A review of life-cycle assessment of windows

Buildings and building components, particularly windows, have been the subject of numerous studies seeking to recognize and quantify the environmental impacts caused by processes in the product life cycle.

By Salazar, James,Sowlati, Taraneh
Publication: Forest Products Journal
Date: Wednesday, October 1 2008

Life-cycle assessment (LCA) has provided a systematic scheme for evaluation and comparison of environmental burdens of complex production processes. The design of windows LCAs reflects two distinct considerations, the comparison of manufacturing effects caused by various frame materials as well as the justification of increased manufacturing effects for improved performance during use. In all LCAs that considered frame materials, wood had lower embodied energy than the market alternatives, PVC and aluminum (Citherlet et al. 2000, Asif et al. 2002, Menzies and Muneer 2003, Recio et al. 2005). Salazar and Sowlati (2008) did note that contemporary frame systems, PVC, aluminum clad wood, and fiberglass, are all comparable in cradle-to-gate emissions and that the primary determinant of a life cycle advantage stems from a longer service life and lower replacement frequency. The inclusion of use-phase energy amounts showed the relative insignificance of frame production (Entec 2000) and justified the increased resource use during manufacturing to achieve improved energy performance during occupancy (Citherlet et al. 2000, Asif et al. 2002, Kiani et al. 2004, Syrrakou et al. 2005).

---

Building construction, use and demolition contribute significantly to the resource use and waste generated globally. Building operations in the United States consume 39 percent of all primary energy (US Dept. of Energy 2006), and 71 percent of electricity (US Dept. of Energy 2006). Construction and demolition uses 40 percent of all raw materials globally (Roodman and Lenssen 1995), while generating 40 percent of landfill materials (Bright et al. 2003). To reduce the environmental impacts of buildings, green building initiatives, including certification schemes and eco-labeling, have emerged and grown in acceptance. Recently, efforts have been made to incorporate life-cycle assessment (LCA) results into green building certification and rating systems (Trusty and Horst 2002, US Green Building Council 2007). Life-cycle assessment is a methodology to account for all significant material and energy usage and subsequent environmental impact that result from a product’s life cycle.

A building is a unique and complex system comprising different components. Windows are one of the important elements of a building and account for 10 to 25 percent of a building’s exposed surface (Recio et al. 2005). They perform multiple functions in a building envelope, acting as an interface to transmit light, circulate air, and provide outdoor view. While windows are available in different designs and sizes, their main components include the frame, sash, and insulated glazed unit (IGU). Window frames can be made from wood, PVC, aluminum, fiberglass, wood composites, or a combination of them. The glazed unit, which is the translucent component of a window, can be single, double or triple glazed, based on the number of panes of glass incorporated into the window. A variety of coatings on the glazing surface, such as low-E coating, and inert gases between panes, such as krypton and argon, are available. Windows can be characterized as fixed, those that do not open, or operable, those that open. As windows are an important element in the design of a building, the focus of many previous studies was to understand and evaluate the environmental consequences of different window types.

Life-cycle assessments of windows vary significantly in the decisions they seek to inform and hence the scope and design of such analyses are also heterogeneous (Chevalier et al. 2003). The varying scopes, data sources, functional units, assumptions, and findings of the windows LCA literature will be discussed here to understand previous efforts and gain wisdom from their analyses and results.

Life-cycle assessment of windows

Life-cycle assessment is a quantitative technique for assessing the resource use and resulted environmental impacts of a product from “cradle to grave.” It considers all life stages of a product from resource extraction and commodity manufacturing to secondary manufacturing, use, maintenance and end of life. Based on ISO standards, LCA consists of four elements: goal and scope definition, life-cycle inventory (LCI), lifecycle impact assessment (LCIA), and interpretation (ISO 14040, ISO 14044).

The first step in an LCA, the goal and scope definition, is to identify the purpose of the study, its boundary, the required precision of data, and the type of information needed for decision making. The analyst determines the resource consumption and emissions caused by a product’s life cycle during the life-cycle inventory stage. Then, the potential human health and environmental burdens are determined in the life-cycle inventory assessment. Interpretation is the last stage of an LCA in which the results are analyzed in order to reach conclusions, explain limitations, and provide recommendations (ISO 14,040, ISO 14,044).

