Attachment_I_GSA_LCA_Project_Final_Report.pdf

LIFE CYCLE ASSESSMENT OF

FEDERAL PROCUREMENT

In Reference to: Solicitation No. XX001-XX994

Prepared by:

Facilities Solutions Group, LLC, 1900 Oracle Way, Suite 400, Reston, VA 20190

Economic Input/Output and Process Life Cycle Analysis of

Federal Procurement With Environmental and Other Impacts 12/9/2011

Facilities Solutions Group, LLC 2

Life Cycle Assessment of Federal Procurement

Dec. 9, 2011

In Reference to:

SOLICITATION NO. XX001-XX994

Report for

Prepared by Facilities Solutions Group, LLC 1900 Oracle Way, Suite 400 Reston, VA 20190

This report includes data that shall not be disclosed outside the Government and shall not be duplicated, used, or disclosed-in whole or in part-for any purpose other than to evaluate this report.

If, however, a contract is awarded to this offeror as a result of-or in connection with-the submission of this data, the Government shall have the right to duplicate, use, or disclose the data to the extent provided in the resulting contract. This restriction does not limit the Government's right to use information contained in this data if it is obtained from another source without restriction.

© 2011 Facilities Solutions Group, LLC

Facilities Solutions Group, LLC 3

ACKNOWLEDGMENTS

I would like to thank all those individuals that supported and participated in the project and who went beyond the requirements of contract to ensure a quality product and definitive report for the GSA’s efforts to meet its objectives in Executive 13514 and other requirements: Michael O’Connor, Jeffrey

Ponder, Chris Keller, Diana Lemmon, and Susan Forth from Facilities Solutions Group; Drs. H. Scott

Matthews and Deanna Matthews from Carnegie Mellon University; Dr. Jeffrey Morris from Sound

Resource Management Group; Deborah Dunning from Sphere E; and Eva Schmincke and Christoph

Koffler from PE International. A special thanks to Brennan Conaway from the GSA FAS for his patience, guidance, and support through the entire project management process.

Jonathan M. Estes, Project Manager, FSG

Facilities Solutions Group, LLC 4

Table of Contents

ACKNOWLEDGMENTS 3

SUMMARY 6

GOALS & APPROACH 9

GOALS 9

APPROACH 9

TASK DESCRIPTIONS 11

TASK 1--ECONOMIC INPUT-OUTPUT LIFE CYCLE ANALYSIS 11

TASK 2--INCORPORATION OF PROCESS-LEVEL DATA 17

TASK 3--INCORPORATION OF CONSUMER-USE PHASE 21

TASK 4--INCORPORATION OF END-OF-LIFE 22

TASK 5 – OTHER FEDERAL GOVERNMENT SEGMENTS 23

TASK 6 – PRIORITIZATION BY OTHER ENVIRONMENTAL IMPACT CATEGORIES 23

TASK 7 – STRUCTURAL PATH ANALYSIS 26

LIFE CYCLE INVENTORY 32

ECONOMIC INPUT-OUTPUT LCA 32

PROCESS LCA 32

LIFE CYCLE IMPACT ASSESSMENT 34

ECONOMIC INPUT-OUTPUT LCA 34

PROCESS LCA 34

HISTORICAL FRAMEWORK FOR INTERFACE OF ECONOMIC LCA & PROCESS LCA 34

METHOD 36

Facilities Solutions Group, LLC 5

ANALYSIS & CONCLUSIONS 38

ECONOMIC INPUT-OUTPUT LCA 38

PROCESS LCA 43

SCREENS 43

CONCLUSIONS 47

RECOMMENDATIONS 50

ECONOMIC INPUT/OUTPUT LCA 50

PROCESS LCA 56

APPLICABLE APPENDICES 58

APPENDIX A: I-O LIFE CYCLE DATA (FROM AUGUST 2010 MONTHLY REPORT) 58

APPENDIX B: FEDERAL SPEND DATA 63

APPENDIX C: CAPITAL FLOW TABLE 64

APPENDIX D: DEVELOPMENT OF CRITERIA AIR POLLUTANT DATA 66

APPENDIX E: TRACI 2.0 CHARACTERIZATION FACTORS 68

Facilities Solutions Group, LLC 6

SUMMARY

Facilities Solutions Group, LLC and its core team of LCA experts provided the General Services

Administration FAS tools and report for Life cycle Assessment (LCA) of Federal Procurement which serves to help the GSA for achieving its priorities in its FY 2010-2015 Strategic Sustainability

Performance Plan including greening its supply chain, eliminating waste in its activities and operations, and using its influence to reduce the environmental impact of the Federal government. Our tools and report help to answer fundamental questions that could strategically direct GSA’s activities including:

1. What are the environmental impacts of the products purchased by the Federal Government, considering the entire product life cycle?

2. Where in the product life cycle do the environmental impacts occur?

3. Which products have the greatest overall environmental impact, considering purchasing volumes?

4. How can GSA make the most significant reductions in the environmental impact of its portfolio within budget constraints?

5. How do the results vary for different agencies?

The data from our analysis could ensure that GSA activities have the maximum positive effect on the environment. For FSG, the LCA of Federal Procurement Project is not only an important endeavor for the

GSA and its property portfolio and employees to meet its objectives, but it will also serve as a new standard of excellence for other Departments and Agencies to emulate and serve to help meet the

GSA’s overriding mission: To use expertise to provide innovative solutions for our customers in support of their missions and by so doing, foster an effective, sustainable, and transparent government for the American people.

