EPA/600/R-02/041
July 2002
Report on the International Workshop on
Electricity Data for Life Cycle Inventories
Held at the Breidenbach Research Center,
Cincinnati, Ohio, October 23-25, 2001
http ://www . sylvatica. com/Electricity Workshop .htm
Summary by
Mary Ann Curran,
US Environmental Protection Agency
Cincinnati, Ohio, 45268
Margaret Mann
National Renewable Energy Laboratory
Golden, Colorado, 80401
Gregory Norris
Sylvatica, Inc.
North Berwick, Maine, 03906
Workshop Co-Hosted by
National Risk Management Research Laboratory
US Environmental Protection Agency
Cincinnati, Ohio, 45268
and
National Renewable Energy Laboratory
US Department of Energy
Golden, Colorado 80401
National Risk Management Research Laboratory
^ Office of Research and Development
™ U.S. Environmental Protection Agency
National Renewable Energy Laboratory Cincinnati, Ohio, 45268
-------
Notice
The views expressed in these Proceedings are those of the individual authors and do not
necessarily reflect the views and policies of the U.S. Environmental Protection Agency (EPA).
Scientists in EPA's Office of Research and Development have prepared the EPA sections, and
those sections have been reviewed in accordance with EPA's peer and administrative review
policies and approved for presentation and publication. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use.
11
-------
Foreword
The U.S. Environmental Protection Agency is charged by Congress with protecting the
Nation's land, air, and water resources. Under a mandate of national environmental laws, the
Agency strives to formulate and implement actions leading to a compatible balance between
human activities and the ability of natural systems to support and nurture life. To meet this
mandate, EPA's research program is providing data and technical support for solving
environmental problems today and building a science knowledge base necessary to manage our
ecological resources wisely, understand how pollutants affect our health, and prevent or reduce
environmental risks in the future.
The National Risk Management Research Laboratory is the Agency's center for
investigation of technological and management approaches for reducing risks from threats to
human health and the environment. The focus of the Laboratory's research program is on
methods for the prevention and control of pollution to air, land, water and subsurface resources;
protection of water quality in public water systems; remediation of contaminated sites and
ground water; and prevention and control of indoor air pollution. The goal of this research effort
is to catalyze development and implementation of innovative, cost-effective environmental
technologies; develop scientific and engineering information needed by EPA to support
regulatory and policy decisions; and provide technical support and information transfer to ensure
effective implementation of environmental regulations and strategies.
This publication has been produced as part of the Laboratory's strategic long-term
research plan. It is published and made available by EPA's Office of Research and Development
to assist the user community and to link researchers with their clients.
E. Timothy Oppelt, Director
National Risk Management Research Laboratory
in
-------
Abstract
A three-day workshop was held in October 2001 to discuss life cycle inventory data for
electricity production. Electricity was selected as the topic for discussion since it features very
prominently in the LCA results for most product life cycles, yet there is no consistency in how
these data are calculated and presented. Approximately 40 people attended all or part of the
meeting to discuss issues of data modeling and collection. Attendees included recognized
experts in the electricity generation and life cycle assessment fields.
Five main topics of discussion were identified before the meeting began: 1) Modeling the
response of the energy supply system to demand (i.e. marginal v. average data); 2) Defining the
breadth and width of system boundaries to adequately capture environmental flows and data that
are needed for impact modeling; 3) Allocating environmental burdens across co-products that
come from the same process; 4) Modeling new and non-traditional technologies in which the
data are highly uncertain; and 5) Including transmission and distribution in modeling of
electricity generation. Breakout groups addressed the first four topic areas in individual
discussion groups and reported the results in a plenary session on the last day of the workshop (it
was decided during the meeting to include "transmission and distribution" in other discussions).
Several ideas were advanced by agreement in the break out groups' discussions, for example:
• The workgroup on marginal data made an important distinction in terminology by
defining "marginal," "attributional," and "consequential" modeling; it further
recommended that LCI databases be developed in such a way that they support both
attributional and consequential modeling, and it cited the need for case studies of
consequential modeling of the electricity system in order to shed light on many of the
current questions surrounding this rather new and unfamiliar approach in LCI.
• The workgroup on boundaries created a first cut at listing environmental emissions that
should be included in the inventory.
• The workgroup on new & non-traditional technologies noted that despite difficulties that
arise in conducting LCAs on renewables, due to uncertain operating data, any database
on electricity must be flexible enough to include different stressors.
• Access to unaggregated data was recognized as desirable by all the workgroups in order
to meet most of the data needs.
A key success of the workshop was the creation of the larger network of LCA and electricity
production experts that will provide a good foundation for continued discussions.
IV
-------
Contents
Foreword iii
Abstract iv
1. Introduction 1
1.1 Background 1
1.2 Workshop Attendees 2
1.3 Identifying the Issues 2
2. Summaries of the Discussions on the Issue Areas 4
2.1. Deliberations & Conclusions from the Breakout Group on
Marginal Versus Average Modeling 4
2.2 Deliberations & Conclusions from the Breakout Group on
Boundaries: Flows & Activities 7
2.3 Deliberations & Conclusions from the Breakout Group on
Co-Product Allocation 10
2.4 Deliberations & Conclusions from the Breakout Group on
New & Non-Traditional Technologies 12
3. Conclusions 15
References 15
Appendices 16
• Workshop Agenda 16
• Attendees 18
• Introduction and Overview ("Issues Paper") 22
• Pre-Workshop Responses 31
• Suggested Reading 59
• Summary of Feedback on Issues 60
• Response to Review Comments 81
• Presentation Slides and Material 94
-------
Exhibits
Exhibit 1. System Boundary for Energy Supply Systems 9
Exhibit 2. "Minimum" List of Environmental Flows for Energy Supply Systems 10
VI
-------
Chapter 1
Introduction
Data collection for life cycle inventories
(LCIs) remains a critical factor in the
successful completion of a life cycle
assessment (LCA). Access to reliable data
continues to be a significant barrier to the
advancement and use of LCAs in
environmental management.
Over the years, LCA practitioners have been
left to their own means to collect and model
inventory data as they have conducted
studies for clients. However, these data are
the property of the practitioner and not
typically made available to the public, or
they must be purchased. Furthermore, since
different modeling assumptions can be
made, there is no consistency in how these
data are calculated and presented in different
LCAs.
While most LCI data are specific to a
particular study and its goal, there are data
that are common in all LCIs, namely
electricity, transportation and waste
management. Electricity use, especially,
features very prominently in the total LCA
results for a majority of product life cycles.
Therefore, the benefits of public LCI data on
electricity generation would be high for
those who undertake LCAs and for those
who draw conclusions based on LCAs.
1.1 Background
Electricity is a major consideration in any
LCA. It is important to accurately calculate
and model resource use and pollutant
releases for activities related to the
generation and distribution of electricity,
such as how and where electricity is
produced, with what input requirements, and
with what pollution and waste cones-
quences. As LCAs are being conducted
more frequently as part of overall
environmental management approaches
within both the public and private sectors, it
is becoming increasingly important that LCI
data become more readily-available. Also it
is vital that data be used consistently
between LCAs in order to lead to more
fairly comparable results and reliable
conclusions.
Modeling of the environmental burdens of
electricity production is far from a simple or
straightforward task. Indeed, the electricity
supply system is among the most complex
of all the industries addressed in an LCA.
This complexity arises from a number of
factors, including:
• the broad geographic scope of power
grids and electricity markets with
power wheeling;
• the dynamics of supply dispatch in
response to demand changes,
overlaid on daily and seasonal
dynamics;
• the wide variation among generation
stations in emissions and inputs per
unit generation across and even
within fuel types;
• the rapid ongoing evolution and
regional variety of the electricity
system and the regulatory
environment in which it operates;
• the rapid and ongoing evolution of
electricity generation technologies,
and uncertainties about future market
penetration of new technologies, and
• the potentially long time frames and
importance of electricity consump-
tion for the life cycles of durable
products.
Existing LCI datasets generally fail to
capture the effects of these complexities. Of
course, all models must be simplifications of
reality to be useful, but the potential effects
-------
of these complexities upon the usefulness of
LCI results from current databases warrants
examination. Another priority issue for
resolution is the lack of consistency in scope
(both of environmental flows and
technospheric flows) among existing
databases for different regions, and even
among alternative databases covering the
same region.
In order to comprehensively address the
issues involved in modeling data for
electricity generation, it was decided to hold
a 3-day workshop with recognized experts in
the electricity generation and life cycle
assessment fields to work together to lead to
agreement that could be used in developing
a uniform/consistent electricity database for
life cycle inventories.
1.2 Workshop Attendees
Approximately 40 people attended all or part
of the meeting. The list of attendees is
located in the Appendix. The breakdown of
representation is approximately as follows:
Industry Experts*
Government Experts*
LCA Practitioners
LCA Researchers
Academia
OtherUSEPA'ers
10%
15%
20%
20%
17.5%
17.5%
* Experts in traditional & non-traditional electricity
generation.
1.3 Identifying the Issues
The following topical areas, referred to as
the "issues," were identified by the
workshop planners and used to organize the
presentations and discussions:
• Marginal v. Average: "Should LCA model
the response of the energy supply system
to demand?" (and consequences for co-
product allocation)
• Boundaries: "How wide and broad should
the boundaries be to capture environ-
mental flows and data that are needed for
impact? "
• New & Non-Traditional: " How should
LCA model new technologies, in which the
data are highly uncertain, and how should
increased demand for new technologies be
accounted for? "
• Co-Product Allocation: "How should
environmental burdens be allocated
across co-products that come from the
same process?"
• Transmission/Distribution: "How should
T&D impacts be included in modeling of
electricity generation? "
Prior to the workshop, a short summary
document, entitled "An International
Workshop on Electricity Data for Life Cycle
Inventories: Introduction and Overview
(August 2001)" was prepared to describe
these issues and what they mean. To help
initiate the thought process and stimulate
responses from the invitees to the meeting, a
series of 25 questions was also posed in the
summary document.
Input on these issues was solicited from
everyone who planned to attend the
workshop as well as those who were
interested in the effort but were unable to
attend. Around ten thoughtpieces and other
background material, such as journal
articles, were submitted. A summary
document was prepared from these
submittals and distributed to everyone
before the workshop ("International
Workshop on Electricity Data for Life Cycle
Inventories: Summary of Feedback on
Issues," 18 October 2001) as well as posted
on the website that was created expressly for
the workshop http://www.sylvatica.com/
El ectri city Workshop. htm.
-------
The first day of the workshop began with an
initial plenary session in which presentations
were made on data sources and a summary
of each issue area. This led to breakout
working groups that were tasked on the
second day of the workshop with discussing
the issues and identifying where there was
either consensus or disagreement.
Workgroup Members
Marginal versus Average Data:
John Abraham, US EPA
JohnBurckle, US EPA (retired), USA
Tomas Ekvall, Chalmers, Sweden
Bill Franklin, Franklin Associates, USA
Patrick Hofstetter, ORISE Post Doc, Switzerland
Greg Keoleian, University of Michigan, USA
Benoit Maurice, EDF, France
Greg Norris, Sylvatica, USA
Philippa Notten, University of Capetown, S. Africa
Scott Properzi, Energi D2, Denmark
John Sheehan, NREL, USA
Tom Tramm, Consultant, USA
Bo Weidema, 2.0 LCA Consultants, Denmark
New and Non-Traditional (NNT) Technologies:
Merwin Brown, NREL, USA
Joyce Cooper, University of Washington, USA
Rolf Frischknecht, ESU Services, Switzerland
Douglas Gyorke, NETL, USA
Marty Heller, University of Michigan, USA
Wolfram Krewitt, ITT, Germany
Ivars Licis, US EPA
LynnManfredo, SAIC, USA
Maggie Mann, NREL, USA
Jonathan Overly, University of Tennessee, USA
Boundaries & Co-Product Allocation:
Jane Bare, US EPA
Bill Barrett, NRC Post Doc, USA
Jamie Meil, Athena Institute, USA
Michael Overcash, North Carolina State University, USA
Bev Sauer, Franklin Associates, USA
Rita Schenck, IERE, USA
Caroline Setterwall, Vattenfall, Sweden
Tim Skone, SAIC, USA
Ray Smith, US EPA
-------
Chapter 2
Summaries of the Discussions on the Issue Areas
o Electricity LCI data for use in
Summaries of the discussions on Marginal
versus Average, Boundaries, Co-Product °
Allocation, and New & Non-Traditional
were presented to the group in plenary. The
originally-planned discussion on T&D was
folded into the discussions under both
Boundaries and New & Non-Traditional.
The sections below describe the workgroup
sessions.
other, general LCIs.
Using LCI to compare electricity
generation options.
2.1 Deliberations and Conclusions
from the Breakout Group on
Marginal versus Average Modeling
The group began by clarifying its objectives.
They were identified as follows:
• Clarify terminology, define the
meanings of key terms.
• Determine when attributional and
consequential LCI are each
appropriate.
• Characterize the feasibility of
attributional and of consequential
LCI as applied to electricity supply,
in terms of:
o Cost and time.
o Data availability, quality, and
uncertainty.
• Address the issue of clarifying how
the consequential approach might be
applied in practice, with what models
and data.
A fifth objective had initially been identified
for the group, but was never engaged by the
group as being particularly interesting,
important, or clear:
• Determine whether there are
different or equivalent answers to the
above four issues, depending on
whether one is addressing either of
the following two application areas:
Terminology
In defining and clarifying terminology, we
built on the contributions of Tomas Ekvall.
Decisions mean initial disturbances or
changes to some part of the LCI system.
Examples of decisions include whether to
locate a new factory in a given region, or
whether to install a high-efficiency device
rather than a standard-efficiency device, or
whether to pass more stringent building
codes or appliance efficiency standards.
Decisions lead to Consequences, through
whole series or chains of cause-effect
relationships. Other synonyms for
consequences include effects and outcomes.
Example consequences of interest here
would include emissions from electricity
generation, and investments in particular
new kinds of power generating capacity.
Both decisions and consequences can have
the properties of timing, duration, and
magnitude. It is magnitude which leads to
the definition of "marginal."
Marginal disturbances or perturbations are
infinitesimal disturbances; e.g., installing
one new end-use is a small but not an
infinitesimal disturbance. A marginal
disturbance is in theory infinitesimal, but in
practice it is small enough to be
approximated as an infinitesimal
disturbance. This requires that the response
to the disturbance be proportional to the
magnitude of the disturbance.
Marginal consequences are the response of
the system to a marginal disturbance. For
example, the marginal consequences of a
-------
very small increase in electricity demand
may include slight increase in air pollutant
emissions and fuel consumption.
The workgroup's discussion moved from
using the term "marginal versus average"
to "consequential versus attributional."
Prior authors have used terminology to
differentiate "marginal versus average" LCI,
and they have also labeled the options as
"retrospective versus prospective" LCI. The
breakout group determined that the central
distinction being considered by this breakout
group was one best described as
differentiating "attributional" versus
"consequential" LCI. Attributional and
consequential LCIs are modeling methods
which respond to different questions:
• Attributional LCIs attempt to answer
"how are things (pollutants,
resources, and exchanges among
processes) flowing within the chosen
temporal window?" while
• Consequential LCIs attempt to
answer "how will flows change in
response to decisions?"
Finally the group noted that retrospective
LCIs are LCIs about prior situations or
changes/decisions which occurred in the
past, while prospective LCIs are about
future situations or changes/decisions. An
LCI can therefore be prospective
attributional (how will things be flowing in
the future?), prospective consequential (how
will a future decision change flows?),
retrospective attributional (how were things
flowing in the past?) and retrospective
consequential (how did a prior decision
change the flows?).
When are attributional and consequential
LCI each appropriate?
The attributional approach to LCI serves to
allocate or attribute, to each product being
produced in the economy at a given point in
time, portions of the total pollution (and
resource consumption flows) occurring from
the economy as it is at a given point in time.
Thus, annual electricity production from
hydropower in the Pacific Northwest would
be assigned or attributed to each of the uses
of kWh of electricity occurring in the Pacific
Northwest during that same year.
The rules used to define which processes are
in or out of the system in attributional
modeling are those based on an observation
of how materials and energy are flowing in
the system at the given point in time. For
example, if concrete is made with 1 kg fly
ash and 1 kg Portland Cement per unit of
concrete output, then the LCI model will
show these flows into and out of the
concrete manufacturing process.
Note that the "given point in time" could be
past, present, or future.
The consequential approach to LCI attempts
to estimate how flows to and from the
environment will change as a result of
different potential decisions. In general, the
system response to changes in output
demand (e.g., increased or decreased
demand for some product) will vary between
the short- and long-term. In the short term,
the response will be changes in output from
existing production capacity (e.g., existing
power plants, factories, etc.) In the long
term, the response will be changes in the
timing, and perhaps the nature, of
investments in new production capacity.
The rules used to define which processes are
in or out of the system in consequential
modeling are those based on an estimation
of how material and energy flows will
change as a result of the potential decisions
or disturbances. In the fly ash example, if
the output of fly ash is constrained -
namely, if it is fixed based on the demand
for electricity - then increases in the demand
for high-fly-ash-concrete will not change the
output of fly ash in the short run. Instead, it
-------
would increase the output of concrete made
100% from Portland Cement. The
consequential LCI model would attempt to
take such output constraints explicitly into
account.
Characterizing the Response of the
Electric Utility System to Demand
Changes
Some members of the breakout group were
familiar with realities of how the electricity
system (at least in the US) currently
responds to changes in demand. Others
were familiar with responses of electricity
systems in Europe. From their input, the
following general facts were captured:
1) When the results over a year are
aggregated, the short-term output responses
to electricity demand changes typically
occur at plants that have the highest variable
cost among those operating at the time of the
demand change.1
2) In the long term, the type of new
capacity added is generally the one which is
estimated (by investment decision makers)
to satisfy the given load shape at the lowest
overall cost.
3) The future is irreducibly uncertain,
while the electricity supply system is
dynamic and evolving. Thus, there are
important levels of irreducible uncertainty
concerning how the electricity supply
system will respond to demand changes,
even if we used the most sophisticated
models available.
1 Note that on an hourly basis there are exceptions.
For example, hydropower is often dispatched to meet
daily peaks rather than base load. Hydro units
respond more reliably than more complex generating
options, so they are scheduled to come on to meet the
daily peaks or to address local environmental
concerns. However, limited water supply means that
there are only so many kilowatt-hours available per
year from a hydro unit, so by the end of the year,
demand changes accruing during the year will not
have affected the output from the hydro unit.
In addition, it is noted that in contrast with
many other products, electricity has the
specific characteristic that it cannot be
stocked directly. At any moment,
production must be equal the sum of
consumption and transmissions losses.
Throughout the day, the load shape varies
greatly due to increasing and decreasing use,
such as lighting at night. To produce
electricity, utilities typically have different
power plants which are able to adapt their
production to the consumption, producing
electricity as base load, (e.g., nuclear
energy), semi-base-load (e.g., coal, gas, fuel
power plant) and peak load (e.g., gas
turbine). This element has to be taken into
account when one tried to characterize the
response of the electric utility system to
demand changes. A "base load use" or a
"peak load use" will not have the same
answer. Rather than using simple
assumptions to characterize electricity
production, LCA practitioners should model
for electricity planning which allows for the
integration of such parameters.
Appropriateness and Feasibility of Each
Method
The participants agreed that, "ideally," LCA
results would inform decision makers about
the consequences of decision options that
they are evaluating. However, there
remained a significant level of concern
about switching from attributional to
consequential LCI modeling. Perhaps this is
because the participants had not, with only
two exceptions, ever undertaken or read the
results of a consequential LCA.
Group participants had the following
questions about consequential LCI:
• Does the change from attributional to
consequential LCA ("A-> C") affect
the results of the LCI? How much?
In what cases, i.e., which product
types, in which geographic regions?
• Does A—»• C alter LCA-based
decisions?
-------
• How easy will consequential LCA
results be to explain to users of the
results?
• How easy will consequential LCA be
to perform?
Recommendations
Based on its deliberations and concerns, the
breakout group concluded with the
following recommendations:
1) LCI databases should be developed in a
way that is technology-based, so that the
data can support either attributional or
consequential modeling. Specifically, they
should:
a) not aggregate over different
technology types within a sector
and
b) not aggregate over markets.
This will require solving issues around the
protection of confidential information, such
as is already faced by developers of
transparent LCI databases.
2) LCI databases should contain ample
meta-data, so that users can make informed
modeling decisions to use the data for either
attributional or consequential modeling.
3) Feasibility studies which apply energy
system models are needed in order to
generate short-, medium- and long-term LCI
results for a modest incremental change in
demand for different regions, and for
different types of end-use, which are
characterized by differences in timing (daily
and seasonal) and duration. Such studies
would provide answers to all four of the
questions posed by group participants about
currently unknown aspects of consequential
modeling of the electricity supply system.
2.2 Deliberations and Conclusions
from the Breakout Group on
Boundaries and Flows
LCA attempts to approximate the
comprehensive treatment of the
environmental, health and resource burdens
associated with product systems. In theory,
this comprehensiveness entails inclusion of
"all significant" burdens (e.g., pollution
releases, resource consumption flows, or
other impacts) from "all" causally-connected
processes. Thus, the system boundary for a
life cycle inventory model requires a series
of choices along two dimensions:
environment and supply chain. The purpose
of the Boundary and Flows Workgroup
(WG) was to discuss the following topics
related to assembly and handling of
electricity LCI data:
1. Which activities and operations
along the supply chain should be
included? That is, how wide and
how broad should the system
boundaries be drawn? (e.g., should
capital equipment be included?
transport of workers to the
production sites? service sector
inputs such as from designers,
lawyers, accountants, advertising,
etc.?)
2. Based on prior LCA and non-LCA
environmental evalua-tions of the
electricity supply system, is there a
set of environmental flows for which
reporting in LCI databases should be
required? Is it possible to define a
recommended set of environmental
flows that would be sufficient to
include in databases?
3. What is the most commonly
accepted system of nomen-clature
for environmental flows?
-------
The workgroup successfully addressed the
first two questions/topics, however, the
scope of the third question was determined
to be too broad and extensive to be covered
within the limited meeting time of the WG.
Boundaries
The participants evaluated which activities
and operations along the supply chain/life-
cycle should be included for energy supply
systems. Particularly, they discussed what
should be included (e.g., should capital
equipment be included? transportation of
workers to the production sites? service
sector inputs such as from designers,
lawyers, accountants, advertising, etc.?).
The consensus was to include infrastructure
only for dedicated resources. For example,
the material used to construct a boiler used
in a coal-fired utility plant should be
included, but the materials used to construct
the cranes that are used to erect boilers and
other plant structures would not be included.
Likewise, impacts from workers traveling to
and from work should be excluded. This is
not a hard-and-fast rule, but more a general
rule-of-thumb to be used in drawing
boundaries for energy supply systems. The
potential impact from infrastructure
operations should always be evaluated, even
on a cursory level, to support the exclusion
with confidence.
Taking a step back, the workgroup also
evaluated the main processes or activities
that should be considered when conducting
an LCI of any energy supply system. The
results of this effort are illustrated in Exhibit
1. Process or activities identified in Exhibit
1 for energy supply systems should not be
excluded without proper process knowledge.
If excluded, the corresponding rationale
should be documented in a transparent
manner and provided with the results of the
LCI. The specific nomenclature for each
process or activity identified in Exhibit 1
may vary from one practitioner to another,
but the intent of each box should be
evaluated for each LCI.
Environmental Flows
The Boundaries and Flows workgroup
evaluated the feasibility of a "default" or
"standard" list of environmental flows for
electricity supply systems. The consensus
of the workgroup and the Workshop
attendees was that a "default" list would
provide the perception that only those
environmental flows were of significant
concern and all others could be excluded.
This is not true. Our ability (as LCA
practitioners) to understand the impacts
from energy supply systems is based on
previous experience (past LCA work) and to
a greater extent, the availability of data to
model the energy supply system. Future
efforts to model the energy supply system
should not be limited to previous
experiences or perceived understandings of
significant environmental flows; rather,
every effort should be made to challenge the
validity and accuracy of scientific
knowledge upon which the conclusions
about energy supply systems are drawn.
Therefore, the workgroup rephrased the
question to ask "is there a minimum list of
environmental flows for energy supply
systems that one should expect to be
included in an LCI?" With some
apprehension, the workgroup developed a
tentative minimum list of environmental
flows to be considered for energy systems;
see Exhibit 2. Exclusion of these
environmental flows should raise concern
towards the comprehensiveness of the LCI
data set.
-------
Exhibit 1. System Boundary for Energy Supply Systems
On site
Resource
Extraction
Transport
Manufacturing ;
Water
Production &
Processing
Generation p-
t
Fleet Operations
Pollution
Control
Construction <
Demolition
"ransnuision -> Distribution
System Boundary
Distributed generation
-------
Exhibit 2. "Minimum" List of Environmental Flows for Energy Supply Systems
Resources
Water (location & type)
Fuel (in ground)
Minerals (in ground)
Biomass (harvested)
Land use (area & location)
Wastes
Solid waste
Radioactive Waste (H, M, L)
Hazardous Waste
Other Releases
radionuclides
Air Emission
CO,
CO
PM(10, 2.5)
CH4
SOX
NOX
NH3
Hg,Pb
VOC (MM)
Dioxin
PAHs
SF6
HFCs
Water Emissions
Chemical oxygen demand
(COD)*
IDS
Total suspended solids
(TSS)
Biological oxygen demand
(BOD) (5, 7, 10)*
Flow
Temperature change,** or
thermal loading in energy units
NH3 (as N)
Total Kjeldahl nitrogen
(TKN) (as N)
NO3, NO2 (as N)
Polycyclic aromatic
hydrocarbons (PAH's)
Phosphates (as P)
Cu, Ni, As, Cd, Cr, Pb, Hg
* COD and BOD are indicators of water quality rather than flows
** Limitation on temperature depends on the temperature of the river
Next Steps for Boundaries and Flows
Research
The workgroup identified the following next
steps to continue the progress made during
the electricity workshop.
1. Apply the system boundaries and
environmental flows guidance in the
development of the following model
energy supply systems:
a. Coal w/anthracite
b. Coal w/lignite
c. Natural Gas
d. Oil
e. Nuclear
f. Hydro
g. Wind
h. Biomass
i. Geothermal
j. Other
2. Research the potential impact from
the following, and other, non-
traditional environmental flows:
a. Noise
b. Radiation
c. Biological Resources
2.3 Deliberations and Conclusions
from the Breakout Group on Co-
Product Allocation
Co-product allocation arises as an issue
whenever a process produces more than one
useful product. For example, steam turbine
systems may sell both electricity and low
10
-------
pressure steam as useful products. When
co-products are present, practitioners must
determine how much of the burdens
associated with operating and supplying the
multi-output process should be allocated to
each co-product. Practitioners must also
decide how to allocate environmental
burdens across co-products when one is a
waste stream that can be sold for other uses.
The ISO standards for LCA, particularly
ISO 14041 on inventory analysis, provide
methodological guidance on this issue. But
they call for practitioners to attempt to avoid
allocation if possible; and secondly, to
attempt modeling approaches which reflect
the physical relationships between the
process outputs and its inputs. In summary,
proper application of the ISO guidelines on
allocation requires a physical understanding
of the co-product production processes. The
consensus of the workgroup was to follow
the guidance outlined in ISO 14041 for
energy supply systems. The following
highlights some key issues related to
allocation per ISO 14041.
ISO 14041 requires the following procedure
be used for allocation in multifunction
processes:
• Allocation should be avoided,
wherever possible, either through
division of the multifunction process
into sub-processes, and collection of
separate data for each sub-process, or
through expansion of the systems
investigated until the same functions
are delivered by all systems
compared.
• Where allocation cannot be avoided,
the allocation should reflect the
physical relationships between the
environmental burdens and the
functions, i.e., how the burdens are
changed by quantitative changes in
the functions delivered by the
system.
• Where such physical causal
relationships alone cannot be used as
the basis for allocation, the allocation
should reflect other relationships
between the environmental burdens
and the functions.
For allocation in open-loop recycling, ISO
14041 recommends the same procedure but
allows a few additional options. If the
recycling does not cause a change in the
inherent properties of the material, the
allocation may be avoided through
calculating the environmental burdens as if
the material was recycled back into the same
product. Otherwise, the allocation can be
based on physical properties, economic
value, or the number of subsequent uses of
the recycled material. The international
standard does not include information on the
effect of the different methods on the life
cycle modeling, for example the feasibility
of the methods, the amount of work
required, or what type of information that
results from the application of the methods.
A major point which came to light during
the workshop discussions on allocation was
that the choice of allocation method depends
considerably upon whether the LCA is being
performed from an attributional or a
consequential point of view. This point is
demonstrated in the dissertation and
publications of Tomas Ekvall. See, for
example, his 1997 paper with Ann-Marie
Tillman, published in the International
Journal of LCA1 In that paper, they very
helpfully differentiate cause-oriented from
effects-oriented bases for allocation, and
suggest that for LCAs supporting decisions
about the future (e.g., for consequential
LCAs), effects-oriented basis for allocation
is appropriate. System expansion is an
effect-oriented approach, while economic
allocation is a cause-oriented approach.
This issue of the relationship between
consequential/attributional LCA and the
choice of allocation method is also discussed
11
-------
together with a detailed presentation of
system expansion methods for allocation in
a 2001 paper by Bo Weidema in the Journal
of Industrial Ecology.2
2.4 Deliberations and Conclusions
from the Breakout Group on New
& Non-Traditional Technologies
(NNT)
In working on the question of how to
conduct LCAs of NNT technologies for
electricity generation, the group felt it
necessary for the purpose of this discussion,
to distinguish their role as database
developers and not LCA practitioners.
