Material Flows and Stocks in the Urban Building Sector: A Case Study from Vienna for the Years 1990–2015

: Population growth in cities leads to high raw material consumption and greenhouse gas emissions. In temperate climates were heating of buildings is among the major contributors to greenhouse gases, thermal insulation of buildings became a standard in recent years. Both population growth and greenhouse gas mitigation may thus have some influence on the quantity and composition of building material stock in cities. By using the case study of Vienna, this influence is evaluated by calculating the stock of major building materials (concrete, bricks, mortar, and plaster, steel, wood, glass, mineral wool, and polystyrene) between the years 1990 and 2015. The results show a growth of the material stock from 274 kt in the year 1990 to 345 kt in the year 2015, resulting in a total increase of 26%. During the same period, the population grew by 22%. On a material level, the increase of thermal insulation materials like polystyrene and mineral wool by factors of 6.5 and 2.5 respectively were much higher than for other materials, indicating energy efficiency and greenhouse gas mitigation in the building construction sector. The displacement of brickwork by concrete as the most important construction material, however, is rather a response to population growth as concrete buildings can be raised faster. A question for the future is to which extent this change from brickwork to high carbon ‐ intensive concrete countervails the achievements in greenhouse gas reduction by thermal insulation.


1.1.Background
Studies on the anthropogenic metabolism show that in highly urbanized societies, cities are the largest consumers of raw materials and the largest producers of waste and greenhouse gases (GHGs) [1][2][3]. In this metabolism, buildings play a particular role, as the raw materials used for their construction, the waste from their demolition and renovation, as well as the GHGs from their heating and cooling are responsible for the bulk of these resource consumptions and emissions from urban areas [4][5][6]. In countries with temperate and continental climates like Austria with long and cold winters, one major focus of sustainable urban development in the past and present is the reduction of GHGs from heating of buildings. The main reason therefore is that in 1990, which is the reference year for GHG reductions according to the Kyoto Protocol, heating of buildings was, at 17%, the second largest producer of GHG emissions in Austria, in the same range as transport and only topped by energy production and industries [7]. In the Austrian capital Vienna, buildings were the largest producer of GHG emissions in 1990 (29%). As a consequence, GHG mitigation policy in Vienna's building sector focused on a shift in energy carriers (oil to gas), district heating, and a reduction of heat demand [8]. The latter was achieved by higher thermal insulation standards for new buildings and renovation of old buildings with insulation materials and new windows. Even though a higher renovation rate of 3% of the not insulated building stock compared to the observed 1% is claimed to achieve GHG reduction targets [7], the success of these measures is undoubtable. While between 1990 and 2016, the overall GHG emissions in Austria remained at a stable level, the GHG emission reduction in the building sector was -37%. In Vienna, this decrease was even larger with -39%, making the building sector to be only responsible for 18% of GHG emissions in the city, compared to 29% in the year 1990 [9]. These figures are even more impressive when considering that the population of Vienna grew between 1991 and 2017 by 22%, from 1.56 to 1.87 million inhabitants [10]. Both the population growth as well as the GHG emission reduction policy had an impact on the consumption of raw materials and the composition of the material stock in buildings (e.g., a higher share of insulation materials), as well as on the production of waste (e.g. from demolishing old buildings and substituting them by new low-energy buildings). The question, however, is the size of this impact with respect to building material turnover (material inputs and outputs) and the subsequent building material stock of the main building and insulation materials, i.e. concrete, bricks, mortar, and plaster, steel, wood, glass, polystyrene, and mineral wool.
The present study aims to answer this question by determining the development of material flows and stocks of main building materials in the City of Vienna between 1990 and 2015.

State of Research
A number of studies investigated the material stock in buildings at different spatial levels [11], focusing on individual buildings [12,13], cities [14][15][16][17], regions [18], countries [19][20][21][22][23], and even country groups [24,25]. Most studies determine the building stock at a certain point of time, but not the dynamics over a longer time period [12,13,15,18,19,24]. Dynamics of the material stock in buildings for a certain time span are available at higher spatial scale [20,21,23,25] and to a lesser extent for cities. Of these, few are also retrospective [14,16,17], and an even fewer number links their analysis to GHG mitigation policy [26]. There is, however, no study that investigated the dynamics of the material stock in buildings in one of the EU's major cities (i.e. capitals and cities above 500,000 inhabitants), considering the response of the building construction sector in these cities to GHG mitigation policy and thus providing the base for projecting the status quo. By carrying out the case study at hand, this gap will be partially closed.

