You are on page 1of 14

Building and Environment 240 (2023) 110432

Contents lists available at ScienceDirect

Building and Environment


journal homepage: www.elsevier.com/locate/buildenv

Modular construction’s capacity to reduce embodied carbon emissions in


California’s housing sector
Fiona Greer a, b, *, Arpad Horvath a
a
Department of Civil and Environmental Engineering, University of California, Berkeley, USA
b
Center for the Built Environment, University of California, Berkeley, USA

A B S T R A C T

Demand for new housing construction must be balanced with minimizing embodied greenhouse gas (GHG) emissions from building materials, logistical operations,
and construction activities. This study examines the potential embodied GHG benefits that an innovative housing production method, factory-built modular con­
struction, might have in meeting the multifamily housing needs, with California as a case study. A model is developed to quantify the embodied GHG emissions from
materials, transportation, and construction for representative modular housing projects, contrasting results to scenarios where housing is constructed with con­
ventional methods. Results are scaled from single building prototypes to meeting housing demand for lower income residents in California, representing 1.1 million
housing units in all 58 of the state’s counties. Statewide, compared to all housing units being stick-built, most modular types achieve emission reductions of 2–22%,
with potential benefits depending upon structural framing material and factory location. A parametric sensitivity analysis reveals additional key modeling variables:
module size and capacity of the module delivery vehicle. A Monte Carlo analysis is conducted for each county, comparing a random allocation of modular types from
all factory locations to conventionally constructed housing. Counties farthest away from module factories experience GHG emission gains, while all other counties
experience emission benefits of 1–14% (removing one of the modular types increases benefits to 4–20% for all counties). While the results are California-specific, the
modeling framework is applicable to any location and relevant for regions assessing the embodied carbon impacts of rapid housing construction methods.

1. Introduction in each of the state’s counties by 2030 [8]. Any solution aimed at
addressing the housing shortages that California faces must also account
The building sector, including construction and operation, accounts for greenhouse gas (GHG) emissions. Buildings account for a quarter of
for almost 40% of global greenhouse gas (GHG) emissions [1]. Buildings the state’s GHG emissions [9], but that value is an undercount as it does
have both embodied and operational GHG emissions. Operational not include the embodied emissions from materials and construction
emissions come from energy sources used in operating the building. activities.
Embodied emissions come from materials, construction, transportation, This study aims to determine if an innovative housing production
maintenance, and end-of-life activities. As buildings are beginning to method, factory-built modules, might yield potential GHG benefits while
emit fewer operational GHG emissions due to decarbonized energy also meeting the state’s housing needs. The fundamental difference
supplies and energy efficient designs, embodied emissions become a between modular and conventionally constructed buildings is in the
much more significant portion of the building sector’s carbon footprint manner of construction. In the life-cycle phases in a building project,
[2–5]. Since upfront embodied GHG emissions are released at the outset which are outlined in Fig. 1 in the Supporting Information (S.I.), raw
of the building’s construction, coupled with the enormous near-term materials for the building are extracted and processed (A1-A3), building
demand for new and renovated building construction, stakeholders materials are transported from production facilities to site and con­
must seriously consider solutions that minimize the embodied GHG structed into a building (A4-A5), energy and materials are used to
emissions of buildings [1]. operate and maintain (B1–B7), and decommission (C1–C6) the building
California, the fifth largest economy in the world and home to 39 once it has reached its end of life (EOL). Some of the building materials
million people [6,7], faces two crises simultaneously: the immediate and can enter back into circulation after the building’s service life (D). The
near-threats posed by climate change and the extreme shortage of resi­ difference in life-cycle steps between modular and conventional site-
dential housing. According to recent estimates, more than 2.5 million built construction is outlined in Fig. 1. With traditional construction,
housing units should be planned to be built to meet the housing demand building materials are delivered from a material production or storage

* Corresponding author. 410 McLaughlin Hall, Berkeley, CA, 94720, USA.


E-mail addresses: fionagreer@berkeley.edu (F. Greer), horvath@ce.berkeley.edu (A. Horvath).

https://doi.org/10.1016/j.buildenv.2023.110432
Received 30 March 2023; Received in revised form 10 May 2023; Accepted 17 May 2023
Available online 24 May 2023
0360-1323/© 2023 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
F. Greer and A. Horvath Building and Environment 240 (2023) 110432

study building type, material selection, service life, geographic bound­


ary, emission source, and amount of prefabricated material. On average,
the case studies (largely of residential buildings) demonstrated a 15.6%
reduction in embodied carbon and a 3.2% reduction in operational
carbon relative to projects built exclusively with on-site construction
methods. The average reductions found across the case studies are not
definitive as the literature is inconsistent in its methodological choices
[13] and the number of case studies is too small a sample size. Claims
about modular buildings’ lower embodied carbon emissions are influ­
enced by the purported future reusability of wood and steel material
elements within the modular buildings. A complete assessment of a
modular building would also examine any trade-offs between embodied
and operational carbon emissions as a result of design decisions [13].
Fig. 1. Differences in life-cycle phases between prefabricated (modular) Table 1 outlines key information about the identified modular con­
buildings and traditionally constructed buildings. struction LCA studies. Note that the studies with the “3D” designation
focus entirely on self-contained volumetric modules to offer a compar­
facility directly to the site where the building is constructed from the ison with this research’s scope. A brief literature review on panelized
ground up. In modular construction, which is a subset of prefabricated and prefabricated elements, assemblies, and components is included in
construction, self-contained building units are fabricated at an off-site the S.I. An overview of the studies is organized by life-cycle stages.
facility and then installed at the construction site. On-site installation
practices for prefabricated construction depend upon the level of pre­ 1.1.1.1. Product stage. Most of the studies listed in Table 1 include
fabrication of building components, which range from elements (e.g., embodied emissions from raw material extraction and processing. The
precast concrete columns) and assemblies (e.g., panelized wall systems) framing materials include steel, light-framed steel, cross-laminated
to entire modules with exterior/interior walls, roofs, floors, and me­ timber (CLT), dimensional lumber, and concrete. Emissions for the
chanical, electrical, and plumbing (MEP) systems. The most extensive product stage are typically calculated by relating the volumes of mate­
level of prefabrication is modular construction. Adopting the terminol­ rials from the case study building to material-specific emission factors.
ogy defined in Gibb 2001 [10], modular construction refers to the The bill of materials (BOM) for a case study are derived from cross-
fabrication, delivery, and installation of “pre-assembled volumetric sections of exterior/interior walls, floors, and roofs or directly from
units” at a site. The structural framing, roofing, exterior facades, interior 3D models of the building. Emission factors for building materials reflect
partitions, finishes, and fixtures are all assembled into self-contained the impacts associated with harvesting/mining raw materials, trans­
volumetric modules in a factory setting and shipped to site to be inter­ porting the raw material to a processing site, and processing the raw
locked with other modules. materials to its finished product (e.g., mixing cement, aggregate, and
This study examines the potential embodied GHG (carbon) benefits water at a concrete batch plant). Most of the studies used generic, pre-
that an innovative housing production method, factory-built modules, established material-specific emission factors. Only one study utilized
might have in meeting the state’s housing needs. While the results and environmental product declarations (EPDs) to calculate emissions from
discussion are California-specific, the methods and modeling framework materials. EPDs are standardized reports of the life-cycle impacts for a
are applicable to any location. A GHG comparison of different housing specific product or material, oftentimes commissioned by the material’s
types built with factory-made modules relative to conventional, on-site manufacturer. This means that EPDs can reflect local conditions at the
construction methods is presented. Emissions are scaled from proto­ city, state, region, or nation-wide level. Recent trends in industry
typical residential building case studies to the future stock of multi- practice point towards the use of EPDs to evaluate the environmental
family housing in California to gain a comprehensive understanding of impacts from building materials.
potential savings from modular construction. As with LCA studies of conventionally constructed buildings, the
contribution of embodied GHG emissions relative to the entire modular
building’s life-cycle impacts varies by factors such as building type and
1.1. Literature review on modular housing location. A study examined differences in cradle-to-gate GHG emissions
(A1-A4) for two types of modular single-family buildings compared to
1.1.1. Life-cycle assessment studies on modular housing an on-site constructed residential building, finding that raw materials
Prior studies have asserted that modular construction can yield (A1-A2) comprised between 48% and 56% of embodied GHG emissions
environmental benefits relative to conventional construction methods, for the modular buildings and 53% for the on-site building [16].
but those potential benefits need to be quantitatively assessed [11].
Modular projects should be assessed using a range of economic, envi­ 1.1.1.2. Construction process stage. Based upon the studies included in
ronmental, and social criteria so that stakeholders (e.g., contractors, Table 1, most include GHG emissions from transportation and con­
developers, regulatory agencies) can select the optimal construction struction activities in their scope. There is variety in the extent to which
method for a project [12]. Construction stakeholders ranked potentially A4 and A5 emissions are analyzed. Some provide relatively detailed
relevant environmental indices for a sustainable modular building descriptions of off-site construction activities and on-site installation
assessment framework: operational energy efficiency ranked highest, activities [14,17,28], which is important for realistically capturing the
suggesting interest about differences between modular and traditional GHG impacts from the construction process stage. However, some of the
building envelopes [12]. Embodied energy and GHG emissions ranked in studies use simplifying assumptions that might yield unrealistic results.
the middle among the twelve environmental indices, suggesting that Examples include modeling off-site construction impacts as: (1) a per­
stakeholders might not view those factors as a critical difference be­ centage of primary energy used in produced building materials [21]; (2)
tween modular and traditional buildings. the same, in terms of energy consumption, as on-site construction [15];
One study conducted a literature review of 27 case studies that uti­ (3) an assumed emission factor based on a prior, unrelated study [20];
lized life-cycle assessment (LCA) and found a wide range of embodied and (4) a ratio of on-site energy consumption [23]. Recent industry re­
(105–864 kg CO2 per m2) and operational (11–76 kg CO2 per m2) ports have emphasized that the potential waste savings with modular
emissions from prefabricated buildings [13]. The wide range is in part construction could be on the order of 20–90%, but these values are based
attributable to variability in study assumptions and uncertainties in case

