Modeling the life cycle energy and environmental performance of amorphous silicon BIPV roofing in the US

Renewable Energy 28 (2003) 271–293 www.elsevier.com/locate/renene

Modeling the life cycle energy and environmental performance of amorphous silicon BIPV roofing in the US Gregory A. Keoleian ∗, Geoffrey McD. Lewis Center for Sustainable Systems, University of Michigan, Dana Building, 430 E. University, Ann Arbor MI 48109-1115, USA Received 16 June 2001; accepted 25 January 2002

Abstract Building integrated photovoltaics (BIPV) perform traditional architectural functions of walls and roofs while also generating electricity. The displacement of utility generated electricity and conventional building materials can conserve fossil fuels and have environmental benefits. A life cycle inventory model is presented that characterizes the energy and environmental performance of BIPV systems relative to the conventional grid and displaced building materials. The model is applied to an amorphous silicon PV roofing shingle in different regions across the US. The electricity production efficiency (electricity output/total primary energy input excluding insolation) for a reference BIPV system (2kWp PV shingle system with a 6% conversion efficiency and 20 year life) ranged from 3.6 in Portland OR to 5.9 in Phoenix, AZ indicating a significant return on energy investment. The reference system had the greatest air pollution prevention benefits in cities with conventional electricity generation mixes dominated by coal and natural gas, not necessarily in cities where the insolation and displaced conventional electricity were greatest.  2002 Elsevier Science Ltd. All rights reserved. Keywords: Building integrated photovoltaics (BIPV); Life cycle assessment; Energy performance; Air pollution prevention

∗

Corresponding author. Tel.: +1-734-764-3194; fax: +1-734-647-5841. E-mail address: [email protected] (G.A. Keoleian).

0960-1481/03/$ - see front matter  2002 Elsevier Science Ltd. All rights reserved. PII: S 0 9 6 0 - 1 4 8 1 ( 0 2 ) 0 0 0 2 2 - 8

272

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

1. Introduction Currently, only a small percentage of electricity is generated from renewable, potentially sustainable sources worldwide. Hydropower accounted for 17.9% while biomass, geothermal, solar and wind combined account for 1.6% of the worldwide electricity generation in 1998 [1]. Producing electricity from non-renewable coal, natural gas, fuel oil, and nuclear fuel is not sustainable on a long term basis. Environmental impacts associated with these electricity generating systems include release of greenhouse gases, acidification, dispersion of air pollutants such as mercury, formation of smog and ground-level ozone, and the generation of long-lived radioactive waste. Unfortunately, when decisions are being made about new electricity generation capacity, short-term economic factors predominate and these environmental impacts are often of secondary importance. Despite the current high costs of photovoltaics, PV production worldwide is growing at a dramatic rate. The world market for photovoltaics exceeded 200 MW in 1999 compared to 48 MW in 1990 [2]. An elegant application of photovoltaic (PV) technology is in building-integrated designs (BIPV), in which the PV modules become an integral part of the building envelope [3]. BIPV systems perform the traditional architectural functions of walls and roofs (weather protection, structural, and aesthetic) while performing the additional function of generating electricity. BIPV systems displace conventional building materials and utility-generated electricity and do not require additional land area or supplementary support structures. Several different manufacturers are currently supplying BIPV roofing and facade elements. Current design, planning, and implementation of BIPV systems does not adequately consider many life cycle issues related to materials production, manufacturing, use and end-of-life management. These issues include life cycle energy performance, pollution prevention benefits, and related cost savings. A number of investigators besides the authors have studied the life cycle energy performance of PV devices [4–10]. Fewer studies have also explored the pollution prevention benefits of PV [11–13]. Speigel et al. evaluated the power plant air pollutant emissions offset by photovoltaics [14]. Their work, however, did not consider the emissions associated with PV production. Johnson et al. [13] conducted a comparative life cycle assessment of the grid, grid connected PV and stand-alone PV systems. A comprehensive accounting of the full environmental impacts of BIPV in comparison with conventional building materials and fossil/nuclear electricity generating technologies has not been conducted and is the primary focus of this paper. A computer software tool, the photovoltaic-building integrated life cycle design tool, or PV-BILD was developed to compare BIPV systems with conventional systems [15]. This tool is based on a comprehensive set of interconnected modules characterizing environmental, cost, performance, regulatory, and policy factors influencing BIPV systems. The primary function of PV-BILD is to perform comparative assessments between BIPV systems and the functionally equivalent combination of conventional building materials and electricity generating systems. PV-BILD is applicable to any type of grid-tied building integration configuration including walls,

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

273

facades, roofs, light-filtration and screening elements and any type of PV technology such as crystalline, polycrystalline, or amorphous silicon as well as cadmium telluride (CdTe) and copper indium diselenide (CIS) thin films. Life cycle inventory and cost assessments of the displaced conventional building materials and electricity generation are modeled as well. Inclusion of these displaced systems is an integral part of any comprehensive evaluation of a BIPV installation and is necessary to determine the full benefits and any tradeoffs of the system. The purpose of this paper is to present the life cycle assessment (LCA) model for evaluating the environmental and energy performance of BIPV technology. The LCA model is a major component of the PV-BILD tool. This model is applied to evaluate a specific BIPV technology, amorphous silicon PV roofing in the USA. Life cycle energy metrics and pollution prevention factors are defined in order to compare BIPV systems with displaced conventional building materials and centralized electricity generating systems. The energy metrics include electricity production efficiency and energy payback time which are measured for an amorphous silicon PV roofing product at 15 sites in the USA and compared to results for corresponding regional electricity grids. Pollution prevention factors based on the difference between air pollutant emissions from BIPV and conventional systems are computed for these same sites. The air pollutants inventoried include greenhouse gases, NOx, SO2, particulates, volatile organic compounds (VOCs), and mercury. Analysis of results for various regions throughout the USA takes into account both differences in insolation and in the mix of energy sources used to generate electricity.

2. Materials and methods A model for evaluation of the life cycle environmental and energy performance of the BIPV and conventional systems was developed in accordance with International Organization for Standardization (ISO) LCA protocol [16]. Life cycle assessment is an analytical method for characterizing the full environmental consequences of a product system. The system boundary encompasses material production, manufacturing, use and end-of-life management stages. LCA, as defined by ISO, consists of four components: goal and scope definition, inventory analysis, impact assessment, and interpretation of results. After defining the study objectives, audience and initial boundaries, the material and energy inputs and outputs across each life cycle stage are quantified in the inventory stage. Impact assessment characterizes inputs and outputs with respect to human and ecological health and welfare. The PV-BILD model does not include an impact assessment model. The authors presented model equations and metrics for assessing the life cycle energy performance of a photovoltaic system in an earlier publication [15]. This paper extends this previous work in two ways: other environmental parameters besides primary energy are inventoried and the analysis addresses the dual functionality of building integrated photovoltaics.