The goal of the database effort was
determined to be three-fold: 1) provide good
inventory data for each NNT generation
technology, 2) provide guidelines and/or
models that will help practitioners choose
the correct electricity mix, and subsequent
environmental stressors, for their product
life cycle assessment, and 3) ensure
consistency.
Scope of NNT generation
Areas of interest for NNT generation
include:
Future technologies
Renewables
Non-baseload generators
Distributed generation
Future technologies include those that have
the potential to someday contribute
significantly to the grid mix, but do not
currently influence the environmental
impact of common electricity usage. For
these technologies, there is limited operating
data, which is almost never site-specific.
While actual operating conditions are
difficult to predict with certainty, these
technologies are often viewed as being more
environmental benign. Future technologies
may include the second category,
renewables, but will also include generation
options such as fuel cells, microturbines, and
advanced coal.
Difficulties that arise in conducting LCAs
on renewables present challenges for LCI
database developers. For example, in LCI's
that are based on a functional unit of
producing a KWh, operating emissions may
be very low or essentially zero. The
predominant source of emissions associated
with the generating technology may be
construction emissions, which are
problematic to allocate over the functional
unit of KWh.
Additionally, because some impacts can be
very different than those from traditional
generators (e.g., bird kills), the database
must be flexible enough to include different
stressors. Finally, an important driver for
renewables is the avoidance of conventional
generation and impacts. Future discussions
on database development will need to agree
on how avoided impacts are handled.
Non-baseload generators are those that do
not produce power on a continuous or
controllable basis. Examples include some
renewable generators such as wind and
photovoltaics (PV), or power plants that are
used for providing peak energy. With
regard to database development, care must
be taken in data sets when referencing
stressors to a functional unit. The functional
unit can be defined either from the supply-
side as the kWhs that come from the
generator itself, or from the demand-side as
the kWhs that are consumed by the user of
the electricity.
The drivers for distributed generation are the
demand for reliable power, the desire to
avoid down-time costs, and the mitigation of
significant up-front capital expenditures for
large generators and transmission and
distribution infrastructure. Distributed
12
-------
generators (DGs) are typically small, and
may use fossil or renewable fuels. For a
large penetration into the grid, LCA
practitioners may take account, during
impact assessment, of the fact that emission
source locations are distributed over a large
geographic region as well. Additionally,
depending on the reason for a DG
installation, the functional unit may not be
kWh.
Focus of the discussion
In the course of the discussion on NNT
generation, four questions were answered:
• How are data sets constructed for
new technologies, for which there
are higher degrees of uncertainty in
environmental stressors?
• Is there a need to develop a common
future energy scenario that considers
renewable and distributed energy
sources for use in prospective LCAs?
• How should distributed generation
be accounted for in national or
regional energy grid data?
• What percent of the grid mix does a
technology have to supply before we
care about it in our product LCAs?
For the entire database, the group felt very
strongly that all data should be kept as
unaggregated as possible. That is, each set
of data should not represent the cradle-to-
gate inventory for the technology it is
describing. Rather, construction, mining,
transportation, and operation should be
provided in data modules such that a user
can separate them out.
Results of the discussion
The questions posed above were answered
as follows:
1. How are data sets constructed for
new technologies, for which there
are higher degrees of uncertainty
in environmental stressors?
• Use best available mass &
energy & production data.
• Where there are data gaps, make
a conservative expert judgment
for missing data points and
document assumptions (SETAC
working group)
• Include a calculation routine that
allows the users to vary
performance/emissions
parameters.
• Document assumptions, sources
of data, and year in which data
were obtained.
• Be alert to the situation where you
need to input stressors that are
not common to current
generation technologies (e.g.,
bird kill, land use).
Is there a need to develop a
common future energy scenario
that considers renewable and
distributed energy sources for use
in prospective LCAs?
• No. However, there is a need
to provide for the application
of various future energy
scenarios.
• Provide a tool or modules that
describe different energy
mixes/scenarios.
How should distributed generation
be accounted for in national or
regional energy grid data?
• The same way that traditional
generators are accounted for.
• Different transmission and
distribution losses are important.
• Stressors from non-baseload
generation should be discounted to
the percent of time that they
supply electricity to the consumer.
What percent of the grid mix does
a technology have to supply before
13
-------
we care about it in our product
LCAs?
• If you can assemble life cycle
inventory data for a technology,
provide it in the database.
• Use the module/tool described in
2) to give the user an opportunity
to incorporate them into their grid
mix, or they can do it manually.
In addition to the issues described above,
other concerns should be considered in
future related activities. Of key importance
is the incorrectness of using current data for
future technologies. Conclusions regarding
the environmental benefits that could be
achieved with future technologies would be
misguided when significant technological
advancement is possible. Similarly, while
the database is to contain inventory data for
the various technologies, LCAs conducted
for different timeframes will need to take
into account predictions of different grid
mixes.
14
-------
Chapter 3
Conclusions
The workshop successfully met its stated
goal to facilitate the exchange of ideas and
information. As was identified in the issues
paper and follow-up discussions, the
information needs to be established are too
numerous to be fully explored or resolved at
a brief three-day workshop. However, the
hard work of the break out groups lead to
many of the discussions points being
advanced.
The workgroup on marginal data made
an important distinction in terminology
by defining "marginal," "attributional,"
and "consequential." While there was
much discussion and many questions
remained unresolved, the group did
achieve consensus on the following
recommendations:
• LCI databases should be developed
in such a way that they support
both attributional and
consequential modeling.
• There is the strong need for case
studies of consequential modeling
of the electricity system in order to
shed light on many of the current
questions surrounding this rather
new and unfamiliar approach in
LCI.
• The workgroup on Boundaries created a
first cut at a "minimum list" of
environmental emissions that should be
included in the inventory.
• The workgroup on New & Non-
Traditional Technologies noted that
despite difficulties that arise in
conducting LCAs on renewable
generating technologies, due to uncertain
operating data, any database on
electricity must be flexible enough to
include non-traditional stressors (e.g.,
bird kills).
A consistent thread throughout all the
conversations was the desire for having
access to unaggregated data, although the
practicalities involved in this, such as
confidentiality issues, were not discussed.
A key success of the workshop was the
network that was created among experts in
the LCA and electricity production fields.
The establishment of this larger workgroup
will provide a good foundation for continued
discussions.
The workshop conveners are exploring next
steps, and encourage all workshop
participants as well as other interested
parties to please use the workshop website
as a repository for documents, thought
pieces, and links which relate directly to the
topics discussed at the workshop and
summarized in this document.
References
[1] Ekvall T, and Tillman A-M, "Open-Loop
Recycling: Criteria for Allocation
Procedures," Int Journal of LCA , 2(3),
pp!55-162, 1997.
[2] Weidema B., "Avoiding Co-Product
Allocation in Life-Cycle Assessment,"
Journal of Industrial Ecology, 4(3),
ppll-33, 2001.
15
-------
Electricity Data for Life Cycle Inventories
A.W. Breidenbach Research Center
26 W. Martin Luther King Drive
Cincinnati, Ohio, USA 45268
October 23 - 25, 2001
Agenda
L
Tuesday, October 23rd
9:00 - 9:30 Registration (there is no fee to attend this workshop)
9:30 - 12:00 Plenary 1 (Auditorium) Opening Remarks
9:30-10:00 Welcome and Objective of the Workshop, with
Basics ofLCA Practice
Mary Ann Curran, LCA Researcher, US EPA/ORD
10:00- 10:30 Uses and Users of Electricity LCA Information
Bo Weidema (invited)
10:30- 11:45 Electricity Data Sources
Asia-A. Inaba, NIAIST
Europe - W. Krewitt, DLR-ITT
Southern Countries - A. Quiros (invited), EcoGlobal
US - R. Morgan, US EPA
11:45 - 12:00 Plan for Conducting the Workshop
12:00 - 13:00 Lunch (on your own)
13:00 - 15:00 Plenary 2 (Auditorium) Presentation and Discussion of Issues
13:00 - 13:45 Boundaries:
"How wide and broad should the boundaries be to capture environmental
flows and data that are needed for impact? "
Gregory Norris, Sylvatica and Patrick Hofstetter, ORISE
13:45 - 14:30 Marginal v. Average:
"Should LCA model the energy supply system's response to demand? "
(and consequences for co-product allocation)
Tomas Ekvall, Chalmers
14:30-15:00 New & Non-Traditional:
"How should LCA model new technologies, in which the data are highly
uncertain, and how should increased demand for new technologies be
accounted for?"
16
-------
Margaret Mann, National Renewable Energy Laboratory, DOE
15:00 Introduction of Discussion Groups and Room Assignments
18:00 Dinner with Discussion Groups
I
Wednesday, October 24
8:00 -12:00 Resume Discussion Group Meetings:
Parallel groups on the following topics:
Boundaries; New & Non-Traditional, and Marginal v. Average
12:00 - 13:00 Lunch (on your own)
13:00 - 17:00 Making Progress with the Discussions
13:00 - 14:00 Preliminary Report Outs (20 minutes each)
14:00 - 17:00 Continue Original Discussion Groups
Introduction of 2 New Parallel Groups:
Co-Product Allocation & Transmission/Distribution
18:00 Dinner
L
Thursday, October 25
8:00 - 13:30 Plenary 3 (Auditorium) Presentation of Discussion Summaries
8:00-9:00 Boundaries
9:00-10:00 New & Non-Traditional
10:00 - 10:15 Coffee Break
10:15-11:15 Marginal v. Average
11:15-11:45 Allocation
11:45 - 12:15 Transmission and Distribution
12:30 - 13:30 Brown-bag Lunch and Final Discussion, Plans, Next Steps
13:30 Adjourn
17
-------
International Workshop on Electricity Data for Life Cycle Inventories
FINAL LIST OF ATTENDEES
October 23-25, 2001
Cincinnati, Ohio, USA
John P. Abraham
Phone: 513-569-7124
Fax:513-569-7111
E-mail: abraham.john@epa.gov
Jane Bare
Phone: 513-569-7513
Fax:513-569-7111
E-mail: bare.jane@epa.gov
Bill Barrett
Phone: 513-569-7220
Fax:513-569-7471
E-mail: barrett.william@epa.gov
Merwin Brown
Phone: 303-275-4364
Fax: 303-275-3059
E-mail: merwin_brown@nrel.gov
John Burckle
Phone: 513-569-7496
Fax:513-569-7471
E-mail: burckle.john@epa.gov
Joyce Cooper
Phone: 206-543-5040
Fax: 206-685-8047
E-mail: cooperjs@u.washington.edu
Mary Ann Curran
Phone: 513-569-7782
Fax:513-569-7111
E-mail: curran.maryann@epa.gov
Tomas Ekvall, Ph.D.
Phone: +46-31-772 1445
Fax: +46-31 -772 35 92
E-mail: tomas.ekvall@entek.chalmers.se
Gordon Evans
Phone: 513-569-7684
Fax:513-569-7111
E-mail: evans.gordon@epa.gov
William E. Franklin
Phone: 913-649-2225 Ext. 232
Fax:913-649-64944121
E-mail: wfranklin@fal.com
Rolf Frischknecht
Phone: +41 1 940 61 91
Fax: +41 1 940 61 94
E-mail: frischknecht@esu-services.ch
Douglas Grosse
Phone: 513-569-7844
Fax:513-569-7585
E-mail: grosse.douglas@epa.gov
US EPA, National Risk Management Research Laboratory
26 W. Martin Luther King Drive, MS-466
Cincinnati, OH 45268
USA
US EPA, National Risk Management Research Laboratory
26 W. Martin Luther King Drive, MS-466
Cincinnati, OH 45268
USA
US EPA, National Risk Management Research Laboratory
26 W. Martin Luther King Drive
Cincinnati, OH 45268
USA
National Renewable Energy Laboratory
1617 Cole Blvd.
Golden, CO 80401
USA
US EPA, National Risk Management Research Laboratory
26 W. Martin Luther King Drive
Cincinnati, OH 45268
USA
University of Washington
Seattle, WA 98195
Box 352600
USA
US EPA, National Risk Management Research Laboratory
26 W. Martin Luther King Drive, MS-466
Cincinnati, OH 45268
USA
Chalmers University of Technology
Energy Systems Technology Group
Sweden
US EPA, National Risk Management Research Laboratory
26 W. Martin Luther King Drive, MS-466
Cincinnati, OH 45268
USA
Franklin Associates, Ltd.
West 83rd Street, Suite 108
Prairie Village, KS 66208
USA
ESU-Services
Kanzleistrasse 4
Switzerland
US EPA, National Risk Management Research Laboratory
26 W. Martin Luther King Drive
Cincinnati, OH 45268
USA
18
-------
Douglas F. Gyorke
Phone: 412-386-6173
Fax: 412-386-4775
E-mail: douglas.gyorke@netl.doe.gov
Paul Marten
Phone: 513-569-7045
Fax:513-569-7471
E-mail: harten.paul@epa.gov
Martin Heller
Phone: 734-764-2637
Fax: 734-647-5841
E-mail: mcheller@umich.edu
Abby Hill
Phone: 513-569-7100
Fax:513-569-7585
E-mail: hill.abby@epa.gov
Patrick Hofstetter
Phone: 513-569-7326
Fax:513-569-7111
E-mail: hofstetter.patrick@epa.gov
George Huffman
Phone: 513-569-7431
Fax:513-569-7471
E-mail: huffman.george@epa.gov
Gregory A. Keoleian
Phone:734-764-3194
Fax: 734-647-5841430
E-mail: gregak@umich.edu
Dr. Wolfram Krewitt
Phone: +49-711-6862766
Fax: +49-711-6862783
E-mail: wolfram.krewitt@dlr.de
C.C. Lee
Phone: 513-569-7520
Fax:513-569-7471
E-mail: lee.chung@epa.gov
Ivars Licis
Phone:513-569-7718
Fax:513-569-7471
E-mail: licis.ivars@epa.gov
Lynn Manfredo
Phone: 412-386-6839
Fax:412-386-4516
E-mail: manfredo@netl.doe.gov
Margaret Mann
Phone: 303-275-2921
Fax: 303-275-2905
E-mail: margaret_mann@nrel.gov
Benoit Maurice
Phone: 33 1 60 73 74 38
Fax: 33 1 60 73 75 27
E-mail: Benoit.Maurice@edf.fr
Jamie Meil
Phone: 613-722-8075
Fax:613-722-962828
E-mail: jkmeil@fox.nstn.ca
USDOE / National Energy Technology Laboratory
P.O. Box 10940
Pittsburgh, PA 15236
USA
US EPA, National Risk Management Research Laboratory
26 W. Martin Luther King Drive
Cincinnati, OH 45268
USA
Center for Sustainable Systems / University of Michigan
430 East University
Ann Arbor, Ml 48109-1115
USA
US EPA, National Risk Management Research Laboratory
26 W. Martin Luther King Drive
Cincinnati, OH 45268
USA
ORISE Research Fellow at the US EPA
26 W. Martin Luther King Drive, MS-466
Cincinnati, OH 45268
USA
US EPA, National Risk Management Research Laboratory
26 W. Martin Luther King Drive
Cincinnati, OH 45268
USA
Center for Sustainable Systems / University of Michigan
East University, Dana Building
Ann Arbor, Ml 48109-1115
USA
German Aerospace Center (DLR), Inst. of Tech. Thermodynamics
System Analysis and Technology Assessment
Pfaffenwaldring 38-40
Germany
US EPA, National Risk Management Research Laboratory
26 W. Martin Luther King Drive
Cincinnati, OH 45268
USA
US EPA, National Risk Management Research Laboratory
26 W. Martin Luther King Drive
Cincinnati, OH 45268
USA
SAIC
P.O. Box 18689
Pittsburgh, PA 15236
USA
National Renewable Energy Laboratory
1617 Cole Blvd., MS-1613
Arvada, CO 80401
USA
Electricite de France - R&D Division
Site des Ranardieres. Route de Sens - Ecuelles
France
Athena Sustainable Materials Institute
St. John Street, P.O. Box 189
Merrickville , Ontario KOG 1 NO
Canada
19
-------
Rick Morgan
Phone: 202-564-9143
Fax: 202-565-2156
E-mail: morgan.rick@epa.gov
Gregory A. Norris, Ph.D.
Phone: 207-676-7640
Fax:207-676-7647147
E-mail: norris@sylvatica.com
Philippa Notten
Phone: 202-362-6357
Fax: 202-362-6357
E-mail: pnotten@netzero.net
Michael Overcash
Phone:919-515-2325
Fax:919-515-3465
E-mail: overcash@eos.ncsu.edu
Jonathan Overly
Phone: 865-974-3625
Fax:865-974-1838311
E-mail: jgoverly@utk.edu
Frank Princiotta
Phone: 513-569-7391
Fax:513-569-7680
E-mail: princiotta.frank @epa.gov
Scott Properzi
Phone: +4544806614
Fax: +45 4480 6604
E-mail: spx@e2.dk
Beverly J. Sauer
Phone: 913-649-2225 Ext. 228
Fax:913-649-6494
E-mail: bsauer@fal.com
Rita Schenck
Phone: 206-463-7430
Fax:206-279-157019001
E-mail: rita@iere.org
EPA/OAR/OAP - Climate Protection Partnerships Division
1200 Pennsylvania Ave., NW (6202J)
Washington, DC 20460
USA
Sylvatica
Bauneg Hill Road, Suite 200
North Berwick, ME 03906
USA
University of Cape Town, South Africa
5415 Connecticut Ave. #140
Washington, DC 20015
USA
North Carolina State University
Department of Chemical Engineering, Box 7905
Raleigh, NC 27695-7905
USA
The Center for Clean Products and Clean Technologies at the University of
Tennessee
Conference Center Bldg.
Knoxville, TN 37996-4134
USA
US EPA, National Risk Management Research Laboratory
26 W. Martin Luther King Drive
Cincinnati, OH 45268
USA
ENERGIE2
Lautruphoj 5
Denmark
Franklin Associates, Ltd.
4121 West 83rd Street, Suite 108
Prairie Village, KS 66208
USA
Institute for Environmental Research and Education
Vashon Hwy SW
Vashon, WA 98070
USA
Caroline Setterwall
Phone:+46 (0)223 55103
Fax:
E-mail: caroline.setterwall@swedpower.vattenfall.se
SwedPower AB within Vattenfall AB
Nedre Lasarbo 5
Sweden
John Sheehan
Phone: 303-384-6136
Fax: 303-384-6877
E-mail: john_sheehan@nrel.gov
Timothy Skone, PE
Phone: 703-318-4604
Fax: 703-736-0826
E-mail: skonet@saic.com
Ray Smith
Phone:513-569-7161
Fax:513-569-7111
E-mail: smith.raymond@epa.gov
Tom Tramm
Phone: 847-256-3553
Fax:
E-mail: t.tramm@att.net
National Renewable Energy Laboratory
1617 Cole Blvd.
Golden, CO 80401
USA
SAIC
11251 Roger Bacon Drive
Reston, VA20190
USA
US EPA, National Risk Management Research Laboratory
26 W. Martin Luther King Drive, MS-466
Cincinnati, OH 45268
USA
Consultant
518 Washington Ave.
Wilmette, IL 60091
USA
20
-------
Dr. Bo Weidema 2.-0 LCA Consultants
Phone: +45 333 22822 Borgergade 6, 1.
Fax:+45 339 11103
E-mail: bow@lca-net.com Denmark
21
-------
An International Workshop on
Electricity Data for Life Cycle Inventories
Introduction and Overview
Background:
The environmental consequences of electrical energy production frequently account for a major
portion of the total environmental burdens identified in product Life Cycle Assessments, across a
broad variety of product types, and across a range of impact categories. Therefore, accurate,
complete, and up-to-date LCA information and data on electricity production is vital.
Modeling of the environmental burdens of electricity production is far from a simple or
straightforward task. Indeed, the electricity supply system is among the most complex of all the
industries addressed in an LCA. This complexity arises from a number of factors, including:
• the broad geographic scope of power grids and electricity markets with power wheeling.
• the dynamics of supply dispatch in response to demand changes, overlaid on daily and
seasonal dynamics;
• the wide variation among generation stations in emissions and inputs per unit generation
across and even within fuel types;
• the rapid ongoing evolution and regional variety of the electricity system and the
regulatory environment in which it operates;
• the rapid and ongoing evolution of electricity generation technologies, and uncertainties
about their future market penetration, and
• the potentially long time frames and importance of electricity consumption for the life
cycles of durable products.
In the face of the increasing globalization of supply chains, we note at least two major limitations
of currently available LCA data on electricity production:
• limited geographic coverage
• lack of consistency among databases for different regions, and even among alternative
databases covering the same region
This lack of consistency among existing databases would be of only minor concern if there were
a current consensus of approach to LCA modeling of electricity supply systems, and widespread
effort to develop consistent data for all regions of the globe. Our assessment of the present
situation is that there is neither. Fortunately, there appears to be growing recognition of the need
for both.
22
-------
Objectives of the Workshop
The purpose of the USEPA/NREL International Workshop on Electricity Data for LCIs is to
make progress in response to the needs identified above, by providing a forum for:
• exchanging information on the state-of-the-practice of collecting and reporting electricity
life cycle inventory data, and
• identifying technical assumptions and their ramifications in collecting inventory data.
In particular, the workshop is intended to facilitate the exchange of ideas and information on
three fronts: methodological issues, ongoing and planned LCA data development efforts with an
electricity component, and identification of best sources of data and models from which to build
LCA models of electricity supply.
Format of the Workshop
A workshop of 2 and a half days is planned, for the dates 23-25 October, 2001. The location will
be EPA's Andrew W. Breidenbach Research Center, Cincinnati, Ohio, USA. The workshop will
move from initial plenary sessions to breakout working groups. The plenary sessions are
intended to provide all attendees with a common basis of understanding of the life cycle
perspective, the issues that are involved in collecting electricity data, and the specific discussion
topics for the workshop. Workgroups will flesh out specific issues in more detail and work
toward consensus. The topic areas planned for the workgroups are as follows:
• Average versus Marginal Systems Modeling
• Boundaries for Electricity Generation Systems
o Environmental flows and releases
o System definition
• New and Non-Traditional Electricity Generation
• Transmission and Distribution
• Outputs and Co-Product Allocation
Summary of Workgroup Topic Areas
Average versus Marginal Systems Modeling
Current LCA modeling represents an allocation of the total environmental burdens of a macro-
system (e.g., today's economy) to the life cycles of individual products and services. All such
LCAs are structured so that, theoretically, the results could be combined to form a total response.
The goal is to answer the question: "If we were to assign the total environmental burdens caused
by global demand for goods and services across all components of that demand, how much
burden would we assign to each unit of good or service?" Heijungs (1997) referred to this
question as "the attribution problem." Thus, for electricity generation, LCAs assign, or
apportion, the burdens of a region's annual generation equally across each kWh of electricity
produced and consumed. Thus, if the annual generation for a region comes from equal shares of
particular energy sources, for example, hydro, nuclear, and fossil fuel prime movers, then each
23
-------
kWh produced and used in this region would be modeled as an "average kWh," produced from
1/3 hydro, 1/3 nuclear, and 1/3 fossil fuel. This is the approach taken in attributional LCA
modeling.
An evolution is taking place within the field of Life Cycle Assessment, away from models of
"average" systems which support retrospective analyses, towards models of "marginal" systems
which support prospective analyses. In contrast to attributional LCAs, prospective LCAs
explicitly attempt to characterize what the impacts will be of potential decisions. Thus, they are
designed to provide insight about "what will happen if we decide A or B," rather than "which
product is to blame for which burdens."
The processes whose levels of output will be impacted by a decision or a change in demand are
referred to as the "marginal" processes - those producing "at the margin." .
At first blush, the marginal modeling underlying prospective LCA may appear more complex or
data-intensive than the average modeling underlying attributional LCA. In practice, this is not
necessarily the case, and in fact prospective LCA helps take some of the arbitrariness out of
thorny LCI modeling issues such as allocation (Weidema 2001). In most if not all cases, the use
of LCA for decision support appears to call for adopting the prospective approach as far as
possible.
A number of inter-related questions arise in attempting to identify how the energy supply system
actually responds to changes in demand, depending upon characteristics of the demand change
including its location, timing, duration, and magnitude. In order to provide a sound basis for
prospective modeling of the electricity supply system, we must address the following questions:
• What do we know about how the electricity supply system responds to changes in
demand?
• How is this system's response to demand currently modeled, by what models and with
what accuracy?
• How should LCA incorporate these understandings and perhaps the results of these
models in its treatment of electricity?
Boundaries for the Electricity Generation Systems
LCAs attempt to approximate comprehensive treatment of the environmental, health and
resource burdens associated with product systems. In theory, this comprehensiveness entails
inclusion of "all significant" burdens (e.g., pollution releases, resource consumption flows, or
other impacts) of "all" causally-connected processes. Thus, the system boundary for a life cycle
inventory model requires a series of choices along two dimensions: environment and supply
chain. In the case of a life cycle inventory database concerning the electricity supply system, we
note the following boundary decisions which must be made:
24
-------
1. Which environmental flows and other data needed for impact modeling should be
tracked, how, and with what specificity, for processes in the electricity supply system?
How should the cut-off criteria be determined?
2. Which activities and operations along the supply chain should be included? That is how
wide and how broad should the system boundaries been drawn? (e.g., should capital
equipment be included? transport of workers to the production sites? service sector
inputs such as from designers, lawyers, accountants, advertising, etc.?)
Decisions related to establishing specific cut-off criteria to set boundaries for particular processes
in the system under study, are properly left to the goal and scope definition portions of individual
life cycle assessments, or to the protocol development phase of the LCI database projects. This
workshop will seek to pool insights from prior and current LCAs of electricity systems
concerning the broader boundary questions of what sorts of flows and what sorts of processes are
important to retain in general when modeling the electricity supply system
This workgroup will address the multiple 'what is in, what is out?" sorts of questions which are
fundamental to life cycle inventory analysis. The workgroup will address its topics in a pair of
sequential sessions.
The first session on boundaries will address environmental emissions and releases. Key
questions include:
• Based on prior LCA and non-LCA environmental evaluations of the electricity supply
system, is there a set of air emissions for which reporting in LCI databases should be
required ? Is it possible to define a recommended set of air emissions which it would be
sufficient to include in databases? What are the principal data sources for the key air
emissions, and are there important differences among them from country to country?
• Water releases - the same set of questions as posed above for air emissions.
• Additional releases (e.g., radioactive isotopes) - the same set of questions.
• Other impacts (e.g., thermal enrichment of water, land use, etc.) - the same set of
questions.
A second work session will address setting the system boundaries which will be used to
determine which mass and energy flows will be accounted for. Key questions include:
• What inputs besides fuels are essential/important to include, for different types of
generation?
• How have input/output-based LCA analyses been used in the past to shed light on this
question, and what have their findings been?
• What is the significance and suggested treatment of maintenance and repair inputs?
• What is the significance and suggested treatment of supporting infrastructure?
25
-------
New and Non-Traditional Electricity Generation
As mentioned in the introduction, the electricity supply system is dynamic, with old technologies
being slowly replaced by new. As interest in minimizing the environmental impacts of
electricity generation increases, so will the ongoing development and evaluation of innovative
electricity supply technologies. One arena of potentially influential use of LCAs of energy
systems may be in environmentally evaluating and comparing new generation technologies.
Characterizing them for LCA poses a whole new and different set of data and modeling issues.
There are at least three inter-related sets of issues involving LCAs of new and non-traditional
generation. The first is simply how to model the new technologies in LCA. For new
technologies which simply replace other point source generation facilities, this may not be a
challenge. But how shall LCA characterize distributed generation, whether from the average
perspective and from the marginal perspective?
The second set of issues relates to comparative evaluation of the new technologies, such as fuel
cells, from the LCA perspective. LCA evaluations of nascent electrical generation technologies
may inform policy and/or research prioritization among competing options. How can LCAs be
performed in a consistent, holistic, and valid fashion for these systems which are marked by high
degrees of uncertainty and technological volatility, as well as scarcity of data?
The third set of issues relates to the way in which LCAs of product life cycles will tend to treat
new generation, and potentially to influence the demand for new capacity. The treatment of in-
place capacity will probably need to be considered separately from the treatment of demand
which drives new capacity. An example concerns the proper treatment of flow-limited
renewable energy, such as wind power capacity in place. From a prospective point of view, no
change in product demand (whether increase or decrease) will change the amount of electricity
generated by wind power capacity in place - its output is fixed by nature. So how, if at all,
should this wind capacity appear in the results of a prospective LCA.
Transmission and Distribution
The transmission and distribution infrastructure component of the electricity supply system has
traditionally been accounted for in LCA on in terms of the expected average line losses, or loss
of power due to electrical resistance in the system connecting the point of generation to the point
of use. The amount of this loss depends on the length of the transmission, the voltage at which
transmission occurs, the size of the conductor, and the manner in which electricity is transmitted.
Common losses range between two to five percent of power being transmitted (EIRRG 1998).
LCA researchers from Asian Pacific countries have identified additional issues associated with
26
-------
what might be termed "fugitive losses" of electric power, which is un-metered or un-identified
electricity consumption.
In fact, there may be important reasons other than line power losses to include the transmission
and distribution network within the scope of LCAs of the electricity supply system - namely, the
environmental impacts of constructing, maintaining, and operating the systems themselves.