2
F. Greer and A. Horvath Building and Environment 240 (2023) 110432

Table 1
Overview of modular LCA studies by modular material, building typology, region, and life-cycle stage. Region codes: US = United States; CA = Canada; UK = United
Kingdom; AU = Australia; SE = Sweden; TH = Thailand; PK = Pakistan; KR = Republic of Korea; EU = European Union; ZA = South Africa; IN = India.
Source Modular Traditional Building Use Building Year Region Embodied Operational A1, A3 A4, B1 B6 C1–C4
Material Material Type* GHG [kg GHG [kg CO2/ A2 A5 –
CO2/m2] m2/year] B5

[14] Wood Brick Residential/ SF 2010 US 27.7 N/A X X


Low Rise
[15] Wood Wood Residential/ SF 2008 US 91.6 69.2 X X X X
Low Rise
[16] Wood Wood Residential/ SF 2019 CA 92.6–105.3 N/A X X X
Low Rise
[17] Wood Wood Residential/ MF, 3D 2009 CA 104.7 N/A X
Low Rise
[18] Wood Brick Residential/ SF 2011 UK 405–535 N/A X X X
Low Rise
[19] Wood Wood Residential/ SF 2012 US 73.2 N/A X X X
Low Rise
[20] Wood, Steel Concrete Residential/ MF, 3D 2011 AU 578–865 43.4–46.2 X X X X
Mid Rise
[21] Wood Wood Residential/ MF, 3D 2014 SE 139–149 6.3–8.4 X X X X X
Mid Rise
[22] Wood, Steel Brick Residential/ SF 2017 AU 302 N/A X X X
Low Rise
[23] Wood, Steel N/E Residential/ MF, 3D 2016 TH 382–9110 − 6.5 – 30.9 X X X X X
Low Rise
[24] Wood, Steel Brick Residential/ SF, 3D 2021 PK 332 N/A X X X
Low Rise
[25] Concrete, N/A Residential/ MF, 3D 2022 KR 280 N/A X
Steel Mid Rise
[26] Steel, N/A Residential/ SF, 3D 2018 EU 454–647 N/A X X X
Concrete, Low Rise
Wood
[27] Steel N/A Residential/ SF, 3D 2022 EU 560–672 17.2–57.8 X X X X
Low Rise
[28] Cross- N/A Residential/ MF, 3D 2022 SE 263 35.6 X X X X X X
laminated Low Rise
timber (CLT)
[29] Steel N/A Residential/ SF 2019 CA 193–298 105–301 X X X X X
Shipping Low Rise
Container
(SCC)
[30] SCC N/A Residential/ SF 2016 AU 185 9.8 X X X X X
Low Rise
[31] SCC Steel, CLT Residential/ SF 2020 CA, ZA, 45–55 3.8 X X X B4 X
Low Rise IN
[32] SCC N/A Residential/ SF 2020 CN 660–1510 8.7–284 X X X X X X
Low Rise

on a limited number of case studies [33]. sentiment is expressed in another study that used building energy
Transportation emissions are most often modeled as a function of simulation to demonstrate that prefabricated buildings are likely to be
distance between factory and housing site. With some exceptions where more energy efficient than traditionally constructed buildings [35]. As
actual transportation distances between the installation site and factory shown in Table 1, some prior research efforts have examined the
are included [14,28], transportation distances from factories are cradle-to-grave GHG impacts between prefabricated/modular and
assumed to be within city limits [17] or some generic distance [26]. Of traditional construction. Most modular studies that have included the
importance, it is unclear whether any study accounts for return trips for use phase indicate that there are marginal differences, if that, between
delivery vehicles (i.e., the empty flatbed returning to the factory site). factory-built and traditional construction in impacting building opera­
The overall impact of module transportation is context specific, with tional energy [15,21,28]. In those studies, the marginal differences in
some finding that it is one of the more emissions-intensive sources energy use between modular and conventional construction are not due
within the system boundary compared to traditional construction [24], to inherent differences between the two construction methods, but due
while one of the studies found that while material transport was not a to design choices such as material type and building dimensions. More
significant source, worker transportation trips were significant [19]. research efforts should be dedicated to investigating potential differ­
ences between GHG emissions from the use phases of each type of
1.1.1.3. Use stage. There are ongoing efforts, particularly at the U.S. building.
federal level, to leverage the potential to more efficiently achieve energy
efficiency and building decarbonization benefits through the use of 1.1.1.4. End-of-life stage. The EOL implications on GHG emissions are
industrialized construction [34]. Guidelines developed by the National not typically included in most LCA studies on modular housing. In some
Renewable Energy Laboratory suggest that factory-built modules can studies, GHG impacts from the EOL stage are excluded from the system
achieve improved thermal performances relative to on-site construction boundary due to its purported insignificance [27]. EOL is seen by some
by, for example, installing insulation in a quality-controlled environ­ as the key stage that could make prefabricated housing preferable to
ment. Presumably, improved thermal performance would yield lower traditional construction [36]. Modular housing is viewed by some as
GHG emissions from the use phase in factory-built housing. This being easier to disassemble from a GHG perspective, because modular is