274

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

2.1. PV-BILD model architecture A simplified schematic of PV-BILD is shown in Fig. 1. PV-BILD consists of a life cycle inventory model and a life cycle cost model. The methods used for the life cycle cost analysis are shown here for reference only. An in-depth treatment of the environmental economics is presented elsewhere [15]. The life cycle inventory (LCI) model component (the upper half of the figure) contains several modules that quantify material and energy resource inputs as well as waste and pollutant outputs associated with BIPV and conventional electricity generation and building material systems. One module (on the left) characterizes individual BIPV products constructed using various PV technologies (amorphous, polycrystalline, and crystalline silicon, CdTe and CIS thin film) for both roofing and fac¸ ade applications. Another module (on the right) characterizes building materials (fiberglass asphalt shingles, galvanized metal roofing, curtain wall panels, glazing components) displaced by the BIPV system. A third module characterizes conventional electricity generation (the ‘grid’) by North American Electricity Reliability Council (NERC) regions. The final part of the LCI model is a user interface for specifying system parameters. The interface allows selection of a particular BIPV product and inverter, the displaced conventional building material, system geographic location, array size, and system lifetime. PV-BILD currently contains life cycle data (described in detail below) for two United Solar amorphous silicon roofing products (shingle and galvanized steel standing seam), an AES module-level inverter, and the appropriate displaced building materials. The lower half of Fig. 1 illustrates the structure of the life cycle cost component of PV-BILD. The economics module (on the left) uses a significant portion of the user interface to collect data on system capital costs, value of displaced material, electricity price, and interest rate, as well as providing for the input of unit damage

Fig. 1.

PV-BILD model architecture.

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

275

costs for several air pollutant emissions. Output from the life cycle inventory model including kWh of electricity generated and total mass of air pollution prevented are fed into the cost model. The policy module (on the right) collects data on the value of any system subsidies and on compliance costs for carbon emissions, and allows a choice in the CO2 emission regulatory policy (CO2 emissions either regulated or not). Other policy scenarios resulting from Federal Energy Regulatory Commission (FERC) or state Public Utility Commissions (PUC) can be explored using the existing model structure. 2.2. Life cycle inventory analysis Life cycle inventory data were gathered from a variety of sources in accordance with ISO 14040 guidelines. For this demonstration of PV-BILD, the selection of air pollutant emissions and energy were inventoried as shown in Table 1. This study focused on air emissions rather than pollutants released to water and land where data gaps are more prevalent and uncertainties are larger. Air pollutant emissions are also considered to be the most significant environmental concerns for the majority of conventional electricity generating technologies in the USA, with the significant exception of nuclear power. The inventory models were constructed using Ecobalance, Inc.’s Team life cycle assessment software. The associated Deam database was also a source for much of the life cycle data for material production, transportation and conventional electricity generation. TEAM was used to assemble a separate model for each component of the product system. When each component model was complete, the list of overall impacts for that component was exported to the PV-BILD file (in Excel) for use in inventory calculations. Data for United Solar BIPV products and the AES inverter were collected in 1997 and 1998 and are representative of current manufacturing practices. Regional electricity technology mixes, and their impacts, are based on electricity generation data from 1996. The asphalt shingle material data were collected in 1995. Table 1 Data categories reported in this analysis Data category

Subcategories

Air emissions

Criteria air pollutants (CO, Pb, NO2, O3, PM-10, SO2) Greenhouse gases (CO2, CH4, other greenhouse gases) Other toxic air pollutants (Hg)

Energy

Feedstock Process Renewable Non-renewable Total primary energy

276

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

Comparison of a BIPV system and the corresponding conventional system begins with a clear definition of the functions or services that each system is providing. The functions for this investigation are two-fold: building protection for a specified surface area and the provision of AC electricity. The functions can be expressed in units of square meters (m2) of building material and kWh of electricity for a defined service lifetime. The system components required to provide these functions are indicated in Table 2. The photovoltaic building integrated system requires, in addition to the PV itself, balance of system (BOS) components to generate utility-quality AC electricity. The inverter converts the DC output from the PV to AC. The power plant is the principal component of conventional regional electricity generation. Transmission lines and transformers provide the network for distribution. If the BIPV system is feeding into this grid then this component becomes common to both the BIPV and conventional systems. For off grid BIPV applications the distribution component would be eliminated and replaced in most cases with electricity storage devices. A description of the main elements of the inventory model including BIPV products, the inverter, insolation parameters, conventional grid, and conventional roofing are presented here. 2.3. BIPV products Two United Solar BIPV products are currently profiled in PV-BILD: SHR-17 shingles and ASR-128 standing seam metal roofing. These products are manufactured almost identically by United Solar and are triple-junction thin film amorphous silicon photovoltaic cells deposited on stainless steel and encapsulated in a polymer composite. The only differences between the two products are the materials included in the final lamination step. The stabilized conversion efficiency (solar radiation to electricity) of these materials was taken to be 6% based on the manufacturer’s laboratory and field experience. The manufacturer provides a 20 year limited warranty for these BIPV products, consequently model results for environmental performance are based on a 20 year service life. A longer service life for the BIPV system will improve this performance. Data collected from United Solar included a complete bill of materials for both products and energy, resource use, and emissions for the manufacturing facility. The inventory for the standing seam metal roofing product only included the photovoltaic material and not the conventional standing seam roofing product to which it is laminated since this roofing system component is not displaced by the use of United Table 2 Functional unit components System function