Materials and Methods
Material flow analysis (MFA) is used to determine material flows and stocks for buildings in Vienna. Therefore, a stock-flow model is designed, calculating the material stock in buildings for a certain year for which data was available (end of 2013). By using building-specific parameters available from statistics (gross volume GV and useable floor area UFA) and information about the generation of construction and demolition waste (CDW), the material stock of the other years can be calculated and the associated annual material flows of construction materials.

Background
A large number of studies on the materials flows and stocks in building sector have been performed, and many of these have used some form of material flow analysis (MFA) model to estimate the respective material flows [11,20,23,27,28]. For this reason, MFA is also used to develop, illustrate, and calculate the material flows and stocks in buildings in Vienna, using the MFA terms as described by Brunner and Rechberger (2016) [29]. Material in MFA is the umbrella term used for goods, sub-goods and substances. Sub-goods are components of goods (e.g., bricks in masonry work), while substances are chemical elements and compounds (e.g., aluminum in bricks). The spatial boundary of the MFA system refers to the city of Vienna, and the temporal boundary is one year as the MFA is carried out on an annual basis for subsequent years. Material flows ṁ are given in mass per time unit (e.g., kg/yr). For processes (e.g., transport, buildings) and systems (e.g., Vienna), the law of conservation of mass is applied, leading to Equation 1: where ∑ minput is the total mass of input (material) flows minput, ∑ mȯutput is the total mass of output flows mȯutput and mṡtock represents the change in the material stock (net mass flow from or to a stock) located in a process or system.
The stock of a material mstock (e.g., in kg) at time point t is determined either by building a time series of material flows ṁ (Equation 2), or by multiplying a counting unit representing the material i (e.g., volume) by a specific material intensity MI in the counting unit (e.g., kg/volume) at time point t (Equation 3). The MIs used in this study refer to our previous work on buildings in Vienna [15].

System under Investigation
The system investigated refers to buildings in the city of Vienna in the years 1990-2015 [15]. With respect to the building materials considered, a selection is made based on quantitative relevance and their direct and indirect impact on GHG emissions. Based on these restrictions, major building materials (concrete, bricks, mortar, and plaster, iron and steel, and wood) and insulation materials or other materials that have an impact on the energy demand of buildings (glass, mineral wool, and polystyrene) are considered. When talking about data and years, stock data always refers to a point of time, i.e. the status as of the last day of the regarding year (e.g., the stock in the year 1990 is by the 31 December of this year). Contrary to that, flows always refer to a time span (e.g., the material flow in the year 1991 are all materials moved between 1 January to 31 December, 1991).

Overview on the Calculation Procedure
The calculation procedure follows a reverse MFA model. The reason for that is that for the year 2013, data on material stocks in the building sector is available [15]. Reverse in this sense means that starting from the year 2013, the material flows and stocks of the previous years back to the year 1990 are calculated. This is done by disaggregation of the material stock in the year 2013, using statistical data sets on the construction period of buildings, the development of the housing area of the population, and construction and demolition waste flows.