3
F. Greer and A. Horvath Building and Environment 240 (2023) 110432

seen as being easier for disassembly [20,23,27] and apparent reuse of uncertainties in the use and EOL stages. The functional unit for this study
materials [21,28], including concrete, steel, and timber, and waste is embodied GHG emissions per m2 of building area.
savings [18]. California is a relevant case study for analyzing embodied carbon
emissions from wide-scale adoption of prefabricated multifamily,
1.1.2. Scaling approaches multistory housing. The state’s population centers (i.e., the San Fran­
Much of the prefabricated and modular housing literature compares cisco Bay Area and the greater Los Angeles area), where there is most of
single prefabricated or modular buildings to single conventionally built the demand for housing, are largely coupled with existing modular
examples. Assessments that can provide a realistic estimation of the factory locations. However, there are many, largely rural, areas of the
differences in GHG emissions at scale between generalized modular state where housing needs to be constructed, but there are no closely
buildings and conventionally constructed buildings by material type, located factories. The distinction between housing need and proximity
design, and other model parameters provide more context for stake­ to factory locations is relevant for the embodied-carbon context and
holders. One study measures wide adoption of prefabricated construc­ would need to be considered if California were to implement a statewide
tion in all European Union countries using a stock-based approach [37]. policy promoting factory-built housing. California also has a
The authors compare the embodied, operational, and EOL GHG emis­ geographically diverse and changing energy supply. While all retail
sions for light-gauge steel-framed and timber-framed modular residen­ electricity sold in the state must achieve zero carbon emissions by 2045
tial buildings to on-site reinforced concrete residential buildings. Lisbon, [38], each utility has diverse generation assets with varying carbon in­
Berlin, and Stockholm are used as representative climates for all EU tensities [39]. Therefore, emission impacts of each modular factory’s
countries. The stock-based approach accounts for expected rates of new operations, and resulting impacts on the housing projects, will similarly
construction of both residential and office buildings as well as expected vary throughout the state. Finally, California has a rapidly changing
rates of demolition from 2020 through 2050. Five typologies (low, building materials policy landscape; recent legislation requires that
medium, and high-rise residential and medium and high-rise office) for carbon intensity of building materials be reduced [40]. As a result of
steel, wood, and concrete-framed budlings and three different levels of new and future mandates, the proliferation of California-specific EPDs
insulation are explored. The results indicate that by 2050, prefabricated has expanded, further allowing for detailed and representative exami­
buildings can reduce GHGs by 6% (steel-framed) to 4% (timber-framed) nations of the impacts of building projects.
compared to reinforced concrete buildings. The authors stress that Two types of embodied GHG emission inventories are developed for
prefabricated buildings cannot alone help meet the EU’s emission six example building types listed in Table 2. The first emissions in­
reduction goals. The EU study presents a first effort at assessing the ventory is at the single-building scale and focuses only on comparing
implementation of prefabricated buildings at scale, considering spatial emissions from materials used in the modular and conventional housing
and temporal factors. examples. Results from the first emissions inventory are incorporated
The remainder of the literature review, which covers differences in into the second emissions inventory. The second emissions inventory is
modular framing materials and prefabricated assemblies, is presented in calculated for various levels of housing that needs to be built in Cali­
the S.I. document, which also includes some of the calculations and all fornia as per the number of housing units designated in Regional
the data incorporated into this study. Housing Needs Allocation (RHNA) reports for each county. The county
aggregation is selected because California, which is the most populated
1.2. Research objectives state in the United States [41], has 58 counties with diverse populations
(and housing needs), ranging from almost 10 million residents in Los
We assess various case-study multifamily buildings in California that Angeles County to just over 1200 residents in Alpine County [42]. This
are modularly constructed and contrast their embodied GHG emissions second emissions inventory includes emissions from building materials,
to conventionally constructed steel and timber-framed housing. Unlike transportation, and construction activities for every county in the state.
most studies, which focus on single building-to-building comparisons, Fig. 2 shows an overview of the methodological approach for assessing
this study develops a novel approach to assess embodied carbon impacts the embodied GHG emissions from modular housing in California. Key
from implementing modular multifamily housing at scale. The objec­ details on data sources and calculations are presented in the following
tives of this research are to: (1) quantify and compare the embodied subsections. Additionally, a sensitivity analysis is conducted on key
GHG emissions of different housing types built with factory-made modeling variables (e.g., module size, carbon intensity of factory energy
modules relative to conventional, on-site construction methods; (2) supply) in the second emissions inventory to explore which conditions
identify the suitable combination of modular types by housing need and make modular construction yield GHG emission benefits in California.
factory location in California; and (3) identify the relevant parameters The building types in Table 2 are selected for analysis in California
that influence emission benefits. for two key reasons. In California, there are fewer concrete-framed
residential structures because of seismic codes [43]. Steel and
2. Methods timber-framed are more common options; the steel shipping container is
a unique option being implemented in current projects throughout the
This study develops and applies a model to assess the potential GHG state [44,45]. Timber-framed options are further identified in the
benefits from implementing factory-built, modular housing in Califor­ “California Wildfire and Forest Resilience Action Plan” as a suitable
nia. The specific scope of life-cycle impacts investigated is narrowed to
stages A1 through A5 (Fig. 1 in the S.I.). The scope is limited to assessing
upfront embodied GHG emissions for two key reasons. Given the need to Table 2
reduce the amount of GHGs entering the atmosphere now, it is imper­ Descriptions for example buildings used in study.
ative to focus on the near-term emissions associated with constructing Type Frame Story Approximate Gross BOM Data
buildings. Additionally, there is relatively more uncertainty associated Area (m2)
with assessing emission differences between modular and conventional Conventional Timber 3 3700 Representative
buildings for their Use (B6) and EOL (C1–C4) stages compared to their Timber 2 2800 Representative
Product (A1-A3) and Construction Process (A4-A5) stages. It is impor­ Steel 5 4500 Representative
tant to understand and make decisions based upon a complete under­ Modular Steel Shipping 2 700 Actual
standing of the GHG emissions from the whole life cycle of a modularly Container
or a conventionally constructed building. However, it is beyond the Timber 3 3700 Representative
Steel 4–6 37,000 Actual
scope of this research study to comprehensively account for the

4
F. Greer and A. Horvath Building and Environment 240 (2023) 110432

Fig. 2. Methodological approach for assessing the embodied carbon (EC) emissions for modular housing in California.

option for storing carbon [46] and will likely represent a large share of excluded due to lack of availability and specificity of data. There are
new construction in the state. many ways to model material delivery and module transport to reflect
industry practice more accurately. Whether it is traditional on-site
2.1. Calculation steps construction or factory-based, raw materials are either transported
from the supplier to a staging area, or to a wholesale warehouse, or to
The steps for calculating emissions inventories for modular and the project site itself. Given the complexities of these supply chain lo­
conventional housing are very similar. The key differences lie in how gistics and the lack of available data from actual projects, it is sufficient
GHG emissions from transportation and construction activities are to simplify the transportation decisions to just include transportation of
calculated. Table 3 shows the scope of sources included in this study. modules from factories to project sites. Upstream emissions from the fuel
There are sources that could be included in the study, but are used in construction equipment and delivery vehicles (diesel) are also
excluded from the project scope due to lack of specificity about the types
and duration of construction equipment used in on-site and factory-built
Table 3
operations. The GHG emissions from changes in on-site worker trans­
Scope of emission sources for each housing type. Note that it is assumed that
upstream emissions for each housing type are equal. portation are also excluded. As results are presented at scale for the
entire state, any changes in worker transportation impacts would
Housing Type Bill of Materials Transportation Construction
require making assumptions on: (1) the types of vehicles that workers
Factory-Built/ • Standard Upstream + Delivery • Factory drive (e.g., model, age); (2) the distances from where workers are
Modular foundations of modules to manufacturing
traveling to factories versus where they are driving to job sites; and (3)
• Floor housing site • On-site assembly
Construction of modules the total number of trips taken, which would be dependent on how long
• Ceiling workers would be needed on a project.
Finishes The scaling analysis is based upon the number of RHNA units that
• Exterior need to be constructed in every city, town, and unincorporated area for
Windows
all 58 counties within California. The scaling scenario assumes that all
• Wall Finishes
• Floor Finishes unit numbers designated as for “Very Low Income” (<50% of area me­
• Interior Walls/ dian income) and “Low Income” (50–80% of area median income) are
Partitions assumed to be multifamily housing construction. This is around 40% of
• Exterior Walls
all housing units that need to be constructed in the state.
• Roof
Construction
• Roof 2.1.1. Building material-related GHG emissions
Coverings Emissions associated with building materials are calculated by
Conventional • Standard Upstream • On-site determining the masses of specific materials within each building type
foundations construction (i.e., the bill of materials, BOM) and then relating those amounts to
• Floor material-specific EPDs. The prototypical building datasets and EPDs
Construction
used in this study are provided in Table 2 through 6 in the S.I. Quantity
• Ceiling
Finishes takeoffs for materials in each building type are estimated using building
• Exterior information modeling, design codes, 2D construction drawings and
Windows specifications from actual projects, and the RSMeans database.
• Wall Finishes RSMeans, a U.S.-based subscription database for building materials and
• Floor Finishes
construction cost data, also provides BOM data on prototypical designs
• Interior Walls/
Partitions for various residential, commercial, and industrial buildings that vary by
• Exterior Walls structural frame, façade and insultation type, interior wall systems, and
• Roof finishes [47]. Equation (1) is used to calculate the A1 through A3
Construction
emissions:
• Roof
Coverings

