98

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Chapter 5 : Data Acquisition and Management for Life

Cycle Inventory Analysis
Now that the most important elements of the LCA Standard are better understood, we can
begin to think about the work needed to get data for your study. In this chapter, we
introduce the inventory analysis phase of the LCA Standard, as well acquiring and using data
needed for the inventory phase of an LCA or an LCI study. As data collection, management,
and modeling are typically the most time-consuming components of an LCA, understanding
how to work with data is a critical skill. We build on concepts from Chapter 2 in terms of
referencing and quantitative modeling. Improving your qualitative and quantitative skills for
data management will enhance your ability to perform great LCAs. While sequentially this
chapter is part of the content on process-based life cycle assessment, much of the discussion
is relevant to LCA studies in general.

Learning Objectives for the Chapter

At the end of this chapter, you should be able to:
1. Recognize how challenges in data collection may lead to changes in study design
parameters (SDPs), and vice versa
2. Map information from LCI data modules into a unit process framework
3. Explain the difference between primary and secondary data, and when each might be
appropriate in a study
4. Document the use of primary and secondary data in a study
5. Create and assess data quality requirements for a study
6. Extract data and metadata from LCI data modules and use them in support of a
product system analysis
7. Generate an inventory result from LCI data sources
8. Perform an interpretation analysis on LCI results

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

99

ISO Life Cycle Inventory Analysis

After reviewing the ISO LCA Standard and its terminology in Chapter 4, you should be able
to envision the level and type of effort needed to perform an inventory analysis of a chosen
product system. Every study using the ISO Standard has an inventory analysis phase, but as
discussed above, many studies end at this phase and are called LCI studies. Those that
continue on to impact assessment are LCAs. That does not mean that LCI studies have
better inventory analyses than LCAs, in fact LCAs may require more comprehensive
inventory analyses to support the necessary impact assessment.
Figure 5-1, developed by the US EPA, highlights the types of high-level inputs and outputs
that we might care to track in our inventory analysis. As originally mentioned in Chapter 1,
we may be concerned with accounting for material, energy, or other resource inputs, and
product, intermediate, co-product, or release outputs. Recall that based on how you define
your goal, scope, and system boundary, you may be concerned with all or some of the inputs
and outputs defined in Figure 5-1.

Figure 5-1: Overview of Life Cycle Assessment (Source: US EPA 1993)

Inventory analysis follows a straightforward and repeating workflow, which involves the
following steps (as taken from ISO 14044:2006) done as needed until the inventory analysis
matches the then-current goal and scope:

Preparation for data collection based on goal and scope

Data Collection

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

100

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Data Validation (do this even if reusing someone else's data)

Data Allocation (if needed)

Translating Data to the Unit Process

Translating Data to the Functional Unit

Data Aggregation

As the inventory analysis process is iterated, the system boundary and/or goal and scope
may be changed (recall the two-way arrows in Figure 4-1). The procedure is as simple as
needed, and gets more complex as additional processes and flows are added. Each of the
inventory analysis steps are discussed in more detail below, with brief examples for
discussion. Several more detailed examples are shown later in the chapter.
Step 1 - Preparation for data collection based on goal and scope
The goal and scope definition guides which data need to be collected (noting that the goal
and scope may change iteratively during the course of your study and thus may cause
additional data collection effort or previously collected data to be discarded). A key
consideration is the product system diagram and the chosen system boundary. The boundary
shows which processes are in the study and which are not. For every unit process in the
system boundary, you will need to describe the unit process and collect quantitative data
representing its transformation of inputs to outputs. For the most fundamental unit
processes that interface at the system boundary, you will need to ensure that the inputs and
outputs are those elementary flows that pass through the system boundary. For other unit
processes (which may not be connected to those elementary flow inputs and outputs) you
will need to ensure they are connected to each other through non-elementary flows such as
intermediate products or co-products.
When planning your data collection activities, keep in mind that you are trying to represent
as many flows as possible in the unit process shown in Figure 5-2. Choosing which flows to
place at the top, bottom, left, or right of such a diagram is not relevant. The only relevant
part is ensuring inputs flow into and outputs flow out of the unit process box. You want to
quantitatively represent all inputs, either from nature or from the technosphere (defined as
the human altered environment, thus flows like products from other processes). By covering
all natural and human-affected inputs, you have covered all possible inputs. You want to
quantitatively represent outputs, either as products, wastes, emissions, or other releases.
Inputs from nature include resources from the ground, from water, or air (e.g., carbon
dioxide to be sequestered). Outputs to nature will be in the form of emissions or releases to
'compartments' in the ground, air, or water. Outputs may also be classified to 'direct human
uptake' for food products, medicines, etc.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

101

Figure 5-2: Generalized Unit Process Diagram

As a tangible example, imagine a product system like the mobile phone example in Chapter 4
where we have decided that the study should track water use as an input. Any of the unit
processes within the system boundary that directly uses water will need a unit process
representation with a quantity of water as an input and some quantitative measure of output
of the process. For mobile phones, such processes that use water as a direct input from
nature may include plastic production, energy production, and semiconductor
manufacturing. Other unit processes within the boundary may not directly consume water,
but may tie to each other through flows of plastic parts or energy. They themselves will not
have water inputs, but by connecting them all together, in the end, the water use of those
relevant sectors will still be represented. The final overall accounting of inventory inputs
and/or outputs across the life cycle within the system boundary is called a life cycle
inventory result (or LCI result).
The unit process focus of LCA drives the need for data to quantitatively describe the
processes. If data is not available or inaccessible, then the product system, system boundary,
or goal may need to be modified. Data may be available but found not to fit the study. For
example, an initial system boundary may include a waste management phase, but months of
effort could fail to find relevant disposition data for a specific product of the process. In this
case, the system boundary may need to be adjusted (made smaller) and other SDPs edited to
represent this lack of data in the study. On the other hand, data that is assumed to not be
available at first may later be found, which would allow an expansion of the system
boundary. In general, system boundaries are made smaller not larger over the course of a
study.
Step 2 - Data Collection
For each process within the system boundary, ISO requires you to "measure, calculate, or
estimate" data to quantitatively represent the process in your product system model. In LCA,
the "gold standard" is to collect your own data for the specific processes needed, called
primary data collection. This means directly measuring inputs and outputs of the process

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

102

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

on-site for the specific machinery use or transformation that occurs. For example, if you
required primary data for energy use of a process in an automobile assembly line that fastens
a component on to the vehicle with a screw, you might attach an electricity meter to the
piece of machinery that attaches the screw. If you were trying to determine the quantity of
fuel or material used in an injection molding process, you could measure those quantities as
they enter the machine. If you were trying to determine the quantity of emissions you could
place a sensor near the exhaust stack.
If you collect data with methods like this, intended to inventory per-unit use of inputs or
outputs, you need to use statistical sampling and other methods to ensure you generate
statistically sound results. That means not simply attaching the electricity meter one time, or
measuring fuel use or emissions during one production cycle (one unit produced). You
should repeat the same measurement multiple times, and perhaps on multiple pieces of
identical equipment, to ensure that you have a reasonable representation of the process and
to guard against the possibility that you happened to sample a production cycle that was
overly efficient or inefficient with respect to the inputs and outputs. The ISO Standard gives
no specific guidance or rules for how to conduct repeated samples or the number of samples
to find, but general statistical principles can be used for these purposes. Your data collection
summary should then report the mean, median, standard deviation, and other statistical
properties of your measurements. In your inventory analysis you can then choose whether to
use the mean, median, or a percentile range of values.
Note that many primary data collection activities cannot be completed as described above. It
may not be possible to gain access to the input lines of a machine to measure input use on a
per-item processed basis. You thus may need to collect data over the course of time and
then use total production during that time to normalize the unit process inventory. For the
examples in the previous paragraph, you might collect electricity use for a piece of machinery
over a month and then divide by the total number of vehicles that were assembled. Or you
may track the total amount of fuel and material used as input to the molding machine over
the course of a year. In either case, you would end up with an averaged set of inputs and/or
outputs as a function of the product(s) of the unit process. The same general principles
discussed above apply here with respect to finding multiple samples. In this case you could
find several monthly values or several yearly values to find an average, median, or range.
The ISO Standard (14044:2006, Annex A) gives examples of "data collection sheets" that
can support your primary data collection activities. Note that these are only examples, and
that your sheets may look different. The examples are provided to ensure, among other
things, that you are recording quantities and units, dates and locations of record keeping, and
descriptions of sampling done. The most likely scenario is that you will create electronic data
collection sheets by recording all information in a spreadsheet. This is a fair choice because
from our perspective, Microsoft Excel is the most popularly used software tool in support of
LCA. Even practitioners using other advanced LCA software packages still typically use
Microsoft Excel for data management, intermediate analysis, and graphing.
Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

103

Collecting primary data can be difficult or impossible if you do not own all the equipment or
do not have direct access to it either due to geospatial or organizational barriers. This is
often the case for an LCA consultant who may be tasked with performing a study for a client
but who is given no special privileges or access to company facilities. Further, you may need
to collect data from processes that are deemed proprietary or confidential by the owner. This
is possible in the case of a comparative analysis with some long-established industry practice
versus a new technology being proposed by your client or employer. In these cases, the
underlying data collection sheets may be confidential. Your analysis may in these cases only
"internally use" the average data points without publicly stating the quantities found in any
subsequent reports. If the study is making comparative assertions, then it may be necessary
to grant to third-party reviewers (who have signed non-disclosure agreements) access to the
data collection sheets to appreciate the quality of the data and to assess the inventory analysis
done while maintaining overall confidentiality.
Beyond issues of access, while primary data is considered the "gold standard" there are
various reasons why the result may not be as good as expected in the context of an LCA
study. First, the data is only as good as the measurement device (see accuracy and precision
discussion in Chapter 2). Second, if you are not able to measure it yourself then you
outsource the measurement, verification, and validation to someone else and trust them to
do exactly as you require. Various problems may occur, including issues with translation
(e.g., when measuring quantities for foreign-owned or contracted production) or not finding
contacts with sufficient technical expertise to assist you. Third, you must collect data on
every input and output of the process relevant to your study. If you are using only an electric
meter to measure a process that also emits various volatile organic compounds, your
collected data will be incomplete with respect to the full litany of inputs and outputs of the
process. Your inventory for that process would undercount any other inputs or outputs.
This is important because if other processes in your system boundary track volatile organics
(or other inputs and outputs) your primary data will undercount the LCI results.
The alternative to primary data collection is to use secondary data (the "calculating and
estimating" referenced above). Broadly defined, secondary data comes from life cycle
databases, literature sources (e.g., from searches of results in published papers), and other
past work. It is possible you will find data closely, but not exactly, matching the required unit
process. Typical tradeoffs to accessibility are that the secondary data identified is for a
different country, a slightly different process, or averaged across similar machinery. That
does not mean you cannot use it you just need to carefully document the differences
between the process data you are using and the specific process needed in your study. While
deemed inferior given the use of the word secondary, in some cases secondary data may be
of comparable or higher quality than primary data. Secondary data is typically discoverable
because it has been published by the original author who generated it as primary data for
their own study (and thus is typically of good quality). In short, one analyst's primary data
may be another's secondary data. Again, the "secondary" designation is simply recognition