Modeling the window life cycle

The life cycle of a window depends on the materials it is made of and the processes it goes through, and should recognize its functionality, which is reliant on its placement in an enclosed building envelope.

The life cycle of any window begins with the extraction of raw materials from the natural environment. Resource extraction causes impacts to local ecosystems, depletes stocks of nonrenewable resources, requires energy, and causes waste. After extraction, the raw materials are then shipped to commodity manufacturing facilities where they are used to produce standardized inputs for use in window-specific secondary manufacturing. Depending on the frame material, dimension lumber is then milled, PVC resin and aluminum are extruded, or glass fiber and polystyrene are set in a process called pultrusion. The frame pieces are then combined with an IGU that was produced by separating two or more sheets of flat glass with an aluminum bar, filling the space between panes with argon, and sealing the chamber with a polysulphide sealant. After the sealed IGU is set into the frame, the window is shipped to the installation site where it is placed in the structure and undergoes maintenance and replacement until the building is demolished. Figure 1 shows the life cycle of a window and its related product system.

[FIGURE 1 OMITTED]

Calculating the life-cycle inventory values for each life stage requires data from different sources. For commodity manufacturing and extraction, published data are used while firsthand data may be required when none is available. Secondary manufacturing processes are those specific to windows and require firsthand data collection and energy allocations in case of multiple output production. ATHENA’s environmental impact estimator software also incorporates LCI data for the life cycles of several window types with their other databases for inclusion in whole building LCA that the software facilitates. The maintenance and replacement effects are determined through estimations by building product experts and surveys, while simulation software packages (such as DOE2 (1), EnergyPlus (2), Energy-10 (3), and Hot2000 (4)) may be used to generate use-phase energy flows.

Life-cycle assessments of windows have generally been used for two purposes:

1. Compare window frame materials and determine their relative contribution to impacts.

2. Justify resource use and emissions during manufacturing for improved use-phase characteristics of new technology.

The design of LCAs conducted on windows reflects the difference in the nature of these two goals. The first one, comparing frame materials, directs the focus to the processes in the life cycle that are directly related to the frame material. These processes include cradle-to-gate production emissions and resource use, differences in secondary manufacturing processes, maintenance and service life differences, as well as options for recycling at the end of life. The second one, calculating energy payback during use, typically streamline the life-cycle inventory to consider carbon emissions or energy use only and compare this to the expected energy savings or generation during use. Several LCAs consider both issues concurrently by recognizing differences over the entire life cycles of window frame materials including their thermal differences.

Comparative LCAs of window frame materials

Entec (2000) published a comparative LCA on wood and PVC window frames which considered the primary production of wood and PVC from raw materials, frame fabrication, installation, thermal effects during use, as well as landfilling at the end of use. The LCI showed that the PVC window consumed more than 3 times as much coal and oil as the wood window through the production of raw materials as well as producing 7 times as much C[O.sub.2] in that phase. The wood window also acted as a carbon sink, and it was assumed that 32.3 kg of C[O.sub.2] were consumed in tree growth and 7.5 kg were released at the end of its life netting a carbon sink of 25 kg.

Asif et al. (2002) also considered window frame material in their LCA of aluminum, wood, PVC, and aluminum-clad wood window frames. The embodied energy was found for the four frame types with an accelerated aging test. Industry survey was used to gain an understanding of use-phase service life and maintenance expectations. Aluminum production was the most energy intensive (225 MJ/kg) with the use of recycled material only requiring 7 percent of this energy (16 MJ/kg). PVC production was also energy intensive (70 M J/ kg) and caused emissions of hydrocarbons, dioxins, vinyl chloride, phthalates, and heavy metals. Wood frames had the lowest embodied energy (5.2 MJ/kg) and the best thermal characteristics, but required greater maintenance and preservative use. Cladding eliminated maintenance requirements by protecting the wood. The embodied energy of aluminum, PVC, aluminum clad timber, and timber were 6,000 M J, 2,980 MJ, 1,460 M J, and 995 M J, respectively.

