A PARAMETRIC LIFE CYCLE ASSESSMENT MODEL FOR FAÇADE OPTIMIZATION

A PARAMETRIC LIFE CYCLE ASSESSMENT MODEL FOR FAÇADE OPTIMIZATION Alexander Hollberg1 and Jürgen Ruth1 1 Bauhaus-Universität Weimar, Germany Chair of Energy-based Design [email protected]

ABSTRACT This paper presents an extensive study on the determining factors for the optimization of insulation thickness based on life cycle assessment (LCA). The life cycle environmental impact is calculated for a series of different insulation materials and heating systems. Furthermore, we introduce a parametric tool that employs evolutionary algorithms for the LCAbased optimization of the building envelope. An example of its application in the early design stage is shown for retrofitting a single-family house with insulation. The results demonstrate how interdependent the heating system and thermal envelope are and clearly indicate the potential of façade retrofitting, but also prove the fact that more insulation does not necessarily mean more climate protection.

INTRODUCTION In the recent debate on energy saving measures, attention has focussed on operational energy. To minimize energy loss through the building envelope, buildings are usually equipped with more and more insulation. The implementation of these measures causes the embodied energy to gain significance in the life cycle energy balance. As such, the embodied energy has to be taken into account from the very beginning of the design process. In addition to the energy demand, the environmental impact should be analysed in order to be able to meet climate protection objectives in the building sector. A holistic evaluation of the life cycle environmental impact is only possible using LCA. Meijer and Majcen (2012) undertook a detailed study on the optimal insulation thickness based on LCA that compared different indicators for the environmental impact of retrofitting an apartment building. The insulation materials used in the study were cellulose fibre and foam glass and a gas-fired boiler was chosen as the heating system for the next 100 years. New and more efficient heating systems have been developed in recent years and their efficiency is expected to improve further in future. These have significant influence on the amount of emissions during the operational phase (cf. Hollberg & Ruth, 2013). In this paper, we compare not only different insulation materials but also focus on the

effect of different heating systems on the optimal insulation thickness. Decisions made in the early phases of the design process have significant consequences as they lay down general conditions for the subsequent planning process (cf. Steinmann, 1997). As such, they also have the biggest effect on energy demand and environmental impact (cf. Hegger et al., 2007). Optimization based on LCA should therefore be carried out as early as possible. The problem is that detailed information needed for LCA is usually not available in the early design stage. Similarly, most tools for building-related LCA also need detailed information. The parametric approach we present at the end of this paper aims to facilitate optimization in the decisive early stages of the planning process.

SIMULATION Life cycle approach In order to analyse the environmental impact of a design solution over its entire life cycle, all the different stages of a building – production, use and end-of-life – have to be considered. Here, the environmental impact and energy demand of the use stage, which is defined by the operation of the building, is called operational, while the other stages are combined in the embodied energy demand or impact. To differentiate demand and impact between different phases, we use the indices for the for embodied and for life operational phase, cycle. To evaluate the embodied energy demand, the mass of the material is calculated and multiplied by the environmental data of each material. A thermal simulation is used to determine the operational energy demand per year. Multiplying this demand by the service life (or the period under consideration) and adding the embodied energy demand leads to the life cycle energy demand. Eq. 1 shows the procedure for the Total Primary Energy (PET). In an analogous manner, this approach can be used to assess the life cycle environmental impact.



(1)

Reference building For this paper, a typical single-family home in Germany from the 1960s is used as a reference building. The amount of single and two family houses in Germany built before the first Thermal Insulation Ordinance in 1977 is around 10 million (cf. DENA et al., 2012, p25). In most cases, these houses are not insulated and they account for 47 % of the amount of the total end energy demand in the building sector (cf. DENA et al., 2012, p28). For the years 2005-2008 the rate of retrofitting was around 0.8 % per annum (cf. Diefenbach et al., 2010, p71), but the aim of the German government is to raise this rate to 2 % per annum. This shows the high demand for energy efficient retrofitting in the near future.