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

104

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

that it is being "reused" from a previously existing source and not collected new in your own
study. Many credible and peer reviewed studies are constructed mostly or entirely of
secondary data. More detail on identifying and using secondary data sources like LCI
databases is below.
For secondary data, you should give details about the secondary source (including a full
reference), the timestamp of the data record, and when you accessed it. In both cases you
must quantitatively maintain the correct units for the inputs and outputs of the unit process.
While not required, it is convenient to make tables that neatly summarize all of this
information.
Regardless of whether your data for a particular process comes from a primary or secondary
source, the ISO Standard requires you to document the data collection process, give details
on when data have been collected, and other information about data quality. Data quality
requirements (DQRs) are required scope items that we did not discuss in Chapter 4 as part
of the SDP, but characterize the fundamental expectations of data that you will use in your
study. As specified by ISO 14044:2006, these include statements about your intentions with
respect to age of data, geospatial reach, completeness, sources, etc. Data quality indicators
are summary metrics used to assess the data quality requirements.
For example, you may have a data quality requirement that says that all data will be primary,
or at least secondary but from peer-reviewed sources. For each unit process, you can have a
data quality indicator noting whether it is primary or secondary, and whether it has been
peer-reviewed. Likewise, you may have a DQR that says all data will be from the same
geospatial region (e.g., a particular country like the US or a whole region like North
America). It is convenient to summarize the DQRs in a standardized tabular form. The first
two columns of Figure 5-3 show a hypothetical DQR table partly based on text from the
2010 Christmas tree study mentioned previously. The final column represents how the
requirements might be indicated as a summary in a completed study. The indicated values
are generally aligned with the requirements (as they should be!).
Data Quality Category
Temporal
Geospatial
Technological

Requirement

Data Quality Indicator

Artificial trees: 2009 data
Natural trees: 2002-2009 data
Data within 10 years of study
Artificial trees: China
Natural trees: US
Data matches local production
All processes used in study are
Most common production process representative of most common
practices
basis
Figure 5-3: Sample Data Quality Requirements (DQR) Table

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

105

Beyond using primary or secondary data, you might need to estimate the parameters for
some or all of the input and outputs of a unit process using methods as introduced in
Chapter 2. Your estimates may be based on data for similar unit processes (but which you
deem to be too dissimilar to use directly), simple transformations based on rules of thumb,
or triangulated averages of several unit processes. From a third-party perspective, estimated
data is perceived as lower quality than primary or secondary sources. However when those
sources cannot be found, estimating may be the only viable alternative.
Example 5-1: Estimating energy use for a service
Question:
Consider that you are trying to generate a unit process associated with an
internal corporate design function as part of the life cycle "overhead" of a particular product
and given the scope of your study need to create an input use of electricity. Your company is all
located in one building. There is no obvious output unit for such a process, so you could define
it to be per 1 product designed, per 1 square foot of design space, etc., as convenient for your
study.
Answer:
You could estimate the input electricity use for a design office over the course of
a year and then try to normalize the output. If you only had annual electricity use for the entire
building (10,000 kWh), and no special knowledge about the energy intensity of any particular
part of the building as subdivided into different functions, you could find the ratio of the total
design space in square feet (2,000 sf) as compared to the total square feet of the building
(50,000 sf), and use that ratio (2/50) to scale down the total consumption to an amount used
for design over the course of a year (400 kWh). If your output was per product, you could then
further normalize the electricity used for the design space by the unique number of products
designed by the staff in that space during the year.

You could add consideration of non-electricity use of energy (e.g., for heating or cooling)
with a similar method. Note that such ancillary support services like design, research and
development, etc., generally have been found to have negligible impacts, and thus many
studies exclude these services from their system boundaries.
Step 3 - Data Validation
Chapter 2 provided some general guidance on validating research results. With respect to
validating LCI data, you generally need to consider the quantitative methods used and ensure
that the resulting inventories meet your stated DQRs. Data validation should be done after
data is collected but before you move on to the actual inventory modeling activities of your
LCA.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

106

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

As an example of validation, it may be useful to validate energy or mass balances of your

processes. Using the injection molding process example from Step 2, one would expect that
the total input mass of material to be greater than (but approximately equal to) the output
mass of molded plastic. You can ensure that the total mass input of plastic resin, fuels, etc.,
is roughly comparable to the mass of molded plastic (subject to reasonable losses). If the
balances are deemed uneven, you can assess whether the measured process is merely
inefficient or whether there is a problem in your data collection, and thus resample.
You can use available secondary data to validate primary data collection. If you have chosen
to collect your own data for a process that is similar to processes for which there is already
secondary data available, you can quantitatively compare your measured results with the
published data. Again, if there are significant differences then you will need to determine the
source of the discrepancy. You can validate secondary data that you have chosen to use
against other sources in similar ways.
The results of validation efforts can be included in the main text of your report or in an
included Appendix, depending on the level of detail and explanation needed. If you collected
primary data and compared it to similar data from the same industry, the following text
might be included to show this:
"Collected data from the year 2012 on the technology-specific process used in this study
was compared to secondary data on the similar injection molding process from 2005
(Reference). The mean of collected data was about 10% lower than the secondary data.
This difference is not significant, and so the collected data is used as the basis for the
process in the study."
If validation suggests the differences are more substantial, that does not automatically mean
that the data is invalid. It is possible that there are no good similar data sources to compare
against, or that the technology has changed substantially. That too could be noted in the
study, such as:
"Collected data from the year 2012 on the technology-specific process used in this study
was compared to secondary data on the similar injection molding process from 2005
(Reference). The mean of collected data was about 50% lower than the secondary data.
This difference is large and significant, but is attributed to the significant improvements
in the industry since 2005, and so the collected data is still chosen as the basis for the
process in the study."
As noted above, the validation step is where you re-assess whether the quantitatively sound
data you want to use also is within the scope of your DQRs. Many studies state DQRs to use
all primary data at the outset, but subsequently realize it is not possible. Likewise studies may
not be able to find sufficient geospatially focused data. In both cases, the DQRs would need
to be iteratively adjusted as the study continues. This constant refining of the initial goal and

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

107

scope may sound like "cheating", but the purpose of the DQRs is as a single and convenient
summary of the study goals for data quality. It allows a reader to quickly get a sense of how
relevant the study results are given the final DQRs. While not required, you can state initial
goal DQRs alongside final DQRs upon completion of the study.
Step 4 - Data Allocation (if needed)
Allocation will be discussed more in Chapter 6, but in short, allocation is the quantitative
process done by the study analyst to assign specific quantities of inputs and outputs to the
various products of a process based on some mathematical relation between the products.
For example, you may have a process that produces multiple outputs, such as a petroleum
refinery process that produces gasoline, diesel, and other fuels and oils. Refineries use a
significant amount of energy. Allocation is needed to quantitatively connect the energy input
to each of the refined products. Without specified allocation procedures, the connections
between those inputs and the various products could be done haphazardly. The ISO
Standard suggests that the method you use to perform the allocation should be based on
underlying physical relationships (such as the share of mass or energy in the products) when
possible. For example, if your product of interest is gasoline, you will need to determine how
much of the total refinery energy was used to make the gasoline. For a mass allocation, you
could calculate it by using the ratio of the mass of the gasoline produced to the total mass of
all of the products. You may have to further research the energetics of the process to
determine what allocation method is most appropriate.
If physical relationships are not possible, then methods such as economic allocationsuch
as by eventual sale price could be used. ISO also says that you should consistently choose
allocation methods as much as possible across your product system, meaning that you
should try not to use a mass-based allocation most of the time and an energy-based
allocation some of the time. This is because mixing allocation methods could be viewed by
your audience or reviewers as a way of artificially biasing the results by picking allocations
that would provide low or high results. Allocation is conceptually similar to the design space
electricity Example 5-1. Most allocations are just linear transformations of effects.
When performing allocation, the most important considerations are to fully document the
allocation method chosen (including underlying allocation factors) and to ensure that total
inputs and outputs are equal to the sum of the allocated inputs and outputs. It is possible
that none of your unit processes have multiple products, and thus you do not need to
perform allocation. You might also be able to avoid allocation entirely, as we will see later.
Step 5 - Translating Data to the Unit Process
In this step you convert the various collected data into a representation of the output of the
unit process. Regardless of how you have defined the study overall, this step requires you to
collect all of the inputs and outputs as needed for 1 unit output from that process. From

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

108

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Example 5-1, you would ensure that the electricity input matched the unit basis of your
product flow (e.g., per 1 product designed). This result also needs to be validated.
Step 6 - Translating Data to the Functional Unit
The reason why this step is included in the ISO LCA Standard is to remind you that you are
doing an overall study on the basis of 1 functional unit of product output. Either during the
data collection phase, or in subsequent analysis, you will need to do a conversion so that the
relative amount of product or intermediate output of the unit process is related to the
amount needed per functional unit. Eventually, all of your unit process flows will need to be
converted to a per-functional unit basis. If all unit processes have been so modified, then
finding the total LCI results per functional unit is a trivial procedure. From Example 5-1, the
design may be used to eventually produce 1 million of the widgets. The electricity use for
one product design must be distributed to the 1 million widgets so that you will then have
the electricity use for a single widget in the design phase (a very small amount). This result
also needs to be validated.
Step 7 - Data Aggregation
In this step, all unit process data in the product system diagram are combined into a single
result for the modeled life cycle of the system. What this typically means is summing all
quantities of all inputs and outputs into a single total result on a functional unit basis.
Aggregation occurs at multiple levels. Figure 4-4 showed the various life cycle stages within
the view of the product system diagram. A first level of aggregation may add all inputs and
outputs under each of the categories of raw material acquisition, use, etc. A second level of
aggregation may occur across all of these stages into a final total life cycle estimate of inputs
and outputs per functional unit. Aggregated results are often reported in a table showing
total inputs and outputs on per-process, or per stage, values, and then a sum for the entire
product system. Example 5-2 shows aggregated results for a published study on wool from
sheep in New Zealand. The purpose of such tables is to emphasize category level results,
such as that half of the life cycle energy use occurs on farm. Results could also be graphed.
Example 5-2: Aggregation Table for Published LCA on Energy to Make Wool
(Source: The AgriBusiness Group, 2006)
Life Cycle Stage

Energy Use
(GJ per tonne wool)

On Farm

22.6

Processing

21.7

Transportation

1.5

Total

45.7

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

109

Beyond such tables, product system diagrams may be annotated with values for different
levels of aggregation by adding quantities per functional unit. Example 5-3 shows a diagram
for a published study on life cycle effects of bottled water and other beverage systems
performed for Nestle Waters. Such values can then be aggregated into summary results.
Example 5-3: Aggregation Diagram for Bottled Water (Source: Quantis, 2010)

We have above implied that aggregation of results occurs over a relatively small number of
subcomponents. However, a product system diagram may be decomposed into multiple sets
of tens or hundreds of constituent pieces that need to be aggregated. If all values for these
subcomponents are on a functional unit basis, the summation is not difficult, but the
bookkeeping of quantities per subcomponent remains an issue. If the underlying
subcomponent values are not consistently on a per functional unit basis, units of analysis
should be double checked to ensure they can be reliably aggregated.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