Asif et al. (2002) recognized the critical nature of window longevity and performed accelerated aging simulations to test some of the weaknesses inherent in each design. These tests included immersion, dry-wet cyclic, salt spray, humidity and temperature, and UV exposure. Powder coated aluminum frames were unaffected by all tests. PVC suffered discoloration when exposed to extreme temperatures, humidity, and UV light. Wood frames showed warping and cracking under extreme humidity and temperature, but remained unaffected when clad in aluminum. A survey was also distributed to “authorities” in which it was found that aluminum-clad wood windows provided the longest service, 46.7 years, with aluminum second, 43.6 years, wood third, 39.6 years, and PVC providing the shortest service, 24.1 years.

Salazar and Sowlati (2008) provided the most recent treatment of window frame materials in LCA and considered three window types commonly available to the North American residential consumer: PVC, fiberglass, and wood covered with an aluminum cladding. The LCA was a case study based on the production of the three windows by a single representative manufacturer of each type. Average transportation distances, commodity systems, maintenance, and service life estimations were used to complete the life-cycle inventory model. These inventories were grouped into impact categories and scaled based on IMPACT 2002 + v2.1 (Jolliet et al. 2003) characterization and damage factors.

The damage modeling results indicated that the life-cycle impacts were dominated by the combustion of nonrenewable energy resources. Burning fuels caused increased emissions of respiratory inorganics, terrestrial acidification/nutrification impacts, and global warming. The PVC window’s life cycle used the most nonrenewable energy and caused the most damage due to that window’s shorter service life, 18 years vs. 25 years for fiberglass and aluminum clad wood. This is despite the fact that PVC requires less energy to produce than the fiberglass. The impacts of the steel reinforcement required to strengthen the PVC window outweigh the benefits of the PVC over the fiberglass. The wood window was negatively affected by the addition of aluminum cladding, which required greater energy to manufacture than the wood component. The sensitivity analysis revealed that replacing the virgin material in aluminum cladding with recycled content improved the life-cycle impacts of the wood window. Using fiberglass or PVC to clad the wood window also improved the environmental performance by reducing energy consumption. The use of cladding materials other than aluminum also prevented the disposal of aluminum into municipal landfills which reduced the aquatic ecotoxicity of the wood window’s life cycle. Other potential improvements to the impacts of the three windows’ life cycles include improving energy efficiency, particularly during secondary manufacturing.

Justification of energy payback

The first LCA of double-glazed windows was published by Weir and Muneer (1998). The study focused on the stages up to and including the manufacturing of a 1200 mm by 1200 mm tilt and turn window. The inventory analysis considered the inert fill gas, timber sash and frame, aluminum, sealed unit, and manufacturing overhead. The following processes were considered for each classification:

* Inert fill gas: argon, krypton, and xenon isolation

* Timber sash and frame: Scandinavian forestry, primary milling, frame milling

* Aluminum: cradle to gate virgin and recycled material systems, cutting

* Sealed unit: pane manufacture, assembly, filling, and sealing

* Manufacturing: heat and lighting

The study also considered energy expenditure and reported the subsequent greenhouse gas emissions. The results showed that the window required 137.1 MJ of energy, 33.2 MJ from the sash and frame, 6.0 MJ from the sealed unit, 0.2 MJ from aluminum production, and 97.7 MJ from lighting and factory energy requirements. The three inert fills that were tested yielded 94.7 kg of C[O.sub.2] for argon, 207.6 kg for krypton, and 1,094.7 kg for xenon. The use-phase energy simulation revealed that the lower inventory values for clear float glass was outweighed by the use of Low-E coating and the best performing sealed unit constructed of a “transparent insulation material” for the glazed surface.

