ENVIRONMENTAL ENGINEERING SCIENCE

Volume 00, Number 00, 2018

ª Mary Ann Liebert, Inc.
DOI: 10.1089/ees.2018.0233

Comparative Life Cycle Assessment of Milk and Plant-Based Alternatives

Courtney A. Grant and Andrea L. Hicks*
Department of Civil and Environmental Engineering, University of Wisconsin-Madison, Madison, Wisconsin.

Received: May 21, 2018 Accepted in revised form: August 30, 2018

Abstract
Plant-based dairy alternatives are purchased by consumers for environmental, nutritional, or health-related
reasons. Yet, some plant-based alternatives require large quantities of water and materials to produce. Ad-
ditionally, these products are transported long distances to retail stores, such as almond products produced in
California. A comparative midpoint life cycle assessment (LCA) is presented to evaluate the environmental
Downloaded by UNIV OF BATH from www.liebertpub.com at 10/28/18. For personal use only.

impacts, and in particular embodied energy, and embodied water of conventional dairy milk with two plant-
based alternatives: almond milk and soy milk. The scope of the study considers the life cycle stages of
agricultural production, transportation to the processing plant, milk processing, transportation to the retail store
in Chicago, and retail. A volumetric functional unit (FU), liter of milk, and nutritional FU, kilogram of protein,
were used for comparison. Data were compiled from the LCA literature and U.S. patents, and the TRACI,
BEES+, and Cumulative Energy Demand impact assessment methods were used. Monte Carlo simulation was
used for uncertainty analysis associated with all inputs in the LCA model. It was found that the dominant life
cycle stages for dairy, soy, and almond milk are the production and retail phases from a cradle-to-retail
perspective. Dairy milk had the highest impacts in 6 of the 12 impact categories, whereas soy milk had the
highest impacts in 4 categories using a volumetric FU. Almond milk has the highest impacts in noncarcino-
genics largely from the use of zinc fertilizer and water use from irrigation. Almond milk has the highest impacts
in all impact categories with 1 kg of protein, with results at least five times higher than soy milk, and at least two
times higher than dairy milk in all impact categories. In nutrition, almond milk provides disadvantages due to its
low protein content compared with dairy milk. Yet, protein has only one nutritional benefit, and there are other
positive and negative nutritional benefits of these products to consider. The study illustrates that FU choice has
a strong influence on the food product LCA results and interpretations and suggests that 2 or more FUs are
necessary for food LCA studies. It may be desirable in future work to compare additional nutritional FUs, such
as calories, to inform consumer choice and integrate nutritional and environmental assessments of food
products.

Keywords: almond; dairy; food; life cycle assessment; milk; soy

Introduction prevalent in 25% of the U.S. population and cow’s milk al-
lergy is also prevalent in certain ethnicities (Makinen et al.,
C onsumption of dairy milk is decreasing in the United
States—down from 0.9 cups per person per day in 1970
to 0.6 in 2010, with consumers replacing dairy milk with
2016). They are also chosen by consumers for environmental
reasons and have been thought to have reduced environ-
mental impacts when compared with dairy milk (GoD-
plant-based alternatives (Stewart et al., 2013). At the same
airyFree.org, 2017). However, there has not yet been a
time, the U.S. dairy alternatives market is growing rapidly at
comprehensive scientific study done to compare dairy with
13.2% from 2015 to 2018 (Plant-Based Foods Association,
nondairy alternatives from a life cycle perspective.
2016). Figure 1 shows a breakdown of the 2015 milk market
Among the dairy milk alternatives, almond milk and soy
for the three highest selling milk products (Adams, 2017).
milk are the most popular products (Plant-Based Foods As-
Plant-based alternatives are popular among consumers for
sociation, 2016). Recently, there has been discussion over
health-related reasons. For example, lactose intolerance is
prohibiting the use of the word ‘‘milk’’ for nondairy products
with respect to the differences in nutritional contents (Ellef-
son, 2018). For the purposes of this work ‘‘milk’’ will be
*Corresponding author: Department of Civil and Environmental
Engineering, University of Wisconsin-Madison, 1415 Engineering utilized to refer to a dairy milk or nondairy milk substitute, as
Drive, Madison, WI 53706. Phone: ; Fax: ; E-mail: hicks5@ is common practice in the United States. From a nutritional
wisc.edu standpoint, soy milk contains 80–100 calories and 7 g of