With the goal of empowering GSA to become a global leader in managing Federal Procurement to achieve the largest possible reductions in cost and environmental impact, FSG has assembled a team of national/international experts in the areas of economic LCA, process, performance data evaluation, and procurement best practices. This team has a collective knowledge base in the U. S., Europe and others parts of the world, including volunteer service in the most relevant working committees of the

International Organization for Standardization (ISO) and the Life Cycle Initiative of the UN Environment

Program (UNEP).

We built the resulting tool upon the EIO-LCA system developed by Carnegie Mellon University and used globally by 1.5 million or more governmental agencies and non-governmental entities to provide the technical framework and the dataset access point for “Integrated LCA” providing a seamless integration of high level economic LCA with expert vetted process data. The two filters essential to achieve advanced Federal Procurement will bring data into a single tool making both types of metrics accessible for use in evaluating procurement options for their ability to reduce significantly energy and water use as well as the emission of greenhouse gases and toxins to air, soil and water.

Facilities Solutions Group, LLC 7

The processes established through our development of an Excel Spreadsheet with mid-level LCA Data

Availability Information will serve GSA with an effective and efficient means of measuring, verifying and reporting how their annual spend on all types of products and services is exceeding national goals for cost reduction and environmental health. Links to a subset of categories inclusive of available process

LCA information were included. By demonstrating how its annual spend on procurement can be managed sustainably, the U. S. will gain recognition and respect throughout the world as a major contributor to a global green economy.

SUMMARY RECOMMENDATIONS

ECONOMIC INPUT/OUTPUT LCA

The system could be better supported as a hosted web service.

The system could also be improved by archiving several years’ worth of federal spending.

The system would also be improved if designed to “update” estimates based on discovered inventories and/or process-based LCI data.

There is substantial improvement needed in the modeling and data acquisition related to use phase and end of life activities, and also in determining how to reasonably estimate and allocate such impacts back to the products causing them.

An improved model would seek additional inventory data from GSA or other agencies to estimate stocks and flows of purchased assets over time to better connect potential energy use and EOL impacts.

More validation about the publically available federal spend data is needed.

PROCESS LCA

Integrate mid-level LCA Data Availability Information in combination with Economic I/O LCA data together as provided in the Dashboard to determine where to best allocate their available resources (staff and funding) to reduce significantly GSA’s environmental footprint.

Move forward to another step in reducing the Federal Government’s footprint by developing sustainable acquisition strategies and adopting new policies to support these strategies.

Determine which industries to enroll in developing and making available independently verified process LCA data.

Enroll a number of industry organizations to provide education, tools and training on the development and use of LCAs (ISO 14040/44) and EPDs (ISO 14025).

Create product LCA data for some of the categories where currently there is no generic LCA data available as noted in the Task Two LCA Mapping Database.

Enroll manufacturers of frequently-used materials in working with their industry groups to provide representative LCA data for the US context where not yet available and to integrate life

Facilities Solutions Group, LLC 8 cycle thinking into their research programs to improve the materials’ environmental footprints over time.

Apply a systems approach for using brand or product-specific LCA datasets to enhance GSA’s specification documents and procurement practices.

GSA has the potential to provide the critical mass needed to initiate, develop and maintain the implementation of life cycle thinking through EPDs in the US.

Facilities Solutions Group, LLC 9

GOALS & APPROACH

GOALS

The goals of this project were to provide life cycle data and method infrastructure to GSA to support various potential sustainability oriented program evaluations, including as possibilities:

Reducing the environmental impact of Federal procurement totaling around $4 trillion

Eliminating waste in the operations and activities of GSA and other Federal Agencies it serves

Empowering GSA to influence the total environmental impact of the U. S. Federal Government

To achieve this objective, high quality datasets integrated within economic input-output Life Cycle

Assessment (LCA) with the potential to reduce significantly the environmental footprint of Federal procurement were developed, combined, and made accessible to GSA in an interactive Excel spreadsheet. We built upon the EIO-LCA tool developed by Carnegie Mellon University and used globally by 1.5 million or more governmental and non-governmental entities to provide the technical framework for screening federal spending at the aggregate input-output (IO) sector level to determine which types of federal spending should be targeted for opportunities for environmental footprint reduction.

APPROACH

We have designed a plan that will allow GSA to understand the potential for reducing the life cycle impacts of federally purchased goods and services, and optionally, to assess how such programs might be designed and implemented both within the agency and with other agencies whose procurement GSA manages. Our plan was divided into seven major tasks, which are discussed below.

We note that members of the team assembled by FSG have been key contributors to Carnegie Mellon

University’s economic input-output life cycle assessment (EIO-LCA) data and web modeling framework, led by Dr. Scott Matthews, that is considered to be in alignment with global best practices in this area of technology. This framework has been available at no cost on the Internet to researchers, policymakers, and interested parties since 1999 (at http://www.eiolca.net) and has been widely used by experts around the world for educational and commercial purposes.