for the results of the simulation using EnergyPlus in this paper. The input parameters were identical: internal heat gains of 50 Wh/m²d (according to DIN V 18599-10, which equals 2.08 W/m²), infiltration of 0.6 ach and a heating set point of 20°C from 6:00 to 23:00 and 16°C at night. The material properties were chosen based on typical local building materials and a basic energy retrofitting scenario was applied. This included an expanded polystyrene (EPS) insulation of 11 cm in the walls, 12 cm of wood fibre insulation boards (WFIB) in the uppermost ceiling and 20 cm of mineral wool in the roof. The slab was insulated with 6 cm of polyurethane (PUR) foam. Together with the original construction, this results in U-values of 0.27 for the wall, 0.19 for the roof, 0.20 for the ceiling, and 0.34 for the slab. For the simulation, we used climate data for Würzburg in Germany. The results from the different programs are shown in Table 1. The heating demand was simulated as ideal heating load. The deviation of the results computed with EnergyPlus Version 8.1 from those simulated with TRNSYS 17 is 2.7%.

Figure 1: Thermal model of reference building Reference heating systems Several scenarios for heating systems were chosen. The first is a conventional gas-fired condensing boiler with an estimated efficiency of 98 %. The impact indicators for heating with gas are given in the ökobau.dat database. The current alternative for boilers are heat pumps (HPs) which use a heat source to transfer heat to a destination. The efficiency depends on the temperature difference of the source and the destination. The efficiency throughout the year is described as an annual performance factor (APF). Here, the following APFs were assumed: 3.5 for HPs installed within the last few years, 4.8 for new products and 7.0 when used, for example, in combination with a thermal store. HPs can be fuelled by either gas or electricity. Two scenarios for electricity were chosen here: the electricity mix in Germany from the year 2008 and renewable energy provided by wind turbines in Germany. We chose wind, because it contributes greatest to the mix of electricity provided by renewable sources in Germany. According to the report of the Frauenhofer-Institut IWES (2012, p9) it provided 34 % of the renewable energy in 2012, which equals 7.7 % of all energy produced in Germany. Wind energy fluctuates considerably, but the availability is higher in winter (cf. IWES, 2012, p19) when heating is needed and solar power is scarce in northern countries, such as Germany. Thermal simulation Thermal analyses have been carried out with different programs for the reference building (cf. Lichtenheld et al., 2013) and serve as verification

Table 1: Comparison of heating energy demand calculated with different simulation programs TRNSYS 17

ENEV V 18599

ARCHI WIZARD

8037 kWh

8288* kWh

7469 kWh

ENERGY PLUS V8.1 7814 kWh

*calculated without consideration of uncontrolled heat input by the heat source

Environmental data There are a number of existing databanks for the specific environmental data for materials and energy carriers. In Germany the ökobau.dat (BMVBS 2013) is used for LCA and building certification and was employed here. Besides the total non-renewable primary energy (PERNT) and total renewable primary energy (PERT), further environmental indicators are given. The ones that were used here are: - Global Warming Potential (GWP), - Eutrophication Potential (EP), - Acidification Potential (AP), - Ozone Layer Depletion Potential (ODP), - Photochemical Oxidation Potential (POCP) The environmental data is given for different stages of the product. These are defined in the Product Category Rules (PCR) (see Table 2). Here, the production stage (module A1-A3), the operational energy use (B6) and the waste processing at the endof-life (C3) were taken into account. Module D gives credit for benefits beyond the system boundaries. This could be the energy gained when the material is processed in a waste incinerating plant, for example. This means the amount of energy and the ecological impact saved by not having to

took values from the ökobau.dat 2013, which comply with DIN EN ISO 15804. Where such data was not provided, values from the 2011 version were used. The end-of-life scenarios were all taken from the 2013 version. The unit on which the data of the ökobau.dat is based varies, so they were transformed to all be based on 1 kg. The combined values are displayed in Table 3. The values for the energy carriers are shown in Table 4. The environmental impact was evaluated separately for each indicator here since, according to DIN EN ISO 14044, “weighting steps are based on valuechoices and are not scientifically based” (DIN EN ISO 14044, 2006, p43). Further important indicators, such as fresh water and terrestrial ecotoxicity or human toxicity are not provided in the ökobau.dat and therefore were not considered here.