1
 2 GRANT AND HICKS

eggs/nuts) are the exception and have lower production im-

pacts, but higher impacts due to transportation. Weber and
Matthews (2008) suggest that dietary shifts from dairy products
to nuts are the most effective means of lowering a consumer’s
environmental footprint as it achieves more GHG reduction
than buying all locally sourced food.
Life cycle assessment (LCA) has been used to evaluate the
environmental impacts of various food production chains,
such as dairy products (Berlin, 2002; Hospido et al., 2003;
Berlin et al., 2008; Rotz et al., 2010). Many studies have
assessed the environmental impacts of milk production at the
farm stage, but several have also included postfarm gate
stages (Kim et al., 2013; Thoma et al., 2013). A small se-
lection of the recent dairy milk studies is shown in Table 1,
showing a large variability in GHG emissions due to differ-
ences in location and functional unit (FU).
FIG. 1. 2015 milk market (Adams, 2017). Figure 2 shows a comparison of the eutrophication impacts
of several corn production inventories for dairy milk. It is
protein per cup (Krans, 2017). This is similar in protein and important to emphasize the large variability and uncertainty
Downloaded by UNIV OF BATH from www.liebertpub.com at 10/28/18. For personal use only.

about equal to the calories in 1% or 2% dairy milk. Almond in cropping inputs and yields, especially due to location and
milk is growing in popularity for those desiring soy-free al- season, which can affect the overall impact results.
ternatives. It is low in calories, but at the same time, almond Many life cycle assessments have been completed on soy-
milk has a low quantity of protein (*1 g per cup) (USDA, bean products. However, these focus primarily on the pro-
2018). From an environmental perspective, large quantities duction of soybean-derived biodiesel, soybean meal, or
of water are needed to cultivate almonds. Additionally, 80% soybean oil (Kim and Dale, 2005, 2009; Dalgaard et al., 2008;
of almonds are grown in California, relying on long-distance Hu et al., 2008; Huo et al., 2009; Panichelli et al., 2009;
transportation for consumption of almond milk in other U.S. Pradhan et al., 2009). Furthermore, there is one comprehensive
states (Kendall et al., 2015). Weber and Matthews (2008) study on the production of almonds (Kendall et al., 2015).
compared food groups produced and distributed across the Only two comparative LCAs on dairy milk versus nondairy
United States and concluded that greenhouse gas (GHG) milk have been performed and results from these studies are
emissions associated with transport of different food groups also shown in Table 1, showing large ranges in GHG emissions
in the United States, using various transportation modes, are (Birgersson et al., 2009; University of California-Los Angeles
dominated by the production phase despite some food items [UCLA], 2016). Birgersson et al. (2009) compared the envi-
being transported long distances (Weber and Matthews, ronmental impacts of soy milk and dairy milk. Six environ-
2008). However, nonred meat protein sources (chicken/fish/ mental impact categories were evaluated, using a FU of 1 L of

Table 1. Literature Review Summary on Life Cycle Assessment of Milk Products

Milk System
References Location product boundary Functional unit GHG emissions
Rotz et al. (2010) USA Dairy Farm kg of ECM 0.37–0.69 kg/kg of ECM
Thomassen et al. (2008) Netherlands Dairy Farm kg of FPCM 1.4–1.5 kg/kg of FPCM
Wang et al. (2016) China Dairy Farm kg FPCM 1.31–2.08 kg/kg FPCM
Thoma et al. (2013) USA Dairy Grave kg FPCM 1.77–2.4 kg/kg FPCM
McGeough et al. (2012) Canada Dairy Farm kg FPCM 0.92 kg/kg FPCM
Basset-Mens et al. (2009) New Zealand Dairy Farm kg of milk 0.718 kg/kg of milk
Bartl et al. (2011) Peru Dairy Farm kg FPCM 3.18 kg/kg of ECM
Hospido et al. (2003) Spain Dairy Factory L of milk 1.05 kg/L of milk
Eide (2002) Norway Dairy Grave 1000 L of milk 520–600 kg/1000 L of milk
Meneses et al. (2012) Spain Dairy Grave L of milk 1.2–1.35 kg/L of milk
Van der Werf et al. (2009) France Dairy Farm 1000 kg of milk 1082–1037 kg/1000 kg FPCM
Birgersson et al. (2009)a Europe Dairy Grave L of milk 0.9 kg/L of milk
UCLA (2016)a California Dairy Gate L of milk 1.67 kg/L of milk
Clune et al. (2017)b Global Dairy Varies kg of milk 0.94–2.06 kg/kg of milk
University of California-Los California Almond Gate L of milk 0.34 kg/L of milk
Angeles (2016)a
Clune et al. (2017)b Global Almond Varies kg of milk 0.39–0.44 kg/kg of milk
Clune et al. (2017)b Global Soy Varies kg of milk 0.66–1.40 kg/kg of milk
Birgersson et al. (2009)a Europe Soy Grave L of milk 0.215/L of milk
a
Study is a comparative LCA.
b
Study is a literature review.
ECM, energy-corrected milk; FPCM, fat and protein-corrected milk; GHG, greenhouse gas; LCA, life cycle assessment.
 LIFE CYCLE ASSESSMENT OF MILK AND ALTERNATIVES 3