CMU’s EIO-LCA models and their underlying datasets have been externally peer-reviewed, and perhaps more importantly, are very frequently accessed and used by other groups in the US and around the world. Users include academic researchers, state and federal policy makers, students and professors in courses at more than 100 colleges and universities in the US, and decision makers in at least 50 Fortune

500 companies as well as users from more than 100 countries. In short, the Carnegie Mellon EIO-LCA http://www.eiolca.net/

Facilities Solutions Group, LLC 10 data and models are considered the gold standard for IO-LCA modeling in the US and in many other parts of the world.

For the LCA and procurement indexing portions of this project the team included Dr. Jeffrey Morris who led a national team of economists and life cycle experts, including Dr. Matthews, to create a consumer environmental index (CEI) for the Washington State Department of Ecology. To our knowledge this was and still is the first successful indexing methodology using a consumption-based inventory for tracking the environmental footprint of consumer final demand over time. CEI is a comprehensive, consumption-based climate, human health and ecosystems impacts model that is used by Ecology to inventory and track the environmental effects of production, use and disposal of goods and services purchased by

Washington consumers (Morris and Matthews 2010).

Morris also has developed an outcomes-based inventory tool MEBCalcTM (Measuring Environmental

Benefits Calculator) for measuring the environmental footprint of waste management systems. This model is very useful in assessing the end-of-life environmental impacts of products, including environmental impacts over the full life cycle of products and packaging materials (Morris, 1996; Morris, 2005; Morris and Bagby, 2008; Morris, 2010).1 This tool combines the best waste management life cycle data and methodologies from a variety of LCA models, including EPA’s WARM (Waste Reduction Model), EPA/NCSU/RTI’s Municipal Solid Waste Decision Support Tool (MSW DST), NIST’s BEES (Building for

Environmental and Economic Sustainability) model, EPA’s Tool for the Reduction and Assessment of

Chemical and Other Environmental Impacts (TRACI), and Carnegie Mellon University’s EIO-LCA model.

FSG brought extensive knowledge and practical experience in sustainability analytics for government agencies and other entities to this work. Sphere E experts bring global knowledge and broad practical application of process LCA to the team. Their experts include Dr. Eva Schmincke, a Principal in Sphere E as well as a Senior Consultant to PE International, and Deborah Dunning, President of The Green

Standard, an NGO that has provided high quality life cycle based resources to the U. S. marketplace for over 10 years, including an award-winning publication on sustainable product standards, an online training program on green procurement practices and development of an LCA-based Environmental

Product Declaration Program.

The collective professional experience of the FSG-led team of global experts in the areas of IO-LCA, Process LCA, Performance Data Management and Green Procurement provided the GSA with high quality data assessment housed in efficient software tools.

Additional explanatory information on MEBCalc

TM

is available on SRMG’s website at http://www.zerowaste.com/pages/MEBCalc.htm (short backgrounder).

http://www.zerowaste.com/pages/MEBCalc.htm

Facilities Solutions Group, LLC 11

TASK DESCRIPTIONS

TASK 1--ECONOMIC INPUT-OUTPUT LIFE CYCLE ANALYSIS

The purpose of this task was to develop best-available estimates of the annual life cycle impacts from production of goods and services, as well as fuel usage, attributable to federal procurement. We envisioned that the motivation for performing this scoping level assessment was to prepare GSA to more readily comply with Executive Order 13514 and other initiatives to reduce the environmental effects of federal procurement.

TASK 1.1: DEVELOPMENT OF AN ECONOMIC INPUT-OUTPUT LCA (EIO-LCA) MODEL FOR

ESTIMATING THE EMBEDDED ENERGY USE AND ENVIRONMENTAL EFFECTS OF FEDERAL

PURCHASING

For this task, we used the official 2002 Benchmark 428 sector Input Output model of the US Economy disseminated by the Bureau of Economic Analysis (BEA) of the Department of Commerce, already available for use in Carnegie Mellon’s EIO-LCA model. The 2002 U.S. EIO-LCA model allows one to estimate the environmental impacts from producing a certain dollar amount of goods and services output from any of 428 sectors of the U.S. economy. The model is on a “purchaser price” (rather than producer price) basis, consistent with the data on federal spending that is used as model inputs and provides a “cradle to consumer” summary of impacts. This means that the default estimates of impacts from the model include the resource extraction and production of the good and services as well as any warehousing/retailing effects and transportation to the purchaser by various modes. We used BEA-supplied data on GDP price indexes for the gross output of each of the 428 sectors to adjust current year values back to the 2002 benchmark year needed for this model

(http://www.bea.gov/industry/gdpbyind_data.htm).

As required by GSA, we also made adjustments to the 2002 benchmark purchaser-basis model to include infrastructure and capital investments (method summarized in Appendix C). We stipulate, however, that this is not the type of model regularly used by the LCA community because the energy and environmental impacts associated with infrastructure and capital generally have little influence on results and are routinely excluded from process LCA data. Thus, including them can lead to spurious results. For this reason we have demonstrated the differences between these two types of models as part of the uncertainty presentation (discussed below).

We chose to use the 428 sector model because it is the accepted “benchmark” for analysis by the federal government, and because alternative methods that try to extend beyond the 428 sector benchmarks are fraught with significant uncertainties. Considering the impacts from specific goods and http://www.bea.gov/industry/gdpbyind_data.htm

Facilities Solutions Group, LLC 12 services that fall within the 428 sectors of spending is appropriately addressed by adding process-level data as discussed in Task 2. As an example, the 428-sector benchmark IO model has a single electricity sector and only two fruit and vegetable sectors. Methods that cosmetically disaggregate these sectors

(or any other sector) into multiple subsectors rely on average price assumptions, e.g., for different generation methods or types of produce that in our experience cannot be adequately substantiated.