BIPV system

Conventional system

Building protection Electricity generation

BIPV BIPV+inverter+other BOS

Shingles or metal roofing Regional electricity grid

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

277

Solar’s ASR-128. A detailed description of United Solar’s manufacturing process for amorphous silicon PV products (identical to those used in the BIPV products studied here) can be found in Lewis and Keoleian [17]. Using the bill of materials and the Ecobalance Deam database, an inventory of material production impacts was assembled. Data were not available for all of the materials used in these BIPV products, but data for 98% of the mass of the PV shingle and 95% of the PV standing seam metal roof were included. No data were available for TefzelTM, a polymer used in encapsulation, and this one material was responsible for most of the remainder of the uninventoried mass. Data were also unavailable for some of the gases used in the deposition of the photovoltaic structure, but these materials are used in such small quantities that the impacts associated with their production are likely to be small on a mass basis. Measurements of electricity use were taken from each process machine to determine manufacturing process energy requirements on a per module basis. Impacts due to this electricity use were assigned based on the NERC East Central Area Reliability Coordination Agreement (ECAR) region, which contained the Michigan location of the United Solar plant. Complete data for plant overhead energy (lighting, heating, etc.) were not available in this study and so could not be included. Energy for vacuum deposition and encapsulation processes are expected to dwarf overhead energy requirements. An alternative means of evaluating manufacturing process energy (not used here) is to divide the total plant energy requirements by the number of modules produced to find energy burdens on a per module basis. Burdens associated with administrative and research and development activities could also be allocated to the PV system. Burdens associated with transportation of materials to the United Solar plants were also inventoried. This was the case for 95% of the mass of the PV shingle and 97% of the mass of the PV standing seam metal roofing. The photovoltaic material is produced in Troy, MI, the final assembly and lamination steps occur in Tijuana, Mexico. Burdens for transporting products and protective packaging by diesel truck between the United Solar plants were included in the inventory, as were impacts for transporting completed PV shingles from the manufacturing plant to the point of use. The distance used for this final transportation link was from Tijuana to the average of the nearest (Los Angeles) and farthest (Boston) cities included in the model.

2.4. Insolation model

Incident solar radiation (insolation) is the ‘fuel’ for photovoltaic systems and is currently modeled using National Renewable Energy Laboratory (NREL) data for yearly average global horizontal insolation in Wh/m2/day [18]. Table 3 illustrates the cities currently available in PV-BILD, their associated NERC region, and average global horizontal insolation. These data assume no shading or ground reflection.

278

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

Table 3 Cities currently in PV-BILD, with NERC region, global horizontal insolation, and state average cost per kWh City

NERC Region

Insolation (Wh/m2/day)

Atlanta Boston Boulder Chicago Detroit Fort Worth Los Angeles Miami Minneapolis New York City Oklahoma City Philadelphia Phoenix Portland (OR) Raleigh

SERC NPCC WSCC MAIN ECAR ERCOT WSCC SERC MAPP NPCC SPP MAAC WSCC WSCC SERC

4582 3910 4576 3868 3779 4891 4946 4833 3892 3991 4762 3987 5733 3517 4395

2.5. Inverter The primary difference between available inverters is size, both power rating and physical dimensions. The inverter included in PV-BILD is a small Advanced Energy Systems (AES) 250 W model (MI250) produced in Wilton, NH. The efficiency of this inverter was assumed to be 95% based on full load operation. Smaller inverters are expected to have lower burdens than larger, higher power inverters, though larger inverters may have lower burdens on a per-Watt basis. PV-BILD multiplies the unit burdens for the inverter by the number of inverters required for a specified PV array power output. The components used in all inverters are similar, differing mainly in size. Although there are a variety of inverters currently being manufactured, no inventory data were publicly available for any of them. The inventory of the AES unit only accounts for material production burdens since manufacturing burdens for assembly of the inverter were not available. Data included a complete bill of materials and specifications for custom parts, though not all parts of the inverter were included in the inventory calculations. Excluded were parts for which data were unavailable or of suspect quality, namely most of the electronic components (integrated circuits, resistors, capacitors, and inductors). This omission is expected to be significant since production of electronic components is generally resource intensive. The inverter inventory did include all structural parts (aluminum extrusions and covers), some electronic components, and the printed wiring board (PWB) fabrication process. Data for the PWB fabrication process and the electronic components are from the Ecobalance, Inc. EIME database. No transportation burdens were assigned for the inverter components since suppliers and their locations were not

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

279

known. Transportation impacts for the completed inverter from the AES facility in Wilton, NH to the point of use by diesel truck were included. The distance used for this transportation link was from Wilton to the average of the nearest (Boston) and farthest (Los Angeles) cities included in the model. 2.6. Grid electricity Conventional grid electricity generation is currently modeled in PV-BILD on a regional basis for nine regions determined by NERC. The boundaries of the current NERC electricity regions are shown in Fig. 2. The impacts for a given region are an aggregate of the entire region and will not be representative of impacts for an individual utility company, city, or generating facility. Also, electricity generated by the PV system is assumed to displace generation in that NERC region but is not expected to significantly alter the mix of generating technologies in the region. The result of this assumption is that regional electricity generation inventory parameters (e.g. air pollutant emission factors) will not change with the addition of a residential size PV array. This assumption will need to be revisited when PV is deployed on a large scale or when electricity generation is inventoried on a scale smaller than NERC regions. Retail competition in the electric utility business will also be cause to revisit some of the assumptions in PV-BILD, since the displaced electricity may not be from the same location as the PV array. Inventory data on conventional electricity generation are compiled from Ecobal-

Fig. 2.

NERC regions.

280

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

ance, Inc.’s Team LCA software and associated Deam database. These inventory data employ EPA AP-42 emission factors and include precombustion and combustion impacts for the total fuel cycle of the energy sources used to generate electricity. Electricity generation inventory data are grouped by NERC region, each one of which has its own average mix of generating technologies (coal, natural gas, nuclear, hydro) and thus, its own profile of resource use and emissions per kW generated. Table 4 lists the mix of generating technologies (as a percentage) in each NERC region. The Florida Reliability Coordinating Council (FRCC) which prior to 1996 was contained within the Southeastern Electric Reliability Council (SERC) was not modeled separately. The inventory data for electricity generation account for losses in transmission and distribution but do not include the environmental impacts associated with constructing power plants. These impacts are assumed to be small in comparison with the environmental impacts from the combustion of fossil fuels over the plant’s lifetime. For example, the energy consumed in the power plant construction has been found to be ⬍1% of the energy embodied in the fuels feeding the plant over its service life [19,20]. 2.7. Conventional building materials The conventional building materials included in PV-BILD are fiberglass composite asphalt shingles and Galvalume metal roofing. These materials are widely used and representative of the choices available for roofing products. Materials that occur in both conventional and BIPV installations were not included since they have no net effect on the comparison. Materials that fall into this category are roofing felt, nails, and galvanized metal standing seam roofing. The galvanized roofing is in this category since the same material is used by United Solar to produce their standing seam BIPV product. Inventory data on conventional building materials are compiled from the National Institute of Standards and Technology’s (NIST) Bees program [21]. Bees is an acronym for Building for Environmental and Economic Sustainability and is a software Table 4 Mix of generating technologies (as a percentage) in each NERC region in 1996 NERC region