Some environmental concerns raised in connection with electricity transmission and distribution
lines include visual impacts, habitat impacts, noise (from high-voltage and ultra-high-voltage
transmission), and others (e.g., any remaining concern about effects of electrical and magnetic
fields?).
This work group will address both the energy losses associated with transmission and
distribution, as well as the impacts of T&D infrastructure itself.
Outputs and Co-Product Allocation
Co-product allocation arises as an issue whenever a process produces more than one useful
product. For example, steam turbine systems may sell both electricity and low pressure steam as
useful products. When co-products are present, modelers must determine how much of the
burdens associated with operating and supplying the multi-output process should be allocated to
each of co-product. Modelers must also decide on how to allocate environmental burden across
co-products when one is a waste stream that can be sold for other uses.
The ISO standards for LCA, particularly ISO 14041 related to inventory analysis, provide
methodological guidance on this issue. But they call for practitioners to attempt to avoid
allocation if possible; and secondly, to attempt modeling approaches which reflect the physical
realities (i.e. mass basis) of the process in terms of how inputs and releases would be altered if
the levels of output were altered for one or more co-products. In summary, proper application of
the ISO guidelines on allocation requires a physical understanding of the co-product production
processes.
The workgroup on co-products and allocation for the electricity supply sector could provide
considerable value to the worldwide LCA community by providing clarity and consensus on
allocation rules. It could also help by pointing to the data sources which characterize the
geographic details of which plants and plant types in which regions are producing how much of
the economically valued co-products; such information will assist in assessing transportation
distances for other LCAs which include the use of these co-products.
Pre-Workshop Activities
Workshop invitees (those who are able to participate in person as well as all others who are
invited but cannot attend the workshop) are asked to submit their thoughts on the five topic
areas. The workshop organizers will review the submissions and compile the results for
27
-------
distribution before the workshop. Also, suggestions for on-line resources and documents
relevant to the upcoming discussions are requested. Submittals are needed no later than
September 21,2001
The value of the workshop discussions will be greatly increased to the extent that participants are
able to inform themselves about the issues ahead of time, and to refine their understandings
through initial exchange of ideas. The benefits of the workshop will be more widely distributed
by providing open access to the information, resources, and discussion points beyond those who
are able to attend.
To facilitate your submittal of ideas and facilitate workgroup discussions, please use the
following questions to guide your response. Or, if we are not asking the right questions, please
let us know that, too.
Average versus Marginal
1. What are the advantages and disadvantages of both average and marginal approaches to
LCI modeling for electricity supply? When is either approach warranted?
2. When applying average or marginal approach how should the following factors be
accounted for, if it all:
o Short-term changes (occurring hourly, daily, and seasonally) in the electricity
source profile resulting from variations in a plants LCI profile as a result of
varying operating efficiency/output due to local demand?
o Short/Mid-term changes (occurring daily, monthly, or yearly) in the electricity
source profile resulting from changes in the source of electricity due to changing
purchase contracts to plants of a different fuel type and emission profile on a
routine basis (potential affect of deregulation)?
o Long-term changes (occurring in 5 - 10 yrs. or more) in the electricity source
profile due to technology/process improvements?
3. When adjusting for expected changes in the electricity source profile (see Question #2),
what is the appropriate time frame minimum level of uncertainty necessary to model the
change for a site-specific source profile, and regional and national averaged source profiles
(24-hour day, seasonal, other)?
4. Changes to the electricity source profile are occurring and will occur in the future, at
what point do these changes significantly impact the LCI to a the point at which they become
observable to the decision-maker (short-term, mid-term, and/or long-term changes)? Should
this determine the level of rigor and uncertainty in modeling electricity supply?
Boundaries
5. Based on prior LCA and non-LCA environmental evaluations of the electricity supply
system, is there a set of environmental interventions for which reporting in LCI databases
28
-------
should be required? Is it possible to define a recommended set that would be included in all
electricity LCIs?
6. Are there source specific/unique environmental interventions that should be considered
(i.e., not overlooked in a standardized list, see Question #1)? For example, ecological effects
of hydroelectric facilities on river ecosystems.
7. What inputs besides fuels are essential/important to include, for different types of
generation?
8. Can knowledge gained from previous input/output-based LCA analyses be used to
address Question #7?
9. What is the significance and suggested treatment of maintenance and repair inputs?
10. What is the significance and suggested treatment of supporting infrastructure?
New and Non-Traditional
1 1 . How should distributed generation be modeled in attributional and prospective LCAs?
12. How can LCA be usefully and consistently applied to assist comparative evaluation and
to guide design improvement for uncertain or rapidly evolving technologies?
13. What are the key modeling issues and data needs in bio-fueled generation, including
modeling of agriculture or forestry?
14. How should fixed-flow renewable technologies (such as photovoltaic, wind, and run-of-
river hydro) be treated in attributional and prospective LCAs,?
Transmission & Distribution
15. How variable are line losses as a function of user class, region, and other factors?
16. What are the major impacts of T&D infrastructure in place - e.g., land use, habitat, other?
17. Are these impacts expected to be important within the scope of actual life cycle
assessments - e.g., in comparison with impacts of electricity production?
18. What data sources are available for characterizing the impacts of T&D infrastructure?
19. What are the major impacts of construction and maintenance of T&D infrastructure, and
are these expected to be important in the larger context?
Co-Products and Allocation
20. How should co-products and allocation be addressed when modeling electricity supply?
21 . What are the co-products of electricity generation, by plant type?
22. What defines a co-product from electricity generation? For example, recreational service
"output" of hydroelectric facilities. If so, how well-characterized is its value, and how
should it be treated in LCA?
29
-------
23. What are the physical relationships that relate variation in the levels of output among the
co-products to variation in the required inputs and the environmental releases from electric
power production?
24. Which of these co-products currently has market value, which others may have market
value in the future, and should impacts of co-product use be considered? If so, how should
the future market potential be addressed?
25. Other issues?
Bibliography
Heijungs, Reinout, 1997. Economic Drama and the Environmental Stage. Ph.D. Dissertation,
CML, University of Leiden, The Netherlands.
Weidema B P, Frees N, Nielsen A-M, 1999. "Marginal Production Technologies for Life Cycle
Inventories." The InternationalJournal of Life Cycle Assessment 4(l):48-56.
B P Weidema, 2001. Avoiding co-product allocation in life-cycle assessment. Journal of
Industrial Ecology. 4(3): 11-33.
Electric Industry Restructuring Research Group, 1998. Electric Industry Restructuring:
Opportunities and Risks for West Virginia - Interim Report No. 5: Transmission Enhancement
and Expansion. West Virginia University, Morgantown, WV.
30
-------
Pre-Workshop Responses
Received from:
• Bo Weidema, 2.0-LCA Consultants, Denmark
• Tomas Ekvall, Chalmers, Sweden
• Philippa Notten, University of Capetown, South Africa
• Wolfram Krewitt, German Areospace Center, Germany
• Rolf Frischknecht, ESU-services, Uster, Switzerland
• Caroline Setterwall, Vattenfall, Sweden
• Michael Overcash, North Carolina State University, USA
• Ivars J. Licis, Environmental Protection Agency, USA
• Gjalt Huppes, CML, The Netherlands
-------
Thoughts on the five topic areas
By Bo Weidema, 2.0-LCA Consultants
1. What are the advantages and disadvantages of both average and
marginal approaches to LCI modeling for electricity supply? When
is either approach warranted?
Advantage of the average (attributional) approach is that much
statistical information is provided in a form suitable for this
approach. Also, it may be easier to communicate to lay-people
without an economic or system-analytical background.
Advantages of the marginal (prospective) approach is that it
provides results that are meaningful in a decision-making
context, it can reduce data collection efforts substantially
(since only data for the marginal production is needed, not data
for the entire system), and it avoids arbitrariness in setting of
system boundaries, notably in relation to geographical and
technological boundaries as well as in relation to co-product
allocation.
The average (attributional) approach may be warranted when
seeking to allocate blame for past activities. The marginal
(prospective) approach is warranted when analysing the
consequences of a decison, i.e. as a decision-support. The
marginal approach can also be applied to allocate blame for past
activities, by using historical data valid at the time of the
decision that led to the situation that you wish to allocate
blame for.
2. When applying the average or marginal approach, how should the
following factors be
accounted for, if it all:
o Short-term changes (occurring hourly, daily, and seasonally) in
the electricity
source profile resulting from variations in a plant's LCI profile
as a result of
varying operating efficiency/output due to local demand?
o Short/Mid-term changes (occurring daily, monthly, or yearly) in
the electricity
source profile resulting from changes in the source of
electricity due to changing
purchase contracts to plants of a different fuel type and
emission profile on a
routine basis (potential affect of deregulation) ?
o Long-term changes (occurring in 5 - 10 yrs. or more) in the
electricity source
profile due to technology/process improvements?
In a decision-making context, it will most often be the long-term
influence which is relevant, i.e. the influence on the average
long-term marginal. However, if you investigate a device that
operates only at a specific time of day, week or season, it is
relevant to look at the average long-term marginal for this
specific energy supply, i.e. to distinguish peak electricity as a
separate product.
32
-------
3. When adjusting for expected changes in the electricity source
profile (see Question #2), what
is the appropriate time frame and minimum level of uncertainty
necessary to model the
change for a site-specific source profile, and regional and
national averaged source profiles
(24-hour day, seasonal, other)?
Follows from the above answer.
4. Changes to the electricity source profile are occurring and
will occur in the future, at what
point do these changes significantly impact the LCI to a the
point at which they become
observable to the decision-maker (short-term, mid-term, and/or
long-term changes) ? Should
this determine the level of rigor and uncertainty in modeling
electricity supply?
Follows from the above.
New and Non-Traditional
I am not really sure that I understand what is the problem here?
It appears straight-forward to me.
Co-Products and Allocation
20. How should co-products and allocation be addressed when
modeling electricity supply?
According to ISO 14041.
22. What defines a co-product from electricity generation? For
example, recreational service
"output" of hydroelectric facilities. If so, how well-
characterized is its value, and how
should it be treated in LCA?
In attributional LCA, a co-product is defined as one that
contribute to the income of the producer. This definition can
also be used in prospective LCA, although here there is no need
for a sharp definition of co-products, since all outputs to
technosphere, whether co-products or waste for treatment, can be
modeled in the same way.
24. Which of these co-products currently has market value, which
others may have market value in the future, and should impacts of
co-product use be considered? If so, how should the
future market potential be addressed?
Follows from the above answer. Future market potential for co-
products can be assessed by the use of forecasting, as when
collecting data for any other future process.
33
-------
By Tomas Ekvall, Chalmers
Here are brief reflections to some of the questions in the Issues Paper:
Average versus Marginal
Comment: It is important to distinguish between two types of changes in the electricity system:
causes (perturbations) and effects. Short-term perturbations have short-term effects; they can also
have long-term effects. Long-term perturbations have long-term effects; in most cases, they also
have short-term effects.
1. Average modeling results in information on the environmental burdens of the electricity
system. Marginal modeling, in most cases, results in information on the effects on these burdens
of changes that are made in the life cycle. Different information may be relevant in different
cases (Ekvall et al. 2001a; enclosed). What we need is a procedure to identify what information
is relevant in a specific case.
2.1 don't understand the description of short-term changes. To my experience, electricity
production in a power plant is not affected by changes in the local demand since power plants are
connected in a regional, national or even international grid. Short-term changes do occur,
however, due to short-term changes in the demand on the larger geographical scale.
Average modeling is typically based on averages over a year or more. If the changes in the
examples significantly change these average data for the relevant system, an average model
should ideally be based on average data reflecting the new situation rather than the old. An
exception is when we make a comparative LCA were the change takes place in one alternative
only. Then we need two average models, reflecting the old and new situations respectively.
In marginal modeling, the aim is to describe the actual consequences. In the Nordic countries,
hourly and daily changes in the electricity demand - as well as some seasonal and yearly changes
- are managed by utilising the storage capacity of hydropower. In the end most of these changes
affect the marginal base load production. The exceptions are short- and mid-term changes that
occur during peak load periods. But this may vary between different countries.
3.1 don't see the relevance of modeling site-specific changes, since the power plants are
connected in a grid (but this reflection is based on my experience from the Nordic electricity
market). To me the only exception is if you do average modeling and the contract specifies that
the electricity comes from a specific plant. In this case, site specific data should be used
regardless of the time frame of the change.
4. Ideally, average data should reflect the average environmental burdens of the system at the
relevant period of time. If the average is expected to change significantly within the time frame
of the system or decision, this should be taken into account. In practice, this is rarely done.
Marginal data probably change more rapidly over time. For long-term margins, this is because
different investments decisions are at stake at different points in time. Different technologies can
be at the margin during the time frame of a single decision or system. This is part of the reason
why marginal data should be expected to reflect a mix of technologies rather than a single
technology (Mattsson et al. 2001; will hopefully be available at the conference).
34
-------
Co-Products and Allocation
20. This depends on whether you want to model the environmental burdens of the system or the
consequences of changes that are made in the system. In other words, the question is related to
the choice between average and marginal data (Ekvall et al. 200Ib; enclosed).
24. It can be relevant to include impacts of co-product use if you want to model consequences of
changes (Ekvall & Finnveden 2001; enclosed).
- Ekvall etal. 2001a.doc
-Ekvalletal.2001b.doc
- Ekvall & Finnveden 2001 .doc
35
-------
Thoughts on Discussion Questions
By Philippa Notten, University of Capetown
Average vs. Marginal
1. The distinction needs to be made between modeling electricity supply for incorporation into
a product/process inventory, and modeling to support decision-making within the electricity
supply industry itself (e.g. choice of a desulphurisation technology on a coal-fired plant). For
the latter, only the marginal approach would appear valid, consistent with the requirements of
modeling prospective decision systems (Weidema et al., 1999; Wenzel, 1999). Although the
marginal approach is probably the most methodologically defensible for the former as well,
average "historical" type LCIs should not be discounted altogether. These types of LCIs, as
commonly available in LCA databases, are indispensable for screening assessments, to
determine the extent of the contribution of electricity supply to the overall product/process
system inventory, and so to determine whether a more accurate marginal analysis is
warranted. This is perhaps more clearly explained with respect to the foreground /
background convention in LCI modeling (Clift et al., 1998) (where the background system is
defined as the set of processes whose operation is not directly affected by decisions based on
the study, other than the quantity of material (or magnitude of the function) input into the
foreground system). Where electricity supply falls into the foreground system, a marginal
approach will always be warranted, whereas an average approach will often be sufficient
where electricity supply falls into the background system.
The advantage of the marginal approach is its inherently lower uncertainty (both in terms of
its avoidance of using average data and its avoidance of arbitrary allocation rules). A possible
disadvantage of the marginal approach is the lack of readily available data, whilst the
principal advantage of the average approach appears to be that this is what is currently
available in LCA databases (although this may well change, as the value of the marginal
approach is appreciated, and it becomes more common to publish inventories of specific
technologies). However, a possible barrier to this may be that companies are more
comfortable publishing LCI information as national or product wide averages (i.e.
confidentiality issues). The disadvantage of the average approach is its high uncertainty. This
could be improved by the definition of more relevant averages (e.g. regional rather than
national or broad technology type), and better reporting of their variability and uncertainty.
This would allow a more informed determination of whether electricity supply can
appropriately be kept in the background system (i.e. the contribution to uncertainty of the
electricity LCI needs to be evaluated in light of the overall inventory uncertainty).
2. Stochastic modeling approaches can be used to incorporate variability, i.e. the inventory is
presented as a range of probable output rather than a single mid-point. However, to be
meaningful, stochastic models need to be applied to the actual process models underlying the
inventory, where the actual variability in the data samples can be incorporated, and
correlations between inputs can be avoided by modeling the causal mechanisms.
Changes in the source profile can be incorporated by modeling short-, mid- and long-term
scenarios, where these can reflect changes in fuels and technologies (i.e. changes in the grid
mix for the average LCIs, and changes within the particular technology for the marginal
LCIs). Importantly, the stochastic models recommended to include process variability should
36
-------
also include data uncertainty, i.e. the input probability distributions should include
uncertainty due to variability, as well as that due to the nature of the data (e.g. uncertainty in
the quality of future fuel sources). In this way, the uncertainty associated with the future
scenarios can be quantitatively reflected in the inventory, i.e. the fact that the long-term
scenario has much higher uncertainty than the short- and mid-term scenarios can be reflected.
Although estimating the uncertainty associated with the data is inherently subjective, there
are methods which mitigate this to some degree (Weidema and Wesnaes, 1996), and even a
subjective estimate of uncertainty is preferable to representing a highly uncertain future
inventory with false accuracy, as a mid-point LCI or a stochastic model only incorporating
variability would.
A further point regarding incorporating data uncertainty and variability in the LCI using
stochastic modeling is that this may force a marginal approach, or at least, force the
definition of more tightly defined average systems. This is because incorporating the
variability within systems averaging widely different processes can result in such high
uncertainty (i.e. such a wide range in the output) that no significant differences are able to be
discerned between options in a comparative assessment (Notten, 2001).
3. This is probably most meaningfully related to the time-frame of the decision system in which
the electricity supply LCI is to be used. For most mid- to long-term decisions, a seasonal or
annual variability is reasoned to be most meaningful. This is because in such systems the
uncertainty in the inventory is likely to be dominated by the uncertainty associated with
modeling the future system (e.g. uncertainty in the future grid mix), and including a higher
level of variability will be unnecessary. Similarly, in national/regional average LCIs, annual
variability should be sufficient as the variability between the technological systems is likely
to dominate the overall uncertainty / variability.
4. This will depend to a large degree on the particular country or region under consideration, i.e.
in what time-frame is a significantly different generation technology envisaged being added
to the grid? For example, in South Africa, the electrical utility is currently in a position of
over-supply, so in the mid- to long-term additional capacity demand can be met by re-
powering the "mothballed" stations and operating current stations at higher loads. The
electricity profile will therefore not change significantly, as changes in the grid mix merely
result in a shift between older and more modern coal-fired plants, the effects of which are
relatively small (Notten, 2001). However, the point at which a non-coal source is added to
the grid will significantly change the profile, especially from a marginal perspective.
The "level of rigor" and the uncertainty of modeling will certainly be affected, since it is not
possible to model future systems at the same level of detail as existing, demonstrated
systems. The subjective estimation of uncertainty (see comments in 2) plays a much larger
role in the modeling of future systems, since it is not possible to rely on the actual variability
in data samples, as in existing systems.
Boundaries
5. This is difficult as the interventions able to be considered will be constrained by the data
availability (the degree of development/demonstration of the technology, and the scope of the
study), however, a default list of interventions towards which to strive would certainly be
helpful. This would also be very useful in standardizing the use of aggregate interventions,
37
-------
e.g. TSP, IDS. Water-related interventions where found to be particularly problematic when
trying to compare across different LCIs, e.g. use of categories such as "sulfates", "nitrates"
etc, rather than individual components.
Also requiring standardization is how energy resources are defined in fossil fuel-burning
systems. This is required because different systems may burn very different quality fuels, e.g.
modern South African coal-fired power stations burn very poor quality coal (ash contents as
high as 40%, and CVs as low as 14 MJ/kg), thereby "freeing up" coal resources for high
revenue coal products. This ability to extend the life of the coal reserves needs to be reflected
in the inventory (most simply achieved by defining a reference CV for coal reserves, and
adjusting the mass of coal consumed accordingly).
6. Non-stack emissions should not be overlooked (e.g. dust from blasting during mining, and
blown from waste dumps). The impacts associated with solid waste management in coal-
fired systems are typically poorly assessed. Diffuse sources of water pollution (as distinct
from pipe-discharges) are often overlooked. These include surface run-off from waste dumps
and stockpiles, and water collecting in opencast mining pits. Leachate from waste dumps and
stockpiles, as well as seepage from ash dams and pollution containment dams are similarly
neglected, although their significance can be considerable (Notten, 2001).
7. In wet-cooled, coal-fired plants water treatment chemicals were found to be the next most
significant inputs after fuels, particularly in stations using poor quality water sources (Notten,
2001) (this could be a feature unique to South African plants, where water availability
constraints force a high degree of internal recycle and re-use within the water plants).
9. This is not possible to decide without reference to a particular decision situation, where it can
be assessed from a marginal perspective, i.e. if the proposed change is expected to
significantly increase the maintenance (or supporting structures) required, it should be
included.
10. (same as 9).
New and Non-Traditional
11. This points to the need to include transmission and distribution in the inventories of point
source generation facilities, so that these can be compared on a consistent basis to distributed
sources.
12. The need to include a quantitative consideration of uncertainty is critical here. It is essential
to guard against the comparison of incomplete and thus incomparable systems (a comparison
of single-point inventories constructed from inconsistent data sets is likely to be more
misleading than useful). Qualitative LCA methods (e.g. Graedel, 1998) may be of more value
than quantitative methods here.
13.
38
-------
14. Attributional LCIs should not be a problem, other than recognizing that the materials of
construction, life-times and use patterns will play a larger role than in conventional energy
technologies. For prospective LCIs, perhaps these could be viewed as incremental changes?
(discrete steps rather than gradual load changes).
Transmission and Distribution
15.
16. South Africa has a problem with bird mortality (eagles insist on nesting in the pylons).
17. Land use, possibly.
18.
19. Habitat loss, herbicides used in maintaining servitudes.
Co-Products and Allocation
20. Weidema's marginal approach (Weidema, 2001) appears the most meaningful for
prospective LCIs.
No allocation problem regarding the electricity product was encountered within coal-fired
electricity production in South African (since no steam or heat is sold as a co-product).
However a different allocation problem arises due to the modern South African power
stations being designed to burn near discard-quality coal. Allocating burdens to the coal-fuel
supplying the station is found to be very significant for those stations supplied by dual-
producing collieries, which produce a high-quality export coal (requiring significant coal
preparation), as well as a low-quality power station coal. This can be regarded as combined
production, so can be modeled by a marginal analysis (keeping the production of one product
fixed, by varying the other)(Weidema, 2001). For some stations, this combined production is
made more interesting by the fact that the power station coal is made up of a blend of run-of-
mine coal and discard (the waste product from export-quality coal preparation). The
combined production is therefore able to reduce the mass of discard waste (the disposal of
which has significant environmental impacts), as well as avoid the waste of energy resources
discard coal represents. The "avoided" burdens approach is used here, where the power-
station coal is "credited" with the avoided burdens of discard disposal. Very significant for
the electricity profile is that the discard-fuel source is essentially burden-free, i.e. is not
allocated any mining burdens other than the "avoided" burdens.
21.
22.
23.
24. A small volume of coal-ash is sold for use in cement products in South Africa. However, this
volume is so small (less than 5%) it has negligible impact on the LCI.
39
-------
References
1. Clift R., Frischknecht R., Huppes G., and Weidema B. P. (1998) Towards a Coherent
Approach to Life Cycle Inventory Analysis. SETAC-Europe.
2. Graedel T. E. (1998) Streamlined Life-Cycle Assessment. Bell Laboratories, Lucent
Technologies, Prentice Hall.
3. Notten P. (2001) Life Cycle Inventory Uncertainty in Resource Based Industries - A Focus
on Coal-Based Power Generation. Ph.D., University of Cape Town.
4. Weidema B. P. (2001) Avoiding Co-Product Allocation in Life-Cycle Assessment. Journal
of'Cleaner Production 4(3), 11-33.
5. Weidema B. P., Frees N., and Nielsen A.-M. (1999) Marginal Production Technologies for
Life Cycle Inventories. InternationalJournal of Life Cycle Assessment 4(1), 48-56.
6. Weidema B. P. and Wesnaes M. S. (1996) Data Quality Management for Life Cycle
Inventories - An Example of Using Data Quality Indicators. Journal of Cleaner Production
4(3-4), 167-174.
7. Wenzel H. (1999) Life Cycle Assessment in Pollution Prevention: Trends in Method
Development and Simplifications. In Tools and Methods for Pollution Prevention (ed. S. K.
Sikdar and U. Diwekar), pp. 119-129. Kluwer Academic Publishers.
40
-------
Feedback on background paper
US EPA/NREL Workshop on Electricity Data for Life Cycle Inventories
By Wolfram Krewitt
German Aerospace Center, Institute of Technical Thermodynamics, System
Analysis and Technology Assessment, Stuttgart, Germany
General remarks
The issues described in the background paper provide a good summary of most relevant research
issues in the field of LCA/electricity and thus sketch an interesting and exciting research agenda.
From my understanding the list of issues suggest that the scope of LCA is getting much broader,
and thus partly leads to overlaps with other scientific areas. While in general I very much
appreciate this development, I think it is also very important to be clear about the inherent
limitations of the LCA approach. I certainly don't want to say that LCA methodology should not
be developed further, but I think it is necessary also to focus - in a positive sense - LCA
activities to key areas where the LCA methodology has proven it's strength, and on the other
hand to invest efforts in establishing well defined interfaces to other areas (e.g. energy system
modeling) and thus gain from synergies. Let me try to illustrate this position with respect to two
points arising from the background paper:
1) Local scale Impacts
Local scale impacts on ecosystem via land use, alteration of water systems etc. partly are the
dominant impacts for decentralized renewable energy technologies. I think there are some
inherent limitations to the LCA methodology in addressing such very local scale impacts. LCA is
based on the capability of summing up specific parameters over a large number of processes,
over time, and over regions. Recent developments explore the feasibility of site and time
dependent LCA. A key problem of very local scale impacts is that it is not necessarily the
technical characteristics of a facility, but much more the specific environmental conditions at a
given site (soil quality, water regime, topography, ...) which determines the level of impact
resulting from a 'unit' of environmental intervention. Technical parameters on the one hand and
site specific environmental conditions on the other hand are very closely interrelated and cannot
be evaluated independently any more. Summing up environmental interventions to an aggregated
indicator is therefore very difficult, if not impossible. I think it is not without reason that up to
now we do not have a satisfying approach for treating land use adequately in LCA.
I am currently working on a project on 'strategies for an ecologically optimized expansion of
renewable energy sources in Germany'. The project includes LCA of renewable energy
technologies, the integration of LCA results into the development of energy system scenarios,
and has a specific focus on the consideration of aspects related to the conservation of nature
which are traditionally not well covered in LCA. Impacts on local ecosystems, the disturbance of
specific flora/fauna habitats etc. are key impacts for some renewable energies. Because of the
strong site dependency of such impacts, I am more and more convinced that LCA is not the
appropriate tool to deal with such impact categories. We need other tools, most probably GIS
based, to complement information derived from LCA to support sound decision making.
2) Link to energy system modeling
41
-------
Some of the issues addressed in the background paper under the heading of 'average vs. marginal
systems modeling' are typically addressed by people that operate energy system simulation or
optimization models, and are quite far away from the traditional process-chain oriented product
LCA. This by no means says that LCA should not tackle such problems, but - again - we should
be aware of inherent limitations, we have to be aware of what other people have developed with
large efforts over many years, and we should seek to identify most appropriate interfaces to
existing tools.
I recently initiated a project proposal to the EU which tried to combine LCA methodology and
the TIMES-type of energy optimization models. Unfortunately the proposal was rejected,
apparently we were not able to clearly communicate what we wanted to do, but I still think this a
very interesting task. Even if the proposal did not make it, the proposal preparation phase was
quite interesting, as we learned that it is not that easy to bring LCA people and energy system
modelers together. One of the important points is the enormous complexity in quite different
areas. Energy system models include many hundreds of individual processes to generate a
realistic picture of supply and demand patterns over time. It is hardly possible to provide detailed
LCA data for all these processes (do we need them??). On the other hand, in terms of
environmental impacts most of the current energy system models focus on CO2 emissions, and
partly cover SO2, NOX and particles (do we need others??).
So, what I want to say in short: There is both a need and potential for further development of
LCA methodology related to energy supply. But do not try to make LCA a 'universal' tool.
Define reasonable links and interfaces to existing tools which already do a good job for specific
tasks, and benefit from synergies.
Average versus Marginal
(1, 2, 3, 4) It seems that in general a marginal approach is preferable, as it better reflects the
'real' conditions. However, this certainly depends on the question at stake. The marginal change
between status A and status B does not necessarily reflect differences in the characteristics of a
specific product. Marginal impacts resulting from the decision A or B do very much depend on a
large number of decisions taken by other actors in the complex system. Many of these decisions
are not independent. Complex models are required to model the behavior of such a system
(energy system simulation/optimization models). Is this still the scope of LCA?
The average approach seems to be relatively straightforward, but often the average inventory is
extrapolated from individual plants for which a detailed inventory exists, but which do not
necessarily represent the actual mix of different plants operated under different conditions. If
reasonable assumptions are taken, the average approach might be more helpful to point out
specific differences between products.
Most of the short-term and also mid-term changes in the electricity source profile have a regular
pattern, so that a reasonable averaging should be used. I do not see a case in which conditions at
a specific hour are really relevant for an LCA study. Long term changes need to be accounted. In
the past, I mainly used annual averages.
42
-------
Boundaries
(5) The European ExternE project on External Costs of Energy (as well as the joint US/EC Study
on Fuel Cycle Externalities, which was a forerunner of ExternE) started with a screening of a
range of pollutants and related impacts. Based on expert judgement (no formal selection
procedure), a set of some few priority impacts and related pollutants was identified. External
costs (as an aggregated damage indicator) were very much dominated by greenhouse gases, NOX,
SC>2, and particles (and related secondary substances, namely ozone, sulfates and nitrates). This
conclusion was quite robust. Although the small number of key substances was often criticized
for being inappropriate, other LCA studies for energy technologies with a more comprehensive
inventory are also dominated by the same set of pollutants.