5
F. Greer and A. Horvath Building and Environment 240 (2023) 110432


n • The total number of modules that need to be transported is calculated
EM = (Mi × EPDi ) (1) by dividing the total needed building area for a county by the
i=1
average module floor area.

⁃ Where EM are the GHG emissions associated with raw material sup­ Transportation GHG emissions are calculated according to Equation
ply, transportation of raw material to manufacturing facilities, and (3):
manufacturing of raw materials into building components.
j ( )
⁃ Mi is the mass of material i for every relevant material in the building ∑ NM,j
ET,Scaled = × 2D × EFO (3)
up to material n. i=1
CT
⁃ EPDi is the environmental product declaration for material i in the
building.
⁃ Where ET, scaled are the transportation GHG emissions scaled to the
entire state.
A building-specific emission factor is found by dividing the total GHG
⁃ NM is the total number of modules for a given county j
emissions from materials by each building’s total gross area. Material
⁃ CT is the capacity of the delivery vehicle
GHG emissions are then scaled to each county in the state by multiplying
⁃ D is the distance from the module factory to the housing site (2D
the total building area that needs to be constructed based upon RHNA
reflects that an empty flatbed truck will return to the factory site)
determinations (Equation (2)).
⁃ EFO is the direct GHG emission factor for the delivery vehicle
j (
∑ )
EM
EM,scaled = × NU,j × AMF (2) A statewide GHG emission factor for a Class 8 delivery vehicle from
i=1
AB
the California Air Resources Board’s EMission FACtor (EMFAC) model is
used [49]. Note that emissions from producing and distributing the fuel
⁃ Where EM, scaled are the material GHG emissions scaled to the entire used for the delivery flatbed truck are not considered in this study.
state.
⁃ EM are the material GHG emissions for a building example 2.1.3. Construction-related GHG emissions
⁃ AB is the gross area for a specific building Construction GHG emissions are estimated for both on-site and
⁃ NU,j are the scenario-specific number of housing units for each factory-made housing. A detailed construction analysis for the on-site
county j and factory-built methods would entail determining the type, dura­
⁃ AMF is the average area for a multifamily housing unit. tion, and emissions intensity of equipment used. Given the lack of in­
formation on construction equipment types and durations for both on-
The total building area for each county is then calculated by multi­ site and factory-built operations and the high-level scaled approach to
plying the total number projected housing units, which are found in the emissions impact, a detailed accounting of emissions by specific con­
sixth installment of each county’s RHNA report [8], by the average gross struction equipment is beyond the scope of this study. Instead, an
area for a multifamily housing unit. The average area for a multifamily average emission factor for on-site construction of buildings (100 kg CO2
unit in the Western United States from the 2015 U S. EIA Residential eq. per m2) that is derived from a review of building LCA literature is
Energy Consumption Survey [48] is used as an approximation of the used [50]. A later sensitivity analysis explores how on-site construction
average floor area for a multifamily housing unit (828 ft2 or 77 m2 per emissions impact overall results. On-site construction GHG emissions
unit). are calculated according to Equation (4):

2.1.2. Transportation-related GHG emissions


j
∑ ( )
EOS,Scaled = NU,j × AMF × EFOS (4)
For comparative purposes, it is assumed that the distances between i=1
raw material suppliers and either the construction site or the module
factory are the same. Therefore, the only transportation emissions
⁃ Where EOS, scaled are the on-site construction GHG emissions scaled to
considered for this study are those associated with delivering modules
the entire state.
from the factory to the site location. It can be difficult to accurately
⁃ NU,j are the scenario-specific number of housing units for each
model the potentially complicated supply chain relationship differences
county j
between conventionally built and factory-built housing. Material de­
⁃ AMF is the average floor area for a multifamily housing unit
liveries to on-site and factory locations are project specific and depen­
⁃ EFOS is the on-site construction emission factor
dent upon total material quantities needed and project schedule; there is
insufficient data to accurately assess the transportation impacts for all
A statewide GHG emission factor for a crane is calculated using the
housing units for each county.
offroad mobile source exhaust emissions inventory in EMFAC.
Eight existing factory locations (Table 8 in the S.I.) are analyzed to
The construction emissions associated with modular housing in­
calculate the travel distances between building sites and module fac­
cludes both factory emissions and on-site assembly emissions. Factory
tories. The county seat for each of the state’s 58 counties is used as the
GHG emissions are calculated by using electricity consumption data
proxy location in Google Maps to measure the vehicle-travel distance
from an actual modular factory in Los Angeles County. The average
between a factory location and a housing site (all traveling distances
electricity consumption per module is found by dividing the total
from each factory location to each county seat are provided in the S.I.).
amount of electricity consumed in the factory by the total number of
Additional assumptions include:
modules completed within the same time span. The GHG emissions in­
tensity for each of the three factory’s electricity supplies is calculated by
• Modules are assumed to be transported from the factory to the
multiplying the electricity profile mix for each factory’s utility by
housing site on a flatbed truck.
source-specific, life-cycle emission factors [51–53]. See Table 7 in the S.
• The flatbed can accommodate two modules in one delivery that are
I. for a description of each factory location’s electricity consumption per
each sized at 160 ft2 (~15 m2). The module size is determined from
module, electricity utility, and GHG electricity emission factor. Note
construction drawings from an existing modular housing project.
that the emission factors for PG&E and SCE are averages for the last six
One housing unit can comprise multiple modules.
years’ worth (2016–2021) of electricity source profiles to account for
year-to-year changes in electricity generation mixes. The last four years’