produce energy by other means are given as a benefit according to the electricity mix in Germany at the time the data was collected. However, this procedure is questionable, as there is no guarantee that the product will reach the plant and not be disposed of elsewhere or that the composition of the electricity mix will be the same in future. Therefore, we do not consider the benefits from module D in this paper. In the ökobau.dat, the data for a cradle-to-gate (module A1-A3) and an end-of-life scenario (C3) has to be chosen. A specific scenario is given for some materials, but sometimes they have to be taken from other sources (which can lead to issues, if the data collection methods vary). For eight insulation materials, we combined the cradle-to-gate with suitable end-of-life scenarios from the ökobau.dat database. Where possible, we

Table 2: PCR for construction materials (adapted from DIN EN 15804 2012 p.14)

C2

C3

C4

Re-use recovery and recycling potential

C1

Disposal

B7

Waste processing

B6

Transport

B5

Demolition

B4

Operational water use

B3

Operational energy use

B2

Benefits and loads beyond the system boundary D

End-of-Life

Refurbishment

Transport

B1

Replacement

Manufacturing

A5

Repair

A4

Maintenance

A3

Use Stage

Use

A2

Construction

A1

Transport

Construction Stage

Raw material supply

Production Stage

Table 3: Environmental indicators for typical insulation materials per kg

[kg SO2-eqv.]

[kg PO4-eqv.]

[kg C2H4-eqv.]

POCP

[kg R11-eqv.]

EP

[kg CO2-eqv.]

AP

22.9 32 30 100 94 100 200 45

81.077 98.102 93.995 31.570 14.442 24.819 36.700 7.920

0.448 1.992 1.585 2.430 1.650 9.179 22.899 3.704

80.628 96.110 92.410 29.140 12.793 15.640 13.801 4.216

6.066 6.660 7.829 1.800 0.957 1.211 0.250 1.188

7.5 10-8 1.7 10-8 7.4 10-8 3.8 10-9 3.3 10-8 7.4 10-9 1.8 10-9 4.7 10-9

0.00595 0.00709 0.01428 0.00366 0.00673 0.00226 0.00176 0.00176

0.00062 0.00068 0.00149 0.00063 0.00115 0.00027 0.0003 0.00032

0.01526 0.00287 0.00238 0.00042 0.00039 0.00024 0.00029 0.00011

Table 4: Environmental indicators for typical Material

ODP

[MJ]

GWP

[MJ]

PENRT

[MJ]

PERT

Density [kg/m³]

PET

EPS XPS PUR Glas wool Rock wool Foamglass WFIB Cellulose

Table 5: Environmental indicators for energy carriers per kWh [kg C2H4-eqv.]

POCP [kg PO4-eqv.]

EP [kg SO2-eqv.]

AP [kg R11-eqv.]

ODP [kg CO2-eqv.]

GWP

[MJ]

PENRT

[MJ]

Electricity mix Electricity wind Gas

PERT

[MJ]

Energy carrier

PET

10.260 9.148 4.290

1.490 9.010 5.76 10-3

8.770 0.138 4.284

0.623 0.012 0.261

3.07 10-9 4.07 10-11 1.11 10-11

1.03 10-3 2.09 10-5 2.09 10-4

9.92 10-5 2.48 10-6 3.02 10-5

7.62 10-5 4.52 10-6 3.28 10-5

Case study In this case study, the effects of insulating the wall on the life cycle impact were analysed according to the respective determining conditions. The roof, the uppermost ceiling and the slab were insulated according to the basic reference scenario. For the eight different materials, the thermal simulation was carried out while increasing the thickness from 1 to 60 cm at intervals of 1 cm. The U-value of a standard wall converges asymptotically making each added cm of insulation less effective. Therefore, more than 30 cm of insulation is usually not regarded as reasonable. Nevertheless, we chose an increased search space of 60 cm, because we searched for the environmental optimum, irrespective of costs. Multiplying the heating energy demand by the impact factor of the specific heating system provided the operational environmental impact (module B6). The operational impact per year was then multiplied by the reference period, which in this case was 30 years. This time span is regarded as a realistic service life for most insulation materials and is also employed by other LCA-tools (cf. bauteilkatalog.ch, 2013). Furthermore, the embodied impact of the material was added to obtain the life cycle impact. We then determined the optimal insulation thickness by finding the minimum life cycle impact. These vary according to the indicator chosen. The results for the life cycle impact can either be displayed according to the heating system or the material. Figure 2 shows the curve progression of GWPLC for EPS for different heating systems, while Figure 3 displays the progression for a wind-powered HP with an APF of 3.5 using different materials. In each case, the optimal insulation thickness can be found at the lowest point of the curves.