Citherlet et al. (2000) completed an LCA of advanced glazing systems in which they focused on several different options for materials and designs. The variables in window design and material included the number and types of panes to be used, the gas used between panes, spacers between panes, and frame material. The impacts of nonrenewable energy requirements, global warming potential, acidification potential, and photochemical ozone creation were considered in the manufacture and disposal of materials. The windows were analyzed for potential energy savings in their use through the simulated office, classroom, and residential applications in the climates of Glasgow, Lausanne, and Rome. (5) Citherlet et al. (2000) considered energy loss through the window unit throughout its lifetime and concluded that improved thermal insulation outweighed increased production requirements and that windows caused the least environmental impacts when they are made of insulative materials such as wood and multiple panes used with inert gas. One seemingly counterintuitive assumption was made in this study, which was to equate the service life of the window to the longevity of the longest lasting component.

This implies that the window is still usable until the last part has failed. The information in this report was illustrated graphically with no exact figures provided. Kiani et al. (2004) considered the manufacture of fully glazed curtain walls and the use-phase energy effects of tinted and reflective glass. The embodied energy was established for glass manufacturing, with published figures considered ranging from 12 to 31 GJ/ton with sealed unit assembly data adopted from Weir and Muneer (1998). The cradle to grave life-cycle inventory indicated that 21.1 percent of total energy was used to manufacture, assemble, and ship the units with the remainder attributable to energy loss in a 25-year simulated service life in London. The Low-E glass and insulated glazed units reduced the operational energy requirements of the structure by 53 percent, which far outweighed the increased manufacturing energy, a savings of 9,826 GJ against an increase of 1,536 GJ.

Recio et al. (2005) performed an LCA of several window systems common to Spain. Life-cycle inventory values, embodied energy, and C[O.sub.2] emissions were found for the production of the following windows:

* PVC with double glazing, PVC from 100 percent virgin and 30 percent recycled

* Aluminum with double glazing (without break), aluminum from 100 percent virgin and 30 percent recycled

* Aluminum with double glazing (with break), aluminum from 100 percent virgin and 30 percent recycled

* Wood with double glazing

* Wood with single glazing

The embodied energy of the wood window was found to be 74.5 kWh, 253.6 for PVC from virgin material, and 1,981.1 for virgin aluminum. The use of recycled material in the PVC and aluminum frame manufacture reduced the manufacturing energy for those windows to 214 kWh and 1406.5 kWh, respectively, but did not change the relative rankings of the three frame materials, with wood requiring the least energy, followed by PVC and aluminum requiring the most.

Although the wood window was shown to have the lowest embodied energy of the three materials considered, the analysis assumed greater conductivity of the wood frame than the PVC (Wood U-value = 2, PVC U-value = 2.5) that outweighed the benefits in the manufacturing stages. Aluminum was found to have the highest manufacturing energy requirements and the highest conductivity (and subsequent energy use) during the use phase. The total energy requirements over the life cycles for the three window types were 1,780 kWh for virgin PVC, 2,633 kWh for wood, and 3,819 to 4,413 kWh for virgin aluminum, depending on whether a thermal break was used.

Syrrakou et al. (2005) provide a recent example of LCA considerations to justify investment into an emerging but not yet widely commercialized technology, electrochromic devices. Electrochromic devices use a low-voltage switch to tint a tungsten ion film and thus vary light transmission characteristics. While this study was limited by only considering energy use, the authors demonstrated that the increase in manufacturing energy required to produce an electrochromic sealed unit, 49 MJ against 42 M J, was far surpassed by the potential energy savings of the device. In fact, it was estimated that the energy payback period was less than 2 years and that the unit saved 52 percent of total energy requirements, or more than 33 times the energy required for its manufacture (Syrrakou et al. 2005). The study recognized that the current cost of these units, roughly 10 times that of a typical IGU, would have to be reduced to within one order of magnitude difference to be adopted in significant quantities.

Table 1 provides a summary of the previous studies on LCA of windows. The goal of each study, which is divided into comparing frame materials or justifying the energy payback, are shown along with the functional units that were considered, and the processes that were deemed within the product system boundaries. One study, Weir and Muneer (1998), only considered the manufacturing of a finished window and made no comparisons between frame materials.