6
F. Greer and A. Horvath Building and Environment 240 (2023) 110432

worth (2018–2021) of electricity sources profiles for IID are averaged as analyses have been conducted are presented in Table 4. The sensitivity
IID still sourced electricity from coal in 2016 and 2017. analysis method employed involves changing the baseline value of one
On-site assembly-related emissions for modular housing are calcu­ modeling variable at a time in 5% increments from − 95% of the baseline
lated with the assumption that one crane could lift one module into value to +100% of the baseline value. Additionally, 25% increments
place every 30 min. Total modular construction emissions are estimated from +100% to +250% of the baseline value are modeled to account for
using Equation (5): any extreme changes in results. The sensitivity analysis is applied to
j
results for housing in Los Angeles County, which is the most populous
∑ [( ) ( )]
EFB,scaled = ME × EFElec,k + TM × NM,j × EFC (5) county in California and has the highest housing need [8,42]. The results
i=1 of this sensitivity analysis are largely representative for all counties,
although there could be different conclusions for each county (e.g.,
northern counties might see a greater difference in changing certain
⁃ Where EFB, scaled are the factory-built construction GHG emissions
modeling variables).
scaled to the entire state.
A Monte Carlo simulation is then conducted to explore how the
⁃ ME is the electricity consumption needed to make each module
distribution of different modular and conventional housing types in­
⁃ EFElec, k is the electricity emission factor for factory k
fluences overall GHG emissions. A Dirichlet distribution [54] is applied
⁃ TM is the time the crane needs to lift one module into place
to simulate 15,000 randomized iterations of what type of modular
⁃ NM, j is the total number of modules for a given county j
housing would be built and where the modular housing would come
⁃ EFC is the direct emission factor for the crane, accounting for emis­
from for each of the state’s 58 counties. This process is also repeated to
sions from fuel combustion while the crane is in operation.
explore what effect a randomized distribution of different conventional
housing types has on overall GHG emissions benefits of modular hous­
2.1.4. Comparison between conventional and modular housing
ing. Equation (7) is the general equation for estimating the impacts from
The differences in GHG emissions between conventional and
randomized combinations of modular and conventional housing:
modular housing are used to determine whether modular housing ulti­
mately yields GHG benefits. Equation (6) is used to calculate the per­ EC,H = a1 × EH1 + a2 × EH2 + … + an × EHn (7)
centage difference in cumulative GHG emissions between each housing
type:
( ) ⁃ Where EC, H is the GHG emissions from some randomized combina­
EFB − ECB tion of housing types in each county
%ΔCB− FB = × 100% (6)
ECB ⁃ a1, a2, …, an are the percentage amounts of each housing type in each
county which must sum to 100%
⁃ EH1, EH2, …, EHN are the GHG emissions from each housing type in
⁃ Where %ΔCB-FB is the percentage change in GHG emissions of
each county
modular housing relative to conventional housing.
⁃ EFB is the GHG emissions for the factory-built (modular) housing
2.3. Limitations
units
⁃ ECB is the GHG emissions for the conventional housing units
Results are presented for the BOM, transportation, and construction
contributions of GHG emissions for six different types of buildings (three
2.2. Sensitivity analysis steps modular, three conventional) that represent potential building types
that might be constructed in California. Ideally, exact quantities of
Whether modular housing yields GHG benefits relative to on-site materials, delivery distances, and construction activities would be
construction depends on the values of several key modeling variables. known and used in the LCA model. Additionally, it should be noted that
When available, data from county documents, currently operating this analysis does not consider land-use or performance requirements for
modular factories, existing modular buildings, and electric utilities have specific locations when conducting the scaling-up assessment. For
been utilized. Representative data for on-site housing projects come example, a multifamily development in San Francisco might be 15 to 20
from prototypical BOMs provided by RSMeans, an annually updated stories high and due to earthquake constraints might need a reinforced
construction estimating database that also includes BOM data for concrete structural framing system.
various residential, commercial and industrial building designs [47]. A
sensitivity analysis is conducted to explore how changes in model var­ 3. Results
iables impact overall GHG emissions and to address uncertainty with
values used in the LCA model. The variables for which sensitivity Results are presented as a scaled-up assessment for the entire state of
California. In terms of individual buildings, GHG emissions from the
BOM for each case study building are presented in the S.I. At the
Table 4
statewide scale, GHG emissions from the BOM and transportation and
Description of sensitivity analysis conducted on housing in Los Angeles County.
construction activities are shown for a range of scenarios and conditions.
No. Modeling Variable Baseline Value

1 On-site construction emission 100 kg CO2 eq. per m2 of housing 3.1. Single building examples
factor for conventional
construction
2 Crane lifting time per module 30 min
Building emission factors for a select number of case studies are
3 Module floor area 15 m2 presented in Table 5. The BOM (i.e., A1-A3) emission factors are of the
4 Delivery flatbed truck capacity 2 modules per delivery trip same order of magnitude as some of the modular case studies from the
5 BOM emissions intensity for all Baseline BOM emissions for each literature (see Table 1), which provides confidence in the results.
module types module type
Fig. 2 in the S.I. shows cumulative emissions for two of the example
6 BOM emissions intensity for all Baseline BOM emissions for each
conventionally built types conventionally built type buildings (conventionally built, steel-framed versus modular timber-
7 Factory electricity consumption per 2500 kWh per module framed) by building component. The two example buildings shown in
module Fig. 2 in the S.I. are of slightly different size and configuration. Weights
8 Factory electricity emission factor Baseline electricity emission factor for by material type for the two example buildings are shown in Table 9 in
each of the three factory locations
the S.I. For the scaled-up analysis presented in the next section, the per-

7
F. Greer and A. Horvath Building and Environment 240 (2023) 110432

Table 5 Results that examine the GHG impacts when combinations of different
Building emission factors (in metric tons of CO2 eq. per m2) for each example module types from all factory locations are compared to combinations of
building type. conventional buildings are presented later in the Results section.
Type Frame Story Approximate Gross BOM EF [MT At the statewide level and under current modeling assumptions,
Area (m2) CO2 eq. m− 2] steel-framed and timber-framed modules fabricated in every factory
Traditional Timber 3 3700 0.17 location (except for Los Angeles County and Ada County in Idaho) yield
Timber 2 2800 0.15 GHG emission benefits relative to all three conventional examples. The
Steel 5 4500 0.27 results in Fig. 3 in the text and in Fig. 3 and 4 in the S.I. suggest the
Modular Steel Shipping 2 700 0.44 following points:
Container
Timber 3 3700 0.19 • The best-performing conventional option is the two-story timber-
Steel 4–6 37,000 0.16
framed building example.
• The shipping container module is not a beneficial option from an
unit material emission factors from Table 5 are used to assess the GHG embodied GHG emissions perspective.
emissions of the two construction methods in a more meaningful • Both steel and timber-framed modules constructed in Idaho yield
manner. emission reductions relative to conventional steel-framed housing.
However, modules from Idaho are more emissions intensive than
timber-framed conventional housing in California.
3.2. Statewide and county-level assessment of GHG emissions from • Emission reductions for all other factory locations in California are,
conventional and modular housing in general, more modest relative to timber-framed conventional
housing (Fig. 3 in the text, Fig. 4 in the S.I.) than they are for steel-
Results are first presented for all 58 counties in the state, comparing framed conventional housing (Fig. 3 in the S.I.).
three different traditionally constructed buildings to several modular
examples made in existing factory locations. All baseline assumptions Examining results at the county level provides further context about
for modeling variables (e.g., module floor area, crane lifting time, de­ which module types and factory locations will yield emission reductions
livery flatbed truck capacity, factory electricity consumption, on-site relative to on-site construction. Figs. 4 and 5 show emission changes for
construction emissions intensity) are provided in Table 4 in the different module types relative to prototypical, conventional timber-
Methods section. Fig. 3 shows the percentage difference in GHG emis­ framed multi-family units in Humboldt and Alameda Counties, respec­
sions for one of the three conventional examples compared to different tively. In Alameda County, modules made in factories within and near
modular options (the remaining results are in the S.I.). It is assumed that the San Francisco Bay Area yield higher emission reductions compared
all required RHNA units are built as one type of modular building (e.g., to other factory locations. In Humboldt County, where the closest fac­
all are steel-framed modules from Ada County, Idaho or all are timber- tory is 280 miles away, the further away a factory is located, the less
framed modules from a factory in Solano County in California).

Fig. 3. Comparison of changes in statewide embodied carbon (EC) emissions for module combinations (by frame type and factory location) relative to if the number
of multifamily housing units assessed in each county were built with traditional on-site, timber-framed construction. A positive percentage change (indicated in red)
means that the module type’s EC emissions are greater than the traditional example. A negative percentage change means that the module type leads to EC emission
savings. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

8
F. Greer and A. Horvath Building and Environment 240 (2023) 110432

Fig. 4. Comparison of changes in Humboldt County embodied carbon (EC) emissions for module combinations (by frame type and factory location) relative to if the
number of multifamily housing units assessed in each county were built with traditional on-site, timber-framed construction. A positive percentage change (indicated
in red) means that the module type’s EC emissions are greater than the traditional example. A negative percentage change means that the module type leads to EC
emission savings. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

likely for there to be emissions benefits for modular housing. • Delivery flatbed truck capacity is also an indicator for overall GHG
performance. A decrease in delivery flatbed truck capacity yields
3.3. Sensitivity of key modeling variables to overall results GHG emission increases for modules.