Calculation of the Material Stock in the Year 2013
The material stock of the selected materials i in buildings is calculated using the data from Kleemann et al. (2017a) [15], who determined this stock for different building categories j. These categories are distinguished with respect to their use (residential, commercial, industrial, and other buildings) and construction period (before 1919, 1919-1945, 1946-1980, 1981-2000, 2001-2015). This results in 20 building categories in total. In the approach of Kleemann et al. (2017a) [15], the material stock of buildings M was calculated after Equation 4 by multiplying the GV of each building category j (in m³) in the year 2013 by the specific material intensity MIGV,ij (in kg/m³) of each building category j and material i, and summing up all categories.
The GV data, available as GIS data set, came from different municipal departments of the City of Vienna, while the MI was determined based on the analysis of in total 66 buildings covering all 20 general building types identified by Kleemann et al. (2017a) existing. This corresponds to a sample size of 0.03% (considering that Vienna has in total 200,000 buildings) and about 3 buildings per building type. The methods used for the analysis comprised of solely analyzing plan documents (40 buildings), plan documents and sampling of buildings (14 buildings), as well as LCA inventories of new buildings (12 buildings). If for some building types no buildings (for sampling), plan documents, or LCA inventories were available, the mean value of other buildings in the same age category was used and crosschecked with literature data [15]. Even though the database developed by Kleemann et al. (2017a) as well as the calculation approach was well received in literature [30], it has a few shortcomings for the present work. In particular, the sampled buildings used for the specific material intensity MI for construction periods before 1981 were buildings which were demolished and thus were mostly not subject of prior renovation (i.e. thermal insulation and roof top extension). For this reason, they are hardly representative to the entire building stock in this construction period, and to apply Equation 4, the building categories j (of which there were 20 in the works of Kleemann et al., 2017a) have to be extended by subcategories, and the material intensities have to be adjusted. Furthermore, in order to calculate the development of the material stock and thus the material flow of buildings, a time series of the building stock is required. As such a time series is not available for the gross volume GV, but at least for residential buildings for another building specific indicator, namely the useable floor area UFA, the approach of Kleemann et al. (2017a) is adjusted and further developed, distinguishing between residential and other buildings.

Calculation of the Material Stocks and Flows for Residential Buildings 1990-2015
For residential buildings, a time series of the useable floor area can be derived from statistics. However, some adjustments are required, described in the first part of this subsection. In the second part, the conversion of the material intensity MI from gross volume to useable floor area is presented.

Development of the Useable Floor Area
The useable floor area UFA (Nutzfläche in German) is an indicator defined by Austrian and European Standards [31]. For residential buildings, the UFA is the area in m² of the apartments used for living. In multi-family dwellings, areas commonly used by all dwellers like hallways or basements, are excluded from the UFA. This is often not the case in single-family dwellings. Data on the UFA was taken from Statistic Austria's housing censuses of the years 1991, 2001, and 2011 [32][33][34] as presented in the online portal Statcube [35]. The censuses were carried out as surveys, and the data collected refers to the 31 st of December of the previous year (e.g., 31.12.1990). The data, shown in Table 1, unveils that in the census 2001, the construction period was unclear for a total number of 64,867 units, which corresponds to about 7%. For this reason, only the census of the years 1991 and 2011 were used in this study. In order to develop a time series for the UFA, the data from Table 1 has to be annually disaggregated. Furthermore, the renovation status as well as rooftop extensions of buildings have to be considered, as these influence the material intensity MI. This was done by dividing the regarding building categories which are subject to renovation and rooftop extension into subcategories, leading to in total ten categories and sub-categories for residential buildings (see Table 2).
where the different UFA are calculated for time intervals from t = 0 to t = s. UFA0, which equals UFAt for t = 1990, was taken from statistical data as shown in Table 1. This table also gives the UFAt for the year t = 2010. UFAnew,0-t was taken from statistical data on the annually constructed number of residential buildings [35]. This newly constructed useable floor area, however, is only relevant for the residential building categories 9 (constructed between 1981-2000) and 10 (constructed between 2001-2015) as shown in Table 2.
The useable floor area for rooftop extensions UFArooftop,0-t constructed was calculated by multiplying the estimated number of rooftop extensions krooftop,0-t by the average useable floor area of a rooftop extension UFArooftop,k (see Equation 7).
UFArooftop,0-t = UFArooftop,k × krooftop,0-t The UFAdemolished,0-20 = UFA0 + UFAnew,0-20 + UFArooftop,0-20 -UFA20 Thereafter, it was possible to determine the total UFAdemolished between 1990 and 2010. This amount was annually disaggregated. For the disaggregation, it was assumed that the annual demolition follows the amount of debris from buildings annually generated in Vienna between 1990 and 2010. The regarding data, which came from different municipal and national statistics, is presented in the Appendix (Table A2). The existing data gaps, namely missing data for selected years, were filled by linear interpolation.
For the demolished useable floor area in the year 2014, data from Kleemann et al. (2017b) [38] on the demolished gross volume shown in Table 3 was used and divided by a conversion factor which was calculated using the total gross volume GV and the useable floor area UFA (Equation 9).  In order to also consider renovation with respect to the thermal insulation of buildings, the useable floor area UFAt at time t had to be further disaggregated into the useable floor area of rooftop extensions UFArooftop,t, the useable floor area not renovated UFAnot renovated,t and the useable floor area renovated UFArenovated,t. As UFAt and UFArooftop,t had already been calculated, only UFArenovated,t was missing. The data for determining the later came from the Municipal Department MA50, responsible for housing in Vienna [39]. The data was available as living units and useable floor area renovated per decade (1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000) and per year (2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015). Personal communication with MA 50 further revealed that there are two types of renovation with respect to thermal insulation, namely thermal insulation renovation only, and a total renovation. The first is valid for the construction period 1946-1980 and the second for the construction period before 1919, while buildings in the construction period 1919-1945 are renovated under both types (50% each). According to MA50, the renovated living units and thus useable floor area can be divided among the three construction periods proportionally, thus relative to the number of living units of each construction period in the year 1991 (which was 321,750 living units for the construction period before 1919; 101,411 living units for the construction period 1919-1945; and 235,224 living units for the construction period 1946-1980). The concluding useable floor area renovated is shown in the Appendix (Table A1).