Results summary

In all LCAs that considered frame materials, wood had lower embodied energy than the market alternatives, PVC and aluminum (Citherlet et al. 2000, Asif et al. 2002, Menzies and Muneer 2003, Recio et al. 2005). Moreover, PVC suffers from the fact that numerous toxic chemicals are required in its manufacturing and may be released at the end of its life (Asif et al. 2002). The uncertainty of these releases and their effects has caused controversy over any claim that PVC is either a superior or inferior material through LCA. Salazar and Sowlati (2008) did note that contemporary frame systems, PVC, aluminum clad wood, and fiberglass, are all comparable in cradle to gate emissions and that the primary determinant of a life-cycle advantage stems from a longer service life and lower replacement frequency.

The inclusion of use-phase energy amounts has been used to show the relative insignificance of frame production (Entec 2000) and to justify increased resource use in manufacturing with improved energy performance during occupancy (Citherlet et al. 2000, Asif et al. 2002, Kiani et al. 2004, Syrrakou et al. 2005).

Discussion

Each of the LCAs presented in this review used published data for the production of commodity materials and made assumptions regarding life cycle characteristics such as the length of service life. The first-hand data that were gathered also assumed specific boundary conditions and multiple output allocations to relate process flows to the functional unit in question. This complicates the verification of presented results since the boundary and assumptions on allocations must be fully known to understand the window product system and verify the results. While it was assumed that the reviewed studies utilized the same system scope and modeling assumptions in cases of comparison, it is impossible to verify or repeat the models that were created since much of the data and its underlying assumptions were not published. It is unclear if any of the reviewed LCAs were examined by external practitioners as it was not mentioned in any of these papers. Based on ISO standards, external reviews are required for acceptance of comparative LCA results that have marketability implications.

None of the studies except Salazar and Sowlati (in press 2008) discussed the sensitivity of results to system variation and modeling decisions. System variation includes regionally specific climate conditions that affect product weathering and thermal loss and gain during use as well as the locations of manufacturing facilities that affect transportation distances and the source of electricity that was used. Modeling decisions that cause uncertainty include the selection amongst multiple representative datasets as well as the modeling of multiple output processes.

The LCAs in this review ignored the inherent uncertainty and provided results based on the most likely scenarios and most representative datasets, limited the scope to exclude highly variable assumptions such as window longevity and use-phase energy loss by providing cradle-to-gate results only, or established specific case scenarios to clearly define the life-cycle conditions. Each of these strategies limits the usefulness of the LCA results; the ignorance of uncertainty limits the ability to justify changes in practice based on findings, narrowing the scope excludes impacts that may outweigh those from the cradle-to-gate manufacture, and case studies may or may not represent average conditions or the conditions of a different relevant case.

The emergence of publicly available data for LCA seeks to alleviate some of these limitations. LCI data sources such as Ecoinvent and the USLCI database require documentation and the adherence to well-defined system boundaries and allocation methodologies. Their availability makes possible the recreation of LCA studies so that they may be adapted to particular cases and changing industry conditions. The availability of the models themselves allows the inclusion of use-phase energy values and weathering that are specific to an installation in a particular location alongside the cradle-to-gate manufacturing LCI. While some heuristics as to the impacts of different window systems may be drawn from the studies discussed in this review, the high variability of the window product system and the sensitivity of results to this variability make necessary the consideration of windows on a case by case basis and future LCAs of windows should facilitate this.

Literature cited

Asif, M., A. Davidson, and T. Muneer. 2002. Life cycle of window materials–a comparative assessment. Napier Univ., School of Engineering, United Kingdom. www.cibse.org/pdfs/Masif.pdf. Accessed August 23, 2007.

Bright, C., C. Flavin, G. Gardner, M. MacDonald, A. Platt McGinn, D. Nierenberg, P. Sampat, J. Sawin, M. O’Meara Sheehan, L. Starke, and H. Youth. 2003. Worldwatch Inst., State of the World 2003. W.W. Norton and Co., New York.

Chevalier, J.L., H. Krogh, and M. Tarantini. 2003. IEA-SHC Task 27: Environmental performance assessment of glazing and windows: Context, overview, main concerns, www.iea-shc-task27.org/. Accessed August 23, 2007.