Before making any conclusions about the emissions benefits from Fig. 5 through 12 in the S.I. offer more insight into how changing key
modular construction, it is crucial to examine how uncertainty with key modeling parameters impacts GHG emissions using Los Angeles County
modeling variables might change results. Within each county, it is likely for illustrative purposes, assuming 100% utilization of each type of
that modules will come from multiple factory locations. Later results use module.
a randomized method to examine how utilization rates of each modular
and conventional type affect overall GHG emissions. 3.4. County-level GHG rankings
A comparison of all modeling variables for one specific type of
module (timber-framed module made in the San Francisco Bay Area and A randomized allocation of each module type is compared to both
used in Los Angeles County) is shown in Fig. 6. Key take-aways from 100% utilization of timber-framed, conventional housing as well as a
Fig. 6 are that: randomized utilization of all three conventional building types. Using
Los Angeles County first as an illustrative example, on average, after
• Crane lifting time per module and the GHG emission factor for fac­ performing 15,000 randomized iterations, all possible combinations of
tory electricity and factory electricity consumption appear to have modular yield approximately 4% fewer GHG emissions than prototypi­
much influence on overall GHG emissions for this module type cal timber-framed conventional (see Fig. 13 in the S.I.). Removing the
• Changing the emissions intensity for the module’s BOM increases the worst-performing modular type (steel shipping container) means that
overall GHG performance for the module while changing the con­ modular is around 24% better than the timber-framed conventional
ventional example’s BOM decreases overall GHG performance for example.
the module Fig. 7 through Fig. 10 show the rankings of modular GHG perfor­
• There appears to be a sweet spot for module floor area (in LA County mance from worst to best for each county in California.
at least) where a module sized between 60 and 180 ft2 yields emis­
sion reductions. Any module smaller than 60 ft2 and greater than • Fig. 7 shows randomized allocation of all modular types compared to
180 ft2 yields massive emission increases just the conventional timber-framed example. Under this scenario, it
• The on-site emission factor appears to be correlated with whether a appears that except for most Northern California counties, modular
module is indeed better than conventional construction. If a lower achieves modest GHG emission benefits.
on-site construction emission factor is achieved relative to what is • Fig. 8 depicts the same scenario as Fig. 7, except the worst-
modeled under the baseline conditions, on-site appears to perform performing modular type (steel shipping container) has been
better than factory-built from an overall GHG emissions perspective. removed. Modular appears to perform better than conventional
statewide by 25%. Modular demonstrates emission reductions in

9
F. Greer and A. Horvath Building and Environment 240 (2023) 110432

Fig. 5. Comparison of changes in Alameda County embodied carbon (EC) emissions for module combinations (by frame type and factory location) relative to if the
number of multifamily housing units assessed in each county were built with traditional on-site, timber-framed construction. A positive percentage change (indicated
in red) means that the module type’s EC emissions are greater than the traditional example. A negative percentage change means that the module type leads to EC
emission savings. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

counties ranging from 12% in Del Norte County to 29% in Santa conclusions about factory-built housing. Sources of uncertainties in
Clara County modeling approaches and the results of this study include:
• Fig. 9 shows randomized allocation of all modular types compared to
randomized allocation of all conventional types. This is probably a • The use of representative data for building BOMs in California.
realistic depiction of how housing developments throughout the However, the representative data come from RSMeans, and while not
state might source modular types from different companies and entirely specific to California codes and conditions, they are based
factory locations. Under this scenario, except for Del Norte and upon national surveys of actual practices from contractors and
Modoc counties, all other counties achieve modest GHG emission builders.
benefits from modular. • The assumption about upstream logistics (e.g., where raw materials
• Fig. 10 depicts the same scenario as Fig. 9, except the worst- are delivered from) being coequal between the two housing con­
performing modular type (steel shipping container) has been struction methods would need to be further explored.
removed. Modular appears to perform better than all conventional • Simplification of on-site construction processes for both conven­
types in all counties. However, the benefits are more muted tional construction and on-site assembly of modules.
compared to the scenario depicted in Fig. 8. • Specific differences between on-site construction and factory activ­
ities in terms of material waste and duration of equipment use can
There are key takeaways from Fig. 7 through 10 for policymakers in likewise inform this scaled approach.
California. The shipping container module type should not be utilized as • Future work would also need to consider additional material types
its BOM has a relatively higher emissions intensity. Embodied carbon (including mass timber/CLT) with California-specific values for
emissions are lowest for counties that are closely located with factories, biomass.
highlighting the importance of module transportation. Realistically, • Further effort would need to be done to expand the scope of life-cycle
modular achieves modest embodied carbon benefits relative to tradi­ stages for each housing construction method, particularly as it affects
tional construction methods in California. to EOL impacts and the storage of biogenic carbon in biomass-based
structural framing.

3.5. Sources of uncertainties


Uncertainty from all sources could be improved by inclusion of more
data from existing projects. This effort would require further coopera­
Uncertainty plays an important role in determining the potential
tion with general contractors, building designers, and modular factory
embodied carbon emission benefits from modular housing. The margins
companies.
of error in data sources and modeling decisions can render any differ­
ences between modular and conventional housing insignificant. Addi­
tional uncertainty assessments, beyond the Monte Carlo simulation for
allocation of modular types, are necessary to draw any firmer

10
F. Greer and A. Horvath Building and Environment 240 (2023) 110432

Fig. 6. Effects of changing key model variables for timber-framed modules made in Solano County in California used for multi-family developments in Los
Angeles County.

Fig. 7. Comparison of average randomized (n =


15,000 iterations) proportions of different modular
types to prototypical timber-framed conventional
housing. A positive percentage change means that
modular construction results in embodied carbon
(EC) emission gains relative to traditional con­
struction methods. A negative percentage change
indicates emission savings. Statewide, modular de­
creases embodied carbon (EC) emissions by 4%
compared to timber-framed conventional housing.
Percentage changes relative to timber-framed con­
ventional are called out for Sacramento (− 1%),
Alameda (− 3%), Los Angeles (− 4%) Counties, the
county with the lowest EC benefits (Del Norte at
13%), and the county with the highest EC benefits
(Kern at − 6%).

11
F. Greer and A. Horvath Building and Environment 240 (2023) 110432

Fig. 8. Comparison of average randomized (n =


15,000 iterations) proportions of different modular
types, minus the steel shipping container module
type, to prototypical timber-framed conventional
housing. A positive percentage change means that
modular construction results in embodied carbon
(EC) emission gains relative to traditional con­
struction methods. A negative percentage change
indicates emission savings. Statewide, modular de­
creases embodied carbon (EC) emissions by 25%
compared to timber-framed conventional housing.
Percentage changes relative to timber-framed con­
ventional are called out for Sacramento (− 23%),
Alameda (− 27%), Los Angeles (− 24%) Counties,
the county with the lowest EC benefits (Del Norte at
− 12%), and the county with the highest EC benefits
(Santa Clara at − 29%).

Fig. 9. Comparison of average randomized (n =


15,000 iterations) proportions of different modular
types to prototypical conventional housing. A posi­
tive percentage change means that modular con­
struction results in embodied carbon (EC) emission
gains relative to traditional construction methods. A
negative percentage change indicates emission sav­
ings. Statewide, modular decreases embodied car­
bon (EC) emissions by 12% compared to all
conventional housing. Percentage changes relative
to timber-framed conventional are called out for
Sacramento (− 9%), Alameda (− 12%), Los Angeles
(− 13%) Counties, the county with the lowest EC
benefits (Del Norte at 3%), and the county with the
highest EC benefits (Kern at − 14%).