Coal

ECAR ERCOT MAAC MAIN MAPP NPCC SERC SPP WSCC

89.2 47.0 52.1 58.8 72.6 21.0 57.6 57.6 35.3

Natural gas 0.3 36.4 3.1 1.1 0.6 12.3 4.8 23.5 8.2

Heavy fuel oil

Nuclear

0.3 0.2 3.2 0.4 0.5 12.5 3.4 0.6 0.2

9.7 16.1 39.7 38.3 16.0 36.6 29.2 16.0 13.0

Hydro 0.5 0.3 1.9 1.4 10.2 17.5 5.0 2.0 43.3

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

281

package that compares pairs of conventional building materials (asphalt vs fiber cement shingles, for example). The Bees database contains an inventory of the energy, resource use, waste and emissions per unit of building material and is the source of the asphalt shingle data. Since Ecobalance is also the source of the Bees database, these data are comparable with all of the BIPV material data. Asphalt shingles were assumed to have been transported 500 miles by diesel truck, an estimate of the distance from one of the many production plants to the point of use. 2.8. Life cycle metrics 2.8.1. Electricity production efficiency The overall energy performance of an electricity generating system can be assessed from the electricity production efficiency. This efficiency h can be defined by a ratio of the electricity generated, Eout to the total primary energy inputs to the system (excluding the insolation energy in the case of BIPV) Ein. h⫽

Eout Ein

(1)

are the sum of the life cycle The total energy inputs for the BIPV system EBIPVsys in energy inputs for the building integrated photovoltaic and the BOS components including the inverter: ⫽ EBIPV ⫹ EBOS EBIPVsys in in in ,

(2)

(the total life cycle inputs for the BIPV system) is the sum of the inputs where EBIPV in for material production, manufacturing, use, and end-of-life management stages: BIPV ⫽ EBIPV ⫹ EBIPV ⫹ EBIPV EBIPV in mtl mfg ⫹ Euse eol .

(3)

A similar equation can be written for the inverter and other BOS components. In this model, the use phase energy inputs for the building integrated photovoltaic and inverter were assumed to be zero. This is an idealized assumption that neglects any inverter maintenance requirements such as the replacement of electronic components that may have failed. Reliable data, however, are not available for modeling service and replacement processes. The infrastructure for handling the end of life management of the BIPV system is difficult to project 20+ years into the future. Current options include landfill disposal or recycling. Landfill disposal is generally not energy intensive. Recycling energy requirements depend on the extent of materials separation and the material recovery methods. The energy consumption for recycling is often allocated, according to LCA accounting convention, to the product system that utilizes the recovered materials rather than to the source BIPV system. Alternatively, the energy burden could be divided between product systems. Given these considerations, the future energy inputs for end-of-life management of the BIPV system were omitted. is the BIPV electricity generated The net electricity output by this system EBIPVsys out less the BOS efficiency losses:

282

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293 BOS EBIPVsys ⫽ EBIPV out gen ⫺Eloss ,

(4)

where the electricity generated by the BIPV over the time period t is given by EBIPV gen ⫽ qFsoltA

(5)

with a constant conversion efficiency q, average solar flux Fsol, and BIPV area A. Transmission distances for BIPV systems are minimal, thus transmission losses are negligible. The main electricity loss occurs through the inverter, which can also be expressed as a fraction g of the electricity generated: Invertor ⫽ gEBIPV EBOS loss ⫽ Eloss gen

(6)

Inverter losses include conversion losses and losses arising from possible inverter downtime. By substituting these relations into Eq. (1), h becomes h⫽

BIPV BOS BOS BOS EBIPV ⫹ EBIPV ⫹ EBIPV ⫹ EBOS mtl mfg ⫹ Euse eol mtl ⫹ Emfg ⫹ Euse ⫹ Eeol . (1⫺g)qFsoltA

(7)

After neglecting use and end-of-life management energy consumption for the BIPV system as discussed above, Eq. (7) becomes h⫽

BOS BOS EBIPV ⫹ EBIPV mtl mfg ⫹ Emtl ⫹ Emfg . (1⫺g)qFsoltA

(8)

The electricity production efficiency for the conventional grid system can be computed in a similar manner. For this system the life cycle energy inputs are given by ⫽ EFUEL ⫹ EFAC ⫹ EINF EGRID in in in in ,

(9)

where EFUEL accounts for the primary energy for the total fuel cycle, EFAC accounts in in for the life cycle energy inputs for construction, maintenance and demolition of the is the life cycle energy inputs associated with the electricity power plants, and EINF in distribution infrastructure (e.g. power lines). Conventional grid systems such as coal fired power plants or gas turbine facilities have energy inputs associated with conis generally much larger than struction materials including concrete and steel. EFUEL in as discussed above. For example, Spath et al. reported that construction EFAC in accounted for less than one percent of the total CO2 emissions in generating electricity from a coal fired power plant [20]. Similarly, Mann and Spath found that the ratio of power plant construction energy to biomass fuel energy (including feedstock production) for the gasification of hybrid poplar to electricity was 0.08 to 43 [19]. The energy delivered to customers is the net energy generated by a power plant (including plant parasitic losses) less the transmission and distribution losses: TD ⫽ EFAC EGRID out gen ⫺Eloss.

(10)

Substituting Eqs. (9) and (10) into Eq. (1), the electricity production efficiency becomes h⫽

TD EFAC gen ⫺Eloss . FAC ⫹ Ein ⫹ EINF in

FUEL in

E

(11)

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

283

Neglecting EFAC and EINF in in , h becomes independent of time. h⫽

TD EFAC gen ⫺Eloss EFUEL in

(12)

2.9. Pollution prevention The pollution prevention achieved through the displacement of conventional grid electricity and building materials by BIPV technology can be measured using life cycle assessment. The total mass of pollutant i prevented from release into the ¯ i can be calculated by taking environment per unit of electricity delivered (kWh) ⌬M the difference between the total pollutant releases throughout the life cycle of the BIPV and CONV systems. ¯ CONVsys ¯ BIPVsys ¯i⫽M ⫺M ⌬M i i

(13)

¯ CONVsys is the mass of pollutant i released from the conventional system per unit M i of electricity delivered over the service life of the power plant and is calculated from: ¯ CONVsys ⫽ M i