DISCUSSION Theoretical optimum The minimal GWPLC that can be achieved depending on the material, the heating system and the energy carrier was calculated and is illustrated in Figure 4. The results show the great variation of optima depending on the material and heating system performance. The GWPO decreases significantly from gas to the wind-powered HP. The results show clearly that the GWPLC rapidly declines as the GWPO decreases. The corresponding optimal insulation thicknesses are displayed in Figure 5. The thicknesses decrease as the GWPO of the heating system decreases. This results in a lower GWPE, which then lowers the GWPLC again. The optimal thickness of 62 cm of WFIB exceeded the initial search space of 60 cm, but was found in an additional simulation with an extended search space. Notable is the fact that the optimal thickness of WFIB is more than twice as high as that of EPS or other fossil-fuel-based materials. Even though much more material is employed, the GWPLC is lower. In this specific case when using a gas heating system, 7.3 t of GWP could be saved when using WFIB instead of EPS. However, insulation thicknesses of 62 cm clearly cannot be realistically applied in the building practise and just serve as a theoretical environmental optimum here.

Figure 4: Minimal GWPLC depending on the heating system and the material Figure 2: Curve progression of GWPLC for EPS depending on the heating system

Figure 3: Curve progression of GWPLC for wind powered HP 3.5 depending on the material

Figure 5: Thickness for minimal GWPLC depending on the heating system

Practical scenario In the practical scenario of an insulation retrofitting measure, the available space for insulation is limited. We therefore undertook a further study to analyse the effects of insulating the complete thermal envelope with the same material and a thickness of 20 cm. This equals a U-value of around 0.18 for the walls for all insulation materials. Only the insulation with foamglass causes a U-value of 0.33 due to the much higher conductivity. The results for the life cycle environmental impact with respect to the heating system are displayed separately for each indicator in Figure 9 to Figure 14. The figures show that a different material has the lowest impact depending on the indicator under analysis. No material performs very well or badly in all categories. For PENRTLC stone wool performed best, while for GWPLC WFIB achieved the best results. One general conclusion that can be drawn is that wind powered HPs perform best in all categories. For all indicators, other than AP and EP, the HPs powered by the electricity mix also perform better than the gas-fired boiler. This can be explained by looking at the impact of the energy carriers. The very small impact of renewable energy results in a low operational impact. Therefore, less insulation material is needed. Depending on the material and the APF of the HP, the optimal thickness lies between 1 and 6 cm (see Figure 5). These low values demonstrate that in this case it is better to lose more heat during the operation period over 30 years than to “invest” embodied energy in the insulation. This could potentially change in future if insulation materials are produced using renewable energy. Furthermore, it can be stated that the difference in life cycle impact is influenced much more greatly by the choice of heating system than by the choice of insulation material. An exception is POCP, where EPS has by far the highest impact (see Figure 14). However, it has to be kept in mind that the embodied impact of the heating system was not included here. Savings compared to the original house Compared to the original house, the operational energy demand can be lowered from 31189 kWh to around 6630 kWh (varying slightly for each material). For a house with a gas-fired boiler, this results in savings of about 3000 GJ of PENRTLC and 180 t of GWPLC for all eight insulation materials. The savings for PENRTLC and GWPLC for EPS and WFIB are displayed in Table 5. The general insulation of 20 cm exceeds the optimal insulation thickness for wind powered HPs. In the results this is indicated by negative values for the savings. In the case of EPS insulation and a wind-powered HP with an APF of 3.5, the PENRTLC is 114 GJ and the GWPLC 8.2 t higher than in the original house. In this particular case, the objective of protecting the

environment by retrofitting insulation has clearly not succeeded and no change would have been better. Here, only the embodied impact of the insulation material has been accounted for. Any additional impact caused by implementing the retrofitting measure, e.g. new plastering, would reduce the savings or further exacerbate the losses. Table 6: Savings resulting from 20 cm insulation compared to the original house