Similarly, methods that use “detailed item output” data from the BEA (which list about eight thousand commodities) simply show the fraction of production in each of the 428 sectors and do not give specific supply chain production information for the detailed items. While we did not use detailed item output information, we used the definitions of the 8000 commodities to map the federal spend data described below more accurately into the benchmark 428 sectors.

For this task, we used and showed comparative estimates for GSA with the benchmark 2002 purchaser model (single region, US only), with the 2002 model augmented with infrastructure and capital expenditures. We used a consumption-based accounting method. Details on how energy and environmental effects are captured and communicated are provided below in the overview of the web architecture. Even though GSA requires only the delivery of a Microsoft Excel Spreadsheet for this project, the architecture of our proposed system used the existing EIO-LCA website “engine” for using and manipulating the GSA IO-LCA model. All model calculations by sector, etc., were exported from that well-tested model.

To communicate how our work is highly useful to reducing the economic and environmental impacts of

Federal procurement, we provide an overview of the underlying EIO-LCA web architecture:

The EIO-LCA web framework was designed in 1999 to be a repository of IO-LCA models from around the world. Some of these models have been developed by Dr. Matthews and CMU colleagues while others have been contributed by researchers from the US and other countries.

As such the web architecture is designed to support inter-operability, to be extensible, scalable, and open to the public. The EIO-LCA website is built on a UNIX server base using apache, perl, PHP, and mySQL components. A number of core functions have been developed that are used

(and modified) in this project to support the family of hosted IO-LCA models:

Support for uploading new IO matrices and satellite account matrices – this is accomplished via secure file transfer (ssh) to a server hosted at the Pittsburgh Supercomputing Center. Files are stored as flat text files in directories unique to each IO model.

Support for large databases and characterization factors – this is accomplished via mySQL and perl and php-based queries to the database. The databases include the complete list of Toxics

Release Inventory (TRI ) substances and life cycle impact assessment characterization factors such as TRACI.

Support for multiplying economic IO matrices and satellite account matrices – accomplished by compiling open source matrix multiplication C++ code, and execution via perl or php scripts on the EIO-LCA server. Input and output matrices are stored as flat text files.

Facilities Solutions Group, LLC 13

Support for graphing and education – over time various charting and graphing routines have been developed to help present the results of IO-LCA modeling, and is used in this project. Of particular importance is the structural path analysis (SPA) visualization routines developed that is leveraged in Task 7.

TASK 1.2 - FEDERAL SPEND

Any EIO-LCA model uses economic purchases by sector as inputs. For example, the life cycle impacts of producing a $20,000 car are estimated by entering $20,000 into the “automobile production” sector. In this project, the input used is data on federal spending which is publicly available and already disaggregated by 6-digit NAICS code. These federal spend data are available through www.fpds.gov.

These sources also provide spending details by agency and identify MAS and GWAS contracts as required by this project. Since there are many such federal spend databases freely available, we cross-checked several other sources, including the GSA staff database, to ensure that our federal spend data is robust and credible.

The Excel tool developed transforms the 2011 raw federal spend data by IO sector (mapped from NAICS codes), agency, and contract type. It aggregates all spending into categories and adjusts back to the year 2002 using price deflators by economic sector to be compatible with the EIO-LCA data. The system is made to be robust enough to allow forecasting of future federal spend levels for scenario planning and what-if analyses. Doing so will enabled the team to achieve the desired results to the maximum extent practical as the processes, once developed, are extendable. This is part of the delivered Excel

Spreadsheet.

We note that there were many discrepancies in the raw federal spend data. For example, there were many expenses that were part of the fpds.gov data feed that had no or invalid NAICS codes, agency IDs, or contract types. Further about 20% of records when aggregated showed $0 or a negative expense. All such records were removed from the dataset. In the end, of $280 billion of federal spending in year to date 2011 tracked by the fpds.gov feed, $267 billion have been mapped into input-output sectors and used as part of the Excel spreadsheet. As a note, 2010 federal spending data available totaled about

$500 billion. We anticipate 2011 results could be scaled up from those presented here to estimate annual totals. Alternatively the final 2011 federal spend data could be replaced in the existing tool.

TASK 1.3 – ENERGY AND ENVIRONMENTAL SATELLITE ACCOUNTS

Our IO-LCA model combined the economic IO data described in Task 1.1 with sector-level data for various energy and environmental effects. Satellite energy and environmental data are used to augment economic input-output models by finding appropriate data that exists at the level of aggregation of an economic sector (e.g., by NAICS code mapped into the required 428 IO sectors). Data are then normalized by the total output of each sector so that the economic IO model results (in dollars) can be combined with the satellite account data (in impacts per dollar) to estimate impacts for each sector http://www.fpds.gov/

Facilities Solutions Group, LLC 14 along the supply chain. Transforming such data from government agency reports into usable satellite accounts is often the most time consuming part of building EIO-LCA models. Fortunately in this project much of the required satellite data exists already.