110

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Life Cycle Interpretation

Because some studies only include an inventory (LCI), we discuss Interpretation, the final
step for all LCAs and LCIs, now. For those studies (LCAs) that also include an impact
assessment, the procedures for the assessment will be discussed in Chapter 10. There is little
detail provided in the ISO Standard on what must be done in this phase, but interpretation is
similar to the last step of the "three step" method introduced in Chapter 2. The
interpretation phase refers to studying the results of the goal and scope, inventory analysis,
and impact assessment, in order to make conclusions and recommendations that can be
reported. As shown in Figure 4-1, interpretation is iterative with the three other phases. As
this chapter is focused on inventory analysis, much of the discussion and examples provided
relate to interpreting inventory results, but the same types of interpretation can be done with
impact assessment results (to be discussed in Chapter 10).
A typical first task in interpretation is to study your results to determine whether conclusions
can be made based on the inventory results that are consistent with the goal and scope. One
of the most common and important interpretation tasks involves discussing which life cycle
stage leads to the largest share of LCI results. A high-level summary helps to set the stage for
subsequent analyses. For example, an LCA of a vehicle will likely show that the use phase
(driving the car) is the largest energy user, as compared to manufacturing and recycling. An
interpretation task could involve creating a tabular or graphical summary showing the energy
use contributions for each of the stages. The interpretation of results from the study
summarized in Example 5-2 could note that energy use on farms is about equal to that in the
processing stage, and that transportation energy use appears negligible.
Part of your goal statement may have been to do a comparison between two types of
products and assess whether the life cycle energy use of one is significantly less than the
other. If your inventory results for the two products are nearly identical (say only 1%
different) then it may be difficult to scientifically conclude that one is better than the other
given the various uncertainties involved. Such an interpretation result could cause you to
directly state that no appreciable difference exists, or it may cause you to change the system
boundary in a way that ends up making them significantly different.
A key part of interpretation is performing relevant sensitivity analyses on your results. The
ISO Standard does not require specific sensitivity analysis scenarios as part of interpretation,
but some consideration of how alternative parameters for inputs, outputs, and methods used
(e.g., allocation) would affect the final results is necessary. As discussed in Chapter 2, a main
purpose of sensitivity analysis is to help assess whether a qualitative conclusion is affected by
quantitative changes in the parameters of the study. For example, if your general qualitative
conclusion is that product A uses significantly less energy than product B, the sensitivity
analysis may test whether different quantitative assumptions related to A or B lead to results
where energy use of A is roughly equal to B, or where A is greater than B. Any of the latter
two outcomes is qualitatively different than the initial conclusion, and it would be important

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

111

for the sensitivity results to be stated so that it was clear that there is a variable that if
credibly changed by a specified amount, has the potential to alter the study conclusions.
While on the subject of assessing comparative differences, it is becoming common for
practitioners in LCA to use a "25% rule" when testing for significant differences. The 25%
rule means that the difference between two LCI results, such as for two competing products,
must be more than 25% different for the results to be deemed significantly different, and
thus for one to be declared as lower than the other. While there is not a large quantitative
framework behind the choice of 25% specifically, this heuristic is common because it
roughly expresses the fact that all data used in such studies is inherently uncertain, and by
forcing 25% differences, then relatively small differences would be deemed too small to be
noted in study conclusions. We will talk more about modeling and assessing uncertainties in
Chapter 11 on uncertainty.
Interpretation can also serve as an additional check on the goal and scope parameters. This is
where you could assess whether a system boundary is appropriate. As an example, while the
ISO Standard encourages full life cycle stage coverage within system boundaries, it does not
require that every LCA encompass all stages. One could try to defend the validity of a life
cycle study of an automobile that focused only on manufacturing, or only on the use stage.
The results of the interpretation phase could then internally weigh in on whether such a
decision was appropriate given the study goal. If a (qualified) conclusion can be drawn, the
study could be left as-is, if not, a broader system boundary could be chosen, with or without
preliminary LCI results.
Regardless, the real purpose of interpretation is to improve the quality of your study,
especially the quality of the written conclusions and recommendations that arise from your
quantitative work. As with other quantitative analysis methods, you will need to also improve
your qualitative skills, including documentation, to ensure that your interpretation efforts are
worthwhile.

Identifying and Using Life Cycle Data Sources

In support of modeling the inputs and outputs associated with unit processes, you will need
a substantial amount of data. Even studies of simple product systems may require data on 10
different unit processes. While this may sound like a small amount of effort, as you will see
below, the task of finding, documenting, manipulating, validating and using life cycle data is
time consuming. The text above gave a fair amount of additional detail related to developing
your own primary data via collection and sampling efforts. This section is related to the
identification and use of secondary data.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

112

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

One prominent source of secondary data is the thousands of peer-reviewed journal papers
done over time by the scientific community, also known as literature sources. Some of
these papers have been explicitly written to be a source of secondary data, while authors of
other papers developed useful data in the course of research (potentially on another topic)
and made the process-level details available as part of the paper or in its supporting
information. Sometimes the study authors are not just teams of researchers, but industry
associations or trade groups (e.g., those trying to disseminate the environmental benefits of
their products). Around the world, industry groups like Plastics Europe, the American
Chemistry Council, and the Portland Cement Association have sponsored projects to make
process-based data available via publications. It is common to see study authors citing
literature sources, and doing so requires you to simply use a standard referencing format like
you would for any source. Unfortunately, data from such sources is typically not available in
electronic form, and thus there are potentials for data entry or transcription errors as you try
to make use of the published data. It is due to issues like these that literature sources
constitute a relatively small share of secondary data used in LCA studies.
There is a substantial amount of secondary data available to support LCAs in various life
cycle databases. These databases are the main source of convenient and easy to access
secondary data. Some of the data represented in these databases are from the literature
sources mentioned above. Since the first studies mentioned in Chapter 1, various databases
comprised of life cycle inventory data have been developed. The original databases were sold
by Ecobilan and others in the mid-1990s. Nowadays the most popular and rigorously
constructed database is from ecoinvent, developed by teams of researchers in Switzerland
and available either by paying directly for access to their data website or by an add-on fee to
popular LCA system tools such as SimaPro and GaBi (which in turn have their own
databases). None of these databases are free, and a license must be obtained to use them. On
the other hand, there are a variety of globally available and publicly accessible (free) life cycle
databases. In the US, LCI data from the National Renewable Energy Laboratory (NREL)'s
LCI database and the USDA's LCA Digital Commons are popular and free3. Figure 5-4
summarizes the major free and paid life cycle databases (of secondary data) in the world that
provide data at the unit process level for use in life cycle studies. Beyond the individual
databases, there is also an "LCA Data Search Engine," managed by the United Nations
Environmental Programme (UNEP), that can assist in finding available free and commercial
unit process data (LCA-DATA 2013). All of the databases have their own user's guides that
you should familiarize yourself with before searching or using the data in your own studies.

Data from the NREL US LCI Database has been transferred over to the USDA LCA Digital Commons as of 2012. Both datasets can
now be accessed from that single web database.
3

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Database

Approximate
Cost

Number of
processes

Notes

ecoinvent

2,500 Euros
($3,000 USD)

4,000+

Has data from around the world, but majority is

from Europe. Available directly, or embedded
within LCA software.

US NREL LCI
Database

Free
(companies,
agencies pay to
publish data)

600+

US focused. Now hosted by USDA LCA Digital

Commons.

USDA LCA
Digital Commons

Free
(manufacturers
and agencies
pay to publish
data)

300+

Focused on agricultural products and processes.

Geospatially specific unit processes for specific US
states.

ELCD

Free

300+

Relatively few processes, spread across various

sectors. Additional data being added rapidly.

BEES

Free

GaBi

$3,000 USD

113

Focused on building and construction materials.

5,000+

Database made by PE International. Global, but

heavily focused on European data.
Figure 5-4: Summary of Data Availability for Free and Licensed LCA Databases
(Sources provided at end of chapter)

These databases can be very comprehensive, with each containing data on hundreds to
thousands of unique processes, with each process comprised of details for potentially
hundreds of input or output flows. Collecting the various details of inputs and outputs for a
particular unit process (which we refer to as an LCI data module but which are referred to
as "datasets" or "processes" by various sources) requires a substantial amount of time and
effort. This embedded level of effort for unit process data is important because even though
it represents a secondary data source, to create a superior set of primary data for a study, you
might need to collect data for 100 or more input and output flows for the process. Of course
your study may have a significantly smaller scope that includes only 5 flows, and thus your
data collection activities would only need to measure those. The databases do highlight an
ongoing conundrum in the LCA community the nave stated preference for primary data
when substantial high-quality secondary data is pervasive. Another benefit of these databases
is that subsets of the data modules are created and maintained consistently, thus a common
set of assumptions or methods would be associated with hundreds of processes. This is yet
another difference to primary data which could have a set of ad-hoc assumptions used in its
creation.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

114

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Now that the availability of vast secondary data sources has been introduced, we discuss the
data structures typical of these LCI data modules. As with many facets of LCA, there is a
global standard for storing information in LCI data modules, known as EcoSpold. The
EcoSpold format is a structured way of storing and exchanging LCI data, where details such
as flows and allocation methods are classified for each process. There is no requirement that
LCA tools use the EcoSpold format, but given its popularity and the trend that all of the
database sources in Figure 5-4 use this format, it is worth knowing. Instead of giving details
on the format (which is fairly technical and generally only useful for personnel involved in
creating LCA software) we instead will demonstrate the way in which LCI data modules are
typically represented in the database and allow you to think about the necessary data
structures separately.
In the rest of this chapter we consider an LCI of the CO2 emitted to generate 1 kWh of coalfired electricity in the United States. Our system boundary for this example (as in Figure 5-5)
has only three unit processes: mining coal, transporting it by rail, and burning it at a power
plant. The refinery process that produces diesel fuel, an input for rail, is outside of our
boundary, but the effects of using diesel as a fuel are included. We can assume, beyond the
fact that this is an academic example, that such a tight boundary is realistic because these are
known to be significant parts of the supply chain of making coal-fired power. We will
discuss the use of screening methods to help us set such boundaries in Chapter 8.