Calculating Material Stocks and Flows Using the Material Intensity Per Useable Floor Area
After calculating the useable floor area, the material intensity given in kg/m³ gross volume (MIGV) was converted into kg/m² useable floor area (MIUFA). This was done by multiplying the MIGV by the CFGV/UFA as determined after Equation 8. Before doing so, however, the data set for the MIGV must be completed for the building period categories, which are renovated or contain a rooftop extension. For the buildings that underwent a renovation with respect to thermal insulation, it was assumed that the material intensity MI of the materials usually built in in thermal insulation renovations, namely styrofoam, mineral wool, and glass (MIstyrofoam, MImineral wool, and MIglass) is similar to that of residential and commercial buildings built between 2001 and 2015. The material intensity of the other building materials (e.g., concrete, bricks) does usually not change during thermal insulation renovations. For rooftop extension, the material stock of 16 randomly selected roof top extensions were analyzed based on the construction documents (i.e. construction plans). With this completed data set, the material stock Mij,t in each year t considered as well as the annual material inputs and outputs are calculated, using Equation 10, which was derived from Equation 4. The corresponding data for the material intensities MI is presented in the Appendix , Table A3.

Material Stocks and Flows Calculation for Non-Residential Buildings
For non-residential buildings (commercial, industrial, and other buildings), no such information for a time series as the development of the useable floor area, was available. For this reason, the gross volume in the end of the year 2013 (GV2013, see Table 3) was used as a base for calculation.
For the calculation of the development of the building stock of buildings built before 1980, Equation 11 was used. The gross volume demolished (GVdemolished) per year was estimated by using the gross volume of non-residential buildings demolished in the year 2014 (see Table 3) and directly relating it to the amount of debris and concrete waste generated per year (see Appendix Table A1-Relative CDW generation using the year 2014 as reference), thereby assuming a constant share between the demolished gross volume of residential and non-residential buildings The gross volume was calculated on an annual basis, starting with the gross volume in the year 2013 taken from Kleemann et al. (2017a) [15] (see Table 3). (11) For the calculation of the development of the building stock of non-residential buildings built after 1981, Equation 11 was used. The only difference to Equation 10 is that for the buildings of this category, not only the gross volume demolished, but also the gross volume annually added to the stock through construction (GVconstructed) must be considered. GVdemolished was calculated by using the gross volume demolished in the year 2014 (see Table 3) and directly relating it to the amount of debris and concrete waste generated per year (see Appendix Table A1-Relative CDW generation using the year 2014 as reference). The annual gross volume constructed after the year 1990 was calculated based on the total amount of the gross volume constructed for the regarding building period (1981-2000 and 2001-2013) and dividing it by the regarding time span (10 years for the first building period, i.e. 1991-2000; 13 years for the second building period, i.e. 2001-2013). The gross volume of each year was then calculated on an annual basis, starting with the gross volume in the year 2013 (see Table 3). (12) Renovation is neither relevant for other buildings, nor for industrial buildings, but for commercial buildings built before 1981. The annual renovation rate based on the building stock in the year 1990 is assumed to be 1% per year [40].
Using then the approach of Kleemann et al. (2017a) [15], the thereafter calculated gross volumes were multiplied after Equation 4 by the material intensities shown in the Appendix (Table A3).

