.00 (0)	–	–	–	–	–	0€	0€
Mortar	22.7	20.31 (−10.5)	–	–	–	–	–	306€	524€
Marble	4.05	3.40 (−15.9)	–	–	–	–	–	2€	123€
Aluminium	0.78	0.72 (−7.3)	–	–	–	–	–	1€	26€
Total	293	303	296	280	295	303	298	4303€	7685€
% Difference with respect to actual (CAD) value for Total		3.3	0.9	−4.6	0.6	3.4	1.6

Numbers in parenthesis are % differences from the CAD values.

CAD: computer-aided design.

Building B produced 293 m³ of materials as calculated by CAD. The comparison with the predictions of the above six methods showed that the absolute differences were less than 4.6%. Generation rate1 prediction had the highest accuracy (295 m³) with a 0.6% difference and the equation (14) prediction had the lowest accuracy (280 m³) with a −4.6% difference. The average and standard deviation of the absolute value % differences for total materials was 2.5% ± 1.5%. The regression equations predictions were less than 10% from the CAD values for concrete, bricks, iron, glass and aluminium. The exceptions were the predictions for wood (26.8%), ceramic tiles (−28.7%) and insulation (40.9%) from the CAD values. The total recycling cost to be charged to the producer for non-mixed CDW was 4303€ and for mixed CDW was 7685€, that is, an increase of 79% (without including the transport cost to the plants).

Discussion

This study developed LR models for the prediction of 12 materials contained in Greek residential buildings. The main input variables were the number of levels, average area and total area in most of the regression equations. Similar variables were used in many previous studies (Supplemental Table S1). However, the calculation of the Pearson Correlation coefficients revealed that additional variables (e.g. construction year, total height, doors, balconies, rooms and windows) must be considered to achieve a more accurate prediction.

Concrete represents almost half of the total DW generated in Greece (Supplemental Figure S4). Similar percentages for generation of concrete appeared also in Spain (Martínez Lage et al., 2010) and China (Zheng et al., 2017). In all previous studies, concrete was the most abundant DW component, but Ding and Xiao (2014) found that bricks were the most abundant component in China. Inert materials such as concrete, bricks and mortar, contributed over 80% to the total CDW in Greece, similar to Norway (Bergsdal et al., 2007), China (Ding and Xiao, 2014), the United States (Cochran et al., 2007) and Spain (Martínez Lage et al., 2010). Wood was not as abundant as in other countries, where wood is used more extensively as construction material.

Total DW generation rates in Greece ranged between 1.00 and 1.59 tonnes/m² (Table 3), depending on the estimation method used. It is important to realize that the differences in generation rates reflect different definitions of the reference area adopted by each estimation method. For example, the area considered in Generation rate1 is larger than that of Generation rate2 (Table 2). The highest ranked method, Generation rate1 (Table 4, bottom), is comparable with that of Malaysia (1.04 tonnes/m²) (Table 3). All other countries register higher generation rates. If Generation rate2 (1.59 tonnes/m²) is considered for comparison, this is similar only to Bangladesh (1.62 tonnes/m²) (Table 3, Islam et al., 2019). These differences are probably due not only to the different construction techniques and construction materials used in each country, but also and the definition of reference area used in the respective regression models.

The EU objectives to promote the CE were the smart demolition for the removal of hazardous materials and the collection of high-quality materials, the reduction of CDW generation and the increase in reuse, recycle or recovery of CDW to a minimum of 70% by weight until 2020 (European Commission, 2020). The LR models of this study could be significant tools for CDW management. They can be used by professionals and researchers for accurate quantification of total and individual DW. This will result in more effective management plans in a building demolition project, regarding maximizing material reuse, recovery and revenues and minimizing the operating cost and illegal dumping. Finally, they can be used to estimate gate fees that a CDW recycling plant will charge to the producers.

The methodology developed in this study for residential buildings could be extended to other types of buildings (industrial, commercial, public, etc.) for estimation and management of DW in Greece. In a future study, in addition to LR, other types of ML models could be developed and their performance could be compared.

Conclusions

Quantifying DW is necessary for their management, but a difficult process to implement, due to variation and diversity of their component materials. In this study, DW from Greek residential buildings was quantified and classified according to the European LoW. For this purpose, building licenses from three different areas in Greece were used and the 3D building models were created. Using CAD, the total volumes of the individual construction materials of the buildings were calculated, resulting in a total waste generation rate upon demolition of 1590 kg/m² of average top view area.

LR models were developed to estimate individual and total quantities of materials contained in Greek urban residential buildings, which will become waste upon demolition. The main predictors were building characteristics, such as number of levels, total area, average level area and total height, and the construction year. Based on material estimates from CAD and model predictions, six different methods for total DW estimation were developed and compared using the ranking of their respective error sum of squares.