Gas boiler HP Mix 3.5 HP Mix 4.8 HP Mix 7.0 HP Wind 3.5 HP Wind 4.8 HP Wind 7.0

PENRTLC [MJ] EPS WFIB 3013017 3021258 1702835 1710121 1202842 1209764 779771 786385 -114243 -108281 -122111 -116154 -128768 -122817

GWPLC [kg] EPS WFIB 181251 189683 120365 128752 84846 93208 54792 63132 -8296 -2 -8969 -676 -9538 -1245

OPTIMIZATION WITH A PARAMETRIC LCA-TOOL Parametric LCA model in Grasshopper In practical application, it is not possible to undertake an extensive analysis of the kind shown in the case study. Instead, we need an automated computer supported approach to quickly find the best retrofitting solution with respect to environmental impact. For this we implemented the method for LCA as described in the section SIMULATION using a parametric design software package called Grasshopper (cf. Hollberg & Ruth, 2013). This is a graphical algorithm editing software used widely by architects thanks to its intuitive user interface. The parametric model allows the planner to change the design easily and quickly provide a range of variations. The automated generation of variations can also be used to optimize the design. In order to simulate the operational energy demand, the Grasshopper-model was linked with Energy-Plus via a plug-in called ArchSim. This makes it possible to change the geometry parametrically, as well as simulation inputs such as the insulation thickness, the material and the construction components. The environmental data for the chosen heating system and the insulation materials is imported from the ökobau.dat database. The mass is automatically computed within grasshopper and finally the life cycle balance according to the chosen service life is drawn. The indicator for the evaluation can be chosen in Grasshopper as well. Optimization The optimization of the model is achieved using evolutionary algorithms. These algorithms require comparatively little background knowledge about the

problem and are especially suited for complex problems where an analytical solution is not possible or is too intricate (cf. Rechenberg 1994). Since parametric LCA has a wide field of application, evolutionary algorithms represent an adequate optimization tool. Here, we used Galapagos, an evolutionary solver integrated in Grasshopper. For the optimization the heating system was given, as well as the period. In general, two different scenarios are possible. The tool can either find the optimal thickness for the wall, roof, ceiling and slab for a given material for each of them, or the thickness can be defined and the solver searches for the best material. To demonstrate the procedure we applied the parametric life cycle approach to minimize the total environmental impact of retrofitting a single-family house. We searched for the optimal insulation thickness for each of the components forming the thermal envelope (wall, roof, uppermost ceiling and slab). This was undertaken for eight different materials regarding each indicator taking into account the specific heating system. The period for the LCA was set to 30 years. In a first scenario the insulation thickness of walls, roof, ceiling and slab were set as free parameters and the minimum GWPLC was declared as the fitness function. The resulting optimal thicknesses for EPS and WFIB are displayed in Figure 6 and 7. We also tested the scenario of best insulation material for a given thickness and the optimum was quickly found.

Figure 6: Galapagos results for minimal GWPLC using EPS depending on the heating system

Figure 7: Galapagos results for minimal GWPLC using WFIB depending on the heating system

The third scenario we investigated was optimizing the insulation thickness and material at the same time. When we defined the same material for all construction components, Galapagos was also able to find the best combination of thickness and material. However, when we added a different material for each component, we did not always receive the best solution. This may be due to the settings of the solver and will be the subject of further research.

Figure 8: Schema of the Grasshopper definition used to optimize insulation thicknesses

CONCLUSION This paper shows that the optimal insulation thickness cannot be determined by U-value alone. Instead, the complex interactions of the determining conditions have to be considered. One factor of great influence is the heating system. With increasing efficiency, the operational energy demand and the environmental impact decrease. This also causes the optimal thickness to decrease, which saves embodied energy and impact. Thus, there is a two-fold effect on the life cycle impact as a result of both parameters. The energy carrier used to fuel the heating system is another determining factor. This becomes particularly clear when renewable energy is used, in this case wind electricity, provided that it is available all year. The environmental impact of this form of energy is very low and has a very small operational impact. In extreme cases, this can lead to the situation where, regarding only the aspect of life cycle energy and environmental impact, no insulation would be better than the 20 cm of insulation commonly applied. This should be kept in mind in order to avoid the over-dimensioning of insulation for future heating systems, especially as the efficiency of HPs is expected to rise quickly and thermal stores become more widely available for private homeowners. Besides the aim of reducing the operational energy demand, insulation measures are usually undertaken in order to improve the thermal comfort. This was not further considered here, as only the minimisation of energy and environmental impact was analysed.