Overview of Assumptions
The methodology section shows that calculating the material stocks and flows for buildings is based on a number of assumptions. Table 4 provides an overview on these assumptions.

Results and Discussion
The detailed results of the calculations, which are presented and discussed in the following subsections, can be found in tabular form in the Appendix (Table A4-Table A8).

Building Material Inputs
The annual material input between 1991 and 2015 was in the range of 3,000 to 5,000 kt/yr. This corresponds to 1.8-3.1 t/capita/yr. This value is in the range of typical European cities [11]. Driven by the construction activity for residential buildings, the input was highest in the 1990s due to public housing programs and slowed down rapidly by the end of the decade due to much lower construction activity, while it increased again after the year 2010. The material inputs were dominated by concrete, followed by brickwork (bricks, mortar, and plaster) and steel, the latter mainly in association to reinforced concrete. Other materials (such as mineral wool or glass) show a more constant annual input, but at a significant lower rate compared to the prior mentioned (see Figure 1 and Figure 2).

Building Material Outputs
On the output side, bricks were the most important material demolished, followed by concrete and mortar and plaster associated to brickwork. The material output from bricks and mortar and plaster increased gradually over time, while that of concrete only increased since the year 2000 (see Figure 3). Correspondingly, steel from reinforced concrete became an important construction and demolition waste (CDW) fraction, too (see Figure 4). Wood, another important CDW fraction, increased as well, particularly in association with brickwork. The reason for that is the high wood content in brick-based buildings [38].

Building Material Stock Dynamics
The result of the material stock calculation is shown in Figure 5 and the subsequent tables. Between the years 1990 and 2015, the material stock increased from 274 to 345 million tons, corresponding to +26% over 25 years or 1% increase per year (see Table 5). This annual increase is higher than the 0.8% for the city center of Wakayama, Japan, between 1987 and 2004, or the 0.7% for Salford Quays, Manchester, UK, between 1990 and 2004 [16]. From a material perspective, concrete is mainly responsible for this growth, followed by bricks, and steel. The reason therefore is that old buildings demolished mainly consisted of brickwork, while newer buildings replacing the old ones are mainly consisting of concrete [15]. Concrete is used as it is superior to brickwork in terms of durability, stability, and construction time, which is particularly relevant when residential and commercial (including public) buildings have to be built for a growing population. This is contrary to the findings from Eschesur-Alzette, Luxembourg, where a renaissance of brick-buildings was recorded in the 1990s [26].  From a pure quantitative perspective, other building materials do not contribute significantly to the stock increase. Having a look at the relative growth of each building material selectively as shown in Figure 6, however, it becomes clear that, even though having a little share in the overall material stock, the amount of insulation materials in buildings like polystyrene (increase by a factor of 6.5) and mineral wool ((increase by a factor of 2.5) increased at much higher rates if compared to the year 1990 (100%) than other construction materials. This indicates clearly the shift in building construction towards low energy buildings in times when Vienna introduced its first climate protection program [39]. While this shift has reportedly led to a reduction of greenhouse gas emissions in the building sector, the question is to which extend this reduction was rebound by using more CO2 intensive concrete instead of less CO2 intensive brickwork. One option to cope with that is the use of less CO2 intensive building materials, like wood, brickwork, or even cements and concretes with lower embodied CO2 emissions [41]. The latter can be achieved by lower clinker share in cement, alternative raw-mix for instance from the fine fraction of CDW debris, or recycling aggregate from CDW [42]. With respect to the utilization, residential buildings is the one building category with the highest growth in absolute terms, particularly in the 1990s when a large number of residential buildings were constructed. In relative terms, the growth of commercial and other buildings was higher, while there is almost no overall growth in industrial buildings between 1990 and 2015, and even a decline in their material stock between 2000 and 2015 (Table 6). Besides the increase in the materials stock for building materials and buildings utilization, it is worth having a look of the material stock dynamics with respect to population growth. As shown in Figure 7, the material stock per capita increased in the 1990s, while it remained stable in the 2000s. After the year 2010, it decreased, as a consequence of the high population growth and the low construction activity between 1997 and 2010. This development is mainly driven by the residential building sector, as it corresponds to the development of the useable floor area, which was 38 m²/capita in the year 1990, peaked at 41 m²/capita in the year 2001, and decreased to 40 m²/capita until the year 2015 [35]. This means that, while the overall material stock increased between 1990 and 2015 by 26%, the per capita material stock only increased by 5%. As the latter corresponds with the useable floor area of residential buildings that dominate the building stock, it can be concluded that in the 1990s, residential buildings were constructed for a stagnant population that used this activity to increase both, its useable floor area and thus its material stock. In the years after that, low construction activity in the residential sector together with high population growth lead to a phase of minor decoupling population with material consumption growth. This trend, even though desirable from a resource consumption perspective, lead to social impacts of questionable desirability, like steep increasing housing prices [43].