LR models were developed for 12 different DW components and compared against the respective CAD estimates.

Two case studies were conducted to test the accuracy of the models. Depending on the model used, the absolute % differences between model predictions and CAD calculations for total DW were <22.7% with an average of 11.1% ± 7.4% for the first case study and <4.6% with an average of 2.5% ± 1.5% for the second.

The total recycling cost charged to the producer for the first case study was estimated at 21504€ if the construction materials were not mixed and 38071€ if they were mixed, and for the second case study, the total cost was estimated at 4303€ and 7685€, respectively.

Although the results of the study pertain to Greek residential buildings and buildings of similar construction, the methodology could be adopted to other countries with similar construction codes.

Similar research could be conducted for buildings with different usage (offices, public buildings, institutional buildings, etc.) to develop prediction models for DW.

Supplemental Material

sj-docx-1-wmr-10.1177_0734242X231155818 – Supplemental material for Modelling of demolition waste generation: Application to Greek residential buildings

Supplemental material, sj-docx-1-wmr-10.1177_0734242X231155818 for Modelling of demolition waste generation: Application to Greek residential buildings by Vangelis Soultanidis and Evangelos A. Voudrias in Waste Management & Research

Footnotes

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

ORCID iD

Evangelos A. Voudrias

Supplemental material

Supplemental material for this article is available online.

References

Abbasi

el Hanandeh

(2016) Forecasting municipal solid waste generation using artificial intelligence modelling approaches. Waste Management 56: 13–22.

Abdallah

Abu Talib

Feroz

, et al. (2020) Artificial intelligence applications in solid waste management: A systematic research review. Waste Management 109: 231–246.

Akanbi

Oyedele

, et al. (2020) Deep learning model for demolition waste prediction in a circular economy. Journal of Cleaner Production 274: 122843.

Akhtar

Sarmah

(2018) Construction and demolition waste generation and properties of recycled aggregate concrete: A global perspective. Journal of Cleaner Production 186: 262–281.

AutoCAD 2016 software (Autodesk, San Rafael, California, USA).

Banias

Achillas

Vlachokostas

, et al. (2011) A web-based decision support system for the optimal management of construction and demolition waste. Waste Management 31: 2497–2502.

Bergsdal

Bohne

Brattebø

(2007) Projection of construction and demolition waste in Norway. Journal of Industrial Ecology 11: 27–39.

Berthouex

Brown

(2002) Statistics for environmental engineers, 2nd edn. Cham: Lewis Publishers

Bonifazi

Capobianco

Serranti

(2019) Hyperspectral imaging and hierarchical PLS-DA applied to asbestos recognition in construction and demolition waste. Applied Sciences 9: 4587.

10.

Cha

Kim

Moon

, et al. (2017) New approach for forecasting demolition waste generation using chi-squared automatic interaction detection (CHAID) method. Journal of Cleaner Production 168: 375–385.

11.

Cha

Moon

Kim

, et al. (2020) Development of a prediction model for demolition waste generation using a random forest algorithm based on small datasets. International Journal of Environmental Research and Public Health 9: 6997.

12.

Cha

Moon

Kim

(2022) A hybrid machine-learning model for predicting the waste generation rate of building demolition projects. Journal of Cleaner Production 375: 134096.

13.

Cheng

JCP

LYH

(2013) A BIM-based system for demolition and renovation waste estimation and planning. Waste Management 33: 1539–1551.

14.

Cochran

Townsend

Reinhart

, et al. (2007) Estimation of regional building-related C&D debris generation and composition: Case study for Florida, US. Waste Management 27: 921–931.

15.

Coelho

de Brito

(2011) Distribution of materials in construction and demolition waste in Portugal. Waste Management and Research 29: 843–853.

16.

Coskuner

Jassim

Zontul

, et al. (2021) Application of artificial intelligence neural network modeling to predict the generation of domestic, commercial and construction wastes. Waste Management and Research 39: 499–507.

17.

Dai

Huang

(2011) A two-stage support-vector-regression optimization model for municipal solid waste management–A case study of Beijing, China. Journal of Environmental Management 92: 3023–3037.

18.

Ding

Xiao

(2014) Estimation of building-related construction and demolition waste in Shanghai. Waste Management 34: 2327–2334.

19.

Duman

Kongar

Gupta

(2019) Estimation of electronic waste using optimized multivariate grey models. Waste Management 95: 241–249.

20.

Edwards

(1976) The correlation coefficient. In: Edwards

(eds) An introduction to Linear Regression and Correlation. Cham: W. H. Freeman and Company, pp.33–46.