Fig. 10. Comparison of average randomized (n =


15,000 iterations) proportions of different modular
types, minus the steel shipping container module
type, to prototypical conventional housing. A posi­
tive percentage change means that modular con­
struction results in embodied carbon (EC) emission
gains relative to traditional construction methods. A
negative percentage change indicates emission sav­
ings. Statewide, modular decreases embodied car­
bon (EC) emissions by 18% compared to all
conventional housing. Percentage changes relative
to timber-framed conventional are called out for
Sacramento (− 4%), Alameda (− 18%), Los Angeles
(− 18%) Counties, the county with the lowest EC
benefits (Del Norte at − 4%), and the county with
the highest EC benefits (Kern at − 20%).

4. Discussion used to assess the embodied carbon impacts from modular multifamily
housing. Regarding the former, notwithstanding uncertainty and po­
There are two important outcomes of this research for policymakers: tential limitations from this study, the results suggest that modular
(1) context-specific policy guidelines for California on whether factory- housing could potentially yield embodied carbon emission benefits
built housing can achieve embodied carbon emission reductions relative relative to conventionally constructed housing. Based upon the data and
to traditional methods, and (2) a scaled modeling approach that can be modeling assumptions in this study, some of the key factors that

12
F. Greer and A. Horvath Building and Environment 240 (2023) 110432

influence potential benefits of modular housing are the: (1) structural boundary to include operational and end-of-life implications of modular
framing material of both the modular and conventional house; (2) dis­ buildings, including tighter building envelopes and circular design.
tance between the factory and the housing site; (3) module size; and (4)
delivery flatbed-truck capacity. However, absent the incorporation of CRediT authorship contribution statement
more representative data for factory and on-site operations, building
material quantities, and additional data, firm conclusions about Fiona Greer: Writing – original draft, Visualization, Validation,
modular housing are not possible to make. Methodology, Investigation, Formal analysis, Data curation, Conceptu­
In general, the results point towards the need to take a county-by- alization. Arpad Horvath: Project administration, Funding acquisition,
county or region-by-region approach for assessing whether modular Conceptualization, Methodology, Supervision, Validation, Writing –
housing can lead to GHG emission reductions. Factory location and review & editing.
required number of housing units influence overall GHG emissions for
each county. Except for the factory in Idaho, all the factories included in
the study are either located in or around the San Francisco Bay Area, in Declaration of competing interest
the Central Valley, or in Southern California. The results demonstrate
that modular housing does not appear to be as beneficial for counties in The authors declare the following financial interests/personal re­
the northern part of the state due to the long distances from existing lationships which may be considered as potential competing interests:
modular factories. In some instances (Fig. 7), conventional housing Financial support for the research was provided by California Air Re­
appears to be preferable to modular housing in Northern California. This sources Board.
conclusion could change with the opening of factories closer to sources
of demand. Data availability
While the results of this research are largely unique to California, the
modeling framework and approach are applicable to any location. This We have shared data and supporting information in the S.I. section.
framework could be replicated for any location by using location-
specific EPDs, transportation distances, factory outputs and electricity Acknowledgments
sources, and housing demand needs. Broadly, this research highlights
the need to study the embodied carbon implications of the built envi­ We gratefully acknowledge the financial support of the California Air
ronment and how life-cycle assessment is needed for policymakers to Resources Board under contract 21STC018. This article has not been
make informed decisions. reviewed by CARB. The views expressed in this document are solely
It is important to consider how effective modular housing might be in those of the authors and do not necessarily reflect those of the agency.
achieving broader environmental goals (such as mitigating GHG emis­ Neither the authors nor CARB endorse any products or technologies
sions) relative to potential negatives. The production capacities of cur­ mentioned in this publication.
rent modular factories in California are much lower than the number of
housing units that need to be constructed. If more factories need to be Appendix A. Supplementary data
built, or if the capacities of existing factories are expanded, there exists
potential to exacerbate existing human health and environmental justice Supplementary data to this article can be found online at https://doi.
concerns (e.g., increased production of modules could lead to increased org/10.1016/j.buildenv.2023.110432.
traffic from diesel-powered delivery vehicles). Given that the Cali­
fornian factories are currently in locations that are designated as
References
disadvantage communities as determined by the state’s environmental
pollution mapping tool [55], additional analysis should be undertaken [1] UN Environment Programme & GlobalABC, Global Status Report for Buildings and
to assess whether the state’s housing needs and its climate change Construction: Towards a Zero-Emissions, Efficient and Resilient Buildings and
Construction Sector (2022).
mitigation goals can be met just as efficiently with more conventional
[2] M. Röck, et al., Embodied GHG emissions of buildings–The hidden challenge for
housing methods that incorporate low-carbon materials. Future research effective climate change mitigation, Appl. Energy 258 (2020), 114107.
should incorporate these additional concerns into assessing the strate­ [3] W. Zhou, et al., Modelling future trends of annual embodied energy of urban
gies that can meet the state’s needs to provide housing while mitigating residential building stock in China, Energy Pol. 165 (2022), 112932.
[4] M. Hu, A look at residential building stock in the United States-mapping life cycle
GHG emissions. embodied carbon emissions and other environmental impact, Sustain. Cities Soc.
89 (2023), 104333.
5. Conclusions [5] W.L. Ng, M.Y. Chin, J. Zhou, K.S. Woon, A.Y. Ching, The overlooked criteria in
green building certification system: embodied energy and thermal insulation on
non-residential building with a case study in Malaysia, Energy 259 (2022), 124912.
The embodied GHG emissions associated with constructing multi­ [6] M. Winkler, California Poised to Overtake Germany as World’s No. 4 Economy,
family housing using factory-built modules are compared to emissions Bloomberg.com, 2022.
[7] U.S. Census Bureau, QuickFacts: California. https://www.census.gov/quickfacts/
from traditional, on-site construction methods. A model is developed to CA, 2022.
assess GHG emissions associated with raw materials, construction ac­ [8] California Department of Housing and Community Development, Regional housing
tivities, and transportation operations for the future stock of multifamily needs allocation. https://www.hcd.ca.gov/planning-and-community-development
/regional-housing-needs-allocation, 2023.
housing in California. The results demonstrate the utility of using a [9] California Air Resources Board (CARB), Building decarbonization. https://ww2.
scaled, context-specific approach that incorporates both representative arb.ca.gov/our-work/programs/building-decarbonization, 2023.
and actual data in assessing the potential benefits of modular housing at [10] A. Gibb, Pre-assembly in Construction: A Review of Recent and Current Industry
and Research Initiatives on Pre-assembly in Construction, 2001.
a local and regional level. Under the various modeling assumptions and [11] M. Kamali, K. Hewage, Life cycle performance of modular buildings: a critical
scenarios, modular housing could achieve average GHG reductions on review, Renew. Sustain. Energy Rev. 62 (2016) 1171–1183.
the order of 4%–20% for most counties compared to housing units [12] M. Kamali, K. Hewage, A.S. Milani, Life cycle sustainability performance
assessment framework for residential modular buildings: aggregated sustainability
constructed using conventional methods. Limitations of the study
indices, Build. Environ. 138 (2018) 21–41.
include the limited sample size of actual project data. Future research [13] Y. Teng, K. Li, W. Pan, T. Ng, Reducing building life cycle carbon emissions
efforts should expand upon this study’s scaling approach by including through prefabrication: evidence from and gaps in empirical studies, Build.
larger samples of existing project data (e.g., more building projects, Environ. 132 (2018) 125–136.
[14] L.R. Kawecki, Environmental Performance of Modular Fabrication: Calculating the
material waste differences between modular and traditional projects, Carbon Footprint of Energy Used in the Construction of a Modular Home, Arizona
different building materials) and by expanding the scope of the system State University, 2010.