Conclusions
The case study of the construction materials flows and stocks in the city of Vienna between 1990-2015 exemplarily showed some of the dynamics of modern European cities, even though these dynamics are not necessarily in line with initial hypothesis. With respect to the reaction of the construction sector and thus the construction material turnover, one might expect an increase in the material input and stock when population is increasing. However, the example of Vienna showed exactly the opposite trend, namely a strong increase when the population was declining in the 1990s, and a lower increase when the population was growing after the turn of the millennium. However, having a more detailed look also at the most recent statistical data on construction [19], it turns out that the construction sector only reacts to population growth after some time lag. The same was observed in the 1990s, even though the initial population growth in the beginning of this decade was much smaller than after the turn of the millennium. Contrary to that, the reaction to global warming by greenhouse gas mitigation manifested in the thermal insulation of buildings showed a more continuous trend. This has to do with a policy initiated in the 1990s by the climate action program [39] and continued by the Smart City Wien Framework Strategy that at the first time in the history of the city set some ambitious trans-sectoral targets for a sustainable urban future, amongst others by the reduction of greenhouse gas emissions, energy demand, and the consumption of raw materials in the building sector [44]. While some achievements were gained for the first of these objectives by the thermal insulation of buildings, the material stock is still growing, driven by the consumption of raw materials for building construction. Even though the per capita material stock in buildings decreased particularly in the years between 2010 and 2015, the question is not only if this is sufficient for the objectives of the Smart City Wien Framework Strategy, but also whether the causes for this decrease, namely a reduction of the useable floor area in residential buildings, is viable as a sole strategy. Of course, in the sense of sufficiency, a reduction of consumption of useable floor area can make sense, but in a city with a policy of pronounced social standards and equality like Vienna (which is also highlighted in its Smart City Wien Framework Strategy), this reduction must be distributed among as many individuals in society as possible. Thus, other options of resource conservation to reduce the growth of the material stock to a sustainable rate, like the avoidance of demolishing old but still livable buildings or enhanced recycling of construction and demolition waste, should be considered.    Year 1946Year -1980Year 1919Year -1945Year 1946Year -1980Year 1981Year -2000Year 1800Year -1918Year 1800Year -1918Year 1946Year -1980Year 1919Year -1945Year 1946Year -1980Year 1946Year -1980Year 1919Year -1945Year 1946Year -1980Year 1981 Table A4. 4 Results Part A-All buildings (material flows given in kt/yr, material stocks in kt).