MFUEL ⫹ MFAC ⫹ M’INF ⫹ M’BLDGMTL i i i i , GRID Eout

(14)

is the total fuel cycle emission of pollutant i over the service life of where MFUEL i the power plant tFAC servicelife including extraction, fuel processing, distribution, and use are the emissions associated (combustion for fossil fuels) related emissions, MFAC i are the emissions associated with the construction of the power plant facility; M’INF i with the production of the electricity distribution infrastructure that would be required are for the time period equivalent to the service life of the power plant; M’BLDGMTL i the emissions associated with the production of conventional building materials that would be required for the time period equivalent to the service life of the power is the total electricity delivered over the service life of the power plant, and EGRID out plant. The following adjustment is made to account for differences in the service life between the conventional building materials and the power plant facility: tFAC servicelife ⫽ MBLDGMTL , M’BLDGMTL i i tBLDGMTL servicelife

(15)

are the emissions associated with the production of conventional where MBLDGMTL i building materials based on a service life tBLDGMTL servicelife . A similar adjustment can be made for the electricity distribution infrastructure: tFAC servicelife ⫽ MINF · INF M’INF i i tservicelife

(16)

The total pollutant i released from the BIPV system per unit of electricity delivered over the service life of the BIPV system is calculated from: ¯ BIPVsys ⫽ M i

MBIPV ⫹ M⬙Inverter i i , BIPVsys Eout

(17)

284

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

where MBIPV are the emissions associated with production of the BIPV, M⬙Inverter are i i the emissions associated with the inverter used for the time period equivalent to the GRID is the total electricity delivered over the service life of the BIPV tBIPV servicelife, andEout service life of the BIPV. Electricity distribution infrastructure emissions can be allocated to the BIPV system for on-grid applications because this infrastructure is required to ensure a continuous supply of electricity to both utilize (during overare production of) and match (during under-production of) the BIPV output. M⬙INF i the emissions associated with the production of the electricity distribution infrastructure that would be required for the time period equivalent to the service life of the BIPV. Again, an adjustment can be made to account for the differences in the service life between the BIPV and the inverter: tBIPV servicelife ⫽ MInverter , M⬙Inverter i i tInverter servicelife

(18)

where MInverter lifecycle,i are the emissions associated with the production of the inverter based INF .The general on a service life tInverter servicelife. A similar equation can be written for M⬙i ¯ i becomes equation for ⌬M tFAC tFAC servicelife servicelife ⫹ MFAC ⫹ MINF ⫹ MBLDGMTL MFUEL i i i i INF tservicelife tBLDGMTL servicelife ¯ i⫽ ⌬M EGRID out

(19)

tBIPV tBIPV servicelife servicelife ⫹ MInverter ⫹ MINF MBIPV i i i Inverter tservicelife tINF servicelife . ⫺ BIPVsys Eout BIPV INF / EBIPVsys ⫽ tFAC terms cancel out, which is Since EGRID out out servicelife / tservicelife then Mi expected since the infrastructure would be common to both systems. For off-grid term for the BIPV system would be replaced with a MSTORE applications, the MINF i i term to account for emissions related to the electricity storage device:

tFAC tFAC servicelife servicelife ⫹ MFAC ⫹ MINF ⫹ MBLDGMTL MFUEL i i i i INF tservicelife tBLDGMTL servicelife ¯i⫽ ⌬M EGRID out

(20)

tBIPV tBIPV Inverter servicelife STOR servicelife ⫹ M ⫹ M MBIPV i i i tInverter tSTOR servicelife servicelife . ⫺ EBIPVsys out For the analysis of the amorphous silicon BIPV system the following assumptions BIPV ⰇMFAC and thus MFAC drops out; tBLDGMTL were made: MFUEL i i i servicelife ⫽ tservicelife ⫽ Inverter tservicelife; and that both the conventional and BIPV systems require grid infrastructo represent the total fuel cycle ture. Given these conditions and defining M⬙FUEL i emissions of pollutant i for conventional electricity generation over the time period ¯ i becomes: equal to the service life of the BIPV, ⌬M

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

¯i⫽ ⌬M

M⬙FUEL ⫹ MBLDGMTL ⫺MBIPV ⫺MInverter i i i i . BIPVsys Eout

285

(21)

3. Results The following results were all generated with the same ‘reference’ system chosen to be representative of a residential rooftop installation. This system comprised 34 m2 of United Solar SHR-17 PV shingles (an array rated at ca 2kWp) deploy horizontally and AES MI-250 inverters. Global horizontal insolation data in Table 3 were used to calculate electricity generation for the BIPV reference system. A zero tilt application was assumed for modeling purposes even though such applications are not practical for the shingle system. Generally for BIPV systems, a tilt angle equal to the latitude normally maximises annual electricity generation. The analyses are based on a BIPV service lifetime of 20 years. Simulations were conducted using this reference system for all 15 cities currently represented in the PV-BILD tool. 3.1. Electricity produced and energy payback time The total electricity generated in kilowatt-hours (kWh) over a 20 year period by the reference system in 15 cities is presented in Table 5. These results were calculated using Eqs (4–6) and were based on an inverter conversion efficiency of 95%. The electricity generated, which is a function of insolation in these cities, is the amount of conventional grid electricity that would be displaced by the BIPV system. The average electricity generation rate on a per square meter basis can also be computed by rearranging Eqs. (4–6). For example, the reference system electricity output in Table 5 Electricity produced (kWh) by the reference system over its 20 year lifetime, by city City

kWh produced, lifetime

Atlanta Boston Boulder Chicago Detroit Fort Worth Los Angeles Miami Minneapolis New York Oklahoma City Philadelphia Phoenix Portland Raleigh

64820 55320 64740 54720 53460 69190 69970 68370 55060 56460 67370 56410 81110 49760 62180

286

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

Detroit is 0.95×0.06×(3779 Wh/m2/day)×(day/24 h) 9.0) W/m2. Values reported in Table 5 are upper limits for the electricity generated because system downtime for possible maintenance and repairs are not factored into the analysis. Conceivably this maintenance can be scheduled during the nighttime hours to reduce loss in generation. Field data indicate that inverter failures are not uncommon [22]. These results mirror the insolation data from Table 3. Energy payback time values are given in Table 6. These values indicate the number of years it takes the reference system to generate the amount of energy that was required to manufacture the system. These results also mirror the insolation, and since the reference system is used for all cities, the only variable affecting energy production is insolation. Cities with high insolation have shorter energy payback times. 3.2. Electricity production efficiency The electricity production efficiency for the reference BIPV system and the conventional grid are compared in Table 7. The h values for the BIPV system were calculated using Eq. (8) and h values for the conventional grid system were evaluated using Eq. (12). h values for the BIPV system are expected to represent upper limits to the actual efficiencies because of gaps in the inventory analysis, particularly for the inverter manufacturing energy. Nonetheless, the efficiencies of the BIPV system are dramatically greater than the corresponding conventional grid system efficiencies. The conventional grid efficiencies were calculated for each NERC region so cities located in the same region have the same electricity production efficiency. The BIPV system electricity production efficiencies also track the insolation available in each of the cities considered. Since the energy inputs were identical for all Table 6 Energy payback time (years) for the reference system, by city City