The picture will change with an increasing share of renewables, but the above mentioned
pollutants still dominate LCA results for renewables technologies because of the importance of
the conventional energy supply mix.
In the beginning of the project, ExternE carried out a reasonable screening of potential impacts
from heavy metals and some few organics and concluded that for the broad range of fuel chains
analyzed the impacts from these substances are negligible compared to those from the above
mentioned priority pollutants. The current phase of ExternE explores in more detail potential
impacts from heavy metals and organic substances, with a focus on emissions to soil and water.
Results are not yet available.
In addition to air pollutants, the consumption of energy and non-energy resources should be
included in the inventory. A problem for non-energy resources is of course to decide which are
important. New technologies like fuel cells require some exotic materials at currently tiny
quantities, but market introduction might lead to a significant demand in the future and thus lead
to problems. For PV systems new materials are under development. Very difficult to recommend
a default list.
(6) Yes, there are of course source specific interventions, often they are unique for a given plant
at a specific site. As discussed above, I doubt whether LCA is the most appropriate tool to deal
with such impacts in general, as they often depend on site specific environmental conditions
rather than on the facility's technical features. To give some examples, ExternE discusses for
instance effects like increase in real estate value, improved commercial shipping, tourist
attractions, and effects on the scenery in the catchment area resulting from a run-of-river plant at
the river Danube in Austria. For a hydro project in Norway impacts of temperate water into a
fjord on the ferry traffic are described and quantified. Noise impacts as well as visual intrusion
might be important for wind turbines, but again the effect is very site dependent.
(7, 8) This of course very much depends on the technology. Of course for renewables the 'other'
types of impacts are increasingly important (do you expect here a complete list??). Extended
input/output tables are of course helpful to quantify emissions from these inputs, and also to
identify the main source sectors. As an example the figure below shows results from a hybrid-
LCA (process chain analysis linked to I/O analysis): the contribution of different processes
(partly aggregated already in the diagram) to total CO2-emisisons from a 5 kW PV roof
application, and the respective contributions from process chain analysis and I/O analysis.
43
-------
CO2-emissions in g/kWh
176
M
'55
_>
re
7
O
c
're
o
(/)
(/)
0)
o
o
o
£
CD
100
| | Process chain analysis
Input/Output Analysis
Fig. 1: CO2-emissons from a 5 kW PV roof application
(9, 10) Impacts from maintenance and repair in general are relatively low (see figure 1) and
might be well covered by I/O. What do you mean by 'supporting infrastructure'?
New and Non-Traditional
(12) Germany aims at a share of 50% electricity from renewable energy sources in 2050, i.e. we
have to expect a drastic change in our energy system. Current LCAs for emerging technologies
however most often use the current electricity mix as an input to upstream processes. Besides of
the obvious strive for getting the best available data for the relevant new materials (using also
tools like technical learning curves etc.), it is important to use the characteristics of a future
energy system with an adequate share of renewable energy as an input to basic processes in order
to provide a more realistic picture of the emissions resulting from energy supply. It is of course
not easy to agree on a specific future energy scenario, so the door is open for another source of
potential differences between LCA studies.
(13) The agricultural reference system is of major importance. As also transport processes are
important, the spatial pattern of supply of biofuels and the demand for e.g. district heating have
to match.
(14) If we want to compare specific energy technologies to evaluate their potential for solving
specific environmental problems, it is sufficient to look at the impacts normalized to a kWh at
the power plant's gate. If we want to take into account the capacity effect and a given supply
task, we might take into account a back-up technology. I would prefer however to look at the
44
-------
full energy system with a certain share of renewables, which again requires the use of an energy
system model.
Co-Products and Allocation
I just want to give the example from ExternE, where we discussed in detail the allocation of
impacts from a combined heat and power plant between the electricity and heat output (knowing
that this is a problem that has been extensively discussed already in the past):
In ExternE we discussed the following allocation alternatives:
• Allocation based on operational characteristics (operation is heat or electricity oriented ^
electricity or heat as by-product ^ benefits of joint production are assigned to the by-
product. We can also use the ratio of additional fuel input required for producing the By-
product" to the total fuel input required for both electricity and heat production.)
• Allocation based on thermodynamic parameters (energy or exergy content; several other
parameters are discussed in the literature, taking into account various characteristics of the
steam-process, but results achieved with these parameters generally are within the range
covered by the energy content and the exergy indicators)
• Allocation based on the final products' price (Assuming a perfect market, the final products'
price probably is the best indicator for the utility of the product. However, the price of electricity
and heat is highly dependent on the customers demand characteristics and other non-technical
parameters.)
As an example, the table below shows external costs (as an aggregated indicator of
environmental impacts) for heat and electricity from a 520 MW coal fired CHP plant in Germany
for different allocation rules:
mEuro/kWhei MEuro/MJ
Credit for heat,
substituting 2.7
oil fired domestic boilers 5.0
gas fired domestic
boilers
Credit for electricity (-7.8)
Additional fuel input 4.9 0.13
Energy content 2.3 0.64
Exergy 4.6 0.24
Price 3.6 0.38
(note that these numbers are taken from a 1996 ExternE report. External costs in this example are
not based on the most recent ExternE methodology, but they still give a good indication on the
effect of different allocation rules)
45
-------
In ExternE we decided to report results based on exergy-allocation as mandatory, and in addition
(optional) results based on energy content or price allocation to reflect specific conditions under
which the CHP plant might be operated.
I do have some doubts to which extend it is possible to include secondary effects (like
recreational effects from hydro plants, or the example from Norway given above in which the
hydro project leads to effects on ferry traffic). These effects are very site specific and need a case
by case evaluation. Often employment effects are mentioned as a positive effect resulting from
the introduction of renewable energy systems. I recommend to not consider such effects in an
LCA.
46
-------
Input paper for discussion during the International Workshop on
Electricity Data for Life Cycle Inventories
Written by
Dr. Rolf Frischknecht, ESU-services, Uster, Switzerland
10.09.2001
Overview
The present input paper is structured according to the issues paper distributed among the
participants mid of August 2001. It contains some additional questions as well as possible
answers to some of the questions listed. The answers given are not complete nor exhaustive.
Further explanations will be given during the workshop.
I like to emphasise, that I very much appreciate the sophisticated level of the present version of
the issues paper. It covers most of the pertinent and most important methodological topics.
Average versus Marginal Systems Modelling
In order to be able to discuss the issue in a well-structured way, I suggest to first try to link the
way of modelling to the different questions/goals of an LCA (see, e.g., Tab. 5.9 in Frischknecht
(1998)).
I like to precise the first question and add the two following questions (see also Chapter 5 in
Frischknecht (1998)):
What do we know about how the electricity supply system respons to different levels of
changes in demand (from short term changes like unexpected increase in demand during
the day to long term (20 to 40 years) development of electricity consumption)?
- How to separate replacement (of old equipment) from expansion of production when
performing a prospective LCA?
Some of the marginal technologies are rather used to replace old equipment instead of
increasing the production capacity. Hence marginal technology mixes should be
determined carefully.
What is the relation between overall market size of a product/service and its respective
individual products?
Overall market trends are often used to determine marginal technologies. Environmental
and energy policy may be used to limit the overall energy consumption or the overall
environmental pollution. However, LCA is rather suited to help finding an optimal
allocation of "pollution rights" among all products/services on a micro-economic level.
Therefore a relation between macro- and micro-economic perspectives should be
established in one way or another.
Boundaries for the Electricity Generation System
Additionally to the two boundary dimensions "environment" and "supply chain" I would add two
additional ones concerning time and geography:
- When modelling the supply with a certain electricity mix, which geographic area is
adequate to represent the mix? Some LCA experts argue for instance that for the Swiss
electricity supply system the western European integrated electricity network (UCTE) is
the adequate electricity mix due to the intensive trade relations.
- Pollution from electricity supply systems may occur in the very far future (e.g. long-term
radionuclide releases from abandoned uranium milling sites). How shall such emissions
47
-------
be treated? How should we treat the treatment and disposal of radioactive wastes? How
should discounting be applied in LCI of electricity systems (positive, zero, negative
discount rate), see e.g., Hellweg (2000:125ff.)?
- Besides capital equipment, transport of workers to the site etc., research and
administrative divisions should explicitely be addressed when asking about system
boundaries.
New and Non-traditional Electricity Generation
Concerning new electricity generating technologies my major concern is about the correct way
of modelling it, again depending on the goal of the LCA.
Do we need a dynamic model to assess the environmental impacts of (the introduction of) new
technologies in order to answer the question whether these technologies should be favoured or
not? Or, would it be sufficient to model a possible future (steady) state (when the technology is
more or less established)?
The correct treatment of flow-limited sources such as wind or hydroelectric power refers to the
question about the relation between the macro- and microeconomic situations. To my
understanding LCA is a tool which helps to allocate scarce environmental resources and scarce
"pollution rights" among competing products (like the price system is used to efficiently allocate
the scarce traditional production factors human power, capital and land). Hence, macro-
economic limitations like the ones mentioned should not have an influence on the way the LCA
is carried out.
Transmission and Distribution
It should be precised on which level of voltage the losses are reported. On the low voltage level
(380V), losses more than 10% are not unusual. On higher voltage levels, losses are much lower
(below a few percents). Additionally power switching stations (SF6-losses) and operation of high
voltage transmission lines (N2O-emissions) are important.
Outputs and Co-Product Allocation
Concerning co-product allocation I like to refer to my article written in the Int.J. LCA, Vol.5,
No. 2,pp.85-95 or Chapter 7 of my Ph.D. thesis.
Questions
Average versus Marginal
1. The choice whether to use a marginal or an average approach depends on the goal of the
study (see, e.g., Tab. 5.9 in Frischknecht (1998)).
2. See excerpts from Chapter 5 of Frischknecht (1998), especially Subchapter 5.3.
3. dito
4. Decisions which affect the electricity demand within a production site (on the short, long,
or very long term) should adequately be reflected in the LCI model. Hence, only long-
term changes in the electricity supply mix should be considered in a long term decision
situation for instance.
Boundaries
5. An adequate treatment of direct ecosystem impairment caused by hydroelectric power
production, coal, oil and uranium extraction is still missing. I am not sure whether these
impacts may ever be considered adequately. Nuclear waste disposal is another issue
where an adequate assessment within LCA is still missing. Noise seems to get more
48
-------
attention in connection with wind power. Also here, broadly accepted inventory
parameters and impact assessment methods are still missing.
6. If LCI/LCA data should be used in a labelling scheme for green electricity, I recommend
to additionally include local criteria which cannot be considered in standard LCAs. This
is successfully done within the "naturemade star" label for environmentally excellent
electricity from reneweable sources (see http://www.naturemade.org). It covers aspects
like river ecosystems, agricultural techniques, visual impacts of wind power plants etc.
7. Catalysts and especially precious metals are important due to the high environmental
loads per gram. Capital equipment is (obviously) relevant for new renewables such as
photovoltaics and wind power but also for hydroelectric power plants. I have no idea
about the environmental relevance of research and administration (flights). For
hydroelectric power plants care must be taken not to neglect methane emissions if
substantial amounts of biomass have been drowned by the artificial lake.
8. We have little experience on input/output-based LCAs. We applied it to roughly estimate
the contribution of prospection (assuming that 50% of the amount of money spent for
prospection is used for computer equipment). It was a very minor contribution and we
therefore excluded prospection from the analysis.
9. Reliable data is rare to answer this question.
10. Concerning conventional fossil fueled power plants it seems rather negligible (except for
land use aspects). For other sources (see answer to question 7.) it is the main contribution.
New and Non-Traditional
11.1 am not sure whether I understand the question. I see no (principal) difference in analysis
depending on the degree of distribution of a power plant technology.
12. Comparative evaluation and design improvement are two completely different LCA
goals. While the first needs data about the current situation (in order to improve it), the
latter requires an analysis of a future situation where production processes and
technologies needed for a new power plant type are optimised. The question can then be
answered whether it is worthwhile (from an environmental point of view) to develop such
a technology at all.
13. Bio-fuels are included in the natural cycle of chemical elements. I consider the adequate
modelling of nutrients and trace elements cycles as important aspects. Furthermore, the
influence of agriculture/forestry on biodiversity and carbon balance (carbon fixing in the
topsoil of forests) may also be important.
14. No special treatment. Reasons are given above.
Transmission & Distribution (T&D)
15. Losses are very much dependent on the level of voltage (corresponding to your term
"user class"?).
16. One major impact is of course due to the losses (and therefore depending on the
environmental quality of the power plant mix). Land use, influence on biodiversity etc.
may also be relevant. However, these aspects cannot yet adequately be considered. We
should not forget the emissions of SF6 in power switching units and the production of
N2O of high voltage power lines.
17.1 expect electricity losses and SF6-emissions to be important (or at least not negligible).
18. The Okoinventare von Energiesystemen" (Frischknecht et al. 1996) provides an overview
of the Swiss network. It excludes SF6-emissions, however. Any utility should have data
on their T&D infrastructure (and operation).
19. See above.
Co-products and Allocation
49
-------
20. To my opinion there is no general rule or approach that is acceptable for all parties
involved nor sensible for all situations (see Frischknecht 2000).
21. coal: electricity, heat, gypsum (building industry), fly ash (cement production)
oil: electricity, heat, gypsum (?)
natural gas: electricity, heat
coke oven gas: disposal of "residual gas" of coke production, electricity, heat
blastfurnace gas: disposal of "residual gas" of iron melting, electricity, heat
nuclear power: electricity, heat
hydroelectric power plant: electricity, flood protection, irrigation, recreation, fishing
photovoltaics: electricity, weather protection (if integrated in slope roofs, fa9ades)
combined heat and power units (motors, fuel cells): electricity, heat
biomass: electricity, heat
waste incineration: treated wastes, heat, electricity, slags (cement industry)
biogas from manure: treated manure, heat electricity
biogas from organic waste: compost, heat, electricity
biogas from sewage sludge treatment: conditioned sludge, heat, electricity
geothermal power plant: electricity, heat, therapeutic baths (e.g. the Blue Lagoon,
Iceland)
General comment: in many cases, heat is not used but emitted to air and water.
22.1 suggest (as a pragmatic solution) to limit "real" co-products in relation to a certain share
of total proceeds (e.g. >10%).
23.1 think this is not the right question. Think about a spark ignition engine, which at the
same time produces electricity and heat. One may operate the engine in a way that only
electricity or only useful heat is being produced. However, investment calculations have
been made on the basis to use both products as much as possible. It makes therefore no
sense to vary one of the products in order to determine the environmental releases to be
attributed to this product. The CHP plant is a joint production unit because of economic
reasons. Hence, economic considerations may overrun mere physical considerations.
24. Current market values are delicate without considering the development of the overall
markets. However, I do not see too many difficulties here, except maybe power plants
burning blast furnace and coke oven gases.
Other issues
25. The issue of discounting future activities and environmental releases and related to that of
an appropriate time horizon. See also our paper on health impacts due to ionising
radiation (Frischknecht et al. 2000).
26. The issue of an appropriate geographical boundary when analysing electricity supply
mixes.
27. The issue of rather using LCI data of entire decision units (the utility as a whole;
divisions within a utility) rather than of their individual production technologies, when
dealing with electricity supply mixes.
References
Frischknecht R., 2000. Allocation in Life Cycle Inventory Analysis, in Int.J.LCA, Vol. 5 No. 2,
pp. 85-95
Frischknecht R., 1998. Life Cycle Inventory Analysis for Decision-Making; Scope-dependent In-
ventory System Models and Context-specific Joint Product Allocation, Ph.D.-thesis Nr. 12599,
Swiss Federal Institute of Technology (ETH), Zurich
50
-------
Frischknecht R., A. Braunschweig, P. Hofstetter, P. Suter, 2000. human health damages due to
ionising radiation in life cycle impact assessment, in Environ.Ass.Rev., Vol. 20, No.2, pp. 159-
189.
Frischknecht R., P. Suter (Eds.) et al., 1996. Okoinventare von Energiesystemen, 3rd updated
edition, Bundesamt fur Energie, Bern, ISBN 3-9520661-1-7
Hellweg S., 2000. Time- and Site-Dependent Life-Cycle Assessment of Thermal Waste
Treatment Processes, Ph.D.-thesis Nr. 13999, Swiss Federal Institute of Technology (ETH),
Zurich
51
-------
Submittal of ideas
Caroline Setterwall, Vattenfall
Average versus Marginal
Systems to be studied in a life cycle perspective have several life phases: construction phase,
operation phase, demolition/dismantling phase and use phase. If you are studying an existing
system the construction phase is historical and the demolition phase will come in the future
whereas system operation and product use is happening all through the life time of the system.
The real electricity supply probably varies through the life cycle and the mix of power generation
systems as well.
Vattenfall always uses data describing today's technologies also for historical and future system
processes. I.e. our LCA results never mirror the real environmental impact caused by the studied
system throughout the life cycle. The results reflect a potential impact under certain defined
circumstances.
If we know the electricity supplier in a life cycle process we try to find out the electricity
generation mix (percentage hard coal based generation, nuclear, hydro etc.) of this supplier. If
that is impossible we use the average generation mix per year of the country or the region (TEA
Statistics). We use today's mix for both historical and future processes. Our attitude is that a
summation of LCA study results of every system should not extremely exceed the actual
environmental impact. You shouldn't burden one electricity needing system for the fact that
marginal electricity generation is needed in a electricity generation system to secure deliveries to
all consumers.
If the goal of a LCA study is to describe an extension of a system, i.e. a future production
increase or a planned system which will need so much electricity that the electricity balance in a
region or country will be influenced a marginal approach is appropriate.
The approach is however dependent on the goal of the LCA study and you should always try to
be more specific and detailed in your descriptions and calculations of subsystems contributing
the most to the environmental impact of the studied system. Sometimes it is appropriate to model
different scenarios to deliver a diversified picture of a systems environmental impact.
Boundaries etc.
Generally can be said that for power systems using a fuel, handling of this fuel throughout the
life cycle is crucial with respect to environmental impact whereas for other power systems the
construction phase is more important. Vattenfall has the ambition to study those sub-processes in
detail, which contribute largely to the environmental impact of the system, whereas other sub-
processes are studied with a lower degree of exactitude.
For all kinds of power systems Vattenfall makes inventories of the system's construction phase,
fuel production phase, operation phase including reinvestments and waste handling of fuel
residues and demolition phase (for hydro power no demolition is considered, but a higher degree
of reinvestment instead).
52
-------
Vattenfall has been working with LCA since 1993 and has till now studied specific existing
plants, most of them owned by Vattenfall. The following power systems have been studied in an
LCA perspective:
• Nuclear power (2 plants with together 7 reactors, BWR and PWR, 2 mines, 2 enrichment
sites)
• Hydro power (3 stations representative for the generation in a Swedish river (the Lule river,
Vattenfall is now studying a second river))
• Oil-condensing (1 plant, reserve power)
• Oil-based gas turbines (1 plant, reserve power)
• Natural gas with gas and steam turbines (1 prospected plant, with gas delivered from
different sites)
• Coal-condensing (3 plants in Denmark with coal deliveries from different sites)
• Biofuelled combined heat and power (1 existing plant fired with either energy forest (salix)
or forest residues)
• Fuel cells (2 plants fuelled with natural gas)
• Solar cells (based on a Dutch study, adapted to Swedish conditions)
The construction phase is inventoried with respect to major construction materials and
transports of those materials. Data for fabrication of components (generators, turbines,
transformers etc) is often hard to get and is therefore often neglected but manufacturing of the
raw materials for these components is included. Mostly the amounts of the following
construction materials and processes are inventoried and followed to the cradle:
Steel
Concrete
Copper
Aluminum
Other metals
Plastics
Rock wool
Wood (mould wood)
Ground work (blasting mass handling)
For certain power systems there are special materials, which are important for instance catalysts
in fuel cells or solar cell material. Till now Vattenfall has neglected electronics due to lack of
manufacturing data but we will eventually start an inventory of electronics in different plants to
find out the relative impact of such components.
Data about fuel extraction and processing is obtained from fuel suppliers or from literature. All
steps are inventoried including transports and storage.
Data concerning the operation phase is retrieved from the plants' environmental management
system (ISO 14001 or EMAS) or from the environmental report, which is sent to the authorities.
These data include site-specific emissions to air, water and ground, consumption of bulk
chemicals and fuels and generated waste. Reinvestments are considered (often a percentage of
the construction phase). Following parameters are often included:
Fuel amounts
Use of bulk chemicals and cooling media
53
-------
Use of land
Emissions from fuel use
Radioactive emissions for nuclear power
Site specific process emissions (for example ammonia from NOx-reduction measures)
Emissions of cooling media
Emissions of greenhouse gases due to overflooding of land in connection with dam construction
(hydro power)
Fuel residues
Wastes
Production of inputs is included but not the construction of the production sites.
Transport emissions and fuel use are included all through the life cycle but fabrication of
vehicles and roads etc is excluded.
In the demolition phase assumptions are made regarding recycling degrees and waste-handling
options of different waste fractions based on today's technologies.
Impact on biotopes is an important issue in several activities connected to electricity generation,
hydro power operation, mines etc. Vattenfall has developed a special method to describe biotope
changes quantitatively: The Biotopmethod.
In the last years Vattenfall predominately has used the LCA methods described in ISO TR 14025
about Environmental Product Declarations (EPD) (you'll find the Swedish requirements based
on ISO TR 14025 in English at http://www.environdec.com/eng/doc/pdf/e_epd_msrl9992.pdf).
The utilities in Sweden have developed Product-specific Requirements, which are adapted for
electricity and heat LCAs for EPDs (you'll find them in English at
http ://www. environdec. com/psr/e_psr9801 .pdf).
Till now Vattenfall has two third party certified EPDs, Hydro Power Electricity from the Lule
River and Electricity from the Nuclear Power Plant at Forsmark
(http://www.environdec.com/eng/registrations.asp).
54
-------
Thoughts on the 5 topics
By Michael Overcash, North Carolina State University, USA
Average vs Marginal
The increase accuracy of a marginal approach is probably not easily used since the errors in
other aspects are often large.
In a marginal analysis there is still debate over what is the marginal plant. Just because others
are using the average of plants it is unclear that the incremental plant is coal, nuclear, hydro, etc.
It is kind of like who was in line first and not how is the whole system operating.
Boundaries
Not sure there is a lot here. The boundary should be as transparent as possible and then let
people decide how big.
New and Traditional
There is already a good deal of work on new sources. The goal is to make these transparent
Transmission
We use 1.8% transmission losses as the high voltage case for main power delivery. I am not
sure what the losses are under transformers and then local delivery. There is a real difference
between residential and industrial.
Co-products and allocation
We use mass and try to break down processes into enough detail so that the emergence of a co-
product is clear and can be allocated (the micro-allocation approach).
55
-------
Ivars J. Licis
U.S. Environmental Protection Agency
I have already mentioned some of this to May Ann C., but just for the record:
MACRO vs MICRO
I see the issues you and this workshop raise as falling into two major classes, the Macro and the
Micro class of issues.
The Macros deal with the purpose of the "data", or the whole LCA of electricity activity, (step
one in LCA 101) and the setting"boundaries", the Micros start to dabble (OK, focus) into the
refinements, including such things like the geographic differences, average vs. marginal, et al.
While both classes contain topics of interest, the Macros seem to need defining earlier (maybe
with some iteration). Within the framework of one workshop, I would guess we could not tackle
more than the Macro, but who knows. Mary Ann has indicated that significant work on the
Macro level has been done by others earlier. This will be helpful.
MOVING TARGET
There is another consideration that can bedevil the best of our intentions. We are tackling the
electricity issue at a time that it is far from a stationary target and , if anything, it will be in an
accelerating mode. I would attribute this to things like deregulation, distributed generation,
environmental pressures and regulations coming into force in the short term, expected rise in
fossil fuel prices (maybe with the exception of coal-if you do not have to count environmental
requirements and carbon taxes) rise in population and a threat of global climate impact (believed
by many as probably the largest environmental danger if not THE largest danger).
We should not be looking at how electricity has been made up to now, but rather how it will be
made in the coming years. This puts a different spin on it and what and how data may need
to be gathered.
With this in mind, deliberating what time period this is aiming for has major implications (under
Macro). Additionally, including, new technology can not be given a secondary level of inclusion
for anything but the shortest time horizons ( I realize nobody said they would, but I found some
handwriting between the lines) which may or may not be enough to arrive at a usable end
product.
On the other hand, getting a fast start on gathering "some" specific data can be useful, most of
all for the purpose of moving up on the learning curve of what happens when you actually start
beating the virtual bushes for it. It may not be the that such data gets actually used but serves a
way to get a lot more insight into what is really needed, what can be had and what to do about
thedifference and to a data base design.
My feeling is that we don't know enough about the design of this activity to just go out and do it.
By no means am I suggesting this is not worthwhile. I am suggesting that this may be more like
the war against terrorism vs. Operation Rolling Thunder, and the Macro end will demand more
attention than I think we have allowed for it.
PURPOSE OF LCA DATA
My understanding is that Mary Ann feels that this data will not be used for the purposes of
comparing energy sources (and it is her workshop). My prejudice is that, coming from the EPA,
56
-------
it HAS to be able to be used for that. When the EPA says that this has more impact than that,
where is the impact being considered?
My impression is that it is at least at some national level, not at a given facility or even industry
sector, and the rest of the country is not involved. As soon as we start looking at electricity, we
are looking at ALL the industrial and domestic sectors. It is closer to considering "what if
everybody took this as EPA's advice?" . That is what this integration, multi-media, multi-impact,
LCA' ing is about, I believe.
As soon as more than one energy technology for making electricity is assessed, some kind of
defacto comparison has been made. Even if we personally do not let the two sets of information
come close to each other, this comparison is unavoidable. There is uncertainty about new
technologies and how to model them, that is true. There may be equal uncertainty about the
existing collection of grandfathered power plants if we look to a future with the things mentioned
above. I do not believe we can take a snapshot of "NOW" in the electricity sector and use it even
if it's a lot more convenient. To me, this illustrates the need for the workshop to air out these
issues at the Macro end.
HOW TO HANDLE "IN-PLACE CAPACITY"
"In-place capacity" should not be a separate issue, Energy is just about never where most people
live (Iceland may be the big exception). Coal and oil are shipped large distances. So is
electricity.
What's common here? The cost and impacts of the processes to do so. The same goes for wind,
solar, wave, biomass, and even geothermal. If something is "closer", it probably costs less and
probably requires less energy loss to get to the user AND maybe less of an impact. That's one
part that needs to be determined, distance and means of transport, vs. impact.
A little heard of issue is number of people these activities require and the impact that generates.
One could consider driving to work, the "work station" and its support and the infrastructure that
supports all of that and the electricity part of the above.
Lastly, the energy beast (and electricity in particular) is different (from, say, diapers) because
each and every supporting resource represents the use of energy and its impacts.
This seems to make a tightly intertwined web or tangle. Each little portion appears minuscule by
itself but there are a lot of them, they are not all of equal impact and therefore may be significant.
Without some specific information to the contrary, deciding to leave out areas is tricky if not
dangerous (Example: maintenance-a significant activity in boil-and-burn processes, still our most
popular technology). We need to ASSESS what actually goes there or even better, have a
certified expert at the workshop with info in hand.
57
-------
QUESTIONS, MODELS AND DATA IN LCA
NOTE FOR THE INTERNATIONAL WORKSHOP ON ELECTRICITY
DATA FOR LIFE CYCLE INVENTORIES
Gjalt Huppes
CML, Leiden University, Netherlands
huppes@cml.leidenuniv.nl
www. 1 ei denuni v. nl/cml/s sp/
Contents:
Summary
1 Questions, models and data
2 Questions
3 Models
4 Data
5 Unit processes and systems
6 Conclusions and recommendations
Summary
Data become information when they are gathered adequately for a purpose, as one envisaged
application. As several applications exist each requiring their own and different data, a first
step is to specify the purpose for which data are gathered. A first choice is that data are to be
used for improved decision making, by indicating effects of choices. This quite common
assumption implies that we use data of the past to indicate future effects, involving some sort
of causality as incorporated in a model structure. Depending on the question we want to
answer, different causalities are involved. Causalities can be arranged systematically only in
a model, so we should have models. The model, fitted with appropriate data, can indicate
future effect. Here the focus is on LCA-type of models, a model with a very simple structure.
Although depicting economic processes and their relations, they lack main economic
mechanisms like supply and demand relations. Some bandwidth still exists in the group of
simple LCA models.
There is a basic level of LCA model, where average yearly scores suffice to operate the
model, and to give answers on a specific group of questions. More sophisticated questions
require more sophisticated models. These make sense only if they also incorporate more
sophisticated data. Especially questions related to system dynamics are scientifically
interesting and have great practical importance. However, they lack a standardised model
framework to systematically guide data development. Optimisation models, the non-dynamic
ones, may be more easily operationalised but require an explicit normative framework: what
is to be optimised? Given questions on choices we want to specify effects for, and models for
answering them, then the requirements on data can be specified for making these models
operational.
For long term development, setting up data bases on unit processes should be prime focus,
kept separate as much as possible from model and methods dependent processing of such
data, and from aggregating them into (sub)system using some always contentious model.
Viewing current results from EGRID, a few conclusions can be drawn: keep unit process data
pure, absolutely separated from applications requiring allocation, subtractions etc, and make
them complete, including capital goods, maintenance and other overheads. They then may
serve as many purposes as possible, including a role in the causal analysis of LCA modelling.