FIG. 2. Eutrophication im-

pacts of corn production us-
ing various data sources.
Downloaded by UNIV OF BATH from www.liebertpub.com at 10/28/18. For personal use only.

drink, including land use, energy use, eutrophication, acidifi- their inclusion in the scope of the study (ISO 14040). Shown by
cation, climate change, and ozone layer depletion. However, the black arrows in Fig. 3, the system boundaries include the
the study was European based, with production of soybeans material and energy flows upstream and downstream of each
occurring in Brazil, and the study has neglected to include the stage. For a consistent comparison across products, agriculture
water footprint. UCLA (2016) compared almond milk with production upstream emissions include the emissions during
dairy milk using economic input–output LCA (EIO-LCA) in manufacturing of material and energy inputs (including all
the categories of water use and GHG emissions. While the nutrients, pesticides, fuels and energy, and water) and trans-
previous studies suggest that soy milk or almond milk may portation fuels. Production of agricultural machinery, capital
have significant environmental benefits over dairy milk using a equipment, and transport vehicles were excluded due to general
volumetric FU, this current study provides the first of its kind LCA practice. Machinery manufacturing consumes 45.9 MJ/kg
cradle-to-gate LCA that compares the environmental impacts and is unlikely to have a major impact on the results of this
of almond milk and soy milk and dairy milk in a U.S. context. analysis (Aguirre-Villegas et al., 2015). Downstream emissions
This is a critical research gap to address to provide a com- include both fossil and biogenic GHG emissions resulting from
prehensive suite of LCA results relevant for consumer decision production. Biogenic emissions are the result of crop produc-
making in a U.S. context, enabling guidance on making en- tion (for all products), which does not account for soil carbon
vironmentally responsible choices in food consumption. dioxide (CO2) sequestration, and animal digestion (specific to
dairy milk) (Aguirre-Villegas et al., 2015). In the milk pro-
Materials and Methods cessing stage, water, sugar, electricity, and energy/fuel inputs
are tracked (however, cleaners and detergents used are assumed
Goal, scope, and system boundary to be negligible) and emissions from energy combustion are
Several methodologies exist to perform LCA studies, included.
including the ISO 14040 (ISO, 1997) guidelines for process- Consumer use (i.e., use of refrigeration at household and any
based LCA and EIO-LCA. Following the ISO 14040 stan- milk losses) and disposal of milk packaging is considered to be
dards, the first goal is to quantify the environmental impacts, outside of the system boundaries. The scope is limited to as-
embodied energy, and embodied water of dairy milk as sume that the lifetime in the household is similar for each milk
compared with two plant-based alternatives, almond milk and type, and that everything is consumed. There have been nu-
soy milk, from raw material production through transporta- merous LCA studies that have focused exclusively on pack-
tion to a retail store in Chicago, making this study a partial aging systems for beverages, including milk (Falkenstein et al.,
cradle-to-gate LCA. Chicago was chosen as a major metro- 2010; Williams et al., 2011; Meneses et al., 2012). A primary
politan area of the United States. The second is to utilize both focus of these studies is exploring new packaging designs to
a volumetric and nutritional FU to compare environmental extend the shelf life of milk while reducing food waste gen-
impacts, discussed further in the Functional Unit section. The erated at the household level. The goal of this study was more
scope of the study considers the life cycle stages of agricul- holistic, and therefore, did not address specifics of packaging
tural production, transportation to a processing plant, milk and waste disposal. However, packaging was included as a
processing, transportation to retail store, and refrigeration at a material input in the processing stage.
retail store, as illustrated in Fig. 1. More detailed life cycle
diagrams of the various life cycle stages can be found in the
Functional unit
Supplementary Data (Supplementary Figs. S1–S3).
According to ISO 14040, comparative LCA studies shall Various FUs have been used in the literature to compare
perform an analysis of material and energy flows to determine the environmental impacts of food products and systems.