Figure 5-5: Product System Diagram for Coal-Fired Electricity LCI Example

To achieve our goal of the CO2 emitted per kWh, we will need to find process-level data for
coal mining, rail transportation, and electricity generation. In the end, we will combine the

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

115

results from these three unit processes into a single estimate of total CO2 per kWh. This way
of performing process-level LCA is called the process flow diagram approach.
We will focus on the US NREL LCI database (2013) in support of this relatively simple
example. This database has a built-in search feature such that typing in a process name or
browsing amongst categories will show a list of available LCI data modules (see the
Advanced Material at the end of this chapter for brief tutorials on using the LCA Digital
Commons website, that hosts the US LCI data, as well as other databases and tools).
Searching for "electricity" yields a list of hundreds of processes, including these LCI data
modules:

Electricity, diesel, at power plant

Electricity, lignite coal, at power plant

Electricity, natural gas, at power plant

Electricity, anthracite coal, at power plant

Electricity, bituminous coal, at power plant

The nomenclature used may be confusing, but is somewhat consistent across databases. The
constituents of the module name can be deciphered as representing (1) the product, (2) the
primary input, and (3) the boundary of the analysis. In each of the cases above, the unit
process is for making electricity. The inputs are various types of fuels. Finally, the boundary
is such that it represents electricity leaving the power plant (as opposed to at the grid, or at a
point of use like a building). Once you know this nomenclature, it is easier to browse the
databases to find what you are looking for specifically.
Given the above choices, we want to use one of the three coal-fueled electricity generation
unit processes in our example. Lignite and anthracite represent small shares of the
generation mix, so we choose bituminous coal as the most likely representative process and
use the last data module in the list above (alternatively, we could develop a weighted-average
process across the three types that would be useful). Using similar brief search methods in
the US NREL website we would find the following unit processes as relevant for the other
two pieces of our system:

Bituminous coal, at mine

Transport, train, diesel powered

These two processes represent mining of bituminous coal and the transportation of generic
product by diesel-powered train.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

116

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Figure 5-6 shows an abridged excerpt of the US NREL LCI data module for Electricity,
bituminous coal, at power plant. The entire data module is available publicly4. Within the US
NREL LCI database website, such data is found by browsing or searching for the process
name and then viewing the "Exchanges". These data modules give valuable information
about the specific process chosen as well as other processes they are linked to. While here
we discuss viewing the data on the website, it can also be downloaded to a Microsoft Excel
spreadsheet or as XML.
It is noted that this is an abridged view of the LCI data module. The complete LCI data
module consists of quantitative data for 7 inputs and about 60 outputs. For the sake of the
example in this section, we assume the abridged inventory and ignore the rest of the details.
Most of the data modules in databases have far more inputs and outputs than in this
abridged module; it is not uncommon to find data modules with hundreds of outputs (e.g.,
for emissions of combustion processes). If you have a narrow scope that focuses on a few
air emission outputs, many of the other outputs can be ignored in your analysis. However if
you plan to do life cycle impact assessment, the data in the hundreds of inputs and/or
outputs may be useful in the impact assessment. If your study seeks to do a broad impact
assessment, collecting your own primary data can be problematic as your impact assessment
will all but require you to broadly consider the potential flows of your process. If you focus
instead on just a few flows you deem to be important, then the eventual impact assessment
could underestimate the impacts of your process. This is yet another danger of primary data
collection (undercounting flows).

Data from the NREL US LCI database in this chapter are as of July 20, 2014. Values may change in revisions to the database that cannot
be expressed here.
4

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Flow

Category

Type

Unit

Amount

bituminous coal, at mine

root/flows

ProductFlow

kg

4.42e-01

transport, train, diesel

powered

root/flows

ProductFlow

t*km

4.61e-01

electricity, bituminous coal, at

power plant

root/flows

ProductFlow

kWh

1.00

carbon dixoide, fossil

air/unspecified

ElementaryFlow

kg

9.94e-01

117

Comment

Inputs

Transport from mine

to power plant

Outputs

Figure 5-6: Abridged LCI data module from US NREL LCI Database for bituminous coal-fired
electricity generation. Output for functional unit italicized. (Source: US LCI Database 2012)

Figure 5-6 is organized into sections of data for inputs and outputs. At the top, we see the
abridged input flows into the process for generating electric power via bituminous coal.
Recalling the discussion of direct and indirect effects from Chapter 4, the direct inputs listed
are bituminous coal and train transport. The direct outputs listed are fossil CO2 emissions
(which is what results when you burn a fossil fuel) and electricity. Before discussing all of the
inputs and outputs, we briefly focus on the output section to identify a critical component of
the data module the electricity output is listed as a product flow, with units of 1 kWh.
Every LCI process will have one or more outputs, and potentially have one or more product
flows as outputs, but this module has only one. That means that the functional unit basis for
this unit process is per (1) kWh of electricity. All other inputs and outputs in Figure 5-6,
representing the US NREL LCI data module for Electricity, bituminous coal, at power plant are
presented as normalized per 1 kWh. You could think of this module as providing energy
intensities or emissions factors per kWh. Thinking back to the discussion above on data
collection, its unlikely that the study done to generate this LCI data module actually
measured the inputs and outputs needed to make just 1 kWh of electricity at a power plant
it is too small a value. In reality, it is likely that the inputs and outputs were measured over
the course of a month or year, and then normalized by the total electricity generation in kWh
to find these normalized values. It is the same process you would do if you were making the
LCI data module yourself. We will discuss how to see the assumptions and boundaries for
the data modules later in this chapter.
We now consider the abridged data module in more detail. In Figure 5-6, each of the input
flows are a product flow from another process (namely, the product of bituminous coal
mining and the product of train transportation). The unit basis assumption for those inputs
is also given kg for the coal and ton-kilometers (t*km) for the transportation. A tonLife Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

118

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

kilometer is a compound unit (like a kilowatt-hour) that expresses the movement of 1 ton of
material over the distance of 1 kilometer. Both are common SI units. Finally the amount of
input required is presented in scientific notation and can be translated into 0.442 kg of coal
and 0.46 ton-km of train transport. Likewise, the output CO2 emissions to air are estimated
at 0.994 kg. All of these quantities are normalized on a per-kWh generated basis. The
comment column in Figure 5-6 (and which appears in many data modules) gives brief but
important notes about specific inputs and outputs. For example, the input of train
transportation is specified as being a potential transportation route from mine to power
plant, which reminds us that the unit process for generating electricity from coal is already
linked to a requirement of a train from the mine.5
Now that we have seen our first example of a secondary source LCI data module, Figure 5-7
presents a graphical representation of the abridged unit process similar to the generic
diagram of Figure 5-2. The direct inputs, which are product flows from other man made
processes, are on the left side as inputs from the technosphere. The abridged unit process
has no direct inputs from nature. The direct CO2 emissions are at the top. The output
product, and functional unit basis of the process, of electricity is shown on the right. All
quantitative values are representative of the functional unit basis of the unit process.

Figure 5-7: Unit Process Diagram for abridged electricity generation unit process

Returning to our example LCA problem, we now have our first needed data point, that the
direct CO2 emissions are 0.994 kg / kWh generated. Given that we have only three unit
processes in our simple product system, we can work backwards from this initial point to get
estimated CO2 emissions values from mining and train transport. Again using the NREL
LCI database, Figure 5-8 shows abridged data for the data module bituminous coal, at mine. The
The unabridged version of the module has several other averaged transport inputs in ton-km, such as truck, barge, etc. Overall, the
module gives a "weighted average" transport input to get the coal from the mine to the power plant. Since we are only using the abridged
(and unedited) version, we will otherwise undercount the upstream CO2 emissions from delivering coal since we are skipping the weighted
effects from those other modes.
5

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

119

output and functional unit is 1 kg of bituminous coal as it leaves the mine. Two important
inputs are diesel fuel needed to run equipment, and coal. It may seem odd to see coal listed
as an input into a coal mining process, but note it is listed as a resource and as an elementary
flow. As discussed in Chapter 4, elementary flows are flows that have not been transformed
by humans. Coal trapped in the earth for millions of years certainly qualifies as an elementary
flow by that definition! Further, it reminds us that there is an elementary flow input within
our system boundary, not just many product flows. This particular resource is also specified
as being of a certain quality, i.e., with energy content of about 25 MJ per kg. Finally, we can
see from a mass balance perspective that there is some amount of loss in the process, i.e.,
that every 1.24 kg of coal in the ground leads to only 1 kg of coal leaving the mine.
Flow

Category

Type

Unit

Amount

Coal, bituminous, 24.8 MJ per kg

resource/ground

ElementaryFlow

kg

1.24

Diesel, combusted in industrial boiler

root/flows

ProductFlow

8.8e-03

Comment

Inputs

Outputs
Bituminous coal, at mine
root/flows
ProductFlow
kg
1.00
Figure 5-8: Abridged LCI data module from US NREL LCI Database for bituminous coal mining.
Output for functional unit italicized. (Source: US LCI Database 2012)

Figure 5-9 shows the abridged NREL LCI data module for rail transport (transport, train, diesel
powered). The output / functional unit of the process is 1 ton-km of rail transportation service
provided. Providing that service requires 0.00648 liters of diesel fuel and emits .0189 kg of
CO2, both per ton-km.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

120

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Flow

Category

Type

Unit

Amount

root/flows

ProductFlow

6.48e-03

air/unspecified

ElementaryFlow

kg

1.89e-02

Comment

Inputs
Diesel, at refinery
Outputs
Carbon dixoide, fossil

transport, train, diesel

root/flows
ProductFlow
t*km 1
powered
Figure 5-9: Abridged LCI data module from US NREL LCI Database for rail transportation. Output
for functional unit italicized. (Source: US LCI Database 2012)

To then find the total CO2 emissions across these three processes, we can work backwards
from the initial process. We already know there are 0.994 kg/kWh of CO2 emissions at the
power plant. But we also need to mine the coal and deliver it by train for each final kWh of
electricity. The emissions for those activities are easy to associate, since Figure 5-6 provides
us with the needed connecting units to estimate the emissions per kWh. Namely, that 0.442
kg of coal needs to be mined and 0.461 ton-km of rail transport needs to be used per kWh
of electricity generated. We can then just use those unit bases to estimate the CO2 emissions
from those previous processes. Figure 5-8 does not list direct CO2 emissions from coal
mining, although it does list an input of diesel used in a boiler6. If we want to assume that we
are only considering direct emissions from each process, we can assume the CO2 emissions
from coal mining to be zero7, or we could expand our boundary and acquire the LCI data
module for the diesel, combusted in industrial boiler process. Our discussion below follows the
assumption that direct emissions are zero.
Figure 5-9 notes that there are 0.0189 kg of CO2 emissions per ton-km of rail transported.
Equation 5-1 summarizes how to calculate CO2 emissions per kWh for our simplistic
product system. Other than managing the compound units, it is a simple solution: about 1
kg CO2 per kWh. If we were interpreting this result, we would note that the combustion of
coal at the power plant is about 99% of the total emissions.
0.994 kg CO2 /kWh + 0.442 kg * 0 + (0.461 ton-km / kWh)*(0.0189 kg CO2 / ton-km) =
0.994 kg CO2 / kWh + 0.0087 kg CO2 / kWh = 1.003 kg CO2 / kWh

(5-1)

The estimated CO2 emissions for coal-fired electricity of 1 kg / kWh was obtained relatively
easily, requiring only three steps and queries to a single database (US NREL LCI). As always
This particular input of "diesel, combusted in industrial boiler" may not be what you would expect to find in an LCI data module, since it is a
description of how an input of diesel is used. Such flows are fairly common though.
7 Also, the unabridged LCI data modules list emissions of methane to air, which could have been converted to equivalent CO emissions.
2
Doing so would only change the result above by about 10%.
6

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

121

one of our first questions should be "is it right?" We can attempt to validate this value by
looking at external references. Whitaker et al (2012) reviewed 100 LCA studies of coal-fired
electricity generation and found the median value to be 1 kg of CO2 per kWh, thus we
should have reasonable faith that the simple model we built leads to a useful result. Of
course we can add other processes to our system boundary (such as other potential
transportation modes) but we would not appreciably change our simple result of 1 kg/kWh.
Note that anecdotally experts often refer to the emissions from coal-fired power plants to be
2 pounds per kWh, which is a one significant digit equivalent to our 1 kg/kWh result.
Process-based life cycle models are constructed in this way. For each unit process within the
system boundary, data (primary or secondary) is gathered and flows between unit processes
are modeled. Since you must find data for each process, such methods are often referred to
as "bottom up" studies because you are building them up from nothing, as you might
construct a building on empty land.
Beyond validating LCI results, you should also try to validate the values found in any unit
process you decide to use, even if sourced from a well-known database. That is because
errors can and do exist in these databases. It is easy to accidentally put a decimal in the
wrong place when creating a digital database. As an example, the US NREL LCI database
had an error in the CO2 emissions of its air transportation process, of 53 kg per 1000 ton-km
(0.053 kg per ton-km) for several years before it was fixed. This error was brought to their
attention because observant users noted that this value was less than the per-ton-km
emissions for truck transportation, which went against common sense. Major releases of
popular databases are also imperfect. It is common to have errors found and fixed, but this
may happen months after licenses have been purchased, or worse, after studies have been
completed. These are additional reasons why despite being of high quality, you need to
validate your data sources.