58
-------
Suggested Reading
Dones R., Menard M., and Gantner U. "Choice of Electricity-Mix for Different LCA
Applications." 6th LCA Cases Studies Symposium. SETAC-Europe. 1998.
Dones R., Gantner U., Hirschberg S., and Doka G., Knoepfel I. "Environmental Inventories for
Future Electricity Supply Systems for Switzerland", PSI Report No. 96-07. 1996.
Ekvall T. and Finnveden G."Allocation in ISO 14041 - A Critical Review." J. of Cleaner
Production, Vol 9(2001). pp 197-208.
Ekvall T, Molander S., and Tillman A.-M. "Marginal or Average Data - Ethical Implications."
1st Int'l Conference on Life Cycle Management, Copenhagen, Denmark, Aug 2001. pp 91-93.
Ekvall T. "System Expansion and Allocation in Life Cycle Assessment: With Implications for
Wastepaper Management." PhD Thesis. Technical Environmental Planning. Chalmers
University of Technology. Gothenburg, Sweden. 1999.
Frischknecht R. "Allocation in Life Cycle Inventory Analysis for Joint Production," IntJofLCA
5(2), pp 85-95. 2000.
Frischknecht R. "Life Cycle Inventory Analysis for Decision-Making; Scope-dependent
Inventory System Models and Context-specific Joint Product Allocation: Section 5. LCA for
Decision-Making: The Advantage of Marginal Consideration." Excerpt (pp 47-79) from PhD.
Thesis Nr. 12599, Swiss Federal Institute of Technology (ETH), Zurich, 1998.
LCA: Livscyklusvurdering af Dansk El og Kraftvarme. "Life Cycle Assessment of Danish
Electricity and CHP." Summary. October 2000.
Michaelis P. "Royal Commission Environmental Pollution Study On Energy And The
Environment: Paper prepared as background to the Study Life Cycle Assessment of Energy
Systems." Centre for Environmental Strategy, University of Surrey. March 1998.
"Vattenfall's Life Cycle Studies of Electricity." Vattenfall, Stockholm, Sweden. October 1999.
59
-------
International Workshop on Electricity Data for Life Cycle Inventories
Summary of Feedback on Issues
Prepared by
Mary Ann Curran, US EPA
Timothy Skone, SAIC
Alina Martin, SAIC
18 October 2001
Introduction
In preparation for the workshop, invitees were asked to submit their views and ideas on five key
issues that relate to data collection for electricity. Written comments were submitted by nine
individuals.
North Carolina State University, USA
University of Capetown, South Africa
Consultants
2.-0 LCA Consultants, Denmark
ESU Services, Switzerland
Orion Corp, New Zealand
Electricity Suppliers
EDF, France
Vattenfall, Sweden
Researchers
ITT, Germany
US EPA
Also, several published pieces were submitted for consideration (listed in the bibliography).
The responses have been compiled and are summarized below. Each issue area as presented in
the original Issues Paper (August 2001) is presented in italics at the beginning of each section for
easier reference.
The following themes were repeated throughout the comments:
1. Focus is on LCI issues, to avoid confusion with impact assessment.
2. The approach for data collection of electricity depends on the objective of the study.
3. Transparency is essential in drawing boundaries, etc.
60
-------
Footnote: It is important to note that while at times the term LCA may be used, the focus of this
discussion piece and the follow on workshop is to discuss life cycle inventories.
General Remarks
Before discussing the responses that were given on the key issues, it appears that respondents fall
into two distinct camps when thinking about the modeling and application of the electricity life
cycle data. Some respondents, such as the power generators, are interested in creating an LCA
that focuses on the power generating facility, while others focus on the need to model electricity
production data that can be used in any product LCA. One respondent worded it this way:
"The distinction needs to be made between modeling electricity supply for incorporation into a
product/process inventory, and modeling to support decision-making within the electricity
supply industry itself (e.g., choice of a desulphurisation technology on a coal-fired plant)."
We may need to address both of these limits or any other variations in between, but the important
thing is to keep these objectives separate and clear.
It is very important to be clear about the inherent usefulness of the LCA approach and to focus
on where it has proven its strength. At the same time, LCA should be interfaced with other areas
(e.g. energy system modeling) and thus gain from synergies. Energy system models include
many hundreds of individual processes to generate a realistic picture of supply and demand
patterns over time. It is hardly possible to provide detailed LCA data for all these processes (do
we need them?). On the other hand, in terms of environmental impacts, most of the current
energy system models focus on CO2 emissions, and partly cover SO2, NOX and particles (do we
need others?). There is both a need and potential for further development of LCA methodology
related to energy supply for energy systems into a product/process inventory and modeling to
support decisions-making within the electricity supply industry.
Average versus Marginal Systems Modeling
Current LCA modeling represents an allocation of the total environmental burdens of a macro-
system (e.g., today's economy) to the life cycles of individual products and services. All such
LCAs are structured so that, theoretically, the results could be combined to form a total
response. The goal is to answer the question: "If we were to assign the total environmental
burdens caused by global demand for goods and services across all components of that demand,
how much burden would we assign to each unit of good or service?" Heijungs (1997) referred
to this question as "the attribution problem. " Thus, for electricity generation, LCAs assign, or
apportion, the burdens of a region's annual generation equally across each kWh of electricity
produced and consumed. Thus, if the annual generation for a region comes from equal shares of
particular energy sources, for example, hydro, nuclear, and fossil fuel prime movers, then each
kWh produced and used in this region would be modeled as an "average kWh, " produced from
1/3 hydro, 1/3 nuclear, and 1/3 fossil fuel. This is the approach taken in attributional LCA
modeling.
An evolution is taking place within the field of Life Cycle Assessment, away from models of
"average" systems that support retrospective analyses, towards models of "marginal" systems
that support prospective analyses. In contrast to attributional LCAs, prospective LCAs explicitly
attempt to characterize what the impacts will be of potential decisions. Thus, they are designed
61
-------
to provide insight about "what will happen if we decide A or B, " rather than "which product is
to blame for which burdens. " The processes whose levels of output will be impacted by a
decision or a change in demand are referred to as the "marginal" processes - those producing
"at the margin."
At first blush, the marginal modeling underlying prospective LCA may appear more complex or
data-intensive than the average modeling underlying attributional LCA. In practice, this is not
necessarily the case, and in fact prospective LCA helps take some of the arbitrariness out of
thorny LCI modeling issues such as allocation (Weidema 2001). In most if not all cases, the use
of LCA for decision support appears to call for adopting the prospective approach as far as
possible.
A number of inter-related questions arise in attempting to identify how the energy supply system
actually responds to changes in demand, depending upon characteristics of the demand change
including its location, timing, duration, and magnitude. In order to provide a sound basis for
prospective modeling of the electricity supply system, we must address the following questions:
• What do we know about how the electricity supply system responds to changes in
demand?
• How is this system's response to demand currently modeled, by what models and with
what accuracy?
• How should LCA incorporate these understandings and perhaps the results of these
models in its treatment of electricity?
Respondents recognize this as a critical issue that needs resolution. They agree that the choice is
crucial and can significantly affect the results of an LCA.
The purpose or use of electricity data in an LCA is also crucial to the issue of average and
marginal modeling. The purpose of electricity data can be grouped in to one of two categories:
1. Conducting an LCA of an energy system (e.g., power plant, distribution system)
2. Conducting an LCA of a product/process (i.e., electricity data is a system input)
The European LCI terminology of "background LCI
data" and "foreground LCI data" can also be related to
the two cases above. In the first case, electricity LCI
data would be considered "foreground LCI data"
because it is directly affected by decisions based on the
study. Where as the second case represents electricity
LCI data as "background LCI data" because it is not
directly affected by decisions based on the study, other
than the quantity of material inputted into the
foreground system. The distinction between the
purpose of electricity data in an LCA is key to many
62
In the Journal of LCA, The
SETAC-Eur Workgroup on Data
Availability defines marginal as:
short term (variation in demand in a
short term, calculated with no
change of the production facility)
or long term (increase or decrease
of the electricity consumption on a
long term, calculated with possible
to change of the production
facility). And if anyone can
interpret this for me, I will be
grateful.
-------
discussions concerning the use and appropriateness of average and marginal modeling
techniques.
Responses to the question of average versus marginal modeling can be broadly grouped into
three types of responses:
1. Applicability of average and marginal modeling of energy systems based on the scope
and purpose of an LCA
2. Effect of average and marginal modeling on estimating uncertainty in LCI data
3. Appropriateness (or lack their of) accounting for marginal changes in "background" LCI
data that would result from a product/process under consideration.
Each topic is summarized below.
Average -vs- Marginal Topic #1: Applicability of average and marginal modeling of
energy systems based on the scope and purpose of an LCA.
The applicability of using average or marginal modeling for electricity data is dependant on the
purpose and scope of the LCA. Based on comments received from the respondents, the selection
is not always clear but the following should be considered.
The purpose or use of the electricity data; Background or Foreground LCI data.
"Where electricity supply falls into the foreground system, a marginal approach will always be
warranted, whereas an average approach will often be sufficient where electricity supply falls
into the background system."
The scope of the LCA; short, mid, or long-term (e.g., days, months, years, etc.).
"Ideally a marginal approach is preferable in order to reflect 'real' conditions, they also
expressed doubt that detailed data, such as specific hourly information, is relevant for LCA's,
and that annual averages are straight-forward and sufficient. Most of the short-term and also
mid-term changes in the electricity source profile have a regular pattern, so that a reasonable
averaging should be used. In national/regional average LCIs, annual variability should be
sufficient as the variability between the technological systems is likely to dominate the overall
uncertainty/variability. For product-type LCA's, modeling complex behaviors goes beyond the
scope of LCA."
"There may be instances where long-term marginal data is relevant. For example, an LCA of
a device that operates only at a specific time of day, week or
season, will need to factor in long-term marginal data for this
specific energy supply, i.e. to distinguish peak electricity as a
separate product."
The overall impact of the electricity contribution to the LCA results.
"Often in electricity production, older technologies are used (i.e brought on-line) to supplement
mainstream power production during times of high demand. This is more cost effective than
increasing the production capacity by building new facilities.
63
-------
Unless a decision is a huge electricity consumer, a marginal approach may be less clear since a
given site at which a life cycle decision is to be made, may not be supplied now or in the future
by one electricity source. The marginal or prospective approach is actually more complex (than
the actual calculation of a hypothetical marginal MJ) because the assumptions are less rigorous
or certain. Will the marginal MJ be from electricity technology source A or maybe B? With
deregulation and new/non-traditional generation and the generally long construction/permitting
times, the marginal source may not even occur before an actual life cycle decision has been made
and then remade."
Q1: Agree or Disagree with the respondents?
Q2: Are marginal data that add detail to the inventory worth the cost?
Q3: How has electricity data been historically modeled?
Q4: How are life cycle decisions actually affecting electricity use at the site or
at the product use sites; is it increasing use or decreasing?
Q5: Are refinements from adopting a marginal approach (as compared to the
average) providing improvements in accuracy that are less than the
errors elsewhere in a system under analysis?
Average -vs- Marginal Topic #2: Effect of average and marginal modeling on estimating
uncertainty in LCI data.
The general consensus of the respondents indicated that marginal modeling provides the ability
to quantitatively estimate the uncertainty in LCI data, were as with average modeling
(specifically in the prospective case) uncertainty can only be best estimated qualitatively with
little premise. Therefore, to best justify the uncertainty in the results of an LCA, marginal
modeling would be the preferred approach. The following respondent excerpts are provided to
add context (and opposition) to this conclusion as it relates with stochastic (probability)
modeling approaches commonly used in LCA's of energy systems.
"Changes in the source profile can be incorporated by modeling short-, mid- and long-term
scenarios, where these can reflect changes in fuels and technologies (i.e. changes in the grid mix
for the average LCIs, and changes within the particular technology for the marginal LCIs).
Importantly, the stochastic (probability) models recommended to include process variability
should also include data uncertainty, i.e. the input probability distributions should include
uncertainty due to variability, as well as that due to the nature of the data (e.g. uncertainty in the
quality of future fuel sources). In this way, the uncertainty associated with the future scenarios
can be quantitatively reflected in the inventory, i.e. the fact that the long-term scenario has much
higher uncertainty than the short- and mid-term scenarios can be reflected. Although estimating
the uncertainty associated with the data is inherently subjective, there are methods which
mitigate this to some degree (Weidema and Wesnaes, 1996), and even a subjective estimate of
uncertainty is preferable to representing a highly uncertain future inventory with false accuracy,
as a mid-point LCI or a stochastic model only incorporating variability would."
64
-------
"Stochastic modeling approaches can be used to incorporate variability, i.e. the inventory is
presented as a range of probable output rather than a single mid-point. However, to be
meaningful, stochastic models need to be applied to the actual process models underlying the
inventory, where the actual variability in the data samples can be incorporated, and correlations
between inputs can be avoided by modeling the causal mechanisms."
"A further point regarding incorporating data uncertainty and variability in the LCI using
stochastic modeling is that this may force a marginal approach, or at least, force the definition of
more tightly defined average systems. This is because incorporating the variability within
systems averaging widely different processes can result in such high uncertainty (i.e. such a wide
range in the output) that no significant differences are discernable between options in a
comparative assessment"
"The marginal approach has its advantages and disadvantages. While the uncertainty of the
system being modeled is inherently lower (both in terms of its avoidance of using average data
and its avoidance of arbitrary allocation rules), the necessary data is not always readily available.
Furthermore, companies are more comfortable publishing LCI information as national or product
wide averages to protect confidentiality. The uncertainty of the average approach could be
improved by the use of improved reporting along regional rather than national averages, or broad
technology types, etc. This would allow a more informed determination of whether electricity
supply can appropriately be kept in the background system (i.e. the contribution to uncertainty of
the electricity LCI needs to be evaluated in light of the overall inventory uncertainty)."
Q1: Agree or Disagree with the respondents?
Q2: Does marginal modeling decrease the uncertainty in LCI results, or only
improve the transparency of the degree of uncertainty?
Q3: Is the level of detail as stochastic modeling necessary for "background"
and/or "foreground" electricity LCI data?
Q4: What is an acceptable level of uncertainty in both "background" and
"foreground" electricity LCI data?
Average -vs- Marginal Topic #3: Appropriateness (or lack their of) accounting for
marginal changes in "background" LCI data that would result from a product/process
under consideration.
When conducting an LCA (especially a prospective LCA) of a product/process, the
implementation or use of the product or process in question can effect the background electricity
LCI data, especially if the product or process consumes a significant amount of energy in relation
to the local energy grid. Responses to this topic were varied and mixed based on the
appropriateness of including these changes (i.e., assessing a product/process with the associated
environmental impacts) and the additional effort to account for them correctly.
The following example was given to demonstrate this view:
65
-------
"Consider the theoretical case where a steel company wants to construct an electric arc furnace
for re-melting of steel scrap. Such an industrial plant can require several hundred GWh per year
of base-load electricity. An LCA is performed to investigate the environmental implications of
locating the plant in different countries. Based on the ethical rule that good systems should be
supported, the LCA practitioner decides to use average data for Norwegian electricity production
in the study. This is based, to more than 99%, on hydropower. However, the electricity in
Norway, Sweden, Denmark, and Finland is freely traded on a common Nordic market. Except
for grid losses, the consequences of using electricity in Norway are essentially the same as using
electricity in Denmark or Finland. If the investment is made in Norway, more electricity will be
used within that country and less Norwegian electricity will be available in the other Nordic
countries. Despite the large electricity demand of the plant, the start-up of a new electric arc
furnace in Norway would still have a marginal effect, and no more, on the production of base-
load electricity in the Nordic countries. The electricity production that is affected by a marginal
change in the base-load demand is, in the short run, based on coal combustion. Hence, the short-
run consequence of locating the plant in Norway is that more electricity is produced in coal-
power plants."
As another respondent stated, "Advantages of the marginal (prospective) approach is that it can
reduce data collection efforts substantially (since only data for the marginal production is
needed, not data for the entire system), and it avoids arbitrariness in setting of system
boundaries, notably in relation to geographical and technological boundaries as well as in
relation to co-product allocation. The average (attributional) approach may be warranted when
seeking to allocate blame for past activities. The marginal (prospective) approach is warranted
when analysing the consequences of a decision, i.e. as a decision-support. The marginal
approach can also be applied to allocate blame for past activities by using historical data valid at
the time of the decision that led to the situation that you wish to allocate blame for."
"If we know the electricity supplier in a life cycle process, we try to find out the electricity
generation mix (percentage hard coal based generation, nuclear, hydro etc.) of this supplier. If
that is impossible, we use the average generation mix per year of the country or the region (IEA
Statistics). We use today's mix for both historical and future processes. Our attitude is that a
summation of LCA study results of every system should not extremely exceed the actual
environmental impact. You should not burden one electricity needing system for the fact that
marginal electricity generation is needed in an electricity generation system to secure deliveries
to all consumers."
Q1: Agree or Disagree with the respondents?
Q2: Can we identify regional grid mixes? And are these appropriate for LCA?
For what purpose?
Q3: Is it appropriate to account for environmental impacts (changes in LCI
data) resulting from background changes in electricity supply as a result
of a the product/process under review?
66
-------
Boundaries for the Electricity Generation Systems
LCAs attempt to approximate comprehensive treatment of the environmental, health and
resource burdens associated with product systems. In theory, this comprehensiveness entails
inclusion of "all significant" burdens (e.g., pollution releases, resource consumption flows, or
other impacts) of "all" causally-connected processes. Thus, the system boundary for a life cycle
inventory model requires a series of choices along two dimensions: environment and supply
chain. In the case of a life cycle inventory database concerning the electricity supply system, we
note the following boundary decisions which must be made:
• Which environmental flows and other data needed for impact modeling should be
tracked, how, and with what specificity, for processes in the electricity supply system?
How should the cut-off criteria be determined?
• Which activities and operations along the supply chain should be included? That is,
how wide and how broad should the system boundaries been drawn? (e.g., should
capital equipment be included? transport of workers to the production sites? service
sector inputs such as from designers, lawyers, accountants, advertising, etc. ?)
Decisions related to establishing specific cut-off criteria to set boundaries for particular
processes in the system under study are properly left to the goal and scope definition portions of
individual life cycle assessments, or to the protocol development phase of the LCI database
projects. This workshop will seek to pool insights from prior and current LCAs of electricity
systems concerning the broader boundary questions of what sorts of flows and what sorts of
processes are important to retain in general when modeling the electricity supply system.
This workgroup will address the multiple 'what is in, what is out? " sorts of questions, which are
fundamental to life cycle inventory analysis. The workgroup will address its topics in a pair of
sequential sessions.
The first session on boundaries will address environmental emissions and releases. Key
questions include:
• Based on prior LCA andnon-LCA environmental evaluations of the electricity supply
system, is there a set of air emissions for which reporting in LCI databases should be
required? Is it possible to define a recommended set of air emissions that would be
sufficient to include in databases? What are the principal data sources for the key air
emissions, and are there important differences among them from country to country?
• Water releases - the same set of questions as posed above for air emissions.
• Additional releases (e.g., radioactive isotopes) - the same set of questions.
• Other impacts (e.g., thermal enrichment of water, land use, etc.) - the same set of
questions.
A second work session will address setting the system boundaries which will be used to
determine which mass and energy flows will be accounted for. Key questions include:
67
-------
What inputs besides fuels are essential/important to include for different types of
generation?
How have input/output-based LCA analyses been used in the past to shed light on this
question, and what have their findings been?
What is the significance and suggested treatment of maintenance and repair inputs?
What is the significance and suggested treatment of supporting infrastructure?
Respondents comments can be divided into two categories: Environmental Flows and Activities
& Operations. Relevant excerpts are provided below for context.
Environmental Flows
"Of course, the interventions to be considered are constrained by the data availability (the degree
of development/demonstration of the technology, and the scope of the study), however, a default
list of interventions towards which to strive would certainly be helpful. This would also be very
useful in standardizing the use of aggregate interventions, e.g. TSP, TDS. Water-related
interventions where found to be particularly problematic when trying to compare across different
LCIs, e.g. use of categories such as "sulfates", "nitrates" etc., rather than individual
components."
"Also requiring standardization is how energy resources are defined in fossil fuel-burning
systems. This is required because different systems may burn very different quality fuels."
"It would be valuable to know other significant inputs, such as water, chemicals for the treatment
of water, and inputs related to repair/maintenance. At that point, one can decide what to report.
In addition, these other inventory parameters must be highly transparent."
"The European ExternE project on External Costs of Energy (as well as the joint US/EC Study
on Fuel Cycle Externalities, which was a forerunner of ExternE) started with a screening of a
range of pollutants and related impacts. Based on expert judgement (no formal selection
procedure), a set of some priority impacts and related pollutants were identified. External costs
(as an aggregated damage indicator) were very much dominated by greenhouse gases, NOX, SO2,
and particles (and related secondary substances, namely ozone, sulfates and nitrates). This
conclusion was quite robust. Although the small number of key substances was often criticized
for being inappropriate, other LCA studies for energy technologies with a more comprehensive
inventory are also dominated by the same set of pollutants."
"The picture will change with an increasing share of renewables, but the above mentioned
pollutants still dominate LCA results for renewables technologies because of the importance of
the conventional energy supply mix."
"Non-stack emissions should not be overlooked (e.g. dust from blasting during mining, and
blown from waste dumps). The impacts associated with solid waste management in coal-fired
systems are typically poorly assessed. Diffuse sources of water pollution (as distinct from pipe-
68
-------
discharges) are often overlooked. These include surface run-off from waste dumps and
stockpiles, and water collecting in open cast mining pits. Leachate from waste dumps and
stockpiles, as well as seepage from ash dams and pollution containment dams are similarly
neglected, although their significance can be considerable."
"Local scale impacts on ecosystem via land use, alteration of water systems etc. partly are the
dominant impacts for decentralized renewable energy technologies. In past LCA's,the
methodology does not treat land use adequately in LCA and is limited in addressing such very
local scale impacts. While recent developments explore the feasibility of site and time dependent
LCA, a key problem of local scale impacts is that it is not necessarily the technical
characteristics of a facility, but much more the specific environmental conditions at a given site
(soil quality, water regime, topography, ...) which determines the level of impact resulting from a
'unit' of environmental intervention. Technical parameters on the one hand and site specific
environmental conditions on the other hand are very closely interrelated and cannot be evaluated
independently any more. Summing up environmental interventions to an aggregated indicator is
therefore very difficult, if not impossible."
"If LCI/LCA data should be used in a labelling scheme for green electricity, I recommend to
additionally include local criteria which cannot be considered in standard LCAs. This is
successfully done within the "naturemade star" label for environmentally excellent electricity
from renewable sources (see http://www.naturemade.org). It covers aspects like river
ecosystems, agricultural techniques, visual impacts of wind power plants, etc."
"An adequate treatment of direct ecosystem impairment caused by hydroelectric power
production, coal, oil and uranium extraction is still missing. Nuclear waste disposal is another
issue where an adequate assessment within LCA is still missing. Noise seems to get more
attention in connection with wind power. Also here, broadly accepted inventory parameters and
impact assessment methods are still missing."
"For hydroelectric power plants, care must be taken not to neglect methane emissions if
substantial amounts of biomass have been drowned by the artificial lake."
"Pollution from electricity supply systems may occur in the very far future (e.g. long-term
radionuclide releases from abandoned uranium milling sites). How shall such emissions be
treated? How should we treat the treatment and disposal of radioactive wastes? How should
discounting be applied in LCI of electricity systems (positive, zero, negative discount rate)?"
Q1: Is a list of environmental flows, and other data needed for LCA,
already available?
Q2: Where should the boundaries be drawn for electricity generation
with respect to environmental flows?
Q3: What is the most commonly accepted system of nomenclature for
environmental flows?
Activities and Operations
69
-------
The SETAC Workgroup on Data Availability stated that "Good LCA data on energy production
comprise data on extraction, refining, transport and storage of fuels, electricity production,
distribution and consumption. The construction and demolition of power plants, as well as
processing and recycling of fuel wastes, are all part of an LCA for electricity production. For
nuclear, hydro, wind, and solar power the production of the equipment/facility has the largest
impact in the environment."
It seemed that not all respondents understood what was meant by infrastructure support.
However, one comment addressed capital equipment, transport of workers to the site, research
and administrative divisions, etc., saying these areas should be explicitly addressed when
defining system boundaries.
"Catalysts and especially precious metals are important due to the high environmental loads per
gram. Capital equipment is (obviously) relevant for new renewables such as photovoltaics and
wind power but also for hydroelectric power plants."
Q: In a product LCA, are infrastucture support activities negligible?
New and Non-Traditional Electricity Generation
As mentioned in the introduction, the electricity supply system is dynamic, with old technologies
being slowly replaced by new. As interest in minimizing the environmental impacts of electricity
generation increases, so will the ongoing development and evaluation of innovative electricity
supply technologies. One arena of potentially influential use ofLCAs of energy systems may be
in environmentally evaluating and comparing new generation technologies. Characterizing
them for LCA poses a whole new and different set of data and modeling issues.
There are at least three inter-related sets of issues involving LCAs of new and non-traditional
generation. The first is simply how to model the new technologies in LCA. For new
technologies that simply replace other point source generation facilities, this may not be a
challenge. But how shall LCA characterize distributed generation, from the average perspective
and from the marginal perspective?
The second set of issues relates to comparative evaluation of the new technologies, such as fuel
cells, from the LCA perspective. LCA evaluations of nascent electrical generation technologies
may inform policy and/or research prioritization among competing options. How can LCAs be
performed in a consistent, holistic, and valid fashion for these systems, which are marked by
high degrees of uncertainty and technological volatility, as well as scarcity of data?
The third set of issues relates to the way in which LCAs of product life cycles will tend to treat
new generation, and potentially to influence the demand for new capacity. The treatment ofin-
place capacity will probably need to be considered separately from the treatment of demand that
drives new capacity. An example concerns the proper treatment of flow-limited renewable
70
-------
energy, such as wind power capacity in place. From a prospective point of view, no change in
product demand (whether increase or decrease) will change the amount of electricity generated
by wind power capacity in place - its output is fixed by nature. So how, if at all, should this wind
capacity appear in the results of a prospective LCA?
Due to issues such as deregulation, distributed generation, environmental pressures and
regulations coming into force in the short term, expected rise in fossil fuel, population growth
and a threat of global climate impact, we are tackling the electricity issue at a time that it is far
from a stationary target. In addition to looking at how electricity is being made now, we should
also look at how it will be made in the coming years. This puts a different spin on it and what
and how data may need to be gathered.
Respondents expressed the following views:
"Many countries are setting goals to increase the use of renewable energy sources. For example,
Germany aims at a share of 50% electricity from renewable energy sources by 2050, resulting in
a drastic change in their energy system. Current LCAs for emerging technologies however most
often use the current electricity mix as an input to upstream processes. Besides the obvious
strive for getting the best available data for the relevant new materials (using tools like technical
learning curves, etc.), it is important to use the characteristics of a future energy system with an
adequate share of renewable energy as an input to basic processes in order to provide a more
realistic picture of the emissions resulting from energy supply. It is of course not easy to agree
on a specific future energy scenario, so the door is open for another source of potential
differences between LCA studies. Relevant excerpts from respondents are provided below for
context."
"The major opportunity is to establish the life cycle inventory with all inputs that track along the
supply chain back to natural resources. Transparent descriptions of boundaries and possible
multiple outputs should be made. The intersection with a current electricity power grid (either
measured as average or marginal) is likely to be a secondary life cycle issue. These new
electricity sources have already been studied in several life cycle reports."
"Boundaries for the inventories for new technologies must be consistent and include activities
such as transmission and distribution and agricultural activities. The inventories of point source
generation facilities must be inclusive so that they can be compared on a consistent basis to
distributed sources."
"The need to include a quantitative consideration of uncertainty is critical here. It is essential to
guard against the comparison of incomplete systems, i.e. incomplete inventories. A comparison
using incomplete or inconsistent data sets is likely to be more misleading than useful.
Qualitative LCA methods (such as Graedel's Streamlined LCA approach) may be of more value
than quantitative methods here."
71
-------
"The application of attributional LCIs to new technologies should not be a problem, as long as it
is recognized that materials of construction, life-times and use patterns will play a larger role in
the analysis than for conventional energy technologies. For prospective LCIs, perhaps these
could be viewed as incremental changes approached through discrete steps rather than load
changes?"
"If we want to compare specific energy technologies to evaluate their potential for solving
specific environmental problems, it is sufficient to look at the impacts normalized to a kWh at
the power plant's gate. If we want to take into account the capacity effect and a given supply
task, we might take into account a back-up technology. I would prefer however to look at the
full energy system with a certain share of renewables, which again requires the use of an energy
system model."
"There is uncertainty about new technologies and how to model them; that is true. There may be
equal uncertainty about the existing collection of grandfathered power plants if we look to a
future with the things mentioned above. I do not believe we can take a snapshot of "NOW" in
the electricity sector and use it even if it is a lot more convenient. To me, this illustrates the need
for the workshop to air out these issues at the Macro end."
"Energy is just about never generated from natural resources to use where most people live
(Iceland may be the big exception). Coal and oil are shipped large distances; so is electricity. If
something is "closer", it probably costs less and probably requires less energy loss to get to the
user AND maybe less of an impact. The cost and impacts of transportation should be addressed.
The same goes for wind, solar, wave, biomass, and even geothermal."