The original data sources for the most prominent satellite accounts we used are from the U.S.

Department of Energy’s Manufacturing Energy Consumption Survey, the U.S. Census Bureau’s Annual

Survey of Manufactures and the U.S. Environmental Protection Agency‘s annual Inventory of

Greenhouse Gas Sources and Sinks, amongst others.

For the required satellite data, we used existing data from the 2002 CMU EIO-LCA model posted on the

Internet in all cases except for conventional pollutants which was created for this project. For the pre-existing EIO-LCA data, documentation on the underlying sources, allocation methods of data by sector, etc., were compiled on the EIO-LCA website as a single document (http://www.eiolca.net/docs/full-document-2002-042310.pdf) but have generally been used in previous peer-reviewed articles. In the case of several of the satellite accounts, e.g., land use, highly credible peer-reviewed journal papers were provided as the source of this data (Blackhurtst 2010).

We included data on hazardous waste generation and land use. Hazardous waste data transformations are in the EIO-LCA data document above. In the case of land use in hectares, we referenced a journal paper as the source (Costello 2011).

Satellite Account Type Data Sources Completed

Required

Greenhouse gases considered by the 2007 Intergovernmental Panel on Climate Change 4th Assessment Report (CO2, CH4, N20, HFC- 23, HFC-134, and SF6) in kilograms

EIO-LCA Documentation X

Criteria pollutants (O3, PM10, PM2.5, CO, NOx, SO2, and Pb) in kilograms

US EPA National Emissions Inventory Database

X

Toxic releases as defined by the U.S. EPA Toxics Release Inventory in kilograms

EIO-LCA Documentation X

Non-renewable energy (Coal, Natural gas, Petroleum) in Terajoules (TJ)

EIO-LCA Documentation X

Use of other energy resources (Biomass, Waste) in Terajoules

EIO-LCA Documentation X

Fossil-based and non-fossil based electricity use in TJ and converted to kilowatt-hours

EIO-LCA Documentation X

Water consumption (gallons, cubic meters) EIO-LCA, as published in Blackhurst (2010).

X

Optional

Hazardous Waste (RCRA Subtitle C) EIO-LCA Documentation X http://www.eiolca.net/docs/full-document-2002-042310.pdf http://www.eiolca.net/docs/full-document-2002-042310.pdf

Facilities Solutions Group, LLC 15

TASK 1.4 – CHARACTERIZATION FACTORS

To help in making more accessible presentations of the satellite account results, which can collectively be overwhelming, we used characterization factors from the latest version of TRACI (2.0) that was published as a peer-reviewed journal paper

(http://www.springerlink.com/content/y575g782233k57r7/) and publicly released in late 2010. TRACI

2.0 is a significant upgrade to the original TRACI impact assessment tool, which was released in 2002 and which has been used in several government agencies such as NIST and EPA for assessing life cycles and purchasing decisions. We note in our table below the required and optional TRACI factors that are used and the correct equivalence units from the latest TRACI 2.0 (the RFP noted the 9 year old TRACI units).

For the Excel spreadsheet models, we chemically converted several of the equivalence factors to more accessible units, such as kilograms of sulfur dioxide equivalents rather than moles of hydrogen equivalents for environmental impacts from acidification.

Factors Units Completed

Required

Global warming potential (GWP) kilograms CO2 equivalents X

Acidification potential kilograms H+ moles equivalents

X

Smog formation potential kilograms O3 equivalents X

Human health carcinogenic effects potential human cancer morbidity cases per kilogram emitted

X

Human health non-carcinogenic effects potential human non-cancer morbidity cases/ kilogram emitted

X

Human health respiratory effects potential kilograms PM10 equivalents

X

Optional

Eutrophication air potential kg N equivalents X

Eutrophication water potential kg N equivalents X

Ozone depletion potential kg CFC-11 equivalents X

Ecosystem toxicity potential potentially affected fraction of species (PAF)

X

TASK 1.5 – PRIORITIZATION

To generate the requested deliverable by GSA, we exported the resulting EIO-LCA results from Task 1.1 for $1 million of final demand into each of the 428 sectors for all satellite accounts and for the TRACI characterization factors. For each category in the federal spend we mapped its NAICS code with the http://www.springerlink.com/content/y575g782233k57r7/

Facilities Solutions Group, LLC 16 corresponding 2002 IO sector. We then multiplied the amount of spend in each category (in millions) by the per-million satellite and TRACI characterized results and export to an MS Excel Spreadsheet. This enabled us to rank the 428 IO sectors on both a per million dollar spent and total federal spending basis with respect to each of 55 environmental and energy impact indicators (the satellite accounts and TRACI impact characterization categories). These 55 indicators represent a substantial aggregation of many hundreds of pollutants, energy use, and land use and water use data series. However, evaluating 55 different rankings still remains a formidable problem for GSA. Thus, Task 6 below reports on our development of a prioritization tool that reduces the 55 indicators to a single vector indicator and then reports results for the specific indicators that drive the ranking for the single overall indicator.

TASK 1.6 – FUELS-USE PHASE

The model developed as described in Tasks 1.1-1.5 included the resource extraction, production, and transport to market impacts of federally purchased goods and services. This phase of goods and services life cycles is typically termed the “upstream” phase to distinguish these upstream impacts of making goods and services from the impacts of using those goods and services and then managing packaging and product discards when the packaging and products have served their useful purpose.