Details for Other Databases

The discussion above was focused on the US NREL LCI Database, which contains only
process data for US-based production, yet there are other considerations both for data
access and metadata for the other databases. As noted in Figure 5-4, the ecoinvent database
is far more geospatially diverse. While generally focused on Europe, data can be found in
ecoinvent for other regions of the world as well. This fact creates a new challenge in
interpreting available process data modules, namely, determining the country of production
basis assumption for the data. While examining the metadata can be useful, ecoinvent and
other databases typically summarize the country used within the process naming convention.
For example, a process you might find within ecoinvent might be called electricity, hard coal, at
power plant, DE, where the first part is the process name formatted similar to the NREL
database, and at the end is an abbreviated term for the country or region to which that

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

122

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

process is representative. Figure 5-10 summarizes some of the popular abbreviations used
for country basis within ecoinvent.
Country or Region

Abbreviation

Country or Region

Abbreviation

Norway

NO

Japan

JP

Australia

AU

Canada

CA

India

IN

Global

GLO

China

CN

Europe

RER

Germany

DE

Africa

RAF

United States

US

Asia

RAS

Netherlands

NL

RU

Hong Kong

HK

France

FR

Russian Federation
Latin America and the
Caribbean
North America

RLA
RNA

United Kingdom
GB
Middle East
RME
Figure 5-10: Summary of abbreviations for countries and regions in ecoinvent

Ecoinvent has substantially more available metadata for its data modules, including primary
sources, representative years, and names of individuals who audited the datasets. While
ecoinvent data are not free, the metadata is freely accessible via the database website. Thus,
you could do a substantial amount of background work verifying that ecoinvent has the data
you want before deciding to purchase a license.
A particular feature of ecoinvent data is its availability at either the unit process or system
process level. Viewing and using ecoinvent system processes is like using already rolled-up
information (and computations would be faster), while using unit processes will be more
computationally intensive. This will be discussed more in Chapter 9.

LCI Data Module Metadata

Our example using actual LCI data modules from the US NREL LCI database jumped
straight into extracting and looking at the quantitative data. However, all LCI data modules
provide some level of metadata, which is information regarding how the data was collected,
how the modules were constructed, etc. Metadata is also referred to as "data about data".
The metadata that we care about for our unit processes are elements such as the year the
data was collected, where it was collected, whether the values are single measurements or
averages, and whether it was peer reviewed. To understand metadata more, we can look at
the metadata available for the processes we used above. The US NREL LCI Database has
three different metadata categories as well as the Exchanges information shown above.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

123

Figure 5-11 shows metadata from the Activity metadata portion of the US NREL LCI
database for the Electricity, bituminous coal, at power plant process used above. This metadata
notes that the process falls into the Utilities subcategory (used for browsing on the website)
and that it has not yet been fully validated. It applies to the US, and thus it is most
appropriate for use in studies looking to estimate impacts of coal-fired electricity generation
done within the United States. Note that this does not mean that you can only use it for that
geographical region. A process like coal-fired generation is quite similar around the world;
although factors such as pollution controls may differ greatly by region. However, since
capture of carbon is basically non-existent, if we wanted to use this process to estimate CO2
emissions from coal-fired generation in other regions it might still be quite useful.
The metadata field for "infrastructure process" notes whether the process includes estimated
infrastructure effects. For example, one could imagine two parallel unit processes for
electricity generation, where one includes estimated flows from needing to build the power
plant and one does not (such as the one referenced above). In general, infrastructure
processes are fairly rare, and most LCA study scopes exclude consideration of infrastructure
for simplicity.
Name

Electricity, bituminous coal, at power plant

Category

Utilities - Fossil Fuel Electric Power Generation

Description

Important note: although most of the data in the US LCI database

has undergone some sort of review, the database as a whole has not
yet undergone a formal validation process. Please email comments to
[email protected].

Location

US

Geography Comment

United States

Infrastructure Process

False

Quantitative Reference
Electricity, bituminous coal, at power plant
Figure 5-11: Activity metadata for Electricity, bituminous coal, at power plant process

Figure 5-12 shows the Modeling metadata for the coal-fired generation unit process. There is
no metadata provided for the first nine categories of this category, but there are ten
references provided to show the source data used to make the unit process. While a specific
"data year" is not dictated by the metadata, by looking at the underlying data sources, the
source data came from the period 1998-2003. Thus, the unit process data would be most
useful for analyses done with other data from that time period. If we wanted to use this
process data for a more recent year, we would either have to look for an LCI data module

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

124

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

that was newer, or verify that the technologies have not changed much since the 1998-2003
period.
LCI Method
Modelling constants
Data completeness
Data selection
Data treatment
Sampling procedure
Data collection period
Reviewer
Other evaluation
Sources

U.S. EPA 1998 Emis. Factor AP-42 Section 1.1, Bituminus and Subbituminus Utility
Combustion
U.S. Energy Information Administration 2000 Electric Power Annual 2000
Energy Information Administration 2000 Cost and Quality of Fuels for Electric
Utility Plants 2000
Energy Information Administration 2000 Electric Power Annual 2000
U.S. EPA 1998 Study of Haz Air Pol Emis from Elec Utility Steam Gen Units V1
EPA-453/R-98-004a
U.S. EPA 1999 EPA 530-R-99-010
unspecified 2002 Code of Federal Regulations. Title 40, Part 423
Energy Information Administration 9999 Annual Steam-Electric Plant Operation and
Design Data
Franklin Associates 2003 Data Details for Bituminous Utility Combustion
Figure 5-12: Modeling metadata for Electricity, bituminous coal, at power plant process

Finally, Figure 5-13 shows the Administrative metadata for the Electricity, bituminous coal, at
power plant process. There are no explicitly-defined intended applications (or suggested
restrictions on such applications), suggesting that it is broadly useful in studies. The data are

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

125

not copyrighted, are publicly available, and were generated by Franklin Associates, a
subsidiary of ERG, one of the most respected life cycle consulting business in the US. The
"Data Generator" is a significant piece of information. You may opt to use or not use a data
source based on who created it. A reputable firm has a high level of credibility. A listed
individual with no obvious affiliation or reputation might be less credible. Finally, the
metadata notes that it was created and last updated in October 2011, meaning that perhaps it
was last checked for errors on this date, not that the data is confirmed to still be valid for the
technology as of this date.
Intended Applications

"

Copyright

false

Restrictions

All information can be accessed by everybody.

Data Owner
Data Generator

Franklin Associates

Data Documentor

Franklin Associates

Project
Version
Created

2011-10-24

Last Update
2011-10-24
Figure 5-13: Administrative metadata for Electricity, bituminous coal, at power plant process

Our metadata examples have focused on the publicly available US NREL LCI Database, but
other databases like ELCD and ecoinvent have similar metadata formats. These other
databases typically have more substantive detail, in terms of additional fields and more
consistent entries in these fields. Since these other data sources are not public, we have not
used examples here.
You should browse through the available metadata for any of the databases that you have
access to, so that you can better appreciate the records that may exist within various
metadata records. Remember that the reason for better appreciating the value of the
metadata is to help you with deciding which secondary data sources to use, and how
compatible they are with your intended goal and scope.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

126

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Referencing Secondary Data

When you use secondary data as part of your study it must be appropriately referenced, as
with any other source. Referencing data sources was first mentioned in Chapter 2, but here
we discuss several important additions for referencing data from LCA databases. As an
example, the US NREL LCI database explicitly suggests the following referencing style for
use of its data modules:
When referencing the USLCI Database, please use the following format: U.S. Life Cycle
Inventory Database. (2012). National Renewable Energy Laboratory, 2012. Accessed
November 19, 2012: https://www.lcacommons.gov/nrel/search
However, this is the minimum referencing you should provide for process data. First of all,
you can not simply reference the database. You need to ensure that the specific unit process
from which you have used data is clear to the reader, for example if they would like to
validate your work. That means you need to explicitly reference the name of the process
(either obviously in the text or in the reference section). In the US NREL database and other
sources, there may be hundreds of LCI data modules for electricity. Thus, the danger is that
in the report you loosely reference data for coal-fired electricity generation as being from
"the NREL database", but do not provide enough detail for the reader to know which
electricity process was used. Unfortunately, this is a common occurrence in LCA reports.
This situation can be avoided by explicitly noting the name of the process used in the
reference, such as:
U.S. Life Cycle Inventory Database. Electricity, bituminous coal, at power plant unit process
(2012). National Renewable Energy Laboratory, 2012. Accessed Nov. 19, 2012:
https://www.lcacommons.gov/nrel/search
A generic reference to the database, as given at the top of this section, may be acceptable if
the report separately lists all of the specific processes used in the study, such as in an
inventory data source table listing all of the processes used.
You will likely use multiple unit processes from the same database. You can either create
additional references like the one above for each process, or use a combined reference that
lists all processes as part of the reference, such as:
U.S. Life Cycle Inventory Database. Electricity, bituminous coal, at power plant; bituminous coal,
at mine; transport, train, diesel powered unit processes (2012). National Renewable Energy
Laboratory, 2012. Accessed Nov. 19, 2012: https://www.lcacommons.gov/nrel/search
The greater the number of similar processes, the greater the need to specify which specific
data module you used in your analysis. This becomes especially important if you are using
LCI data modules from several databases.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

127

A final note about referencing is that the LCA databases are generally not primary sources,
they are secondary sources. Ideally, sources would credit the original author, not the database
owner who is just providing access. If the LCI data module is taken wholesale from another
source (i.e., if a single source were listed in the metadata), it may make sense to also
reference the primary source, or to add the primary source to the database reference. In this
case the reference might look like one of the following:
RPPG Of The American Chemistry Council 2011. Life Cycle Inventory Of Plastic
Fabrication
Processes:
Injection
Molding
And
Thermoforming.
http://plastics.americanchemistry.com/Education-Resources/Publications/LCI-ofPlastic-Fabrication-Processes-Injection-Molding-and-Thermoforming.pdf. via U.S. Life
Cycle Inventory Database. Injection molding, rigid polypropylene part, at plant unit process
(2012). National Renewable Energy Laboratory, 2012. Accessed November 19, 2012:
https://www.lcacommons.gov/nrel/search
U.S. Life Cycle Inventory Database. Injection molding, rigid polypropylene part, at plant unit
process (2012). National Renewable Energy Laboratory, 2012. Accessed November 19,
2012: https://www.lcacommons.gov/nrel/search (Primary source: RPPG Of The
American Chemistry Council 2011. Life Cycle Inventory Of Plastic Fabrication
Processes:
Injection
Molding
And
Thermoforming.
http://plastics.americanchemistry.com/Education-Resources/Publications/LCI-ofPlastic-Fabrication-Processes-Injection-Molding-and-Thermoforming.pdf)
As noted in Chapter 2, ideally you would identify multiple data sources (i.e., multiple LCI
data modules) for a given task. This is especially useful when using secondary data because
you are not collecting data from your own controlled processes. Since the data is secondary,
it is likely that there are slight differences in assumptions or boundaries than what you would
have used if collecting primary data. By using multiple sources, and finding averages and/or
standard deviations, you could build a more robust quantitative model of the LCI results. We
will discuss such uncertainty analysis for inventories in Chapter 10.