In this paper we have shown that the parametric LCA tool is an effective way of lowering the impact of retrofitting measures and helps to optimize the design towards the most environmentally beneficial solution. The parametric approach permits quick analysis of a large number of variations and is therefore applicable in the important early design stages. We showed the application of the tool for an insulation retrofitting project where the geometry is fixed. A future objective will be to also apply this method to the design of new buildings. Furthermore, the embodied impact of the heating system components will be integrated. Future research will also investigate the computer-aided generation of designs based on environmental life cycle performance.

NOMENCLATURE index for embodied; index for operational; index for life cycle; Total primary energy Total renewable primary energy Total non-renewable primary energy Global Warming Potential for a time horizon of 100 years; Eutrophication Potential; Acidification Potential; Ozon Layer Depletion Potential; Photochemical Ozone Creation Potential;

ACKNOWLEDGEMENTS This study was carried out as part of the research project FOGEB, funded by the Thuringian Ministry for Economics, Labour and Technology and the European Social Funds (ESF). We would also like to thank Timur Dogan for providing the ArchSim plugin.

REFERENCES bauteilkatalog.ch, 2013. Bauteilkatalog. Available at: http://www.bauteilkatalog.ch/ch/de/Bauteilkatalo g.asp [Accessed March 5, 2013]. BMVBS, 2013. ökobau.dat. Available at: http://www.nachhaltigesbauen.de/baustoff-undgebaeudedaten/oekobaudat.html [Accessed January 2, 2014]. DENA et al., 2012. Der dena-Gebäudereport 2012. Statistiken und Analysen zur Energieeffizienz im Gebäudebestand., Diefenbach, N., Cischinsky, H. & Rodenfels, M., 2010. Datenbasis Gebäudebestand. DIN

EN 15804, 2012. Nachhaltigkeit von Bauwerken – Umweltproduktdeklarationen –. , (April).

DIN EN ISO 14044, 2006. DIN EN ISO 14044 Umweltmanagement – Ökobilanz – Anforderungen und Anleitungen. Hegger, M. et al., 2007. Energie Atlas: Nachhaltige Architektur, Birkhäuser. Hollberg, A. & Ruth, J., 2013. Facade optimization based on life cycle demands. In 8th ENERGY FORUM on Advanced Building Skins. IWES, 2012. Windenergie Report Deutschland 2012, Fraunhofer-Institut für Windenergie und Energiesystemtechnik. Lichtenheld, T., Schneider, S. & Klüber, N., 2013. Einsatzgrenzen von Bilanzierungsund Simulationstools für die Energieanalyse in der frühen Planungsphase von Gebäuden. In Bauphysiktage Kaiserlautern. Masea, 2014. Masea Datenbank. Frauenhofer-Institut für Bauphysik. Available at: www.maseaensan.de [Accessed January 10, 2014]. Meijer, A. & Majcen, D., 2012. Is there an optimal insulation thickness? An LCA-based study on the environmental performance of insulation materials and natural gas consumption. In A. Ventura & C. de la Roche, eds. International Symposium on Life Cycle Assessment and Construction. pp. 188–196. Rechenberg, I., 1994. Evolutionsstrategie Stuttgart: frommann-holzboog.

’94,

Steinmann, F., 1997. Modellbildung und computergestütztes Modellieren in frühen Phasen des architektonischen Entwurfs. Bauhaus-Universität Weimar.

Figure 9: Total non-renewable primary energy

Figure 12: Acidification Potential

Figure 10: Global Warming Potential

Figure 13: Eutrophication Potential

Figure 11: Ozone Layer Depletion Potential

Figure 14: Photochemical Oxidation Potential