Task 1.6 provides the impacts of energy use during the use phase of the goods and services life cycle.

Completing assessments of the “use phase” of federally purchased fuels, e.g., for fleet vehicles, stationary combustion, etc., required the incorporation of process LCA data.

The RFP listed gasoline, diesel, natural gas, and coal as potential fuels to be studied. The natural gas may be used in CNG vehicles, for building space heating, cogeneration, etc. Thus to estimate fuel use phase impacts from federal spending on these fuels, representative specific technology pathways were assessed. Note that the federal spend database does not provide detail on purchases of specific fuels for use in specific combustion technologies.

Since there are hundreds of technology pathways, and thus relevant process LCA data sources (e.g., a natural gas turbine, a natural gas boiler with given efficiency, etc) for each fuel, we streamlined this effort to fit with the “scoping assessment” nature of this task. We mapped each of the fuel uses to several top technologies used for each fuel assumed by the federal government.

For example we mapped:

Federally purchased gasoline to process LCA data for (1) light trucks and (2) cars

Federally purchased diesel to process LCA data for (1) heavy trucks, (2) small and (3) medium diesel generators in the US

Federally purchased natural gas to (1) CNG vehicles, (2) small, (3) medium and (4) large industrial boilers

Federally purchased coal to (1) steam turbines (2) industrial boilers and (3) coal stoves

Facilities Solutions Group, LLC 17

In partnership with PE International, suppliers of the globally used GaBi life cycle inventory tool, we extracted data from the Ecoinvent database for the specific processes above.

As the federal spend gave only expenditure on fuels on a physical unit basis, to be consistent with the remainder of the IO-LCA model used we assumed price to physical unit conversions (e.g., $/gallon or

$/MJ of natural gas) based on average prices for the year 2011. Gasoline and diesel results were further averaged to generate an overall average of petroleum based fuels. The numbers vary by technology, and we created averages of the pathways by fuel for incorporation into the base input-output screening tool. The fuel end use data were only added to the following sectors of the model: coal mining, natural gas distribution, and petroleum refining (gasoline and diesel fuel average).

TASK 2--INCORPORATION OF PROCESS-LEVEL DATA

As discussed in Task 1, IO-LCA models can only act as first-order screening tools to help identify sources of large impacts or emissions. Therefore, a robust Federal Procurement Program must identify the

NAICS categories for which quality process LCA data is available for use in informing data uncertainties in the economic LCA data provided. While economic LCA and process LCA have been widely accepted and used since the ISO series of 14000 standards for products were approved in the early 21st Century, the interface of these two levels of LCA had not been integrated successfully at the time this contract was awarded.

In a practical sense, the FSG LLC assisted by Sphere E LLC has broke new ground by overcoming to some measure the challenge of finding ways to integrate process LCA data availability mapping document into the EIO-LCA Dashboard. We determined that mid-level process LCA data such as that found in industry average product LCAs and Environmental Product Declarations could be used most productively to inform GSA how to handle categories where the EIO-LCA Reports in Task One have identified considerable uncertainty about the range of one or more environmental impacts of a particular category.

We established an LCA mapping process structured to identify process LCA data that will show LCA data availability for the 129 categories selected, based on the magnitude of their annual spend by GSA and their related environmental impacts as identified in the Task One EIO-LCA Report and the level of interest of GSA in covering specific categories for products they purchase frequently. This is summarized in the Pivot Table illustration below.

Facilities Solutions Group, LLC 18

Lack of Independently Verified LCA Data from Manufacturers

Given the absence until recently of government regulation or strong private sector drivers for LCA data that has been verified as complete, credible and transparent by independent experts, it is not surprising that most manufacturers in the U. S. have either not engaged in developing such data or have chosen not to make it available to outside parties.

Now that the Federal Trade Commission is requiring publicly traded companies to disclose the environmental risks of the product and services they produce and/or use, there is increased pressure for all types of companies to develop and make available process LCA data.

Private sector drivers like the LEED Pilot Credit for building products selected with an ISO-compliant LCA or LCA-based Environmental Product Declaration (EPD) provide an incentive for developing and posting

LCAs and EPDs. Increasingly, U. S. marketplace expectations include independent verification that the

LCAs were developed in alignment with ISO standards 14040 and 14044, and that EPDs were developed in alignment with Standard 14025.

Conceptual Basis for Reviewing and Understanding Task Two

Facilities Solutions Group, LLC 19

“Informed environmental policy analysis requires life cycle assessment (LCA) of products and processes.

LCA is a systematic tool to estimate the environmental consequences of alternative materials, designs, manufacturing processes, product use impacts, and end of life alternatives. Specifically, LCA is a technique to assess the environmental aspects and potential impacts associated with a product, process, or service, by compiling an inventory of relevant energy and material inputs and environmental releases and connecting them to potential environmental impacts Interpreting the results to help decision-makers make a more informed decision.” Source: Report by Mary Ann Curran Systems Analysis

Branch National Risk Management Research Laboratory Cincinnati, Ohio, 2006.

LCA-based Information Is Best Understood as Series of LCA Levels

Best practices as they have evolved have led to the understanding that a number of different levels of

LCA data scope can be recognized, each serving a different yet somewhat related purpose. To think of any LCA data as comparable is to limit the use of LCA for decision making.