13
F. Greer and A. Horvath Building and Environment 240 (2023) 110432

[15] D. Kim, Preliminary Life Cycle Analysis of Modular and Conventional Housing in [35] F.H. Ismail, M. Shahrestani, M. Vahdati, P. Boyd, S. Donyavi, Climate change and
Benton Harbor, MI, 2008. the energy performance of buildings in the future–a case study for prefabricated
[16] M. Kamali, K. Hewage, R. Sadiq, Conventional versus modular construction buildings in the UK, J. Build. Eng. 39 (2021), 102285.
methods: a comparative cradle-to-gate LCA for residential buildings, Energy Build. [36] R.E. López-Guerrero, S. Vera, M. Carpio, A quantitative and qualitative evaluation
204 (2019), 109479. of the sustainability of industrialised building systems: a bibliographic review and
[17] M. Al-Hussein, J.D. Manrique, D. Mah, North Ridge CO2 Analysis Report: analysis of case studies, Renew. Sustain. Energy Rev. 157 (2022), 112034.
Comparison between Modular and On-Site Construction, Univversity of Alberta, [37] V. Tavares, J. Gregory, R. Kirchain, F. Freire, What is the potential for
Edmonton, Canada, 2009. prefabricated buildings to decrease costs and contribute to meeting EU
[18] J. Monahan, J.C. Powell, An embodied carbon and energy analysis of modern environmental targets? Build. Environ. 206 (2021), 108382.
methods of construction in housing: a case study using a lifecycle assessment [38] California Energy Commission, SB 100 joint agency report. California Energy
framework, Energy Build. 43 (2011) 179–188. Commission. https://www.energy.ca.gov/sb100, 2023.
[19] J. Quale, M.J. Eckelman, K.W. Williams, G. Sloditskie, J.B. Zimmerman, [39] E. Grubert, J. Stokes-Draut, A. Horvath, W. Eisenstein, Utility-specific projections
Construction matters: comparing environmental impacts of building modular and of electricity sector greenhouse gas emissions: a committed emissions model-based
conventional homes in the United States, J. Ind. Ecol. 16 (2012) 243–253. case study of California through 2050, Environ. Res. Lett. 15 (2020), 1040a4.
[20] L. Aye, T. Ngo, R.H. Crawford, R. Gammampila, P. Mendis, Life cycle greenhouse [40] C. Holden, California AB2446 | 2021-2022 | regular session. LegiScan. https://legi
gas emissions and energy analysis of prefabricated reusable building modules, scan.com/CA/text/AB2446/id/2607014, 2022.
Energy Build. 47 (2012) 159–168. [41] H. Johnson, E. McGhee, M. Cuellar Mejia, California’s population. Public Policy
[21] A. Dodoo, L. Gustavsson, R. Sathre, Lifecycle carbon implications of conventional Institute of California. https://www.ppic.org/publication/californias-population/,
and low-energy multi-storey timber building systems, Energy Build. 82 (2014) 2023.
194–210. [42] U.S. Census Bureau, California remained most populous state but growth slowed
[22] A.W.A. Hammad, A. Akbarnezhad, Modular vs conventional construction: a multi- last decade. Census.gov. https://www.census.gov/library/stories/state-by-state
criteria framework approach, in: ISARC. Proceedings of the International /california-population-change-between-census-decade.html, 2021.
Symposium on Automation and Robotics in Construction, vol. 34, IAARC [43] A.P. Gursel, A. Shehabi, A. Horvath, Embodied energy and greenhouse gas
Publications, 2017. emission trends from major construction materials of US office buildings
[23] J.J. Bukoski, P. Chaiwiwatworakul, S.H. Gheewala, The life cycle assessment of an constructed after the Mid-1940s, Build. Environ. 110196 (2023).
energy-positive peri-urban residence in a tropical regime, J. Ind. Ecol. 21 (2017) [44] J. McKnight, Shipping containers used to build LA housing complex for the
1115–1127. homeless. Dezeen. https://www.dezeen.com/2021/07/29/shipping-containers-lo
[24] H. Pervez, Y. Ali, A. Petrillo, A quantitative assessment of greenhouse gas (GHG) s-angeles-housing-complex-homeless-nac-architecture-bernards/, 2021.
emissions from conventional and modular construction: a case of developing [45] E. Ikeme, City awarded $12.3 million to build dignity village. Alameda Sun. https
country, J. Clean. Prod. 294 (2021), 126210. ://alamedasun.com/news/city-awarded-123-million-build-dignity-village, 2022.
[25] H. Jang, Y. Ahn, S. Roh, Comparison of the embodied carbon emissions and direct [46] Cal EPA, California’s Wildfire and forest resilience action plan: a comprehensive
construction costs for modular and conventional residential buildings in South strategy of the governor’s forest management task force. https://wildfiretaskforce.
Korea, Buildings 12 (2022) 51. org/wp-content/uploads/2022/04/californiawildfireandforestresilienceactionplan
[26] V. Tavares, N. Lacerda, F. Freire, Embodied energy and greenhouse gas emissions .pdf, 2021.
analysis of a prefabricated modular house: the “Moby” case study, J. Clean. Prod. [47] Gordian, RSMeans data - North America’s leading construction cost database.
212 (2019) 1044–1053. Gordian. https://www.gordian.com/products/rsmeans-data-services/, 2023.
[27] V. Tavares, F. Freire, Life cycle assessment of a prefabricated house for seven [48] EIA. Energy Information Administration (EIA), About the Residential Energy
locations in different climates, J. Build. Eng. 53 (2022), 104504. Consumption Survey (RECS) Table HC1.1 Fuels used and end uses in U.S. homes by
[28] A. Al-Najjar, A. Dodoo, Modular multi-storey construction with cross-laminated housing unit type. https://www.eia.gov/consumption/residential/data/2015/hc
timber: life cycle environmental implications, Wood Mater. Sci. Eng. 1–15 (2022). /php/hc10.13.php, 2015 (2018).
[29] C. Dara, C. Hachem-Vermette, G. Assefa, Life cycle assessment and life cycle [49] CARB, EMFAC. https://arb.ca.gov/emfac/emissions-inventory/0736f434eef49afe
costing of container-based single-family housing in Canada: a case study, Build. 1042ae7089aed89058368c06, 2022.
Environ. 163 (2019), 106332. [50] A. Säynäjoki, J. Heinonen, S. Junnila, A. Horvath, Can life-cycle assessment
[30] H. Islam, G. Zhang, S. Setunge, M.A. Bhuiyan, Life cycle assessment of shipping produce reliable policy guidelines in the building sector? Environ. Res. Lett. 12
container home: a sustainable construction, Energy Build. 128 (2016) 673–685. (2017), 013001.
[31] M. Bertolini, L. Guardigli, Upcycling shipping containers as building components: [51] CEC, Power content label. California Energy Commission. https://www.energy.ca.
an environmental impact assessment, Int. J. Life Cycle Assess. 25 (2020) 947–963. gov/programs-and-topics/programs/power-source-disclosure/power-content-labe
[32] D. Satola, et al., Comparative life cycle assessment of various energy efficiency l, 2023.
designs of a container-based housing unit in China: a case study, Build. Environ. [52] Idaho Power, Our energy sources. https://www.idahopower.com/energy-environ
186 (2020), 107358. ment/energy/energy-sources/, 2023.
[33] MBI, Industry analysis. Modular building institute. https://www.modular.org [53] A. Horvath, J. Stokes, Life-cycle energy assessment of alternative water supply
/industry-analysis/, 2023. systems in California. https://uc-ciee.org/ciee-old/downloads/Life-cycleHorvath.
[34] S. Pless, et al., The Energy in Modular (EMOD) Buildings Method: A Guide to pdf, 2011.
Energy-Efficient Design for Industrialized Construction of Modular Buildings, [54] NumPy Developers, numpy.random.dirichlet — NumPy v1.24 Manual. http
2022. s://numpy.org/doc/stable/reference/random/generated/numpy.random.dirich
let.html, 2022.
[55] OEHHA, CalEnviroScreen. OEHHA. https://oehha.ca.gov/calenviroscreen, 2022.

14

You might also like