Energy payback time, years

Atlanta Boston Boulder Chicago Detroit Fort Worth Los Angeles Miami Minneapolis New York Oklahoma City Philadelphia Phoenix Portland Raleigh

4.24 4.97 4.24 5.02 5.14 3.97 3.93 4.02 4.99 4.87 4.08 4.87 3.39 5.52 4.42

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

287

Table 7 Electricity production efficiency for the reference system and NERC region, by city City

Atlanta Boston Boulder Chicago Detroit Fort Worth Los Angeles Miami Minneapolis New York Oklahoma City Philadelphia Phoenix Portland Raleigh

Electricity production efficiency BIPV system

NERC region

4.72 4.03 4.71 3.98 3.89 5.04 5.09 4.98 4.01 4.11 4.90 4.11 5.91 3.62 4.53

0.26 0.30 0.36 0.25 0.23 0.26 0.36 0.26 0.25 0.30 0.26 0.26 0.36 0.36 0.26

of these systems and the only other determinant of electricity production efficiency for a BIPV system is its energy output, this is not a surprising result. It is useful to compare the BIPV systems against the conventional electricity generation represented by NERC region data. Since the conversion of fossil fuels into electricity via combustion is only about 30% efficient and there are further losses in transmission and distribution of centrally generated electricity, current conventional generation cannot approach the efficiency of the distributed BIPV systems considered here. 3.3. Pollution prevention benefits Air pollution prevention benefits provided by the reference system in six of the 15 cities in PV-BILD are presented in Tables 8 and 9. These cities were chosen to be diverse in conventional generation mix and insolation. Table 8 indicates the mass of air pollution emissions avoided. Table 9 shows the avoided mass per kWh of electricity generated by the reference BIPV system. Pollution prevention values were determined using Eq. (21) based on the following simplifications. The emissions related to construction of the conventional grid power plant facilities and electricity generating infrastructure were omitted due to lack of data and the service lives of the conventional building materials, BIPV product and inverter were assumed to be BIPV Inverter equal (tBLDGMTL servicelife ⫽ tservicelife ⫽ tservicelife). The air pollution prevention occurs mainly in the region of the displaced conventional grid electricity although some air pollution emissions for both the BIPV and conventional systems are distributed spatially outside this region. Air pollution emis-

288

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

Table 8 Total mass of air pollutant emissions avoided by the reference system over a 20 year period (g) Air emission

Carbon dioxide (CO2 as C) Carbon monoxide (CO) Lead (Pb) Mercury (Hg) Methane (CH4) Nitrogen oxides (NOx as NO2) Nitrous oxide (N2O as N) Particulates (unspecified) Sulfur oxides (SOx as SO2)

Mass of air pollutants avoided (g) Boston

Boulder

Detroit

Phoenix

Portland

Raleigh

2.42E+07 6.55E+03 1.11E+00 3.24E⫺01 6.34E+04 6.14E+04

3.27E+07 7.48E+03 9.26E+00 6.11E-01 9.97E+04 1.03E+05

6.60E+07 1.35E+04 2.83E+01 1.31E+00 1.85E+05 2.22E+05

4.19E+07 9.99E+03 1.37E+01 7.76E⫺01 1.27E+05 1.33E+05

2.44E+07 5.18E+03 5.20E+00 4.61E⫺01 7.50E+04 7.63E+04

5.23E+07 1.14E+04 1.93E+01 9.86E⫺01 1.47E+05 1.69E+05

4.38E+02 4.53E+04 8.08E+04

6.95E+02 1.07E+05 1.71E+05

1.47E+03 2.57E+05 3.88E+05

8.94E+02 1.42E+05 2.22E+05

5.12E+02 7.45E+04 1.25E+05

1.13E+03 1.87E+05 2.87E+05

Table 9 Total mass of air pollutant emissions avoided by the reference system (g/kWh) Air emission

Carbon dioxide (CO2 as C) Carbon monoxide (CO) Lead (Pb) Mercury (Hg) Methane (CH4) Nitrogen oxides (NOx as NO2) Nitrous oxide (N2O as N) Particulates (unspecified) Sulfur oxides (SOx as SO2)

Mass of air pollutants avoided per unit electricity generated (g/kWh) Boston

Boulder

Detroit

Phoenix

Portland

Raleigh

4.37E+02 1.18E⫺01 2.01E⫺05 5.85E⫺06 1.15E+00 1.11E+00

5.06E+02 1.16E⫺01 1.43E⫺04 9.45E⫺06 1.54E+00 1.60E+00

1.23E+03 2.52E⫺01 5.30E-04 2.45E⫺05 3.46E+00 4.15E+00

5.16E+02 1.23E⫺01 1.69E⫺04 9.57E⫺06 1.56E+00 1.64E+00

4.90E+02 1.04E⫺01 1.04E⫺04 9.27E⫺06 1.51E+00 1.53E+00

8.41E+02 1.83E⫺01 3.11E⫺04 1.59E⫺05 2.36E+00 2.72E+00

7.92E⫺03 1.07E⫺02 2.75E⫺02 1.10E⫺02 1.03E⫺02 1.82E⫺02 8.19E⫺01 1.65E+00 4.81E+00 1.75E+00 1.50E+00 3.01E+00 1.46E+00 2.64E+00 7.26E+00 2.73E+00 2.51E+00 4.61E+00

sions occur at various sites including material production for the BIPV system, BIPV and inverter manufacturing and distribution routes. In addition to power plant locations for grid electricity, air emissions occur at material production and roofing manufacturing sites. Conventional grid electricity and related air pollutants emissions were modeled for each NERC region based on the mix of electricity sources within the region. The PV-BILD model assumed that BIPV electricity generated at any site within the region would displace an equivalent quantity of composite grid electricity in that region while in reality the actual displacement may occur for one specific plant in the region. Consequently, a BIPV system deployed in Denver, Colorado may displace electricity from a coal fired power plant in the area rather than displace composite kWh gener-