"The correct treatment of flow-limited sources such as wind or hydroelectric power refers to the
question about the relation between the macro- and microeconomic situations. To my
understanding, LCA is a tool which helps to allocate scarce environmental resources and scarce
"pollution rights" among competing products (like the price system is used to efficiently allocate
the scarce traditional production factors human power, capital and land). Hence, macro-
economic limitations like the ones mentioned should not have an influence on the way the LCA
is carried out."
Again, it was pointed out that the "correct" way of modelling new generating technologies
depends on the goal of the LCA. Do we need a dynamic model to assess the environmental
impacts of (the introduction of) new technologies in order to answer the question whether these
technologies should be favoured or not? Or, would it be sufficient to model a possible future
(steady) state (when the technology is more or less established)?
Q1: Can LCA's be conducted on new technologies for which production
data are not available?
Q2: Is there a need to develop a common future energy scenario that
considers renewable and distributed energy sources for use in
prospective LCA's?
Q3: How should distributed generation be accounted for in National or
Regional energy grid data?
-------
Transmission and Distribution
The transmission and distribution infrastructure component of the electricity supply system has
traditionally been accounted for in LCA in terms of the expected average line losses, or loss of
power due to electrical resistance in the system connecting the point of generation to the point of
use. The amount of this loss depends on the length of the transmission, the voltage at which
transmission occurs, the size of the conductor, and the manner in which electricity is transmitted.
Common losses range between two to five percent of power being transmitted (EIRRG 1998).
LCA researchers from Asian Pacific countries have identified additional issues associated with
what might be termed "fugitive losses" of electric power, which is un-meteredor un-identified
electricity consumption.
In fact, there may be important reasons other than line power losses to include the transmission
and distribution network within the scope ofLCAs of the electricity supply system - namely, the
environmental impacts of constructing, maintaining, and operating the systems themselves.
Some environmental concerns raised in connection with electricity transmission and distribution
lines include visual impacts, habitat impacts, noise (from high-voltage and ultra-high-voltage
transmission), and others (e.g., any remaining concern about effects of electrical and magnetic
fields?).
This work group will address both the energy losses associated with transmission and
distribution, as well as the impacts of T&D infrastructure itself.
Transparency was raised as an important factor in this area. The inclusion of traditional
transmission losses should be made transparent. An inventory table should include the notes
regarding whether or not transmission losses are included and what percentage loss was actually
used. It was reported that on the low voltage level (380V), losses more than 10% are not
unusual. On higher voltage levels, losses are much lower (below a few percents). Additionally
power switching stations (SFe-losses) and operation of high voltage transmission lines (TS^O-
emissions) are important. Also, illegal losses should be analyzed, but in places, the magnitude of
this is still unclear. Relevant excerpts from respondents are provided below for context.
"The existence of significant transmission system infrastructure effects leading to environmental
emissions is really unclear. Such effects are more likely to be the focus of separate studies."
"T&D relates to impacts that are expected to be important within the scope of actual life cycle
assessments. For example, South Africa has a problem with bird mortality when eagles insist on
nesting on the pylons. Other considerations include habitat loss, and herbicides used in
maintaining land. Also, we should not forget the emissions of SFe in power switching units and
the production of N2O of high voltage power lines."
Q1: Agree or Disagree with the respondents comments?
Q2: Can losses from T&D process be accounted for confidence (i.e.,
the level of uncertainty does not invalidate the use of the results)?
Q3: What types of environmental interventions should be considered
when modeling T&D systems?
-------
Co-Products and Allocation
Co-product allocation arises as an issue whenever a process produces more than one useful
product. For example, steam turbine systems may sell both electricity and low pressure steam as
useful products. When co-products are present, modelers must determine how much of the
burdens associated with operating and supplying the multi-output process should be allocated to
each co-product. Modelers must also decide on how to allocate environmental burden across
co-products when one is a waste stream that can be sold for other uses.
The ISO standards for LCA, particularly ISO 14041 related to inventory analysis, provide
methodological guidance on this issue. But they call for practitioners to attempt to avoid
allocation if possible; and secondly, to attempt modeling approaches which reflect the physical
realities (i.e. mass basis) of the process in terms of how inputs and releases would be altered if
the levels of output were altered for one or more co-products. In summary, proper application of
the ISO guidelines on allocation requires a physical understanding of the co-product production
processes.
The workgroup on co-products and allocation for the electricity supply sector could provide
considerable value to the worldwide LCA community by providing clarity and consensus on
allocation rules. It could also help by pointing to the data sources which characterize the
geographic details of which plants and plant types in which regions are producing how much of
the economically valued co-products. Such information will assist in assessing transportation
distances for other LCAs which include the use of these co-products.
The SETAC Workgroup on Data Availability stated that "In many cases energy has a big
influence on the results of LCA, which is the main reason why allocation methods must be
chosen and reported carefully. The chosen allocation method has to be transparent and suited to
the purpose of the study... The allocation methods that can be applied are the energy, exergy and
price method. In the working group report these methods are briefly discussed. The main
methods for allocation used today seem to be either the exergy or the energy method."
Some respondents simply stated that allocation should be performed following ISO 14041.
ISO 14041 requires the following procedure be used for allocation in multifunction processes:
• Allocation should be avoided, wherever possible, either through division of the
multifunction process into sub-processes, and collection of separate data for each sub-
process, or through expansion of the systems investigated until the same functions are
delivered by all systems compared.
• Where allocation cannot be avoided, the allocation should reflect the physical
relationships between the environmental burdens and the functions, i.e., how the
burdens are changed by quantitative changes in the functions delivered by the system.
• Where such physical causal relationships alone cannot be used as the basis for
allocation, the allocation should reflect other relationships between the environmental
burdens and the functions.
For allocation in open-loop recycling, ISO 14041 recommends the same procedure but allows a
few additional options. If the recycling does not cause a change in the inherent properties of the
material, the allocation may be avoided through calculating the environmental burdens as if the
74
-------
material was recycled back into the same product. Otherwise, the allocation can be based on
physical properties, economic value, or the number of subsequent uses of the recycled material.
The international standard does not include information on the effect of the different methods on
the life cycle modeling, for example the feasibility of the methods, the amount of work required,
or what type of information that results from the application of the methods.
The following comments regarding allocation highlight alternatives to the ISO 14001 guidance.
"Allocation problems can rarely be eliminated through subdivision. When it is possible, it is an
adequate procedure if decisions based on the LCA results has a significant effect on the
internally used function(s) but a small effect on the production volume of exported functions. In
other cases, it is too time-consuming and/or does not result in accurate and comprehensive
information about the environmental consequences of our actions. This conclusion is clearly an
adjustment compared to ISO 14041.
An allocation problem can be avoided through system expansion as long as there is an alternative
way of generating the exported functions and data can be obtained for this alternative production.
It is an adequate way of dealing with allocation when an action will affect an exported function,
if the data uncertainties are not too large, and if the indirect effects are important enough to be
significant for a decision. This conclusion is different compared to the recommendations in ISO
14041. The application of system expansion gives accurate results only when it is based on
accurate data on the effects on the production of exported functions and on the indirect effects of
changes in the exported functions. In many case studies so far, the system expansion has been
based on inaccurate data or assumptions.
Allocation based on physical, causal relationships is possible for multifunction processes where
the functions are physically independent of each other, if the internally used function is
marginally affected or if the environmental burdens can be represented by a linear and
homogeneous, mathematical function of the functions produced. It is an adequate allocation
method if decisions based on the LCA results have a significant effect on the internally used
function(s) but a small effect on the production volume of exported functions. In other cases, it
is too time-consuming and/or does not result in accurate and comprehensive information about
the environmental consequences of our actions.
System expansion, subdivision and allocation based on physical, causal relationships apparently
have a potential for providing accurate information on the consequences of our actions.
However, further research is required to realize this potential. Other allocation procedures
presented in ISO 14041 do not result in information about the consequences of our actions.
Hence, they should be applied only when a more accurate approach does not provide information
that is significant for any decision that may be based on, or inspired by, the LCA results.
In the light of these conclusions, the following, revised recommendations, are proposed if the
purpose of LCA is to increase our ability to anticipate the environmental consequences of our
actions:
• When the choice of allocation approach is expected not to be important for any
decision which is based on, or inspired by, the LCA results: we recommend that the
most easily applicable allocation method be used.
75
-------
• When the allocation can be important for a decision, but the possible effects on the
production of exported functions are expected not to be important: we recommend
that allocation be avoided through subdivision, that allocation be based on the
physical, causal relationships between the functions and environmental burdens, or
that an adequate approximation thereof be used.
• When the production volume of internally used and exported functions are
proportional and effects on the production of exported functions can be important for
a decision, but the indirect effects of a change in the production of exported functions
are expected not to be important: we recommend that all of the environmental
burdens of the multifunction process be allocated to the product investigated.
• When the indirect effects can be important for a decision: use system expansion or an
adequate approximation thereof.
A conclusion from our analysis is that when the production volume of the different functions
cannot be independently changed, system expansion, or an approximation thereof, is the only
approach that gives comprehensive information of the environmental consequences of our
actions. When the effects on the exported functions can be significant, but the uncertainties
regarding the indirect effects are very large, the LCA practitioner should either develop different
scenarios for the indirect effects, or clearly state that a course of action may have important but
unknown indirect effects on other life cycles."
"Allocation remains a requirement of a LCI and thus a clear picture of the byproduct or co-
product is needed. While the actual use of byproduct is often an economic decision, a LCI can
reflect a whole range of potential use. Again, a transparent description is essential. For example,
the North Carolina State University uses a mass basis and then tries to break down processes into
enough detail so that the emergence of a co-product is clear and can be allocated (the micro-
allocation approach)."
"The marginal approach of allocating avoided or additional burdens, only, appears the most
meaningful for prospective LCIs of energy systems." The following case study illustrates the
benefits of this allocation procedure.
No allocation problem regarding the electricity product was encountered within coal-fired
electricity production in South Africa (since no steam or heat is sold as a co-product).
However a different allocation problem arises due to the modern South African power
stations being designed to burn near discard-quality coal. Allocating burdens to the coal-
fuel supplying the station is found to be very significant for those stations supplied by
dual-producing collieries, which produce a high-quality export coal (requiring significant
coal preparation), as well as a low-quality power station coal. This can be regarded as
combined production, so can be modeled by a marginal analysis (keeping the production
of one product fixed, by varying the other). For some stations, this combined production
is made more interesting by the fact that the power station coal is made up of a blend of
run-of-mine coal and discard (the waste product from export-quality coal preparation).
The combined production is therefore able to reduce the mass of discard waste (the
disposal of which has significant environmental impacts), as well as avoid the waste of
energy resources discard coal represents. The "avoided" burdens approach is used here,
where the power-station coal is "credited" with the avoided burdens of discard disposal.
76
-------
Very significant for the electricity profile is that the discard-fuel source is essentially
burden-free, i.e. is not allocated any mining burdens other than the "avoided" burdens."
"In attributional LCA, a co-product is defined as one that contributes to the income of the
producer. This definition can also be used in prospective LCA, although here there is no need
for a sharp definition of co-products, since all outputs to technosphere, whether co-products or
waste for treatment, can be modelled in the same way."
With respect to forecasting future co-product allocations in a prospective LCA, "future market
potential for co-products can be assessed by the use of forecasting, as when collecting data for
any other future process."
The following discussion provides an overview of the lessons learned and approach used by the
European Commissions ExternE project (purpose is to determine the externalities associated
with fuel cycles). The allocation of impacts were discussed in detail for a combined heat and
power plant between the electricity and heat output.
"In the ExternE project the following allocation alternatives were discussed:
• Allocation based on operational characteristics (operation is heat or electricity
oriented ^ electricity or heat as by-product ^ benefits of joint production are
assigned to the by-product. We can also use the ratio of additional fuel input required
for producing the "by-product" to the total fuel input required for both electricity and
heat production.)
• Allocation based on thermodynamic parameters (energy or exergy content; several other
parameters are discussed in the literature, taking into account various characteristics
of the steam-process, but results achieved with these parameters generally are within
the range covered by the energy content and the exergy indicators)
• Allocation based on the final product's price (Assuming a perfect market, the final
product's price probably is the best indicator for the utility of the product. However, the
price of electricity and heat is highly dependent on the customers demand characteristics
and other non-technical parameters.)
As an example, the table below shows external costs (as an aggregated indicator of
environmental impacts) for heat and electricity from a 520 MW coal fired CHP plant in
Germany for different allocation rules:
77
-------
Credit for heat, substituting
oil fired domestic boilers
gas fired domestic boilers
Credit for electricity
Additional fuel input
Energy content
Exergy
Price
mEuro/kWhei
2.7
5.0
4.9
2.3
4.6
3.6
MEuro/MJ
(-7.8)
0.13
0.64
0.24
0.38
(note that these numbers are taken from a 1996 ExternE report. External costs in this example are
not based on the most recent ExternE methodology, but they still give a good indication on the
effect of different allocation rules)
In ExternE, it was decided to report results based on exergy-allocation as mandatory, and in
addition (optional) results based on energy content or price allocation to reflect specific
conditions under which the CHP plant might be operated.
"Current market values are delicate without considering the development of the overall markets.
However, I do not see too many difficulties here, except maybe power plants burning blast
furnace and coke oven gases.
Suggested co-products:
• coal: electricity, heat, gypsum (building industry), fly ash (cement production)
• oil: electricity, heat, gypsum
• natural gas: electricity, heat
• coke oven gas: disposal of "residual gas" of coke production, electricity, heat
• blast furnace gas: disposal of "residual gas" of iron melting, electricity, heat
• nuclear power: electricity, heat
• hydroelectric power plant: electricity, flood protection, irrigation, recreation, fishing
• photovoltaics: electricity, weather protection (if integrated in slope roofs, fa9ades)
• combined heat and power units (motors, fuel cells): electricity, heat
• biomass: electricity, heat
• waste incineration: treated wastes, heat, electricity, slags (cement industry)
• biogas from manure: treated manure, heat, electricity
• biogas from organic waste: compost, heat, electricity
• biogas from sewage sludge treatment: conditioned sludge, heat, electricity
78
-------
• geothermal power plant: electricity, heat, therapeutic baths (e.g., the Blue Lagoon,
Iceland)
General comment: in many cases, heat is not used but emitted to air and water.
Q1: Is there is a general rule or approach for allocation that is
acceptable for all parties involved and applicable for all situations?
ISO 14001?
Q2: If not for all situations, which method of allocations is best suited
for energy systems, and why?
Q3: Is there an accepted practice for determining when to include or
exclude a co-product (mass, energy, exergy, etc.)?
Other Issues — Transportation
According to the SET AC Workgroup on Data Availability, a good understanding of the technical
aspects of transportation systems is necessary to enable the proper use of LCA data for transport.
Inventory data for transport systems should be based upon a life cycle perspective. The final use
of fuels in transportation is much more important than oil extraction and fuel production. In the
final use the most important parameters are fuel consumption and the loading factor. Variation
in energy intensity per km is caused by the choice of transportation, i.e. ship, rail, road, or air,
and traffic congestion: fuel production (extraction, transport, refining, storage, transport),
transportation (engine type, fuel type, exhaust gas cleaning) and transport performance (vehicle
type/size, load, return trip, traffic conditions).
Waste Management
A good understanding of waste management is necessary to enable the proper use of LCA data
for waste treatment. Inventory data for waste should also be based on a life cycle perspective.
This means that emissions and resources from transportation and waste treatment are included
and described separately. Waste treatment is a complex chain of processes. The structure of the
chain depends on the waste source, country, waste treatment, transportation, etc. Providing a
simple guideline for data availability and quality for waste is difficult. There are various good
publications and case studies from different countries available on the web (see www-address of
EPA, ERRA and EU). These can be used as good information sources for models for LCA
waste data. Also a lot of information can be found in the proceedings of the international
workshop organized on LCA and treatment of solid waste (AFR-report 98, Swedish
Environmental Protection Agency). Environmental authorities in different countries have also
produced similar data sets.
Waste management can be subdivided into three waste modules: waste generation (municipal,
trading, construction, industrial, regulated), waste collection, and waste treatment (landfill,
biological/decomposition, incineration, recycling).
Conclusions
79
-------
The goal on consensus of modeling electricity supply is really a long-term issue. At the
workshop, we need to limit our sights to describing how an average or representative profile
should be created The availability of worldwide data is a smaller issue as there many data
sources now available that are useful for electricity LCI profiles.
The first step is to develop a smaller core of parameters that are technically clear and generally
of environmental interest. These parameters can meet the immediate needs of credible
information. Then a series of second and third rounds of information development can be
undertaken.
Bibliography
Tomas Ekvall, Sverker Molander, and Anne-Marie Tillman, "Marginal or Average Data -
Ethical Implications," printed in the proceedings of 1st Int. Conf. Life Cycle Management.,
Copenhagen, August 2001, pp. 91-93.
Tomas Ekvall, Sverker Molander, and Anne-Marie Tillman, "Normative Moral Philosophy and
Methodology for Life Cycle Assessment," prepared for the workshop, October 2001.
Tomas Ekvall, " System Expansion and Allocation in Life Cycle Assessment: With Implications
for Wastepaper Management," PhD thesis, Chalmers University of Technology, Gothenburg,
Sweden, 1999.
Tomas Ekvall and Goran Finnveden, "Allocation in ISO 14041 - A Critical Review," Printed in
Journal of Cleaner Production, Vol. 9 (2001), pp. 197-208.
Rolf Frischknecht, "Allocation in Life Cycle Inventory Analysis for Joint Production," J. of
LCA, 5(2), pp 1-11,2000.
Rolf Frischknecht,, "Chapter 5: LCA for Decision-Making: The Advantages of Marginal
Considerations," PhD thesis, Swiss Federal Institute of Technology (ETH), Zurich, Switzerland,
1998.
Peter Michaelis, "Life Cycle Assessment of Energy Systems," paper prepared for the Royal
Commission Environmental Pollution Study on Energy and the Environment, March 1998
(http://www.rcep.org.uk/studies/energy/98-6067/michaelis.html)
80
-------
Response to Review Comments
February 28, 2002
Tom Tramm
As I indicated in my voicemail, the report looks great. Attached are a few nit-picky suggestions.
Please call for more detail or other examples.
Section 2.1 - Marginal vs. Average
When are attributional and consequential LCI each appropriate?
Paragraphs 2 and 5 incorporate an example involving cement made of fly ash and clinker. We
may have used this example in our discussions, but it doesn't look right in print. It would be
more realistic if it referred to concrete made from fly ash and Portland Cement. Fly ash is used
to replace Portland Cement in concrete mixtures, usually on a one-for-one basis. Clinker often
refers to bottom ash that would not normally be used in concrete production although it does
have value in other applications.
Response: Reword paragraphs 2 & 5 as follows:
The rules used to define which processes are in or out of the system in attributional modeling are
those based on an observation of how materials and energy are flowing in the system at the
given point in time. For example, if cement concrete is made with 1 kg fly ash and 2 kg clinker 1_
kg Portland Cement per unit of cement concrete output, then the LCI model will show these flows
into and out of the cement concrete manufacturing process.
The rules used to define which processes are in or out of the system in consequential modeling
are those based on an estimation of how material and energy flows will change as a result of the
potential decisions or disturbances. In the fly ash example, if the output of fly ash is constrained
- namely, if it is fixed based on the demand for electricity - then increases in the demand for
high-fly-ash-cement concrete will not in the short run change the output fly ash. Instead, it
would increase the output of cement concrete made 100% from clinker Portland Cement. The
consequential LCI model would attempt to take such output constraints explicitly into account.
Characterizing the Response of the Electric Utility System to Demand Changes
Paragraphs 2 and 3 (Facts 1 and 2) are a little too precise. In both cases there are exceptions that
were discussed and ought to be acknowledged, even if they are probably insignificant to LCA
results.
For example, hydro plants drawing from impoundments are often dispatched to meet daily peaks
rather than base-loaded. Limited water supply means that there are only so many kilowatt-hours
available per year from a dam but you can usually have the power any time you want it. Hydro
units respond more reliably than more complex generating options, so they are scheduled to
come on to meet the daily peaks or address local environmental concerns. Every system has
special cases like these but they would not affect LCA results.
81
-------
The sentence in paragraph 2 would be more accurate if it said: "... output responses typically
occur at plants...." Since the exceptions are inconsequential, we should not be obliged to
discuss them in this report.
Paragraph 3 talks about capacity additions being the ones that will be the most profitable. There
are enough other factors in these decisions that maximizing profitability is seldom possible. For
instance, siting considerations usually preclude the most profitable plants. Some plants will be
added to assure reliable power supply in specific areas. Wind turbines and solar panels will be
installed to help satisfy a relatively small demand for renewable energy. It is enough to
recognize that only profitable plants will be built. Too many considerations are involved in
these decisions to make a simple statement.
Paragraph 3 would be more accurate if it read: "In the long term, the type of new capacity added
is generally the one which is estimated to satisfy the given load shape at the lowest overall cost."
This characterization is historically more accurate and will be more palatable to representatives
of state public utility commissions.
Response: Replace this section as follows:
Some members of the breakout group were familiar with realities of how the electricity system
(at least in the US) currently responds to changes in demand. Others were familiar with
responses of electricity systems in Europe. From their input, the following general facts were
captured:
1) In the short term, output responses typically occur at plants which have the highest variable
cost among those operating at the time of the demand change.
2) In the long term, the type of new capacity added is generally the one which is estimated (by
investment decision makers) to be the most profitable for the given load shape to satisfy the given
load shape at the lowest overall cost.
3) The future is irreducibly uncertain, while the electricity supply system is dynamic and
evolving. Thus, there are important levels of irreducible uncertainty concerning how the
electricity supply system will respond to demand changes, even if we used the most sophisticated
models available.
Section 2.2 - Boundaries and Flows
Boundaries
The first paragraph cites the "high-lift used to load the coal into the feed hopper at the utility
plant" as an example of a component that is not part of the dedicated infrastructure. We need to
use a stronger example. In most cases, coal-handling equipment has no other function and
should really be counted as part of the dedicated infrastructure. However, the cranes used to
erect the coal-handling equipment, boilers, etc. are usually used at many construction sites. The
example should be changed to something like "cranes used to erect the boilers and other plant
structures."
Response: Replace this example as follows:
82
-------
The participants evaluated which activities and operations along the supply chain/life-cycle
should be included for energy supply systems. Particularly, they discussed what should be
included (e.g., should capital equipment be included? transportation of workers to the
production sites? service sector inputs such as from designers, lawyers, accountants, advertising,
etc.?). The consensus was to include infrastructure for only dedicated resources. For example,
the material used to construct a boiler used in a coal-fired utility plant should be included, but the
materials used to construct the high lift used to load the coal into the feed hopper at the utility
piaft^cranes that are used to erect boilers and other plant structures would not be included.
Likewise, impacts from workers traveling to and from work should be excluded. This is not a
hard-and-fast rule, but more a general rule-of-thumb to be used in drawing boundaries for energy
supply systems. The potential impact from infrastructure operations should always be evaluated,
even on a cursory level, to support their exclusion with confidence.
Benoit Maurice
I find your document really good and have add some comments on 2 paragraphs: Characterizing
the Response of the Electric Utility System to Demand Changes and Environmental Flows
Under "Characterizing the Response of the Electric Utility System to Demand Changes," add the
following:
1) Compare to others products like steel or car, electricity has a specific characteristic: it cannot
be stocked : at any moment, production must be equal to the sum of consumption and
transmissions losses. Every day, the load shape has big variations, due to specific use like light
for example. To produce electricity, utilities should have different power plants which are able to
adapt their production to the consumption, this means to produce electricity as base load, (e.g
nuclear energy), semi-base-load (e.g coal, gas, fuel power plant..) and peak load (e.g. gas
turbine). This element has to be taken into account when one try to characterize the response of
the electric utility system to demand changes. A « base load use » or a « peak load use » will not
have the same answer. To characterize the production, LCA practitioners should use rather than
simple arguments, model developed for electricity planning which integrate most of the time
such parameters.
Response: Add a new paragraph at the end of the section, as follows:
In addition, it is noted that in contrast with many other products, electricity has the specific
characteristic that it cannot be stocked directly. At any moment, production must be equal the
sum of consumption and transmissions losses. Throughout the day, the load shape varies greatly
due to increasing and deer easing use, such as lighting at night. To produce electricity, utilities
typically have different power plants which are able to adapt their production to the
consumption, producing electricity as base load, (e.g., nuclear energy), semi-base-load (e.g.,
coal, gas, fuel power plant) and peak load (e.g., gas turbine). This element has to be taken into
account when one tried to characterize the response of the electric utility system to demand
changes. A "base load use " or a "peak load use " will not have the same answer. Rather than
using simple assumptions to characterize electricity production, LCA practitioners should model
for electricity planning which allows for the integration of such parameters.
83
-------
Under Environmental Flows, add the following flows « As, Ni, and Pb » to air emissions. These
substances are widely study in different research programs.
Concerning water emissions, the list is too long : I would limit the metal emissions to Pb and Hg.
BOD (5,7,10) is certainly also too much. Most of the time, BOD 5 is only available.
Furthermore, BOD like COD are indicators of water quality rather than a real flow.
Most of the time, limitation on temperature depend of the temperature of the river: for example,
administration may authorize to emit water with a delta of temperature of 8°C to temperature of
the river. Then the temperature of emission of water change from summer to winter.
Response: Change Exhibit 2 as follows:
Exhibit 2. "Minimum" List of Environmental Flows for Energy Supply Systems
Resources
Water (location & type)
Fuel (in ground)
Minerals (in ground)
Biomass (harvested)
Land use (area & location)
Wastes
Solid waste
Radioactive Waste (H, M, L)
Hazardous Waste
Other Releases
radionuclides
Air Emission
CO,
CO
PM(10, 2.5)
CH4
SOX
NOX
NH3
Hg,Pb
VOC (MM)
Dioxin
PAHs
SF6
HFCs
Water Emissions
Chemical oxygen demand
(COD)*
TDS
Total suspended solids
(TSS)
Biological oxygen demand
(BOD) (5, 7, 10)*
Flow
Temperature change,** or
thermal loading in energy units
NH3 (as N)
Total Kjeldahl nitrogen
(TKN) (as N)
NO3, NO2 (as N)
Polycyclic aromatic
hydrocarbons (PAH's)
Phosphates (as P)
Cu, Ni, As, Cd, Cr, Pb, Hg
* COD and BOD are indicators of water quality rather than flows
** Limitation on temperature depends on the temperature of the river
Concerning the comment on the number of water emissions, lack of availability as the sole basis
for this comment is not enough to make the change. More discussion on this point may be
needed.
Bev Saur and Bill Franklin
We have reviewed the workshop summary document and made some comments and suggested
revisions (see attached document). Overall we found it to be a good summary of the activities
and outcomes of the workshop. We feel very strongly, however, that the strong link between the
purpose of the workshop and the work plan for the US LCI database (DB) project must not be
overlooked.
84
-------
As you are certainly aware through your participation in the advisory committee to the US LCI
DB project, much of the work on characterizing US electricity generation, transmission, and
distribution (at least for traditional generating technologies and some of the better established
emerging technologies such as wind and solar) has already been identified as a top priority for
the database project. It would be most efficient and provide the most benefit to potential users if
efforts to develop LCI data for electricity generation are done in collaboration with the US LCI
DB project rather than independently.
For non-traditional technologies such as fuel cells that are not significant contributors to national
grids, NREL is most likely the organization with the best knowledge of NNT technologies and
would be the best source of data. These data may not initially be included in the US LCI DB
project; however, we would recommend using the US LCI DB protocol where possible in
developing NNT data so that they are compatible for future incorporation in the database.
Similarly, other areas that are of interest to the electricity industry, such as the response of the
system to daily and seasonal dynamics in supply and demand, but of lesser usefulness to the US
LCI DB project, may be appropriate to evaluate independently, again using the US LCI DB
protocol. Incidentally, the Protocol was reviewed independently by Patrick Hofstetter and others.
We feel that it is imperative to avoid the duplication of effort and potential incompatibilities in
methodology, data format, etc. that would inevitably result from independent efforts to develop
LCI electricity data. The US LCI DB project already has the commitment and support of DOE,
the Navy, GSA, and private industry, and NREL has established and is maintaining the website
for dissemination of information on the US LCI DB project. Independent efforts to develop LCI
electricity data would be detrimental to the success and usefulness of the US LCI DB.
Perhaps it would be useful for the next steps section of the report to include a reference to the US
LCI DB project because it is now moving into Phase II and the number one priority is the fuels
and energy database development. For your information I am sending that section of the NREL
report, which is now at NREL for review and will very shortly
be forwarded to full advisory committee for comments.
Bev and Bill attached the draft document with comments, which are summarized here:
1. Insert "life cycle" before inventory in the Abstract.
Response: Accept change.
2. Delete "life" before LCI (redundant) on page 1.
Response: Accept change.
3. Delete an excess space and insert a missing quotation mark on page 2.
Response: Accept change.
4. Insert "who" before "were unable to attend." on page 3.
85
-------
Response: Accept change.
5. Additional wording on page 5: Marginal disturbances or perturbations are infinitesimal
disturbances; e.g., installing one new end-use that causes an incremental increase in demand
for electricity.
Response: Accept change.
6. Two changes on page 7:
Correct misspelling of "been" to "be" and
Invert "for only" to read "The consensus was to include infrastructure only for dedicated
resources."
Response: Accept change.
1. Change "process" to "processes" on page 8.
Response: Accept change.
8. Page 9: Exclusion of these environmental flows should raise concern towards about the
comprehensiveness of the LCI data set, and
under Exhibit 2: (comment: may want to define some terms such as PAH's, TKN, etc.)