21.

European Commission (2015) Closing the loop–An EU action plan for the circular economy. Available at: https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:52015DC0614 (accessed 30 November 2022).

22.

European Commission (2020) Construction and demolition waste. Available at: https://environment.ec.europa.eu/topics/waste-and-recycling/construction-and-demolition-waste_en (accessed 05 December 2022).

23.

Eurostat (2022) Generation of waste by waste category, hazardousness and NACE Rev. 2 activity. Available at: https://ec.europa.eu/eurostat/databrowser/view/env_wasgen/default/table?lang=en (accessed 30 September 2022).

24.

Fatta

Papadopoulos

Avramikos

, et al. (2003) Generation and management of construction and demolition waste in Greece—an existing challenge. Resources, Conservation and Recycling 40: 81–91.

25.

Golbaz

Nabizadeh

Sajadi

(2019) Comparative study of predicting hospital solid waste generation using multiple linear regression and artificial intelligence. Journal of Environmental Health Science and Engineering 17: 41–51.

26.

Guerra

Bakchan

Leite

, et al. (2019) BIM-based automated construction waste estimation algorithms: The case of concrete and drywall waste streams. Waste Management 87: 825–832.

27.

Hellenic Statistical Authority (2022) Generation of waste by waste category, hazardousness and NACE Rev. 2 activity. Available at: https://www.statistics.gr/en/statistics/-/publication/SOP06/- (accessed 02 December 2022).

28.

Hoang

Ishigaki

Kubota

, et al. (2020) Waste generation, composition, and handling in building-related construction and demolition in Hanoi, Vietnam. Waste Management 117: 32–41.

29.

Huang

Shi

Tanikawa

, et al. (2013) Materials demand and environmental impact of buildings construction and demolition in China based on dynamic material flow analysis. Resources, Conservation and Recycling 72: 91–101.

30.

Zhou

Vanhullebusch

, et al. (2022) Smart building demolition and waste management frame with image-to-BIM. Journal of Building Engineering 49: 104058.

31.

Islam

Nazifa

Yuniarto

, et al. (2019) An empirical study of construction and demolition waste generation and implication of recycling. Waste Management 95: 10–21.

32.

Kalcher

Praxmarer

Teischinger

(2017) Quantification of future availabilities of recovered wood from Austrian residential buildings. Resources, Conservation and Recycling 123: 143–152.

33.

Kartam

Al-Mutairi

Al-Ghusain

, et al. (2004) Environmental management of construction and demolition waste in Kuwait. Waste Management 24: 1049–1059.

34.

Kern

Amor

Angulo

, et al. (2018) Factors influencing temporary wood waste generation in high-rise building construction. Waste Management 78: 446–455.

35.

Kern

Ferreira Dias

Kulakowski

, et al. (2015) Waste generated in high-rise buildings construction: A quantification model based on statistical multiple regression. Waste Management 39: 35–44.

36.

Kim

CWF

(2014) Hazardous materials in buildings. Indoor and Built Environment 23: 44–61.

37.

Kim

Hong

Park

, et al. (2017) An estimation framework for building information modeling (BIM)-based demolition waste by type. Waste Management and Research 35: 1285–1295.

38.

Korhonen

Honkasalo

Seppälä

(2018) Circular economy: The concept and its limitations. Ecological Economic 143: 37–46.

39.

Lafayette

KPV

da Paz

DHF

, et al. (2018) Analysis of generation and characterization of construction and demolition waste on construction sites in the city of Recife, Brazil. Revista Materia 23(3).

40.

Llatas

(2011) A model for quantifying construction waste in projects according to the European waste list. Waste Management 31: 1261–1276.

41.

Chen

Peng

, et al. (2015) Benchmarking construction waste management performance using big data. Resources, Conservation and Recycling 105: 49–58.

42.

Lou

Webster

, et al. (2021) Estimating construction waste generation in the greater bay Area, China using machine learning. Waste Management 134: 78–88.

43.

Mah

Fujiwara

(2016) Construction and demolition waste generation rates for high-rise buildings in Malaysia. Waste Management and Research 34: 1224–1230.

44.

Mália

De Brito

Pinheiro

, et al. (2013) Construction and demolition waste indicators. Waste Management and Research 31: 241–255.

45.

Marinova

Deetman

van der Voet

, et al. (2020) Global construction materials database and stock analysis of residential buildings between 1970–2050. Journal of Cleaner Production 247: 119146.

46.

Martínez Lage

Martínez Abella

Vázquez Herrero

, et al. (2010) Estimation of the annual production and composition of C&D Debris in Galicia (Spain). Waste Management 30: 636–645.