Description of the Various Levels of LCA Data &Their Respective Uses

The levels of LCA that the FSG-Sphere E Team sees as relevant to the work being conducted for GSA are:

Macro Level: EIO-LCA Data

Mid-Level: Process LCA Industry Average Data Availability

Micro Level- Process LCA Manufacturer Specific Data

Description of Data Mapping in Task Two Deliverables

While initially, the FSG-Sphere E Team thought we would provide actual data that could be used to inform the EIO-LCA Reports provided in the Task One Deliverables, after studying the categories in those reports we determined that it would of more value to GSA if we provided information on what LCA data was available and from which source for each of the 125 categories selected.

If instead we had attempted to provide the actual data, that would have been of limited usefulness to

GSA and other Federal Agencies as the data would have been on Level Three, data with such a different level of granularity from the EIO-LCA data that it could not have been used effectively. This strategy led to a focus on the NAICS code categories that resulted in 90% of the annual spend of GSA, and categories which overall had contributed to 90% of the GHG emissions.

This resulted in the first 63 categories selected to which we added the 66 categories in GSA’s “wish list’ of product categories that had been purchased frequently, totaling 129 categories in total.

For each of the 129 NAICS code categories, we provided the following information in our LCA Mapping

Excel Document, an Excel Document encompassing 129 data sheets:

NAICS Code

Facilities Solutions Group, LLC 20

Geographical region of data source

Reference year for data source

LCA provider- known quality LCA databases

Dataset name

Foreground data source

Link to database home page

Basis for use/Free or License

While not required, we developed a second "Key Materials Map" showing the data availability of typical raw material used, believing this additional list would empower GSA to invite manufacturers from the prioritized industries to develop process LCA data and provide it to all interested parties to communicate effectively the environmental performance of their products.

Uses of Task Two Deliverables by GSA and Other Federal Agencies

In its early days LCA was primarily used for product comparisons, for example to compare the environmental impacts of disposable and reusable products. Today its applications include government policy, strategic planning, marketing, consumer education, process improvement and product design. It is also used as the basis of eco-labeling and consumer education programs throughout the world.

Source- Life Cycle Assessment by M. Demmers and H. Lewis

Supporting GSA in Meeting Requirements of E. O. 13514 with Excellence

By using mid-level LCA data in the Task Two Report in combination with macro level data in the Task

One Report together as provided in the Dashboard, GSA will be positioned to determine where to best allocate their available resources (staff and funding) to reduce significantly GSA’s environmental footprint.

As both the EIO-LCA Report and the Process LCA Report now each focus on the same categories – ones resulting in 90% of GSA’s annual spend and the product’s with 90% of GHG emissions, GSA has a new resource to advance its ability to address efficiently requirements set forth in Section 13 of E. O. 13514 and to provide support to other Federal Agencies by providing this new resource to the Section 13

Workgroup comprised of many of the agencies it serves.

Prioritizing NAICS Code categories to reduce the Federal Footprint

GSA and other Federal Agencies will be able to use the Task Two LCA Map to move forward to meet a second goal of reducing the Federal Government’s footprint by prioritizing the categories on which to focus while developing sustainable acquisition strategies and adopting new policies to support these strategies.

By using the Task Two LCA Maps in combination with the Task One Report, GSA will be empowered to determine which industries they might most efficiently enroll in developing and making available

Facilities Solutions Group, LLC 21 independently verified process LCA data. Ideally, this would be those that already have invested in LCA studies.

It could also entail enrolling a number of industry organizations to provide education, tools and training on the development and use of LCAs and EPDs. This would support expansion of the dashboard by further integration of macro and mid level LCA data on some of the categories where currently there is no quality LCA data available as noted in the Task Two LCA Mapping Database.

Greening the supply chain by using the Materials Map

By using Task Two and Task One reports together, GSA staff can see where best to apply their available staff time and funding to advance sustainable acquisition by supporting manufacturers of selected categories to consider material selection procedures leading to environmentally more benign products.

The “Key Materials Map” provides GSA with an overview of what material related data sets are available for the selected databases. Such data sets describe the material flows connected to the provision of basic raw materials to the industrial system. While a number of the selected product categories have not yet been subject to LCA studies and no micro level data is available, this map shows that the basic data for carrying out such assessments is available.

In a further initiative to reduce the environmental footprint of the Federal Government, GSA might enroll those manufacturers that are using materials generally connected to high-level environmental impacts in working with their industry groups to enhance their product development by assessing their products’ resource and energy efficiency.

TASK 3--INCORPORATION OF CONSUMER-USE PHASE

While the results of Task 1 provide only “cradle to gate” estimates of life cycle impacts, the results of

Task 2 point to “cradle to grave” data where that is available in LCA-based eco-labels. We note that the majority of the items in the federal spend (by number, not impact) have no significant consumer use phase so this task was limited in scope. Obvious examples of energy-intensive procured goods are computers, light fixtures, and vehicles. Data on product energy use came from DOE surveys from the

“EndUse” laboratory at Lawrence Berkeley National Laboratory.

Task 3 thus explored the extent to which existing data and model results can consider use phase effects beyond those associated with direct use of fuels or electricity, e.g., from use of fuels or electricity during their lifetime.