Additional Considerations about Secondary Data and Metadata

Given the types and classes of data we are likely to find in life cycle studies, we introduce in
this subsection a few more considerations to ensure you are finding and using appropriate
types of data to match the needs of your study. These considerations are in support of the
data quality requirements.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

128

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Temporal Issues
In creating temporal data quality requirements, you will set a target year (or years) for data
used in your study. For example, you might have a DQR of "2005 data" or "data from 20052007" or "data within 5 years of today". After setting target year(s) you then must do your
best to find and use data that most closely matches the target. It is likely that you will not be
able to match all data with the target year(s). When setting and evaluating temporal DQRs,
the following issues need to be understood.
You may need to do some additional work to guarantee you know the basis year of the data
you find, but this is time well spent to ensure compatibility of the models you will build. You
will need to distinguish between the year of data collection and year of publication. In our
CBECS example in Chapter 2, the data were collected in the year 2003 but the study was not
published by DOE until December 2006 (or, almost 2007). It is easy to accidentally consider
the data as being for 2006 because the publication year is shown throughout the reports. But
the data were representative of the year 2003. If your temporal DQR was set at "2005", you
might still be able to justify using the 2003 CBECS data, but would need to assess whether
the electricity intensity of buildings likely changed significantly between 2003 and 2005. The
same types of issues arise when using sources such as US EPA's AP-42 data, which are
compilations of (generally old) previously estimated emissions factors. Other aspects of your
DQRs may further help decide the appropriateness of data newer or older than your target
year.
The same is true of dates given in the metadata of LCI data modules. You don't care about
when you accessed the database, or when it was published in the database. You care about
the primary source's years of analysis. Figure 5-12 showed metadata on the coal-fired
electricity generation process where the underlying data was from 1998-2003, and which was
put in the US LCI database in 2011. An appropriate "timestamp" for this process would be
1998-2003.
While on the topic of temporal issues, we revisit the point about age of data in databases.
The US LCI database project started in the mid-2000s. Looking at the search function in
that database, you can find a distribution of the "basis year" of all of the posted data
modules. This is a date that is not visible within the metadata, but is available for
downloaded data modules and summarized in the web server. Figure 5-14 shows a graph of
the distribution of the years. In short, there is a substantial amount of relatively old data, and
a substantial amount of data where this basis year is not recorded (value given as '9999').
Half of the 200 data modules updated in 2010 are from an update to the freight
transportation datasets. These could be key considerations when considering the suitability
of data in a particular database.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

129

Figure 5-14: Frequency Distribution of Data Years in US NREL LCI Database

(as of August 15, 2013)

Geospatial Issues
You must try to ensure that you are using data with the right geographical and spatial scope
to fit your needs. If you are doing a study where you want to consider the emissions
associated with producing an amount of electricity, then you will find many potential data
sources to use. The EIA has data that can give you the average emissions factors for
electricity generation across the US. E-GRID (a DOE-EPA partnership) can give you
emissions factors at fairly local levels, reflecting the types of power generation used within a
given region. The question is the context of your study. Are you doing a study that inevitably
deals with national average electricity? Then the EIA data is likely suitable. Or are you doing
a study that needs to know the impact of electricity from a particular factory's production?
In that case you likely want a fairly local data source, e.g., from E-GRID. An alternative is to
leverage the idea of ranges, presented in Chapter 2, to represent the whole realm of possible
values for electricity generation, including various local or regional averages all the way up to
the national average.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

130

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Uncertainty and Variability

Sadly, in the field of LCA there are many practitioners who actively or passively ignore the
effects of uncertainty or variability in their studies. They treat all model inputs as single
values and generate only a single result. The prospect of uncertainty or variability is lost in
their model, and typically then that means those effects are lost on the reader of the study.
How can we support a big decision (e.g., paper vs. plastic?) if there is much uncertainty in
the data but we have completely ignored it? We are likely to end up supporting poor
decisions if we do so. We devote Chapter 11 to methods of overcoming and structuring
uncertainty in LCA models.

Chapter Summary
Typically, the most time consuming aspect of an LCA (or LCI) study relates to the data
collection and management phase. While the LCA Standard encourages practitioners to
collect primary data for the product systems being studied, typically secondary data is used
from prior published studies and databases. Using secondary data requires being
knowledgeable and cognizant of issues relating to the sources of data presented and also
requires accurate referencing. Data quality requirements help to manage expectations of the
study team as well as external audiences pertaining to the goals of your data management
efforts. Utilization of effective LCI data management methods leads to excellent and wellreceived studies.

References for this Chapter

BEES LCA Tool, website, http://ws680.nist.gov/Bees/Default.aspx, last accessed August
12, 2013.
ecoinvent website, www.ecoinvent.ch, last accessed August 12, 2013.
ELCD LCA Database, website, http://lca.jrc.ec.europa.eu/lcainfohub/, last accessed
August 12, 2013.
Environmental Protection Agency. 1993. Life Cycle Assessment: Inventory Guidelines and
Principles. EPA/600/R-92/245. Office of Research and Development. Cincinnati, Ohio,
USA.
Gabi Software, website, http://www.gabi-software.com/, last accessed August 12, 2013.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

131

LCA-DATA, UNEP, website, http://lca-data.org:8080/lcasearch, last accessed August 12,

2013.
Quantis, "Environmental Life Cycle Assessment of Drinking Water Alternatives and
Consumer Beverage Consumption in North America", LCA Study completed for Nestle
Waters
North
America,
2010,
http://www.beveragelcafootprint.com/wpcontent/uploads/2010/PDF/Report_NWNA_Final_2010Feb04.pdf,
last
accessed
September 9, 2013.
The Agribusiness Group, "Life Cycle Assessment: New Zealand Merino Industry Merino
Wool
Total
Energy
Use
and
Carbon
Dioxide
Emissions",
2006,
http://www.agrilink.co.nz/Portals/Agrilink/Files/LCA_NZ_Merino_Wool.pdf,
last
accessed September 1, 2013.
US NREL LCI Database, website, http://www.nrel.gov/lci/, last accessed August 12, 2013.
U.S. Life Cycle Inventory Database. Electricity, bituminous coal, at power plant, bituminous coal, at
mine, and transport, train, diesel powered unit processes (2012). National Renewable Energy
Laboratory, 2012. Accessed August 15, 2013: https://www.lcacommons.gov/nrel/search
USDA LCA Digital Commons, website, http://www.lcacommons.gov, last accessed August
12, 2013.
Whitaker, Michael, Heath, Garvin A., O'Donoughue, Patrick, and Vorum, Martin, "Life
Cycle Greenhouse Gas Emissions of Coal-Fired Electricity Generation: Systematic Review
and Harmonization", Journal of Industrial Ecology, 2012. DOI: 10.1111/j.15309290.2012.00465.x

End of Chapter Questions

Objective 1. Recognize how challenges in data collection may lead to changes in
study design parameters (SDPs), and vice versa
1. Using the US NREL LCI Database (from the USDA Digital Commons) or another LCI
database, search or browse amongst the available categories. For each of the following
broadly defined processes in the list below, discuss how many different LCI data
modules are available and qualitatively discuss what different assumptions have been
used to generate the data modules.
a. Refining of petroleum
b. Generating electricity from fossil fuel
Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

132

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

c. Truck transportation
Objective 2. Map information from LCI data modules into a unit process
framework AND
Objective 7. Generate an inventory result from LCI data sources
2. Redo the Figure 5-5 example shown in Equation 5-1, but include the diesel, combusted in
industrial boiler process (referenced as an input in the bituminous coal mining process) within
the system boundary. What is your revised estimate of CO2 emissions per kWh? How
different is your updated estimate?
3. Redo the Figure 5-5 example but include within the system boundary refining of the
diesel used in the coal mining and rail transportation processes. Assume you have LCI
data that there are 2.5 E-04 kg fossil CO2 emissions per liter of diesel fuel refined. How
is your revised estimate of fossil CO2 emissions per kWh compared to Equation 5-1?
Objective 3. Explain the difference between primary and secondary data, and when
each might be appropriate in a study
4. Explain the difference between primary and secondary data. Provide an example of
when each would be appropriate for a study.
Objective 4. Document the use of primary and secondary data in a study
5. The data identified in part 1c above would be secondary data if you were to use it in a
study. If you instead wanted primary data for a study on trucking, discuss what methods
you might use in order to get the data.
Objective 5. Create and assess data quality requirements for a study
6. If you had data quality requirements stating that you wanted data that was national (US)
in scope, and from within 5 years of today, how many of the LCI data modules from
Question 1 would be available? Which others might still be relevant? Justify your
answer.
Objective 6. Extract data and metadata from LCI data modules and use them in
support of a product system analysis
7. Using an LCI database available to you, search for one LCI data module in each of the
following broad categories - energy, agriculture, and transportation. For each of the
three, do the following:
a. List the name of the process.
Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

133

b. Identify the functional unit.

c. Draw a unit process diagram.
d. Try to do a brief validation of the data reported.
e. Comment briefly on an example LCA study that this process might be
appropriate for, and one where it would not be appropriate.
f. Show how to appropriately reference the LCI data module in a study.
Objective 8. Perform an interpretation analysis on LCI results
8. Write an interpretation analysis for the various results expressed in Figure 5-5, and your
results from Questions 3 and 4 above.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

134

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Advanced Material for Chapter 5

The advanced material in this chapter will demonstrate how to find and access LCI data
modules from various popular databases and software tools, and how to use the data to
build simple models like the main model presented in the chapter related to coal-fired
electricity.
Not all databases and software tools are discussed; however, access methods are generally
very similar across tools. For consistency, we will demonstrate how to find the same process
data as used in the chapter so that you can learn about the different options and selections
needed to find equivalent data and metadata across tools. Specifically, we will demonstrate
how to find data from the US LCI database by using the LCA Digital Commons Website,
SimaPro (a commercial LCA tool) and openLCA (a free LCA tool).
The databases and tools use different terminology, categories, etc., to organize LCI data, but
can all lead to the same data. Seeing how each of the tools categorizes and refers to the data
is an important concept to understand.