Citherlet, S., F. Di Guglielmo, and J.B. Gay. 2000. Window and advanced glazing systems life cycle assessment. Energ. Buildings 32(3): 225-34.

Entec. 2000. Life cycle assessment of polyvinyl chloride and alternatives, final report, www.defra.gov.uk/ENVIRONMENT/consult/pvc/. Accessed July 6, 2007.

ISO 14040. 2006. International Organization for Standardization. Environmental management–life cycle assessment–principles and framework.

–. 2006. International Organization for Standardization. Environmental management–life cycle assessment–requirements and guidelines.

Jolliet, O., M. Margni, R. Charles, S. Humbert, J. Payet, G. Rebitzer, and R. Rosenbaum. 2003. IMPACT 2002+: A new life cycle assessment methodology. Inter. J. Life Cycle Assess. 8(6):324-330.

Kiani, M., K. Ip, and A. Miller. 2004. Evaluate the whole life environmental impact of glass in a fully glazed commercial building. Presented at the World Sustainable Building Conf., Tokyo, 2005.

Menzies, G. and T. Muneer. 2003. A life cycle assessment of multi-glazed windows. NorDan UK, Ltd. Retrieved on August 23, 2007 from http://www.nordan.co.uk/features/features_assets/life_cycle.pdf

Recio, J.M.B., R.P. Narvaez, and P.J. Guerrero. 2005. Estimate of energy consumption and C[O.sub.2] emission associated with the production, use and final disposal of PVC, aluminium, and wooden windows. Dept. de Projectes d’Engineyeria, Universitat Politecnica de Catalunya, Environmental Modelling Lab., Barcelona, Spain. www.aboutpvc.org/ recursos/12recurso.pdf. Accessed August 23, 2007.

Roodman, D.M. and N. Lenssen. 1995. A building revolution: How ecology and health concerns are transforming construction. Worldwatch Pap. 124, Worldwatch Inst. www.worldwatch.org/node/866. Accessed August 23, 2007.

Salazar, J. and T. Sowlati. 2008. Life cycle analysis of windows for the North American residential market case study. Scand. J. Forest Res. 23(2):121-132.

Syrrakou, E., S. Papaefthimiou, and P. Yianoulis. 2005. Environmental assessment of electrochromic glazing production. Sol. Energ. Mat. Sol. C. 85(2):205-240.

Trusty, W. and S. Horst. 2002. Integrating LCA tools in green building rating systems. In: Proc. USGBC Greenbuilding Inter. Conf. and Expo, Austin. www.athenasmi.ca/publications/publications.html. Accessed April 11, 2007.

U.S Dept. of Energy (DOE). 2006. Buildings Energy Data Book. http:// buildingsdatabook.eren.doe.gov/default.asp?id-view_book&c=5. Accessed April 23, 2007.

U.S. Green Building Council (USGBC). 2007. USGBC’s Life Cycle Assessment Group Produces First Set of Recommendations. Press release dated January 26, 2007. www.usgbc.org/News/PressReleaseDetails.aspx?ID=2892. Accessed June 26, 2007.

Weir, G. and T. Muneer. 1998. Energy and environmental impact analysis of double-glazed windows. Energy Conservation Manag. 39(3/4): 243-256.

(1) http://doe2.com/.

(2) www.eere.energy.gov/buildings/energyplus/.

(3) www.nrel.govYouildings/energy 10.html.

(4) www.sbc.nrcan.gc.ca/software and tools/hot2000_e.asp.

(5) Historical weather averages for these locations are available at www.weather base.com.

James Salazar

Taraneh Sowlati *

The authors are, respectively, Intern, Athena Sustainable Materials Inst., Ottawa, Ontario, Canada (james.salazar@athenasmi.org); and Assistant Professor, Dept. of Wood Sci., Univ. of British Columbia, Vancouver, British Columbia, Canada (taraneh.sowlati@ubc. ca). This paper was received for publication in February 2008. Article No. 10455.