Project members previously developed a model for the California Air Resources Board that incorporates use-phase energy for 25 products and followed a similar method in the contract for GSA. However, in discussions with GSA during this project there was insufficient data to reasonably explore the impacts of

Facilities Solutions Group, LLC 22 product use at the federal level. As an example, the federal spend data available lists only gross dollar amounts and not specific dollar amounts and/or purchases of numbers of pieces of equipment, and thus no reasonable estimate was possible to assess how many products (e.g., computers) might have been purchased in a given contract given the wide levels of aggregation. Further discussion of available federal data on such products is presented in the EOL section below.

This is not to suggest that use of products is not considered in the model. In fact, purchases of fuels and electricity are already tracked – however the connection of such uses back to the source products is not currently available.

TASK 4--INCORPORATION OF END-OF-LIFE

To indicate how end-of-life (EOL) product impacts could be incorporated into our LCA model and analyses we developed impacts estimates for a number of environmental categories for 4 EOL options for managing discarded desktop computers, laptop computers and televisions sets. This completes the life cycle model of these product categories in the federal spend by adding end-of-life (EOL) estimates.

The 4 EOL pathways examined are reuse, recycling, disposal via landfill either with or without use of collected landfill gas to generate energy, and disposal via incineration with energy recovery.

To illustrate the impacts of EOL management choices we obtained a sample of data on EOL choices for laptop computers. The sample encompassed 6,326 laptops with purchased values totaling $14.4 million, an average purchase price of $2,270. More than 75% of these laptops were auctioned, transferred to another agency or donated to a state. Assuming that these laptops were all put back into service after an average refurbishment cost of 10% of their original purchase value, their reuse represented a substantial avoidance of environmental impacts from production and distribution of new laptops that would have been incurred if a similar number of laptops had been purchased new rather than reused.

For example, the avoided energy use for production and retail distribution of new laptops amounts to over 48 terajoules. The avoided greenhouse gas emissions amount to nearly 3,200 metric tons of carbon dioxide equivalents. Avoided particulate emissions total over 5 metric tons of PM10 equivalents.

If these laptops had been broken down into their constituent materials and those materials recycled, the environmental benefits would have been smaller. For example, energy savings would have been under

1.5 terajoules, greenhouse gas reductions just under 74 metric tons, and avoided particulate emissions just over one fifth of a metric ton.

If these laptops had been landfilled or incinerated, even with energy recovery from the incinerated plastic materials in laptops, there would have either been an increase in environmental impacts in some cases or at least a much lower level of avoided environmental impacts. For example, there would have been avoidance of 0.02 terajoules of energy production in the case of waste-to-energy incineration or usage of 0.01 terajoules for landfilling; emissions of nearly 6 metric tons of greenhouse gases under

Facilities Solutions Group, LLC 23 waste-to-energy incineration and 0.6 metric tons under landfilling; and avoidance of .06 kilograms of atmospheric emissions of particulates for waste-to-energy disposal or emissions of .02 kilograms in the case of landfilling.

The importance of EOL pathways is also exemplified by a result from the study of the environmental impacts of consumer spending in Washington State (Morris and Matthews, 2010). That LCA study revealed that material recycling and composting rather than disposing of products and packaging at the end of their useful lives can reduce carbon emissions from spending, even when energy is recovered from the disposal process, by as much as reducing consumer spending on gasoline and diesel fuels by up to 40%. That is, while the overall EOL footprint from consumer spending is small in most cases compared with the footprint of product manufacturing and use, the reduction in environmental impacts from choosing best EOL management practices can be substantial.

TASK 5 – OTHER FEDERAL GOVERNMENT SEGMENTS

The Federal spend database that is used in this project already lists detail on procurement by the

Federal agency whose budget covers each purchase. Disaggregating this by the top 10 agencies (as requested in the RFP) or to all of the measured entities in the federal government merely represents a database re-sorting of the results that is provided inclusive through Task 4. However the query engine interface provided is able to select between about 60 department or agency level spending as directed by GSA (e.g., overall for a department like DoD, as well as drill down of just Navy expenses). It also allows summaries for all federal spending, all defense spending, and all non-defense spending. It was provided in the MS Excel Spreadsheet as a query report. Likewise, the Federal spend database provided detail on Multiple Award Schedules (MAS) and Government-wide Acquisition Contracts (GWACs). The exported results from the Tasks above were thus prioritized based on these two additional award and acquisition types.

TASK 6 – PRIORITIZATION BY OTHER ENVIRONMENTAL IMPACT CATEGORIES

As indicated for Task 1.5, our Excel model ranks the 428 IO sectors in terms of both one million dollars and total federal spending, as well as agency and contract type subsets of federal spending, for each of

55 energy and environmental impact indicators. In order to simplify analysis of so many different rankings, one could prioritize these upstream life cycle results by averaging the 55 rankings for each IO sector. This would tend to indicate which sectors ranked high across all impacts versus those sectors tending to rank much lower in energy and environmental impacts. However, the information that is lost when relying solely on simple rankings is the extent to which a sector with a higher ranking exceeds impacts for a sector with a lower ranking. In other words, is the difference between a sector ranking

Facilities Solutions Group, LLC 24 higher in global warming impacts versus one ranking lower, two pounds or two million pounds of carbon dioxide equivalents?