Section 1 - Accessing Data via the US LCA Digital Commons

The LCA Digital Commons is a free, US government-sponsored and hosted web-based data
resource. Given that all of its data are publicly available, it is a popular choice for
practitioners. Thus, it is also a great resource for learning about what LCI data looks like,
how to access it, and how to build models.
The main purpose of the Digital Commons is to act as a resource for US Department of
Agriculture (USDA) agricultural data and, as a result, accessing the home page (at
https://www.lcacommons.gov/discovery) will filter access to those datasets. However, the
US LCI database previously hosted by NREL (at http://www.nrel.gov/lci/), and mentioned
extensively in Chapter 5, is also hosted via the Digital Commons website (at
https://www.lcacommons.gov/nrel/search). Given its comprehensiveness, most of the
discussion in this book is related to use of the NREL data. The examples provided below are
for accessing the NREL data source, which has slightly different metadata and contents than
the USDA data but a similar method for searching and viewing.
The LCI data modules on the Digital Commons website can be accessed via searching or
browsing. Brief overviews are provided for both options, followed by how to view and
download selected modules. Before following the tutorial below, you should consider
registering for an account on the Digital Commons website (you will need separate accounts
for the USDA and NREL data). While an account is not required to view all of the data, it is
required if you wish to download the data. You can copy and paste the data from a web
browser instead of downloading but this sometimes leads to formatting errors.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

135

Browsing for LCI Data Modules on the Digital Commons (NREL)

Figure 5-15 shows the NREL Digital Commons home page, where the left hand side shows
how the data modules are organized, including dataset type (elementary flows or unit
processes), high-level categories (like transportation and utilities), and year of data8.

Figure 5-15: Excerpt of LCA Digital Commons Website Home Page

Clicking on the + icon next to the categories generally reveals one or more additional subcategories. For example, under the Utilities category there are fossil-fired and other
generation types. Clicking on any of the dataset type, category/subcategory or year
checkboxes will filter the overall data available. The "order by" box will sort the resulting
modules. Filtering by (checking) Unit processes and the Fossil fuel electric power generation category
under Utilities, and ordering by description will display a subset of LCI data modules, as
shown in Figure 5-16. A resulting process module can be selected (see below for how to do
this and download the data).

Figure 5-16: Abridged View of LCA Digital Commons Browsing Example Results
8

The examples of the NREL US LCI Database in this section are as of July 2014, and may change in the future.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

136

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Searching for an LCI data module via keyword

The homepage has a search feature, and entering a keyword such as electricity and pressing the
Go button on the right hand side, as shown in Figure 5-17, will return a list of data modules
within the NREL LCI database that have that word in the title or category, as shown in
Figure 5-18.

Figure 5-17: Keyword search entry on homepage of NREL engine of LCA Digital Commons Website

Figure 5-18: Abridged Results of electricity keyword search

Figure 5-18 indicates that the search engine returns more than 100 LCI data modules
(records) that may be relevant to "electricity". Some were returned because electricity is in
the name of the process and others because they are in the Electric power distribution data
category. When searching, you can order results by relevance, description, or year. Once a set

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

137

of search results is obtained, results can be narrowed by filtering via the options on the left
side of the screen. For example, you could choose a subset of years to be included in the
search results, which can help ensure you use fairly recent instead of old data (as discussed
along with Figure 5-14). You can also filter based on the LCI data categories available, in this
case by clicking on the + icon next to the high-level category for Utilities, which brings up all
of the subcategories under utilities. Figure 5-19 shows the result of a keyword search for
'electricity', ordered by relevance, and filtered by the Utilities subcategory of Fossil fuel electric
power generation and by data for year 2003. The fifth search result listed is the same one
mentioned in the chapter that forms the basis of the process flow diagram example.

Figure 5-19: Abridged Results of electricity keyword search, ordered and filtered

Selecting and viewing an LCI data module

When you have searched or browsed for a module and selected by clicking on it, the module
detail summary is displayed, as in Figure 5-20.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

138

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Figure 5-20: Details for Electricity, bituminous coal process on LCA digital commons

The default result is a view of the Activity tab, which was shown in Figure 5-11. The
information available under the Modeling and Administrative tabs was presented in Figure 5-12
and Figure 5-13. Finally, an abridged view of the information available on the Exchanges tab
was also shown in Figure 5-6. Not previously mentioned is that the module can be
downloaded by first clicking on the shopping cart icon in the top right (adjacent to the
"Next item" tag). This adds it to your download cart. Once you have identified all of the data
you are interested in, you can view your current cart (menu option shown in Figure 5-21)
and request them all to be downloaded (Figure 5-22).

Figure 5-21: Selection of Current Cart Download Option on LCA Digital Commons

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

139

Figure 5-22: Cart Download Screen on LCA Digital Commons

After clicking download, you will be sent a link via the e-mail in your account registration.
As noted, the format will be an Ecospold XML file. For novices, viewing XML files can be
cumbersome, especially if just trying to look at flow information. While less convenient, the
download menu (All LCI datasets submenu) will allow you to receive a link to a ZIP file
archive containing all of the NREL modules in Microsoft Excel spreadsheet format (or you
can receive all of the modules as Ecospold XML files). You can also download a list of all of
the flows and processes used across the entire set of about 600 modules.
A spreadsheet of all flows and unit processes in the US LCI database (and their
categories) is on the www.lcatextbook.com website in the Chapter 5 folder.
When uncompressed the Electricity, bituminous coal, at power plant module file has four
worksheets, providing the same information as seen in the tabs of the Digital
Commons/NREL website above. The benefit of the spreadsheet file, though, is the ability
to copy and paste that values into a model you may be building. We will discuss building
spreadsheet models with such data in Section 4 of this advanced material.

Section 2 Accessing LCI Data Modules in SimaPro

As mentioned in the chapter, SimaPro is a popular commercial software program specifically
aimed at building quantitative LCA models. Its value lies both in these model-building
support activities as well as in being able to access various datasets from within the program.
Commercial installations of SimaPro cost thousands of dollars, but users may choose
commercial databases (e.g., ecoinvent) to include in the purchase price. Regardless of which
databases are chosen, SimaPro has the ability to use various other free datasets (e.g., US
NREL, ELCD, etc.). This tutorial assumes that such databases have already been installed
and will demonstrate how to find the same US NREL-based LCI data as in Section 1.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

140

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

This tutorial also does not describe any of the initial steps needed to purchase a license for
or install SimaPro on your Windows computer or server. It will only briefly mention the
login and database selection steps, which are otherwise well covered in the SimaPro guides
provided with the software.
Note that SimaPro refers to the overall modeling environment of data available as a
"database" and individual LCI data sources (e.g., ecoinvent) as "libraries". After starting
SimaPro, selecting the database (typically called "Professional"), and opening or creating a
new project of your choice, you will be presented with the screen in Figure 5-23. On the left
side of the screen are various options used in creating an LCA in the tool. By default the
"processes" view is selected, showing the names and hierarchy of all processes in the
currently selected libraries of the database. This list shows thousands of processes (and many
of those will be from the ecoinvent database given its large size).

Figure 5-23: Default View of Processes in Libraries When Starting SimaPro

You can narrow the processes displayed by clicking on "Libraries" on the left hand side
menu, which will display Figure 5-24. Here you can select a subset of the available libraries
for use in browsing (or searching) for process data. You can choose "Deselect all" and then
to follow along with this tutorial, click just the "US LCI" database library in order to access
only the US NREL LCI data.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

141

Figure 5-24: List of Various Available Libraries in SimaPro

If you then click the "Processes" option on the left hand side, you return to the original
screen but now SimaPro filters and shows only processes from the selected libraries, as in
Figure 5-25. Many of the previously displayed processes are no longer displayed.

Figure 5-25: View of Processes and Data Hierarchy for US-LCI Library in SimaPro

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

142

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Now that you have prepared SimaPro to look for the processes in a specific database library,
you can browse or search for data.
Browsing for LCI Data Modules in SimaPro
Looking more closely at Figure 5-25, the middle pane of the window shows the categorized
hierarchy of data modules (similar to the expandable hierarchy list in the Digital Commons
tool). However, these are not the same categories used on the NREL LCA Digital
Commons website. Instead, they are the standard categories used in SimaPro for processes
in any library. Clicking on the + icon next to any of the categories will expand it and show its
subcategories. To find the Electricity, bituminous coal process, expand the Energy category then
expand Electricity by fuel, then expand coal, resulting in a screen like Figure 5-26. Several of
the other processes burning coal to make electricity and mentioned in the chapter would also
be visible.

Figure 5-26: Processes Shown by Expanding Hierarchy of Coal-Sourced Electricity in SimaPro

The bottom pane shows some of the metadata detail for the selected process. By browsing
throughout the categories (and collapsing or expanding as needed) and reading the metadata
you can find a suitable process for your model. The tutorial will demonstrate how to view or
download such data after briefly describing how to search for the same process.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

143

Searching for a process in SimaPro

Once libraries have been specified as noted above, clicking on the magnifying glass icon in
the toolbar brings up the search interface as shown in Figure 5-27. You enter your search
term in the top box, and then choose from several search options. If you are just looking for
process data (as in this tutorial) then you would want to restrict your choice of where to look
for the data to only libraries you have currently chosen (i.e., via the interface in Figure 5-24)
rather than all libraries. This will also make your search return results more quickly. Note the
default search only looks in the names of processes, not in the metadata (the "all fields"
option changes this behavior).

Figure 5-27: Search Interface in SimaPro

Figure 5-28 shows the result of a narrowed search on the word "electricity" in the name of
processes only in "Current project and libraries" and sorted by the results column "Name".
Since we have already selected only the US LCI database in libraries, the results will not
include those from ecoinvent, etc. One of the results is the same Electricity, bituminous coal, at
power plant process previously discussed.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

144

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Figure 5-28: Results of Modified Search for Electricity in SimaPro

By clicking "Go to" in the upper right corner of the search results box, SimaPro "goes to"
the same place in the drill-down hierarchy as shown in Figure 5-26.
Viewing process data in SimaPro
To view process data, choose a process by clicking on it (e.g., as in Figure 5-26) and then
click the View button on the right hand side. This returns the process data and metadata
overview shown in Figure 5-29. Similar to the Digital Commons website, the default screen
shows high-level summary information for the process. Full information is found in the
documentation and system description tabs.

Figure 5-29: Process Data and Metadata Overview in SimaPro

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

145

Clicking on the input-output tab displays the flow data in Figure 5-30, which for this process
is now quite familiar. If you need to download this data, you can do so by choosing
"Export" in the File menu, and choosing to export as a Microsoft Excel file.

Figure 5-30: View of Process Flow Data (Inputs and Outputs) in SimaPro

Section 3 Accessing LCI Data Modules in openLCA

openLCA is a free LCA modeling environment (available at http://www.openlca.org/)
available for Windows, Mac, and Linux operating systems. While installation and
configuration can be quite complicated (and is not detailed here), various datasets are
available. The tutorial assumes you have access to a working openLCA installation with the
US LCI database, and discusses how to find the same US NREL-based LCI data as in
Section 1.
After launching openLCA and connecting to your data source you should see a list of all of
your databases, as shown in Figure 5-31. If you do not see the search and navigation tabs,
you may add them under the "Window menu -> Show views option" to add them. If you
have installed the US LCI database, it should be one of the options available.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

146

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Figure 5-31: List of Data Connections in openLCA

Browsing for process data in openLCA

Clicking on the triangle to the left of the folder allows you to open it and see the standard
hierarchy of information for all data sources in openLCA, like in Figure 5-32. This is where
you could see the process data, types of flows, and units.