Response: Accept change.
9. Page 10: "Practitioners must also decide en-how to allocate environmental burden across co-
products when one is a waste stream that can be sold for other uses."
Change i.e. to e.g.: (e.g., mass basis)
Response: Accept change; also omitted the reference to "mass basis" which in fact ISO
explicitly does not call for.
10. Heading for section 2.4 - insert (NNT).
Response: Accept change.
11. Page 12: Reword sentence to: "For example, because operating emissions may be very low
or essentially zero, the predominant source of emissions associated with the generating
technology may be construction emissions, which are problematic to allocate allocating
construction omissions over the functional unit of kWh is problematic."
Response: Accept change.
12. Page 12: "For a large penetration into the grid, LCA practitioners must take into account that
DGs do not produce point-source emissions as do large central generators." What type of
emissions do they produce? "Additionally, depending on the reason for a DG installation,
the functional unit may not be kWh." Example would be helpful.
86
-------
Response: The intent of the statement was to say DG 's are more spread out geographically than
large, point-source generators. The paragraph was rewritten as follows:
The drivers for distributed generation are the demand for reliable power, the desire to avoid
down-time costs, and the mitigation of significant up-front capital expenditures for large
generators and transmission and distribution infrastructure. Distributed generators (DGs) are
typically small, and may use fossil or renewable fuels. For a large penetration into the grid, LCA
practitioners may take account, during impact assessment, of the fact that emission source
locations are distributed over a large geographic region as well. Additionally, depending on the
reason for a DG installation, the functional unit may not be kWh.
13. Page 13: If you do can assemble life cycle inventory data for a technology.
Response: Accept change.
14. Page 14, "discussions-points"
Response: Accept change.
- The workgroup on Boundaries created a first cut at listing a "minimum" list of environmental
emissions that should be included in the inventory.
• The workgroup on New & Non-Traditional Technologies noted that, despite difficulties that
arise in conducting LCAs on renewable generating technologies, due to uncertain operating data,
any database on electricity must be flexible enough to include different non-conventional
stressors, e.g. bird kills, land use.
Response: Accept change, except land use was omitted here since its inclusion is becoming
conventional.
Rolf Frischknecht
2.1 attributional vs consequential:
The third recommendation might be described more precisely as it depends on the time scale on
which the IkWh change occurs (short term, long-term, very long-term). I attach an extract of my
thesis where I elaborate the use of consequential models in decision situations.
Response: Changes were made throughout this section to reflect the reviewer's comments.
2.2 Boundaries:
I am not very happy with the "minimum list" as already stated in the plenary session.
Radionuclide emissions from NPP, reprocessing plants and uranium mines should be added to
the minimum list. Electricity is NOT a resource (but a product like many others as well). SF6 is
missing as an important pollutant in electricity distributing systems. HFCs are missing although
important in cooling equipment for underground coal mines.
87
-------
Response: Radionuclides, SF6 andHFCs were added to the table. Also, electricity was omitted
since this is a list of flows to/from the environment, not flows from other processes in the
technosphere.
2.3 allocation
There was nearly any discussion about the usefulness of a distinction in allocation procedures
between a consequential and an attributional approach. I doubt whether such a distinction is
necessary and meaningful. I attach a paper published in the Int.J.LCA which covers exactly this
topic and where I describe a managerial economics-oriented approach to allocation illustrated
with combined heat and power production.
I would appreciate it very much if you could include the two references attached into the
document:
FrischknechtR., 1998. "Life Cycle Inventory Analysis for Decision-Making; Scope-dependent
Inventory System Models and Context-specific Joint Product Allocation: Section 5. LCA for
Decision-Making: The Advantage of Marginal Consideration." Excerpt (pp. 47-79) from PhD.
Thesis Nr. 12599, Swiss Federal Institute of Technology (ETH), Zurich
Frischknecht R., 2000. "Allocation in Life Cycle Inventory Analysis for Joint Production," in Int
J of LCA 5(2), pp 1-111.
Response: These references are included in the website created for the workshop, as well.
Philippa Notten
I appreciate my comments come too late, but just in case it may be of use/interest to you, I have
attached the summary of what I saw as the important points of the Marginal vs Average group's
discussion that I drew up for Jim Petrie (he sponsored my participation at the workshop).
My only problem with the report write-up is that I don't think the distinction between
"Attributional" and "Consequential" is particularly clear, and was sorry not to see the "snappy"
definitions we argued out appearing, i.e:
Attributional: "How are things flowing in the chosen temporal window?"
Consequential: "How will flows change in response to decisions?"
To my mind, the write-up does not make it clear that these terms were proposed to replace the
"marginal" vs "average" distinction of before (i.e. that an attributional approach requires average
data and arbitrary rules and boundaries, which the consequential approach avoids).
Another small comment is that I would disagree that the distinction between electricity LCI data
for use in other LCIs or for use in comparing electricity generation options is not "particularly
interesting, important, or clear." The "consequential" vs "attibutional" discussion is only
relevant to the former, since a consequence- or decision-oriented approach is the only valid
approach for technology choice studies (but perhaps that is only my opinion). To my mind, of
greater interest is the inevitable mix of average and marginal data that creeps in even when a
consequential approach is taken, due to the lack of applicable data (however, that we did not get
round to discussing).
88
-------
Attachment:
. Marginal vs. Average
A significant portion of the group's time was spent on defining what exactly we were
discussing. It was agreed that "marginal" was not a good word, as what was actually
meant was "focussing on changes or consequences" (be they large or infinitesimal). It
was therefore decided that "consequential" LCA was a better term. The discussion
slipped from discussing merely the use of average or marginal data, to the framing of the
whole LCA problem. The topic was therefore recast as "attributional vs. consequential",
which incorporated the wider issues of boundary definition and allocation practices, in
addition to the type of data used. Once the terminology had been agreed upon a very
large portion of the time went into explaining/defining exactly what was meant by each
approach. The following summary statements were (painstakingly) hammered out:
Attributional: "How are things flowing in the chosen temporal window?"
Consequential: "How will flows change in response to decisions?"
Response: Insert a version ofPippa 'sparagraph in the report to capture the evolution of
the discussion. An appropriate place is at the end of the section on "terminology" (page
5), just before the section "When are attributional and consequential LCA each
appropriate."
The remainder ofPippa's comments are a summary of the discussion, more than a comment for
suggested change.
The key feature of the attributional approach is that it cannot avoid arbitrary rules, e.g. in
defining the system boundary (both spatially and temporally), as well as in deciding
which processes and environmental interventions to include. A notable difference in the
definitions as cast here is that an attributional approach can be used in a prospective LCA
(i.e. LCAs using average data are not only applicable in an retrospective or historic
sense). The consequential approach on the other hand avoids a rule-based approach by
focussing only on the consequences of the change.
Taking a consequential approach was not well-received by all participants, although
strongly supported by a few. It was clear that the tool of LCA is seen by many as a
summary statement of environmental performance, rather than as a decision support tool.
A rule-based approach (as the attributional approach demands) was generally not seen as
a particular problem, as long as a standardised set of procedures (such as laid out in the
ISO standards) was followed. Much of the opposition to the consequential approach
appeared to stem from the two approaches being understood as completely incompatible
approaches. It became clear in discussion that many participants were thinking
consequentially without really realising it (i.e. by looking at the difference between two
attributional LCAs), and that more "converts" to a consequential approach could be won
by focussing on the similarities between the two approaches, rather than giving the
impression that everything that one currently undertands about LCA needs to be "thrown
out the window" when undertaking a consequential LCA.
89
-------
An important aside, not fully explored by the group in the time available, was brought up
by Tomas Ekvall. He presented some interesting reasons why an attributional approach
may sometimes still be more appropriate than a consequential approach, including issues
of fairness and communication (see paper, "Marginal or Average Data - Ethical
Implications", available on website).
The bulk of the group discussions thus did not particularly pertain to the electricity sector
but rather larger LCA methodological issues. In the final group session the discussions
did eventually turn to original topic of electricity data, but discussion got somewhat
"bogged down" in explaining electricity demand curves and in trying to determine the
marginal technology for US power production. It was agreed that different marginal
technologies will be identified depending on the type of load requirement (e.g. peaking or
base load demand), and that a short and long term marginal for a change in base load
demand needs to be identified. Whilst discussions focussed on the typical daily load
curve, it was interesting that Benoit Maurice (EDF) disagreed with the US expert (in the
absence of utility people, the sole input was from a consultant to the US energy industry,
Tom Tramm). Benoit's argument was that in France the load curve changed according to
season (e.g. the nuclear plants are taken down for servicing during summer), so a simple
daily load curve can not be identified. His comments on the complexities in identifying
marginals were more or less echoed by Caroline Setterwall (from Vattenfall, Sweden) in
the report-back session discussion.
Discussion around taking an attributional vs. consequential as it pertains to data
collection and data quality was briefly raised towards the end of discussions. The
important point that average data is inherently of lower quality was finally able to be
raised, which could be clearly demonstrated by trying to determine the average US
electricity mix (the high degree of inter-linking in the US grid means that a state- or
region-based mix is highly arbitrary). Although a strength of the consequential approach
is that it minimises data collection (i.e. data is only required for the processes actually
affected by the decision), concern was expressed that the amount of "meta-data" required
to determine the marginal technologies may be significant.
The only concrete recommendation that could be agreed upon by the group was that any
database that is constructed (electricity or otherwise) should be technology orientated (i.e.
data should not be averaged over technologies and markets). Such a database format suits
both approaches, since for the consequential approach the necessary technology can be
selected, whilst for the attributional approach the desired mix/average can be taken.
Concern was expressed about data confidentiality (i.e. companies are happier to release
data as sector or regional averages).
Greg Keoleian
The report is well written and well-organized. It is very effective in capturing the presentations
and group breakout comments.
I have a few specific comments based on my review:
90
-------
Page 7: "3) Feasibility studies which apply energy system models are needed in order to generate
LCI results for a 1 kWh change in demand for different systems."
A 1 kWh change in electricity demand would be lost in the noise for a grid system. The problem
is that the variability in the baseline for a grid even at a given time in the day is obviously much
greater than 1 kWh. This points to the complexity in the consequential analysis. The addition of
a new manufacturing facility, however, might be a more reasonable load size that would show an
effect.
Response: Reword this bullet as follows: "3) Feasibility studies which apply energy system
models are needed in order to generate LCI results for a 1 kWh noticeable change in demand for
different systems." Changed to "modest incremental" change.
One challenge with the consequential studies lies with identifying which processes need to be
modeled as incremental. For example, if a new natural gas power plant is constructed to supply
a new manufacturing plant do we need to model potential incremental changes in steel
production since steel is used to construct the plant? Experience can help guide the modeling,
but these types of issues need to be addressed.
Response: Since this particular example was not discussed during the workshop, new wording
will not be added. The idea being expressed here is captured in the report elsewhere.
P 12
"bird kills versus fossil fuel consumption and climate change" I would recommend dropping
versus fossil fuel consumption and climate change because these also apply to renewable
systems at least until renewable infrastructure is produced exclusively using renewable energy.
Response: Reword the sentence as follows: "Additionally, because some impacts can be very
different than those from traditional generators (e.g., land use and bird kills versus fossil fuel
consumption and climate change), the database must be flexible enough to include different
stressors.
91
-------
Uses and users of
electricity LCA information
Bo Weidema
2. -0 LCA consultants
-------
Uses of LCA
long term
5 years
from now
now
historical
product legislation
societal action plans
product development
Strategic
peiforpran
product standards
taxes &. subsidies
Tactical
marketing claims
employee and/or supplier
requirements & incentive
ecolabellmg criteria
generic consumer
information
Operational
hot-spot identification
oduct declaration
area
specific
qeneric
-------
Data requirements related to uses
• Attributional LCAs: Statistically representative,
historical averages
• Prospective LCAs: Data representative of the
processes affected by the studied changes in
production volumes
Tactical applications: Data subdivided by plant,
company, geography or technology
Strategic applications: Data based on modelling,
which can explain the causal relationships between
inputs and outputs
-------
Asian Electricity LCI database
At sushi IN ABA and Masayuki SAGISAKA
Research Center for Life Cycle Assessment,
Advances Industrial Science and Technology (AIST), Japan
0. Asian background data acquisition
For building LCA researchers' network and promoting LCA activities in Asian district, AIST is
supporting to implement LCI case studies with financial assistance by the government of Japan. In
FY2000, this project has ranched with five countries (Japan, Korea, Malaysia, Thailand, Taiwan;
alphabetical order). In this year, expanding member countries to Indonesia, Singapore, Vietnam, the
Philippines and Australia, we are compiling the background data for LCA.
The preliminarily acquired data were discussed among the researchers from member countries
last year. In this report, only the summary of the results are reported since primitive data have to be revised
and upgraded and they must make the readers confused as well as their confidential restriction. We believe
these discussing members in Asian region will become key LCA promoters and develop original research
activities in this region for the progress of LCA in the world.
1. Japan[11
Since most industrial processes consume electricity, it is quite important to develop reliable
inventory data for electricity. In Japan, 10 electric companies supply electricity to the various regions.
There is, however, a problem that only a few figures concerning emissions related to electricity have been
reported. So, Matsuno et al. developed process models of power plants for Japanese situation, which
simulate the mass flows and estimate the missing figures of emissions dependent on technical parameters
of the plants and fuels. Life cycle inventories for the electricity grid mixes of the 10 electric companies in
1997 were developed. Emission of C02, S02, NOx, CH4, CO, Non-methane volatile organic compound
(NMVOC), dust (all particulates) and heavy metals (Ni, V, As, Cd, Cr, Hg, Pb, Zn) from power stations as
well as those from fuel production and transport were investigated. Other pollutants into air, emissions to
water, solid wastes, radiation and radioactive emissions from atomic power stations were not included due
to limitation of available data.
Direct C02 emissions related to 1 kWh of electricity distributed by companies ranged from 0.21
to 1.0 kg/kWh (average value: 0.38 kg/kWh). Direct emissions of S02 and NOx from power stations
related to 1 kWh of electricity are 2.5x10-4 and 2.2x10-4 kg/kWh in average, respectively. S02 emissions
-------
calculated in this work were somehow large compared with those reported by electric companies.
Detailed information concerning total sulphur content in oil consumed in each oil-fired power station are
required for exact calculation of S02 emissions from oil-fired power stations. In addition, the ratio of
sulphur that goes into slag in combustion must be investigated further. The average amounts of CO, CH4,
NMVOC and dust emissions were 5.0x10-5, 8.2x10-6, 1.8x10-5 and 6.8x10-6 kg/kWh, respectively.
Heavy metal emissions from power stations were in the order of 10-9 to 10-8 kg/kWh. Detailed
information concerning heavy metal content in oil and coals consumed in fossil fuel power stations are
further required for improved assessment of heavy metal emissions. Contribution of fuel production and
transport to total C02 emission was relatively small. On the other hand, contributions of fuel production
and transport to total S02 and NOx emissions were relatively large. In the case of CO, NMVOC and dust,
emissions in fuel production and transport were predominant to total emissions. Heavy metal emissions
into air during production and transport of fuels were in the order of 10-8 to 10-9 kg/kWh.
Table 1 (a) Emissions into air related to 1 kWh of electricity distributed by each Electric Power Company
CO2 SO2 NOx CO Methane NMVOC Dust
(
HOKKAIDO Electric Co.
TOHOKU Electric Co.
TOKYO Electric Co.
CHUBU Electric Co.
HOKURIKU Electric Co.
KANSAI Electric Co.
CHUGOKU Electric Co.
SHIKOKU Electric Co.
KYUSHU Electric Co.
OKINAWA Electric Co.
Average of 9 Electric Co.
10 kg/kWW
5.4
5.9
3.8
4.6
4.8
3.0
7.9
4.5
4.1
10
4.4
(10 kg/kWW
11
6.2
4.6
3.8
4.5
3.4
5.9
8.0
3.4
22
4.7
UO kg/kWW
8.9
6.6
5.0
4.2
5.0
3.9
8.7
7.5
5.2
15
5.3
(10 kg/kWlx)
1.3
1.3
1.1
1.1
1.1
0.79
2.5
1.2
1.1
2.3
1.2
(10 kg/kWW
7.6
9.6
6.2
7.5
7.0
4.6
12
5.4
5.0
14
6.8
UO kg/kWW
4.2
2.1
1.9
2.4
2.1
1.9
3.6
4.5
1.1
7.6
2.2
UO kg/kWW
9.6
5.8
3.1
3.4
6.1
3.0
7.8
5.8
4.5
13
4.2
1) Average emissions of HOKKAIDO, TOHOKU, TOKYO, CHUBU, HOKURIKU, KANSAI, CHUGOKU, SHIKOKU, KYUSHU electric companies
Table 1 (b) Emissions into air related to 1 kWh of electricity distributed by each Electric Power Company (continued)
Ni V As Cd Cr Hg Pb Zn
(10"7kg/kWh) (107kg/kWh) (10\g/kWh) (10"9kg/kWh) (108kg/kWh) (10*kg/kWh) (10"8kg/kWh) (108kg/kWh)
HOKKAIDO Electric Co. 1 1 19 2 2 10 6 9
TOHOKU Electric Co. 0.7 0.9 7 1 0.8 7 2 3
TOKYO Electric Co. 0.8 1 8 1 1 4 3 2
CHUBU Electric Co. 0.9 1 5 1 0.7 4 2 2
HOKUEIKUElectric Co. 0.9 0.5 8 0.8 1 734
KANSAI Electric Co. 0.6 0.9 8 1 1 433
CHUGOKU Electric Co. 2 0.8 7 1 0.9 8 2 3
SHIKOKU Electric Co. 31811533
KYUSHU Electric Co. 0.4 0.4 8 0.8 1 533
OKINAWA Electric Co. 4 3 6 1 1 12 2 4
Average of 9 Electric Co.* 0.9 0.9 8 1 1 533
*) Average emissions of HOKKAIDO, TOHOKU, TOKYO, CHUBU, HOKURIKU, KANSAI, CHUGOKU, SHIKOKU, KYUSHU electric companies
-------
LCI for electricity of Japanese electricity grid mixes in 1998
Recently, Sugita et al. updated LCI for electricity of Japanese grid mixes using the same
methodology, based on the statistics of 1998. In his work, effect of electricity exchange between electric
companies on LCI was investigated. The results are shown in Tables l(a) and (b).
2. Taiwan
This study conducts a preliminary analysis of LCI of electricity and compares an existing study in Taiwan.
Two sets of inventory data based on the variations of time and methodology are, therefore, constructed or
compared. Conclusively, the resource inputs per unit of electricity use are between 2.05 and 2.35 heat
content unit. The C02 emission is about 0.7kg per kWh of electricity use in Taiwan in 1999. Other than
the resource inputs and C02 emissions, the two databases vary significantly. During the period of
conducting the investigation, the researchers intend to collect the data of air emission, fly ash production,
and nuclear fuel consumption directly from Taipower. For some uncontrollable reasons, these data are not
acquired in time. Therefore, some further extensions should be made in spite of the ending of the
investigation. With the information compiled for the re-evaluation of constructing the fourth nuclear power
plant and the conjunction with a study for a master thesis, the future extension of the investigation appears
to be optimistic.
3. Korea
The first preliminary analysis of the electricity production system in Korea from the point of
view of LCI was carried out in 1995 using the national average data of the Ministry of Environment
(MOE) and the Korea Electric Power Corporation (KEPCO). The Industrial Advancement
Administration (IAA) supported the study to identify the inputs and outputs associated with thermal power
generation since it has the greatest portion among electricity generation by type in Korea and produces
significant amounts of environmental emissions. In addition, a comparison of the environmental
characteristics between different fuels used such as anthracite, bituminous coal, oil, diesel, and liquefied
natural gas (LNG) has been carried out.
The Korean Energy Management Corporation (KEMCO) expressed interest in developing an
LCI database detailing the raw materials use, emissions and solid wastes associated with energy
production, delivery, and use in Korea. In 1996, it was decided to conduct a pilot study that would
consider power generation at one power plant facility (the MokDong Kangseo District Energy Facility in
Seoul). The life cycle inventory with key gross raw material requirements and resulting emissions to
produce and deliver 1 TJ of heat and 1 TJ of electricity were obtained, respectively.
During 1996 - 1997, the establishment of a preliminary national database on electricity that
-------
included not only thermal power generation but also hydro and nuclear power generation was included in
the MOE project. Then, recently, an LCI database to encompass the full Korean electrical energy grid
which is a single super-grid covering the whole country with all generators feeding into it and all
consumers drawing from it has been developed in a MOCIE LCA project. It is found out that the national
average efficiency of production and delivery of electricity in Korea is 36.2%. C02 emissions related to
IkWh of electricity which final user can use is around 0.487 kg/f.u. and the contribution of direct emission
to total C02 is 94%. In the cases of SOx, NOx, and dust the contributions of fuel production and transport
are relatively large, 40%, 27%, and 27%.
4. Thailand
Life Cycle Inventory (LCI) for the electricity grid mix in Thailand was developed for the first
time. The results of the study were based on data obtained from the Electricity Generating Authority of
Thailand and an Independent Power Producer during October 1998 to September 1999, which covered
about 85% of total gross domestic electricity generation.
Total C02, CO, NMVOC, CH4, NOx, N20, and dust emissions from power plants were
54,527,721 ton, 12,338 ton, 2,601 ton, 1,140 ton, 174,421 ton, 1,705 ton, and 9,005 ton, respectively.
Emissions of sulphur dioxide were estimated to have reached 93,161 ton.
Of total carbon dioxide emissions, the amount of emissions from gas-fired power plants was the
highest (62%), followed by those from coal (37%) and fuel oil (0.42%). Of total sulphur dioxide emissions,
the amount of emissions from coal-fired power plants was the highest (78%) while those from gas-fired
power plants were the smallest (9.8%). Of total NOx emissions, the amount of emissions from coal-fired
power plants was the highest (53%) while those from fuel-fired power plants were the smallest (2%).
Direct S02, C02, and NOx emission intensities from power stations were 1.28x10-03, 0.75, and
2.40x10-03 kg-air pollutants/kWh of electricity consumed by users (kg/kWh) in average, respectively. The
average amount of CH4, CO, NMVOC, and particulate emission intensities were 1.57x10-05, 1.70x10-04,
3.58x10-05 and 1.24x10-04 kg/kWh, respectively. The highest S02, C02, and NOx emission intensities
(related to 1 kWh of electricity consumed by users) were 6.41x10-04, 2.58x10-01, and 1.17x10-03,
respectively. The smallest emission intensities of S02, C02 and NOx were 1.20x10-08, 1.25x10-05, and
1.40x10-07 kg/kWh of consuming electricity, respectively.
5. Malaysia
Life cycle inventories for the electricity grid mix of electricity generating power stations in
Peninsula Malaysia in 1999 were developed. The functional unit investigated was 1 kWh net electricity
delivered to consumers in the study area. The scope of the study was limited to the estimation of the
emissions of C02, NOx, S02, CH4, CO, NMVOC, dust, and heavy metals (Ni, V, As, Cd, Cr, Hg, Pb, Zn).
-------
The preliminary calculated weighted average emissions from the grid per kWh net electricity production in
1999 were 5.6xlO-lkg-C02/kWh, 1.3xlO-3kg-S02/kWh, 6.0xlO-4kg-NOx/kWh, 6.6x10-05 kg-CO/kWh,
1.2xlO-5kg-CH4/kWh, 3.8xlO-5kg-NMVOC/kWh, 4.4x10-5 kg-dust/kWh, 4.6xlO-7kg-Ni/kWh,
6.5xlO-7kg-V/kWh, 7.6x10-8 kg-As/kWh, 8.3xlO-9kg-Cd/kWh, 6. lxlO-7kg-Cr/kWh,
4.2xlO-9kg-Hg/kWh, 3.8xlO-7kg-Pb/kWh and 6.9x10-7 kg-Zn/kWh, respectively. The emission
intensities calculated need to be validated and verified with the actual emissions monitoring data for each
of the power stations under study. As of the time of investigation, actual emissions data could not be
obtained from the relevant authorities. Data verification is critical, as the parameters for estimation of
emissions may not reflect the actual situation in Malaysian power stations.
6.Reference
[1] Matsuno Y. and Betz M; Development of Life Cycle Inventories for Electricity Grid Mixes in Japan,
Int. J. LCA, 5 (5) 295-305(2000)
-------
DLR
LCI Electricity Data Sources - the situation in Europe
Wolfram Krewitt
DLR, wolfram.krewitt@dlr.de
Rolf Frischknecht
ESU-services, frischknecht@esu-services.ch
US EPA/NREL Int. Workshop on Electricity Data for Life Cycle Inventories
Cincinnati, USA, October 23-25, 2001
-------
DLR
the situation in Europe
• very diverse: many countries, many actors (industry,
research, public authorities), various technical products,
different interests, poor level of harmonization
• the pioneers:
-Germany: GEMIS 1.0 in 1989
-Switzerland: Ecolnvent 1sted. in 1994
-The Scandinavian countries
• jumping on the bandwagon ...
Germany, Italy, France,...
-------
DLR
Switzerland (I): Ecoinvent 2000, project outline
• Joint effort of LCA-institutes in the ETH-domain and federal
authorities (funding agencies)
• Nine institutes work on one central database
• Database on unit process level
• Full transparency and accessibility (fee foreseen)
• Access todatabase via the web
• Data exchange with relevant software providers such as
ecobilan/PWC, ifu/ifeu, PRe
-------
DLR
Switzerland (II): Ecoinvent 2000, database content
• Energy supply (PSI, ESU-services)
• Building materials and -processes (EMPA Diibendorf)
• Basic chemicals and plastics (both EMPAs and ETHZ)
• Iran s port services (ETHZ)
• Waste treatment services (all)
• G raphical P a p ers (E M P A St.Gallen)
• D etergents (EMPA St. Gallen)
• Agricultural Processes & products (FAL)
-------
DLR
Germany: current situation
• key research institutes developed and use own database systems and
software; in general no public access
• first database publicly available: Global Emission Model for Integrated
Systems (GEMIS); Oko-lnstitut (version 4.07 July 2001)
www.oeko.de/service/gemis/english/index.htm
• specific competence :
- DLR, System Analysis and Technology Assessment, Inst. of Techn.
Thermodynamics: LCI data for fuel cells, solar systems; database:
GaBi www.dlr.de/tt/system
- IFEU (Heidelberg): LCI data for biofuels; database: UMBERTO
www.ifeu.de
- University of Essen: cumulative energy demand for PV, wind (own
database) www.oeve.uni-essen.de
- FfE Munich: cumulative energy demand for fossil systems, heating
systems (own database) www.ffe.de
-------
DLR
Germany: Recent developments & upcoming activities
• Life cycle Inventories of new electricity generation technologies
Joint project (DLR, Univ. Essen, FfE, Univ. Stuttgart) funded by the German Federal
Ministry of Economics and Technology; (9/2001 - 8/2003)
Objectives: detailed inventories, common database, public data access
• 'Ecologically optimized' strategies for expanding renewable energies in
Germany
Joint project (DLR, IFEU, Wupertal Inst.) funded by the German Federal Ministry of
Environment; (6/2001 -12/2003)
Objectives: LCA of renewables, integration of LCA results into energy scenarios,
focus on nature conservation aspects (including hearings with stakeholder groups)
• Generic database systems with public access:
• German EPA: generic database for environmental management
www.umweltbundesamt.de/uba-info-daten/daten/baum/ (only in German)
• Federal Ministry of Economics and Technology: pre-study for a generic 6
database
-------
DLR
Italy
In 1998, the Italian Environmental Agency ANPA commissioned to the
Politecnico di Milano an Italian database on energy, transportation and waste
management systems
reviewed database available via internet since February 2001
www.mirrorsinanet.anpa.it/EcolProd/documenti%20l-LCA
Linked to EPD activities in Italy: EPD guidelines say that firms making their
declarations should use the I-LCA data, if primary data cannot be used
energy part basically adapts the data of Ecolnvent 1996 to Italian conditions
-------
DLR
Sweden
• Vattenfall has in the past 3 years worked with LCA on power systems for
EPDs (Environmental Product Declarations) according to the Swedish
guidelines based on ISO TR 14025. Two third-party certified EPD's, one for
hydro power and one for nuclear power
http://www.environdec.com/eng/registrations.asp
• LCA activities at Chalmers University (e.g. LCAiT www.lcait.com)
-------
DLR
the European Commission's activities
• Environmental and Ecological Life Cycle Inventories for Present and Future
Power Systems in Europe (ECLIPSE) (11/2001 -10/2003)
Ambiente Italia (Coord.), ESU, EdF, Vattenfall, Fortum, DLR, KEMA, Univ. Stuttgart
Objectives: LCI of new decentralised technologies (PV, wind, fuel cells, biomass,
small scale CHP); develop an e-database with public access (hosted by ANPA)
• series of ExternE projects (External costs of energy)
running since 1990, strong focus on environmental impact assessment and
valuation, country reports from all EU countries available, addressing the relevant
major fuel cycles (emissions, impacts, external costs) http://externe.jrc.es/
• European Energy Data Exchange Network (EDEN) (2/2001 - 7/2002)
compile input data and results from energy models (Primes, Poles, Markal, Times)
emissions covered: CO2, SO2, NOX, particles
-------
Some
IILE
Laboratorio
Ambiente Humano
v Vivienda
Alejandro Pablo Arena
-------
On the Positive side.
^Argentina started a privatization process of the
energy sector in the 90f.
$8 Today, the generation market is characterized by
open and free competition.