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

289

ated in the hydropower-dominated Western Systems Coordinating Council (WSCC) regional grid. The model architecture is not limited in analyzing a finer spatial resolution as long as the electricity generation burdens are available. The three cities with the greatest pollution prevention with respect to an individual air pollutant are indicated in Table 10. These results follow closely the proportion of coal and natural gas in each of these cities’ NERC regions, with coal being more important. Detroit (89.5% coal and natural gas) was the only city included from the ECAR region and had the highest mass of avoided air emissions for nearly all species considered. Fort Worth at 83.4% coal and natural gas, Oklahoma City at 88.1%, Minneapolis at 73.2% and Miami at 62.4% (and the highest insolation in the SERC region) were the only other cities represented in these results of the top three cities. The reference BIPV system has the greatest pollution prevention benefit in these cities because they have generating mixes with more air emissions than the other cities included in PV-BILD. This point is important for policymakers in areas with difficulty meeting federal air quality standards. The relative contribution of each term of the pollution prevention equation is presented in Table 11. The total air pollution related to the displaced grid electricity, displaced shingles, BIPV product, and the inverter are calculated for the reference case in Detroit and expressed on a per kWh basis. The pollution prevention achieved ¯ i is also presented and matches the results in Table 9. A comparison of the air ⌬M pollution from conventional grid electricity production and the total air pollution prevention indicates that the displacement of the grid electricity generation is the dominant factor in the pollution prevention equation. The displaced building material has a negligible impact on the total pollution prevention achieved. Air pollution for the BIPV product and in particular the inverter are understated due to the data gaps identified in the methods.

Table 10 Cities with the three greatest amounts (by mass) of air emission pollution prevention Air pollutant

Carbon dioxide (CO2 as C) Carbon monoxide (CO) Lead (Pb) Mercury (Hg) Methane (CH4) Nitrogen oxides (NOx as NO2) Nitrous oxide (N2O as N) Particulates (unspecified) Sulfur oxides (SOx as SO2)

Pollution prevention benefits, ranked by mass of avoided emission Most

Second most

Third most

Detroit Fort Worth Detroit Detroit Fort Worth Detroit Detroit Detroit Detroit

Oklahoma City Oklahoma City Minneapolis Minneapolis Oklahoma City Oklahoma City Oklahoma City Minneapolis Minneapolis

Miami Detroit Miami Miami Detroit Miami Miami Miami Miami

290

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

Table 11 Breakdown of the air emission pollution prevention achieved [as defined by Eq. (21)] for the reference system deployed in Detroit (g of pollutant/ kWh of electricity generated) Pollutant

Grid electricitya

Displaced building materialsb

BIPV productionc

Inverterd

Pollution preventede

Carbon dioxide (CO2, fossil) Carbon monoxide (CO) Lead (Pb) Mercury (Hg) Methane (CH4) Nitrogen oxides (NOx as NO2) Nitrous oxide (N2O) Particulates (unspecified) Sulfur oxides (SOx as SO2)

1.30E+03 2.98E⫺01 6.85E⫺04 2.53E⫺05 3.59E+00 4.41E+00

8.14E⫺02 1.06E⫺04 0.00E+00 0.00E+00 1.38E⫺04 2.18E⫺04

5.12E+01 3.10E⫺02 1.31E⫺04 7.06E⫺07 1.04E⫺01 2.32E⫺01

1.16E+01 1.48E⫺02 2.46E⫺05 1.88E⫺08 2.61E⫺02 2.95E⫺02

1.23E+03 2.52E⫺01 5.30E⫺04 2.45E⫺05 3.46E+00 4.15E+00

2.93E⫺02 5.42E+00 7.80E+00

5.21E⫺06 6.53E⫺03 4.92E⫺04

1.56E⫺03 5.86E⫺01 2.89E⫺01

2.12E⫺04 2.75E⫺02 2.77E⫺02 4.81E+00 2.53E⫺01 7.26E+00

a

:

M⬙FUEL i EBIPVsys out

b

MBLDGMTL i : BIPVsys Eout

c

MiBIPV : BIPVsys Eout

d

:

MInverter i BIPVsys Eout

e

¯ i. :⌬M

4. Conclusions This paper presented the life cycle assessment method for evaluating BIPV energy and environmental performance relative to conventional displaced grid electricity and building materials. The electricity production efficiency was defined in Eqs. (7) and (11) and enables the comparison of the BIPV and conventional systems. In addition, the air pollution prevention achieved by the BIPV system was calculated from Eq. (21). Application of the model equation to the United Solar amorphous silicon roofing product demonstrated the benefits of a specific BIPV system relative to the regional electricity generation grid and displaced roofing materials. The electricity production efficiency (electricity output/total primary energy input excluding insolation) for a reference system ranged from 3.6 in Portland OR to 5.9 in Phoenix, AZ indicating a significant return on energy investment. The energy performance of this BIPV system is dramatically better than conventional electricity generation where electricity production efficiency values ranged from 0.26 to 0.36. A previous study by the authors of an older, less efficient amorphous silicon module (conversion efficiency=5%) indicated the expected benefits of eliminating framing in a BIPV design [23]. This analysis showed that an aluminum frame reduced the electricity production efficiency by about a half. While the results are not directly comparable (this study used global horizontal insolation and the previous study used direct normal insolation in addition to differences in the conversion efficiencies), the BIPV system is expected to show similar advantages over framed modular units. The exact benefits will depend on the energy intensity of the framing materials. The electricity production efficiency provides one measure of sustainability for BIPV technology. The electricity production efficiency defined by Eq. (1) is similar