47.

Oluleye

Chan

DWM

Olawumi

(2022) Barriers to circular economy adoption and concomitant implementation strategies in building construction and demolition waste management: A PRISMA and interpretive structural modeling approach. Habitat International 126: 102615.

48.

Papamanolis

(2005) The main constructional characteristics of contemporary urban residential buildings in Greece. Building and Environment 40: 391–398.

49.

Pappas

Papamichael

Zorpas

, et al. (2022) Modelling key performance indicators in a gamified waste management tool. Modelling 3: 27–53.

50.

Poon

ATW

See

, et al. (2004) Minimizing demolition wastes in Hong Kong public housing projects. Construction Management and Economics 22: 799–805.

51.

Puthussery

Kumar

Garg

(2017) Evaluation of recycled concrete aggregates for their suitability in construction activities: An experimental study. Waste Management 60: 270–276.

52.

Ramachandran

Tsokos

(2021) Mathematical statistics with applications in R, 3rd edn. Cham: Academic Press.

53.

Ram

Kalidindi

(2017) Estimation of construction and demolition waste using waste generation rates in Chennai, India. Waste Management and Research 35: 610–617.

54.

Riga

Karatzetzou

Apostolaki

, et al. (2021) Verification of seismic risk models using observed damages from past earthquake events. Bulletin of Earthquake Engineering 19: 713–744.

55.

Saidani

Yannou

Leroy

, et al. (2017) How to assess product performance in the circular economy? Proposed requirements for the design of a circularity measurement framework. Recycling 2: 6.

56.

Shi

(2006) Estimation and forecasting of concrete debris amount in China. Resources, Conservation and Recycling 49: 147–158.

57.

Shooshtarian

Maqsood

Caldera

, et al. (2022) Transformation towards a circular economy in the Australian construction and demolition waste management system. Sustainable Production and Consumption 30: 89–106.

58.

Solís-Guzmán

Marrero

Montes-Delgado

, et al. (2009) A Spanish model for quantification and management of construction waste. Waste Management 29: 2542–2548.

59.

Song

Wang

Liu

, et al. (2017) Development of a hybrid model to predict construction and demolition waste: China as a case study. Waste Management 59: 350–361.

60.

, et al. (2021) A building information modeling-based tool for estimating building demolition waste and evaluating its environmental impacts. Waste Management 134: 159–169.

61.

Teixeira

González

MAS

Heineck

LFM

, et al. (2020) Modelling waste generated during construction of buildings using regression analysis. Waste Management and Research 38: 857–867.

62.

Theodoridou

Papadopoulos

Hegger

(2011) Statistical analysis of the Greek residential building stock. Energy and Buildings 43: 2422–2428.

63.

Tuladhar

Marshall

Sivakugan

(2020) Use of recycled concrete aggregate for pavement construction. In: Pacheco-Torgal

Ding

Colangelo

, et al. (eds) Advances in Construction and Demolition Waste Recycling. Cham: Woodhead Publishing, pp.181–197

64.

Vardopoulos

Konstantopoulos

Zorpas

, et al. (2021) Sustainable metropolitan areas perspectives through assessment of the existing waste management strategies. Environmental Science and Pollution Research 28: 24305–24320.

65.

Wei

Xue

Yin

, et al. (2013) Prediction of municipal solid waste generation in China by multiple linear regression method. International Journal of Computers and Applications 35: 136–140.

66.

Wang

Duan

, et al. (2016) An innovative approach to managing demolition waste via GIS (geographic information system): A case study in Shenzhen city, China. Journal of Cleaner Production 112: 494–503.

67.

Sandels

Mjörnell

, et al. (2022) Predicting the presence of hazardous materials in buildings using machine learning. Building and Environment 213: 108894.

68.

ATW

Shen

, et al. (2014) Quantifying construction and demolition waste: An analytical review. Waste Management 34: 1683–1692.

69.

Chang

, et al. (2021) Applying artificial neural networks (ANNs) to solve solid waste-related issues: A critical review. Waste Management 124: 385–402.

70.

Zambon

Salvati

(2019) Metropolitan growth, urban cycles and housing in a Mediterranean country, 1910s–2010s. Cities 95: 102412.

71.

Zhang

di Maio

, et al. (2022) An overview of the waste hierarchy framework for analyzing the circularity in construction and demolition waste management in Europe. Science of The Total Environment 803: 149892.

72.

Zheng

Zhang

, et al. (2017) Characterizing the generation and flows of construction and demolition waste in China. Construction and Building Materials 136: 405–413.

73.

Zorpas

Lasaridi

(2013) Measuring waste prevention. Waste Management 33: 1047–1056.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.37 MB