«!£High efficiency, modern natural gas combined cycle
power plants displace liquid and solid fossil fueled
power plants from dispatching in the market
As a consequence, current energy production is
characterized by lower prices and better environmental
performance than in the previous decade.
Alejandro Pablo Arena
-------
fiftr ITT Trr fir rrrr IT rrrrrr
Energy Secretariat: government agency which
produces the country's annual energy balance
Most available information in the annual energy
balances is published in an aggregated form, which
makes the task of calculating energy produced by
different fuels and technologies very hard and, in
some cases, impossible.
•No information about environmental aspects of
energy generation is available in the energy
balances.
Alejandro Pablo Arena
-------
Main Information sources regarding
Electricity Production in Argenti
4 ENRE: National Electricity Regulation Agency.
4 Every power plant is enforced to give to ENRE
an Environmental Management Plan, together with
an Environmental Diagnosis, and the results of the
environmental monitoring.
• Also a weekly report containing information
about events that produced emission levels beyond
the legal requirements is produced and delivered to
ENRE by every generation agent.
Alejandro Pablo Arena
-------
Main Information sources regarding
Electricity Production in Argenti
OTHER Institutions:
«£National Nuclear Energy Commission, etc.
^Energy Economics Institute, Fundacion Bariloche
^Universities
-^Research Institutions
Alejandro Pablo Arena
-------
Electric sector in Argentin
Challenges...
*£ There is a lack of cooperation among the
involved actors (government, universities,
industries).
4 The scarce information available is fragmented,
information channels are not clear, institutions are
weak.
*$ There are no national emission inventories
(public at least).
Alejandro Pablo Arena
-------
Ongoing and projected Electricity LC
There are different publications about the
environmental impact of the energy sector, none of
them with a LC perspective.
~£ There are no national emission inventories
(public at least), except for the CO2 emissions.
4 There is an ongoing LCI database development
project from the Universidad Tecnologica Nacional
(Mendoza), but due to the current economic crisis
in Argentina there is no funding for the project.
Alejandro Pablo Arena
-------
Ongoing and projected Electricity LC
^ We're interested in participating in
international projects for Electricity
LCI database development:
aparena@lab.cricyt.edu.ar
Alejandro Pablo Arena
-------
OAK RIDGC IMSTITUIC TOR SCICMCC AM& CDLfCATlOSl
United States
I XXl Environmental Protection Agency
*^ -*
Boundaries- Which Stressors Need to
be Captured in an LCI for Electricity?
Patrick Hofstetter12, 1ORISE Research Fellow at U.S. EPA, SAB-STD-NRMRL,
Cincinnati
2Visiting Scientist at Harvard School of Public Health, Boston
International Workshop on Electricity Data for LCI, October 23-25, 2001
Overview
1. Basic principles
lagged impact ca ries
4. I
-------
Different interpretations of the
meaning of "all effects"
Either: Potential impacts from all uses or releases that are
connected with electricity production (=> all elementary
flows need to be included)
Or: Stressors that contribute to more than XX% of all
impacts measured by impact assessment Y (and Z)
But,
DM's may want to include all environmental impacts that:
-burden third parties (or own/employers health), or
-are not yet compensated (non-market damages), or
-are likely to be targeted for eco-taxes, or
-pose potential liability problems, or
-are most efficiently addressed by LCA rather than other tool
-------
Stressors covering more than 95% of
total El'99 impacts
UCPTE-
Mix
0% 20% 40% 60% 80% 100%
land use ll-lll
hard coal
gas
oil
CO2 Carbon dioxide
Ni Nickel
NOx as NO2
Particels
SOx as SO2
Ion Arsenic f
Ion Nickel f
Assumptions:
- LCI was "complete" in terms of covered stressors
- all stressors that are not captured in Eco-lndicator'99 are unimportant
- Eco-lndicators'99 hierarchist's perspective provides the gold standard for damage
modeling
This figure is based on data and tools kindly provided by ESU-Services, Switzerland
-------
Stressors that total
total El'99
Hydro Power
Marginal Gas Power
Photovoltaics
Hard Coal Power
Lignite Power
Nuclear Power
UCPTE-Mix
more tha
impacts
n nrr
1 1
in
v
-nn — '
in n
—
_L
I
I
L_U ^M
0% 20% 40%
J
I
I
n 95% of
MI
i
ii
I
I
ii
II I
i
r-L
60%
~TT
r
n
I
80%
100%
Dland use ll-lll
• land use II-IV
D copper from ore
D tin from ore
• gas from oil
D gas from coal
• hard coal
Dgas
• oil
D CH4 Methane
D CO2 Carbon dioxide
D LT Radio. Rn222 p
• Ni Nickel
• NOx as NO2
• Particels
• Radio. C14 p
D SOx as SO2
D Ion Arsenic f
D Ion Cadmium f
D Ion Nickel f
Assumptions:
- LCI was "complete" in terms of covered stressors
- all stressors that are not captured in Eco-lndicator'99 are unimportant
- Eco-lndicators'99 hierarchist's perspective provides the gold standard for damage
modeling
This figure is based on data and tools kindly provided by ESU-Services, Switzerland
-------
Flagged impact categories
land use (partly state of practice)
ionizing radiation (partly state of practice)
noise (proposals for road noise impacts)
water use
salination
erosion, soil depletion
wildlife impacts of dams
aesthetics
electro magnetic fields
accidental releases (e.g., intermediate materials,
acute effects, uncommon metabolites)
-------
Guidelines from the SETAC Working
Group 'Data availability and data quality'
(Hischieretal. 2001)
Rigid parameter lists for LCIs are not practical;
especially, compulsory lists of measurements
for all inventories are counterproductive.
Instead, practitioners should be obliged to give
the rationale for their scientific choice of
selected and omitted parameters. The
standardized (not; mandatory!) parameter list
established by the subgroup can facilitate this.
-------
Proposed self-commitment
SETAC Working Group 'Data availability
and data quality' (Hischier et al. 2001)
"Included in the inventory were all parameters
that can reasonably be expected to occur in
the processes under study, and that can have
any environmental relevance, especially when
judged with present or foreseeable life cycle
impact assessment methods."
-------
A pragmatic approach by
Braunschweig (1996)
1. Use readily available use and release data for process at hand.
(at least one stressor that may be relevant)
2. Calculate preliminary category indicator scores using impact
assessment method(s) that fit(s) goal and scope definition
3. Calculate for all stressors that are assessed within the chosen
impact assessment method(s) the necessary release/use rate to
contribute more than X% to one indicator score.
4. Use expert judgment, back on the envelope worst-case estimates,
data on older processes or similar processes, etc. to decide which
stressors are worth the effort to get actual data for.
5. Gather the needed data for those stressors you can most easily get it.
6. Redo steps 2 through 5
-------
Summary
1. LCI for decision support relies on relevant
stressors only
2. In most cases only few stressors prove to be
relevant
3. Relevance can be defined in many ways and
depends on the goal of the study
4. LCIs/databases that shall be used for many
different types of decision support lack ONE goal
definition
=> How to define "relevant"?
-------
JJJJJJJJ
J
J
J
J
J
J
Boundaries,
Part 2
gory A. Morris, Ph
ardS
j
j
i
-------
LCI: 2 Families o
roach
lodelina
1. Process modeling
• Classical LCA approach
ior enerav flov/
em
GWJ
;al units
d;
A [ffTiiiBfj
-------
xample
Cuinrnunt sawn nak^ snurrp: finnrsma.
Cluster
r
Allocution arid system mudd
Specific allocation rules
Systuiri iriudul id Curnriirnl
Data
INPUTS
Kriuwri inputs Fruni nature
NiiniL1 AiiiuuriL UnilLuw vuluiHiyh vuluCoiTirriunl
wnnri 1.573 kg 0 0
Knuwn inputs fium luuhnuyphurL- (muturiuh/Fuuls)
A.I 11 M u 111 Unil Luw vul u iHi y h vukiCuin merit
Kiinwn inputs finni tRnhnriKphprE* i
MnniR Ainniint Unit I nw valuil ligh valuCnrnmBnt
blectricityUCFMbHigO.lfre MJ 0 0 winning.
Trunk Inng-itistancn l~5 0.4457 thm 0 JO 175 km full muck, Birpty trunk
counted as 1/3 rull truck (assumed
by PRe): 633 km xOJ001573 tonnt
kni.tnniiR
ElL-ulrimly UCPTE Hiy 2.4
humace oil B 4,3
MJ 0
MJ 0
0
0
^^^^^m
-Mid uthui L-luclric
diving,
-------
econd Approach
:conomic mput/uutput Approac
National I/O model with sector pollution
(kg CO2 per $ output)
Sectors as unit processes
Include e
Int
US parjQ|Q|^g3IBffiS^SB9DBH
Bo
ts
j
j
j
mprehensiveness at expense of
-------
/Output Data
Exhaustive economic censu
Annual updates at lo
iological
Comprehensive, at expense o
j
j
j
-------
j
j
j
-------
Inputs to "Saw
Planing Mills"
Value ($) % of Total
Logging
Sawmills and planing mills, general
Wholesale trade
Electric services (utilities)
Motor freight transportation and warehousing
Railroads and related services
Banking
Petroleum refining
Gas production and distribution (utilities)
Eating and drinking places
Insurance carriers
Metal stampings, n.e.c.
Automotive rental and leasing, without drivers
Advertising
Management and consulting services, testing and research
Sanitary services, steam supply, and irrigation systems
Industrial inorganic and organic chemicals
Veneer and plywood
Automotive repair shops and services
Adhesives and sealants
Water transportation
Communications, except radio and TV
Miscellaneous equipment rental and leasing
Legal services
Woodworking machinery
Abrasive products
U.S. Postal Service
Accounting, auditing and bookkeeping, and miscellaneous :
Real estate agents, managers, operators, and lessors
Fabricated metal products, n.e.c.
Royalties
Industrial and commercial machinery and equipment, n.e.c
Retail trade, except eating and drinking
Security and commodity brokers
Miscellaneous plastics products, n.e.c.
Colleges, universities, and professional schools
Plastics materials and resins
Paperboard containers and boxes
Screw machine products, bolts, etc.
Paints and allied products
Personnel supply services
Water supply and sewerage systems
Hardware, n.e.c.
Business associations and professional membership organ!
Detective and protective services
Engineering, architectural, and surveying services
Air transportation
Lubricating oils and greases
Manifold business forms
Special dies and tools and machine tool accessories
Hariri nnrl prlnp tnnk pvrpnt mnrhinp tnnk nnrl hnnrlQflWQ
6689
1274
621.9
359.2
245.9
214.6
110.4
99.61
73.31
59.91
55.49
53.59
53.2
52.7
48.31
44.2
36.89
35.51
34.6
33.4
33.2
32
30.7
28.3
24.4
23.9
21.3
20.81
20
19.4
17.8
17.61
17.51
17.2
17.09
16.39
15.79
15.61
15.61
13.71
13.71
12.9
12.8
12.51
11.81
10.2
9.497
9.103
8.607
8.299
yam
61.7%
11.8%
5.7%
3.3%
2.3%
2.0%
1.0%
0.9%
0.7%
0.6%
0.5%
0.5%
0.5%
0.5%
0.4%
0.4%
0.3%
0.3%
0.3%
0.3%
0.3%
0.3%
0.3%
0.3%
0.2%
0.2%
0.2%
0.2%
0.2%
0.2%
0.2%
0.2%
0.2%
0.2%
0.2%
0.2%
0.1%
0.1%
0.1%
0.1%
0.1%
0.1%
0.1%
0.1%
0.1%
0.1%
0.1%
0.1%
0.1%
0.1%
n 1%
-------
nput Boundary
f the little
J
J
J
-------
ibility in S
of InoutTvoe
Material / Energy inputs - major
r Equiprner
\R\, [in
UCl.U (E
Material / Energy inputs - minor
Overhead inputs (building, site, etc.)
Service inputs
Personnel-related expenses (travel, hotels
(sp<
i rive:
-------
Air transportation
Electric services (utilities)
Computer peripheral equipment
Wholesale trade
Semiconductors and related devices
Petroleum refining
Other electronic components
Miscellaneous plastics products, n.e.c.
Relays and industrial controls
Gas production and distribution (utilities)
Automotive rental and leasing, without drivers
Hotels and lodging places
Telephone and telegraph apparatus
Aluminum rolling and drawing
Motors and generators
Blast furnaces and steel mills
Sheet metal work
Elect ron tubes
Nonferrous wiredrawing and insulating
Fabricated metal products, n.e.c.
Legal services
Paperboard containers and boxes
Mo tor freight transportation and warehousing
Power, distribution, and specialty transformers
Gaskets, packing, and sealing devices
Metal stampings, n.e.c.
Real estate agents, managers, operators, and lessors
Banking
Eating and drinking places
|Management and consulting services, testing and research labs
% to total
u p s t re a m
em bodied
e n e rg y
12%
12%
10%
10%
9%
6%
5%
4%
3%
3%
2%
2%
2%
2%
1%
1%
1%
1%
1%
1%
1%
1%
1%
1%
1%
1%
1%
1%
1%
1%
c u m u I a t i ve %
12%
24%
34%
43%
53%
59%
64%
68%
71%
74%
76%
78%
80%
81%
83%
84%
85%
85%
86%
87%
87%
88%
89%
89%
90%
9 1 %
9 1 %
92%
92%
93%
-------
^^^^^^^^^^^^^^^^H
Another eye-opener:
mno/
IUU /o
90%
80%
yno/
/ U /o
60% -
50% -
40% -
30% -
20%
10%
no/,
U /o n
30%
8%
5%
4%
53%
F
"FTM" =
:actory-to-mall
D FTM Retail
D FTM Wholesale
D FTM Air
D FTM Truck
D Production
I
1
Embodied energy of different
pre-consumer life cycle stages for computers.
J
J
J
-------
nput Boundary
f x| I *tt I •
f the little i
-------
How many (li
inputs are there?
Numbers of inputs in full model
Q.
C
0
.Q
E
500
450
400
350
300
250
200
150
100
50
0
Median 375 inputs;
Between 300-400
for 90% of industries
0 10 20 30 40 50 60
percent of industries
70
80
90
100
H
-------
Gimmie a break!
Are you saying that inputs of
pencils and lawyers' fees and
air travel by the managers, etc.
make any real difference
in the LCI of a product?
-------
Let's trim out the "little stuff"
and see if anything changes
in the LCI result.
-------
or eac
(fi
exi
industry, rank inputs in terms o1
-breadth model) u
IC02)
leir
in as
inputs a
Direct emissions + £ retained = T(D +
Cr<
j
j
j
-------
Numbers of inputs ("breadth") in full and pruned models
450
400 -
0
1 300
T= 95% retains
Median of 30 inputs;
T= 99% -> med 90 inputs;
T= 90% -> med 16 inputs
Full Model
T = 0.99
T = 0.95
7 = 0.9
0 10 20 30 40 50 60
Percent of sectors
70
80 90
100
-------
How well do the pruned models
Now, each sector is pruned, and j
so are
x^^ I •'• I ^^ x~* ^^ ^"^ x"^'^'^~v t
S\\ I.I 16 S6Ci.G[
uppfy c
n
cons<
ices
:emj
use o
-------
Performance of pruned models:
median cumulative % of 6-tier full-model total
o>
T3
O
O
^p
O
E
O
E
<5
•o
o>
•Full Model
Pruned to 95%
Pruned to 90%
J
J
Direct Tier! Tier 2
Tier3
Tier
Tier4 Tier5 Tier6
I
-------
Pruned model performance varies acros
even with consistent cutoff rule (Here T = 95%)
cts,
to
o
o
5 o
-------
n
Boundary
How many of the little inputs?
or fosi
ler i
rerr
GGGQS found
ir
CJ.OUhli
-------
ry Issues
Geoi
Proper region for electricity su
nd / B;
reoTou
(Grouna
Temporal scope
r^1" x^^ r" x^ \"
[ig-i.6f
ear was IB
-------
Marginal vs. average data
(system expansion vs. allocation)
Tomas Ekvall
Chalmers University of Technology
Gothenburg, Sweden
Int. Workshop on Electricity Data for Life Cycle Inventories
Cincinnati, October 23-25, 2001
-------
Statement 1
Average data describe systems
Marginal data (attempt to) describe
consequences
Energy Conversion CHALMERS
-------
Fuels
etc.
Emissions
A
Electricity
supply
system
T
Ashes etc.
Elec-
tricity
Energy Conversion
CHALMERS
-------
Environmental
burdens (B)
t
Marginal
change
Substantial
change
Complete
change
Point of
operation
Production
volume (Q)
Energy Conversion
CHALMERS
-------
Statement 1'
Allocation (attempts to) describe
systems
System expansion is often required to
describe consequences
Energy Conversion CHALMERS
-------
Raw material
Multifunction
Sub-process
1
process
Product A
(used in the life
cycle investigated)
Sub-process
2
Product B
(used in other
life cycles)
Use of
product B/C
Raw material
extraction
Alternative
production
Product C
-(competing with
product B)
Waste
management
Energy Conversion
CHALMERS
-------
Both can be relevant for...
historic & future oriented LCA
foreground & background system
learning & decision-making context
Energy Conversion CHALMERS
-------
Foreground
Background
Learning
Average
Average /
marginal
Decision
making
Average /
marginal
Average /
marginal
Energy Conversion
CHALMERS
-------
Both have limitations
Relevance: system vs. consequences
Accuracy: subjectivity vs. uncertainty
Energy Conversion CHALMERS
-------
180
160
140
120 -
100
80 -
60
40
20
0
FIN SWE DEN
Country
NOR
DCHP, Ind
DCHP, DH
• Separate power
• Nuclear
DWind
• Hydro
Energy Conversion
CHALMERS
-------
70 -,
60
50
40 -
1
"25
:O
V)
30 -
20
10
0
Gas turbines
Oil condensing
Coal condensing
0 50 100 150 200 250 300 350 400
TWh/year
Energy Conversion
CHALMERS
-------
Suboptimal action
Support good systems
=>
Primary aluminium investment in Norway
=>
Increased electricity production in Denmark
=>
Poor consequences
Energy Conversion CHALMERS
-------
Suboptimal rule
Aim at good consequences
=>
No comparative advantage for Norway
=>
Isolated Norwegian electricity system
=>
Poor electricity system
Energy Conversion CHALMERS
-------
Subjective choice of system
Specified technology or plant
Electricity utility
Regional grid
National grid
nternational grid
Energy Conversion CHALMERS
-------
Uncertain modelling of consequences
Complex and uncertain margins
Consequences beyond the models
Energy Conversion CHALMERS
-------
[TWh]
450-
400-
350-
300-
250-
200-
150-
100-
50-
o
s
^— ^H
••
1
^H
^H
B
^H
•
••
^H
B
•
^H
B
^H
^H
^H
^H
•
^
^H
•
^H
^
•
^H
^H
^
.
1
J
I
1
I
3)
•
•
__
B
•
^m
•
B
^H
•
•
••
•1
^H
^H
"
•
B
^H
:|
J
•
••
^H
^H
•
^H
^H
—
•
••
•
••
^H
1
••
-
1
1
_u
•
••
e
•i
^H
X
2000 2003 2006 2009 2012 2015 2018 2021 2024 2027 2030 2033 2036 2039 2042 2045 2048
D hydro power
D gas turbine
D bio power
• peat power
• oil power
D gas power
Doil chp
• coal chp
Dgas chp
Dbio chp
D peat chp
D waste chp
D coal power
D bio backpress
D nuclear power
Dwind power
gy Conversion CHALMERS
-------
[TWh] 0
2000 2006 2012 2018
Energy Conversion
2024
2030
2036
2042
2048
I hydro power
D gas turbine
• bio power
I peat power
• oil power
D gas power
Doil chp
• coal chp
Dgas chp
Ubio chp
Dpeat chp
• waste chp
• coal power
Ubio backpress
D nuclear power
]wind power
CHALMERS
-------
1,5
0,5
0
-0,5
-1
-1,5
-2
y
DBio
D Electric
DOM
• Coal
DGas
D Other
2000 2006 2012 2018 2024 2030 2036 2042 2048
Energy Conversion
CHALMERS
-------
What consequences are in the model?
Physical cause-effect relations
(Economic relations)
hological relations
Energy Conversion CHALMERS
-------
Full consequences are
utterly uncertain
This should be no barrier
Energy Conversion CHALMERS
-------
Conclusions
• Marginal and average data
give different information
• Allocation and system expansion
give different information
• No clear-cut ranking
• Consider the audience
• Communicate clearly
Energy Conversion CHALMERS
-------
New and Non-traditional
Generators
International Workshop on Electricity
uata for Life Cycle Inventories
Margaret Mann
National Renewable Energy Laboratory
Golden, CO USA
-------
Areas of Interest
Conceptual technologies
Renewables
Non-baseload
Distributed generation
> Key issues - why should we care?
-------
Conceptual Technologies
• Perhaps still in R&D stage
• Little operating data
• Site not yet defined
• Operating conditions difficult to predict
• Often viewed as more environmentally benign
• Power examples: fuel cells, microturbines,
advanced biomass, new PV, new nuclear,
advanced coal
-------
Key Questions for LCAs on
Conceptual Technologies
• How do you compare a new technology to one
that is well defined?
• How do you take into account higher level of
uncertainty?
• How do you define a reliable set of data for an
R&D project?
• How do you take into account lack of site-
specific information?
• To what extent should assessment be qualitative
rather than quantitative?
-------
Renewables: a special case
Many have low, or essentially no, operating emissions
How should construction emissions be allocated to kWhs
generated over lifetime?
Impacts may be different (e.g., land-use and bird kills vs
natural resource consumption and greenhouse gases)
What is the appropriate use of LCA in green power
certification? In regulation?
Key driving factor: avoidance of conventional
environmental impacts - need to standardize treatment of
this in LCA
-------
Non-baseload Generators
• Many new systems do not produce power on a
continuous or controllable basis
• What is functional unit?
• Supply based: kWh from intermittent source
• Demand based: kWh needed to meet load
• How are emissions of supplemental system
allocated?
• How do we compare the emissions/kWh from
intermittent sources to those from baseload
generators?
-------
Distributed Generation
Drivers:
Demand for reliable power
Avoidance of down-time costs
Avoidance of T&D infrastructure
Mitigates large up-front capital expenditures
Potential is significant
Not point-source emitters
Credit for T&D losses; not traditionally assessed to large-
scale generators
May be operated intermittently
What about hybrid systems?
Key: functional unit may not be kWh
-------
Key Issues for LCI
Mix of technologies is not static
In LCI, how do we take into account dynamics of
technologies and grid?
What about the potential for very different grid
mixes in the future?
Different locations will see varying rates of
change - how will this affect relative impacts of
products?
What would be the value of a stochastic model
that gives probabilistic ranges of stressors?
-------
Generalized Questions
Can LCAs be conducted on new technologies for
which production data are not available?
Is there a need to develop a common future energy
scenario that considers renewable and distributed
energy sources for use in prospective LCAs?
How should distributed generation be accounted for
in national or regional energy grid data?
What percent of the grid mix does a technology
have to supply before we care about it in our
product LCAs?
-------
oundaries and Flows
24 October 2001
Cincinnati
Electricity LCI Workshop
-------
i
•Include infrastructure
for dedicated resources
•Add no unit processes
Delete according to
process knowledge
Resource
Extraction
Processing-
Transport-
Manufacturing-
'
On site
Material
Storage &
Handling
Water
Production &
Processing
Generation
Maintenance
Fleet Operations
Pollution
Control
Construction &
Demolition
Disposal
Transmission- >
Distribution*
Distributed generation
1
-------
FLOWS
Not a comprehensive list, but
a minimum list
Resources
•Electricity (location)
•Water (location & type)
•Fuel (in ground)
•Minerals (in ground)
Biomass (harvested)
Land use (area & location)
Wastes
•Solid waste
•Radioactive Waste
(high, low, medium)
•Hazardous Waste
•CO
•PM(10,2.5)
•CH4
•sox
•NOX
•NH,
•VOC (MM)
Dioxin
PAH's
Water
•COD
•IDS
•TSS
•BOD (5,7,1(
•Flow
ATemperature
TKN (as N)
NO3,NO2(asN)
PAH's
Phosphates (as P)
Cu ™
Ni
A o
-------
Rules & Advice
# Follow the ISO
14040 standards
# Don't add unit
processes to system
# Delete unit
processes only
based on process
knowledge
* Get inventory for all
suggested flows, filling
in gaps as needed
* Don't throw away data
(it may someday be
important) but
* Remove from
^m
consideration those
data which are not in
your impact
assessment,
-------
Next Step: Model Examples
# Coal w/ anthracite
Coal w/lignite
# Natural Gas
«j^"^. • •
* Oil
Nuclear
Hydro
* Wind
# Biomass
# Geothermal
# Other
-------
Research Opportunities
* Biological Resources
# Radiation
* Noise
-------
New and Non-traditional
Generators
Merwin Brown
Joyce Cooper
RolfFrischknecht
Douglas Gyorke
Marty Heller
Wolfram Krewitt
Ivars Licis
Lynn Manfredo
Maggie Mann
Jonathan Overly
-------
Group Kept Asking....
• Who are we?
• Database developers
•
• What is the purpose of the LCA?
• Came up with 17 different uses
• Doesn't matter for our purpose today
-------
Questions
Can LCAs be conducted on new technologies for which
production data are not available?
How are data sets constructed for new technologies, for
which there are higher degrees of uncertainty in
environmental stressors?
Is there a need to develop a common future energy scenario
that considers renewable and distributed energy sources for
use in prospective LCAs?
How should distributed generation be accounted for in national
or regional energy grid data?
How should distributed generation be accounted for in
national or regional energy grid data?
What percent of the grid mix does a technology have to supply
before we care about it in our product LCAs?
-------
Answers to Ql
How are data sets constructed for new technologies, for
which there are higher degrees of uncertainty in
environmental stressors?
a. Use best available mass & energy & production data.
b. Where there are data gaps, make a conservative expert
judgment for missing data points and document
assumptions (SETAC working group)
c. Include a calculation routine that allows the users to
vary performance/emissions parameters.
d. Document assumptions, sources of data, and year in
which data were obtained.
e. Be alert to the situation where you need to input
stressors that are not common to current generation
technologies (e.g., bird kill, land use).
-------
Answers to Q2
Is there a need to develop a common
future energy scenario that considers
the use of renewable and distributed
energy sources in grids.
a. No.
b. However, there is a need to provide for
the application of various future energy
scenarios.
c. Provide a tool or modules that describe
different energy mixes/scenarios.
-------
Answers to Q3
How should distributed generation be
accounted for in national or
regional energy grid data?
a. The same way that traditional
generators are accounted for.
b. Different transmission and
distribution losses are important.
-------
Answers to Q4
What percent of the grid mix does a
technology have to supply before we
care about it in our product LCAs?
a. If you can do an LCA on a technology,
provide it in the database.
b. Use the module/tool described in 2) to
give the user an opportunity to
incorporate them into their grid mix, or
they can do it manually.
-------
Other key points
• Database data should be as un-
aggregated as possible.
• Functional unit should always be kWh,
but...
• Database: kWhs from generation technology
• User: kWhs from all sources that generate
electricity being used
• Assessments (and underlying database)
should be done in three separate steps:
construction, operation,
decommissioning. Start-up should also
be considered separately.
-------
JJJJJJJJ
^
j Attributional & Consequential
EPA Worksh
CA of Electricity S
-------
I
Clarify terminology, meanings
Different or similar:
• Electricity LCI for use in other LCIs
• LCI
hen
each ai
time
How-to, implementation, models
-------
Termin
ecision
lerturbatior
sturbancp
in have
Timing and duration
lagnitude
j Marginal perturbal
isequem
;epi
-------
Attributional Approach
j Traditional approach
All
it is
ows
How do things flow, in the system, during the
specified time window?" (past, present, future) J
(Supplier 1
\7 units
1 unit
Process
.Supplier
-------
Consequential Approach
A1
:s to asse
lecision consequences
if
jecrsions f
,rJ- I-
Coi
(A output from
:erm (A capacity investi
>nt,
M:Ino
Supplier 1
-..Constrained
1 more unit
Process
(Supplier 2)
10 more units
-------
Example LCA Decisions involving electricity
iere to DUIIO a new aluminum plant?
Government: new appliance standard
Chemical process modification
• More electricity use. less chemical use
Agreemei
waagm
LCA tells decision
;es of each decision
j
j
j
-------
How DOES the electricity system respond to
changes in demand?
t t( ———-
ts with highest variable cost
Long term: new capacity is one
depends on load shape (hourly, seasonal)
Evolving, dynamic; future is uncertain
j
j
j
1 day, or 1 year
-------
Energy system mo
Dynamic
Causally descriptive
Canb
m n
changes wif
h
ate how sysi
ponds l:o demand
-------
Consequential LCI not familiar to
practitioners -> questions, appi
:ions
a Does C/A effect LCI resu
Which product types, regions?
a Does it alter LCI-based decisions?
easy
a How
aA
y to pertorrr
SfO
H
en?
j
j
j
-------
ommendations
lon't agg
OVI
• Don't aggregate over markets
• -> Solving confidential information issues
Need ample meta-data
Need
un Known
-------
To approxima
one time, who
reled far
•Op<
\ / ( I s~~~- rr-\ ^ s~^ S~^ ^^- T^ f^ * ' \ ^ S~~~- T T f^* \^^ *" \^~ ^~"" F^"^^
V\/G[ K6Q gf |Q i.l iGUQl n. I ig[ 0
Stayed friendly
------- |