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

291

to the net energy ratio which is the electricity output divided by the total fossil fuel energy input. Given that a majority of the total primary inputs into the BIPV system in this study were from fossil sources the two metrics are effectively equal. These metrics show the extent to which fossil fuels can be leveraged into electricity of much greater energy value. This magnification of fossil energy resources is a major attribute of BIPV systems and many other renewable technologies. For comparison one can examine the performance of other renewable systems that have been reported in the literature. Mann and Spath reported a net energy ratio of 15.6 for hybrid poplar grown on short rotation plantations and converted to electricity using a biomass gasification combined cycle system [19]. While the amorphous silicon BIPV system has a lower energy efficiency the biomass system has a significant land area requirement. The power output normalized on a area basis is 0.32 W/m2 for the poplar (based on the area requirements for the plantation) compared to 9.0 W/m2 for the BIPV system (for reference case in Detroit). Net energy ratios can be increased greatly by substituting renewable energy for fossil fuel energy in the production of the renewable technology. Consequently, as renewable energy becomes more widely distributed the net energy ratio becomes even more favorable. Air pollution benefits of the United Solar roofing systems were significant and dominated by the displacement of conventional grid electricity. The benefits from displaced roofing material were small. The analysis of pollution prevention benefits provided by the BIPV system also yielded unexpected results. The reference BIPV system had the greatest air pollution prevention benefits in cities with conventional electricity generation mixes dominated by coal and natural gas, not necessarily in cities where the insolation and displaced conventional electricity were greatest. Detroit had the highest mass of pollution prevention for all air emissions except methane and carbon monoxide. The pollution prevention benefits of displaced conventional building material were negligible compared to the benefits from displacing conventional electricity generation for the reference system. While this LCA focused on the air pollution prevention benefits of BIPV systems, the model equations can also be used to evaluate potential water pollutant and land pollutant releases. For off-grid applications, BIPV systems can have additional energy and environmental advantages over conventional on-grid systems. These advantages can be assessed using the model equations presented here. A detailed assessment would require a life cycle assessment of the electricity generating and transmission infrastructure. The construction of this infrastructure for remote sites is often uneconomical and the energy and environmental consequences of the infrastructure are large when normalized by total power delivered. Analysis of off-grid applications with electricity storage would also have to account for the burdens related to batteries or other storage devices. The design, planning and implementation of BIPV technology is guided by economics in addition to the energy and environmental performance metrics presented here. The life cycle economic model and results are presented in the National Science Foundation Project Report on this research [15].

292

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

Acknowledgements The authors would like to acknowledge the support of the National Science Foundation (NSF) and Lucent Technologies. This work was made possible by Grant No. BES9729268 under the National Science Foundation/Lucent Technologies Industrial Ecology Fellowship program. Dr A. Fredrick Thompson, Program Director, Environmental Technology at NSF served as the Project Officer. We would like to thank Subhendu Guha and Kevin Hoffman at United Solar for their cooperation and for generously supplying data on United Solar PV roofing products. Rob Wills and Mike Kaelin at Advanced Energy Systems were extremely helpful in providing data and documentation on the AES MI-250 microinverter. Life cycle data for steel were provided by the LCA Task Force of the American Iron and Steel Institute through Scott Chubbs at Virterra, Inc. Vince Camobreco at Ecobalance provided invaluable assistance and technical support.

References [1] International Energy Agency, Key World Energy Statistics 2000. 19 February 2000 [htttp://www.iea.org/statist/keyworld/keystats.htm]. [2] Maycock PD. The world PV market : Shifting from subsidy to ‘fully economic’? Renewable Energy World 2000;3(4):59–74. [3] International Energy Agency. In: Sick F, Erge T, editors. Photovoltaics in buildings: a design handbook for architects and engineers. London: James and James; 1996. [4] Aulich HA, Schulze FW, Strake B. Energy pay-back time for crystalline silicon photovoltaic modules using new technologies. In: Proceedings of 18th IEEE PV Specialists Conference. The Institute of Electrical and Electronics Engineers; 1985. p. 1213-1218. [5] Hagedorn GH. Hidden energy in solar cells and photovoltaic power stations. In: Proceedings of 9th EC Photovoltaic Solar Energy Conference, Freiburg, FRG. Boston: Kluwer; 1989. p. 542–5. [6] Schaefer H, Hagedorn G. Hidden energy and correlated environmental characteristics of pv power generation. Renewable Energy 1992;2(2):159–66. [7] Srinivas KS. Investments and production costs of amorphous silicon PV modules. Bern: Swiss Federal Department of Energy, 1991. [8] Palz W, Zibetta H. Energy pay-back time of photovoltaic modules. International Journal of Solar Energy 1991;10:211–6. [9] Prakash R, Bansal NK. Energy analysis of solar photovoltaic module production in India. Energy Sources 1995;17:605–13. [10] Kato K, Murata A, Sakuta K. An evaluation on the life cycle of photovoltaic energy system considering production energy of off-grade silicon. Solar Energy Materials and Solar Cells 1997;47:95–100. [11] Alsema EAE. In: Environmental aspects of solar cell modules. Utrecht University: Utrecht (Netherlands); 1996. p. 4-0. [12] Huber W, Kolb G. Life cycle analysis of silicon-based photovoltaic systems. Solar Energy 1995;54(3):153–63. [13] Johnson AJ, Watt M, Ellis M, Outhred HR. A life cycle assessment of grid, grid connected PV and stand-alone PV power systems for household energy supply. In: Proceedings of 14th European Photovoltaic Solar Energy Conference and Exhibition, Barcelona; 1997. [14] Spiegel RJ, Edward J, Kern C, Greenberg DL. Demonstration of the environmental and demandside management benefits of grid-connected photovoltaic power systems. Solar Energy 1998;62(5):345–58. [15] Lewis GM, Keoleian GA, Moore MR, Mazmanian DL, Navvab M. PV-BILD: A life cycle environ-

G.A. Keoleian, G.McD. Lewis / Renewable Energy 28 (2003) 271–293

[16] [17]

[18] [19] [20] [21] [22] [23]

293

mental and economic assessment tool for building-integrated photovoltaic installations. Center for Sustainable Systems, Report No. CSS99-02R, University of Michigan, 20 December 1999. ISO. Environmental management — Life cycle assessment — Principles and framework. Geneva: International Organization for Standardization, 1997. Lewis G, Keoleian GA. Life cycle design of amorphous silicon photovoltaic modules. Cincinnati (OH): US Environmental Protection Agency, Office of Research and Development, National Risk Management Research Laboratory, 1997. National Renewable Energy Lab Homepage [http://www.nrel.gov/data/solar/by—city], November 1995. Mann MK, Spath PLL. Life cycle assessment of a biomass gasification combined-cycle system. Golden (CO): National Renewable Energy Laboratory, 1997. Spath PL, Mann MK, Kerr DRB. Life cycle assessment of coal-fired power production. Golden (CO): National Renewable Energy Laboratory, 1999. Lippiatt B. BEES (Building for Environmental and Economic Sustainability): balancing environmental and economic performance. National Institute of Standards and Technology (NIST); 1998. Pratt R. DTE Energy, personal communication. March; 2000. Keoleian GA, Lewis GM. Application of life cycle energy analysis to photovoltaic module design. Progress in Photovoltaics Research and Application 1997;5(4):287–300.