Figure 5-32: Hierarchical Organization of Information for openLCA Databases

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

147

If you double click on the "Processes" folder it will display the same sub-hierarchy of
processes (not shown here) that we saw in the NREL/Digital Commons website in Section
1. All of the data for unit processes are contained under that folder. If you click on the
"Utilities" subcategory folder, then the "Fossil Fuel Electric Power Generation" folder, you
will see the Electricity, bituminous coal, at power plant seen above, as shown in Figure 5-33.
Several of the other processes burning coal to make electricity and mentioned in the chapter
would also be visible.

Figure 5-33: Expanded View of Electricity Processes in Fossil Fuel Generation Category

Searching for a process in openLCA

Instead of using the Navigation tab, a search for process data can be done using the Search
tab. Clicking on the search tab brings up the search interface, as shown in Figure 5-34.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

148

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Figure 5-34: Default Search Interface in openLCA

In the first search option, you may search in all databases or narrow the scope of your search
to only a single database (e.g., to the US-LCI database). In the second option, you may
search all object types, or narrow the scope of your search to just "Processes", etc. Finally,
you can enter a search term, such as "electricity". If you choose to search for "electricity"
only in your US LCI database (note you may have named it something different), and only in
processes, and click search you will be presented with the results as in Figure 5-35. Note that
these results have been manually scrolled down to show the same Electricity, bituminous coal, at
power plant process previously identified.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

149

Figure 5-35: Search Results for Electricity in US-LCI Database in OpenLCA

Unlike the other tools, there is no quick and easy way to skim metadata to ensure which
process you want to use.

Viewing process data in openLCA

To view process data, choose a process by double-clicking on it from either the browse or
search interface. This opens a new pane of the openLCA environment and returns the
process data and metadata overview, as shown in Figure 5-36. Similar to the Digital
Commons website, the default screen shows high-level summary information for the process
(not all of the information is shown in the Figure). Additional information is available in the
Inputs/Outputs, Administrative information, other tabs at the bottom of this pane.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

150

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Figure 5-36: Process Data and Metadata Overview in SimaPro

Clicking on the Inputs/Outputs tab displays the flow data in Figure 5-37, which for this
process is now quite familiar.

Figure 5-37: View of Process Flow Data (Inputs and Outputs) in openLCA

If you need to download this data, you can do so by choosing "Export" in the File menu,
but you cannot export it as a Microsoft Excel file.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

151

Section 4 Spreadsheet-based Process Flow Diagram Models

Now that process data has been identified, quantitative process flow diagram-based LCI
models can be built. Amongst the many tools to build such models, Microsoft Excel is one
of the most popular. Excel has many built-in features that are useful for organizing LCI data
and calculating results, and is already familiar to most computer users.
To make these examples easy to follow, we repeat the core example from Chapter 5 (and
shown in Figure 5-5) involving the production of coal-fired electricity via three unit
processes in the US LCI database. The US LCI database is used since it is freely available
and indicative of many other databases (e.g., ELCD). To replicate the structure of the core
model from Chapter 5, we need to manage our process data in support of our process flow
diagram. The following steps illustrate the quantitative structure behind a process-flow
diagram based LCI model.
1) Find all required process data
In the first few sections of the advanced material for this chapter, we showed how to find
the required process data from the US LCI database via several different tools. Using similar
browse and search methods, you can find the LCI data for the other two processes so that
you have found US LCI data for these three core processes:

Electricity, bituminous coal, at power plant

Bituminous coal, at mine

Transport, train, diesel powered

Depending on which tool you used to find the US LCI process data, it may be easy to export
the input and output flows for the functional unit of each process into Excel. If not, you
may need to either copy/paste, or manually enter, the data. Recall that accessing the US LCI
data directly from the LCA Digital Commons can yield Microsoft Excel spreadsheet files.
2) Organize the data into separate worksheets
A single Microsoft Excel spreadsheet file can contain many underlying worksheets, as shown
in the tabs at the bottom of the spreadsheet window. For each of the downloaded or
exported data modules, copy / paste the input/output flows into a separate Microsoft Excel
worksheet. If you downloaded the US LCI process data directly from the lcacommons.gov
website, the input/output flow information is on the "X-Exchange" worksheet of the
downloaded file (the US LCI data in other sources would be formatted in a similar way). The
Transport, train, diesel powered process has 1 input and 9 outputs (including the product output),
as shown in Figure 5-38.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

152

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

Figure 5-38: Display of Extracted flows for Transport, train, diesel powered process from US LCI

3) Create a separate "Model" worksheet in the Microsoft Excel file

This Model worksheet will serve as the primary workspace to keep track of the relevant
flows for the process flow diagram. This sheet uses cell formulas to reference the flows on
the other worksheets that you created from the process LCI datasets.
Beyond just referencing the flows in the other worksheets, the Model worksheet must scale
the functional unit-based results as needed based on the process flow diagram. For example,
in Equation 5-1, results were combined for 1 kWh of electricity from bituminous coal, 0.46
ton-km of train transportation, and from 0.44 kg of coal mining. Since the process LCI data
modules are generally normalized on a basis of a functional unit of 1, we need to multiply
these LCI results by 1, 0.46, or 0.44.
Basic LCI Spreadsheet Example
In this example, a basic cell formula is created on the Model worksheet to add the output
flows of CO2 from the three separate process worksheets. We first make a summary output
result cell for each of the three processes where we multiply the CO2 emissions value from
each worksheet (e.g., the rounded value 0.019 in cell G8 of Figure 5-38) by the functional
unit scale factor listed above. Then we find the sum of CO2 emissions across the three
processes by typing = into an empty cell and then successively clicking on the three scaled
process emissions values.
The Chapter 5 folder has a "Simple and Complex LCI Models from US LCI"
spreadsheet file following the example as shown in the Chapter (which only tracked
emissions of fossil CO2). Figure 5-39 shows an excerpt of the "Simple Model" worksheet in
the file. The same result as shown in the chapter (not rounded off) is visible in cell E8, with
the cell formula =B8+C8+D8.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

153

Figure 5-39: Simple Spreadsheet-Based Process LCI Model

This simple LCI model shows a minimal effort result, such that using a spreadsheet is
perhaps overkill. Tracking only CO2 emissions means that we only have to add three scaled
values, which could be accomplished by hand or on a calculator. However this spreadsheet
motivates the possibility that a slightly more complex spreadsheet could be created that
tracks all flows, not just emissions of CO2.
Complex LCI Spreadsheet Example
Beyond the assumptions made in the simple model above, in LCA we often are concerned
with many (or all) potential flows through our product system. Using the same underlying
worksheets from the simple spreadsheet example, we can track flows of all of the outputs
listed in the various process LCI data modules (or across all potential environmental flows).
This not only allows us a more complete representation of flows, but better prepares us for
next steps such as impact assessment.
In this complex example, we use the same three underlying input/output flow worksheets,
but our Model worksheet more comprehensively organizes and calculates all tracked flows
from within a dataset. Instead of creating cell formulas to sums flows for each output (e.g.,
CO2) by clicking on individual cells in other worksheets, we can use some of Excel's other
built-in functions to pull data from all listed flows of the unit processes into the summary
Model worksheet. An example file is provided, but the remaining text in this section
describes in a bit more detail how to use Excel's SUMPRODUCT function for this task.
The SUMPRODUCT function in Microsoft Excel, named as such because it finds the sum
of a series of multiplied values, is typically used as a built-in way of finding a weighted
average. Each component of the function is multiplied together. For example, instead of the
method shown in the Simple LCI spreadsheet above, we could have copied the CO2

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

154

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

emissions values from the three underlying worksheets into the row of cells B8 through D8,
and then used the function =SUMPRODUCT(B4:D4*B8:D8) to generate the same result.
The "Simple and Complex LCI Models" file has a worksheet "Simple Model (with
SUMPRODUCT)" showing this example in cell E8, yielding the same result as above.
However the SUMPRODUCT function can be more generally useful, because of how Excel
manages TRUE and FALSE values and the fact that the "terms" of SUMPRODUCT are
multiplied together. In Excel, TRUE is represented as 1 and FALSE is represented as 0 (they
are Booleans). So if we have "terms" in the SUMPRODUCT that become 1 or 0, we can use
SUMPRODUCT to only yield results when all expressions are TRUE, else return 0. This is
like achieving the mathematical equivalent of if-then statements on a range of cells.
The magic of this SUMPRODUCT function for our LCI purposes is that if we have a
master list of all possible flows, compartments, and sub-compartments, we can find whether
flow values exist for any or all of them. On the US LCI Digital Commons website, a text file
can be downloaded with all of the nearly 3,000 unique compartment flows present in the US
LCI database. This master list of flows can be pasted into a Model worksheet and then used
to "look up" whether numerical quantities exist for any of them.
A representative cell value in the complex Model worksheet, which has similar cell formulas
in the 3,000 rows of unique flows, looks like this (where cells A9, B9, and C9 are the flow,
compartment, and subcompartment values we are trying to match in the process data):
=E$4*SUMPRODUCT((Electricity_Bitum_Coal_Short!$A$14:$A$65=A
9)*(Electricity_Bitum_Coal_Short!$C$14:$C$65=B9)*(Electrici
ty_Bitum_Coal_Short!$D$14:$D$65=C9)*Electricity_Bitum_Coal_
Short!$G$14:$G$65)
This cell formula multiplies the functional unit scale factor in cell E4 by the
SUMPRODUCT value of:

whether the flow name, compartment, and subcompartment in the unit flows for the
coal-fired electricity process match every item in the master list of flows.

and, if the flow/compartment/subcompartment values match, the inventory value

for the matched flow.

Within the SUMPRODUCT, if the flow/compartment/subcompartment in the unit process

data doesn't match the flow/compartment/subcompartment on the row of the Model
worksheet, the Boolean values are all 0's and the result is 0. If they all match, the Boolean
results are 1, and the final part of the SUMPRODUCT expression (the actual flow quantity)
is returned.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com

Chapter 5: Data Acquisition and Management for Life Cycle Inventory Analysis

155

Figure 5-40: Complex Spreadsheet-Based Process LCI Model

The Chapter 5 folder on the textbook website has spreadsheets with all of the flows and
processes in the US LCI database, as downloaded from the LCA Digital Commons website.
The 'Simple and Complex LCI Models' file has a worksheet 'Complex Model' that
shows how to use the SUMPRODUCT function to track all 3,000 flows present in the
US LCI database (from the flow file above). Of course the results are generally zero for each
flow due to data gaps, but this example model expresses how to broadly track all possible
flows. You should be able to follow how this spreadsheet was made and, if needed, add
additional processes to this spreadsheet model.
Homework Questions for this Section
1. Answer Question 2 from the end of Chapter 5 by using the 'Simple and Complex LCI
Models' spreadsheet introduced in this section.

Life Cycle Assessment: Quantitative Approaches for Decisions That Matter lcatextbook.com