Optimizing the Prefabrication Finishing Level in Modular Construction

Journal: Canadian Journal of Civil Engineering
Manuscript ID cjce-2020-0346.R1
Manuscript Type: Article
Complete List of Authors: Almashaqbeh, Mohammad; University of Illinois at Urbana-Champaign, Department of Civil and Environmental Engineering; The Hashemite University, Department of Civil Engineering El-Rayes, Khaled; University of Illinois at Urbana-Champaign, Civil and Environmental Engineering
Keyword: Modular construction, Prefabrication, Module finishing level, Optimization, Offsite fabrication cost
Is the invited manuscript for consideration in a Special
Issue? : Not applicable (regular submission)
© The Author(s) or their Institution(s)
Canadian Journal of Civil Engineering
Draft
1
Mohammad Almashaqbeh and Khaled El-Rayes
Mohammad Almashaqbeh. Department of Civil and Environmental Engineering, University of
Illinois at Urbana-Champaign, Urbana, IL 61801, USA; Department of Civil Engineering,
Hashemite University, Zarqa 13133, Jordan.
Khaled El-Rayes. Department of Civil and Environmental Engineering, University of Illinois at
Urbana-Champaign, Urbana, IL 61801, USA.
Corresponding author: Mohammad Almashaqbeh (email: [email protected]).
Word count: 6463
Draft
2
Abstract
Prefabricated modules in modular construction projects can have a wide range of finishing levels
that range from partially completed with only structural frame to fully completed with all
structural, wall, mechanical, electrical, finishing and furnishing components. A higher module
finishing level increases the offsite fabrication and transportation cost and decreases onsite
assembly cost and duration while a lower finishing level produces the opposite results. This paper
presents an optimization model that enables construction planners to identify an optimal finishing
level for prefabricated modules in order to minimize the total cost of modular construction projects
that includes all offsite fabrication, transportation, and onsite assembly costs. A case study of a
modular construction project for a healthcare facility was analyzed to illustrate the use of the model
and evaluate its performance. The results of this analysis highlight the original capabilities of the
model in minimizing the total cost of modular construction projects by identifying an optimal
finishing level for each module type in the project from a set of feasible alternatives with varying
building components, weights, and cost rates.
Key words: Modular construction; Prefabrication; Module finishing level; Optimization; Offsite
fabrication cost; Onsite assembly cost; Minimizing modular construction cost.
Page 2 of 28
Draft
3
Introduction
The use of prefabricated modules in the construction industry has expanded in recent years due to
the many reported benefits of modular construction including improved productivity, reduced
costs, shorter duration, and enhanced quality (Azari et al. 2013; MBI 2010; McGraw-Hill
Construction 2011; OSCC 2014; Sage 2017; Smith and Rice 2015; Southern 2016). One of the
main challenges confronting the planners and designers of prefabricated modular construction is
identifying the finishing level of each prefabricated module type (AIA 2019; Azari et al. 2013;
Brown 2014; Garrison and Tweedie 2008; Lawson et al. 2014; Park and Ock 2016). Feasible
alternatives of a module finishing level may include structural frame only [see Fig. 1(a)], structural
frame and walls, or completely finished module with all structural, mechanical, and electrical
systems [see Fig. 1(b)].
The selected module finishing level has a significant impact on the cost of module offsite
fabrication, transportation, and onsite assembly. A higher finishing level increases the offsite
fabrication and transportation cost and decreases onsite assembly cost and duration while a lower
finishing level produces the opposite results. For example, a completely finished module such as
the one shown in Fig. 1(b) requires (a) increased offsite fabrication time and cost (Lawson et al.
2014), (b) higher transportation cost due to the heavier module weight (Smith 2010), and (c) lower
onsite assembly cost and duration due to the completion of most of its components offsite in the
factory (MBI 2010). Accordingly, planners and designers of prefabricated modular construction
need to carefully analyze and optimize the selection of finishing level for each module type in the
project in order to minimize the overall cost of modular construction.
Fig. 1. Examples of feasible finishing levels of modular construction (Images by Stack Modular).
[Colour online.]
Draft
4
Several research studies have been conducted to analyze and optimize the planning of
module assembly and onsite construction activities in modular construction projects. These studies
can be grouped in two main categories that focused on (a) assembly cost of prefabricated modules
and (b) crane selection and operation in modular construction. An example of the first category
studies is a stochastic programming model developed by Hsu et al. (2018) for the optimization of
supply chain in modular construction including module production, storage, and assembly. The
model considered the cost of assembling prefabricated modules on the construction site under
varying daily demand scenarios of modules. Gbadamosi et al. (2019) proposed another model for
optimizing assembly in offsite construction based on the principles of design for manufacture and
assembly (DFMA). The model was designed to minimize assembly time by selecting optimal
building envelope materials based on their ease of assembling, ease of handling, speed of
assembling, and assembly waste. Khalili and Chua (2013) established a system for minimizing
assembly costs of prefabricated buildings by reducing the total number of prefabricated
components at the factory. The system was based on industry foundation class (IFC) and utilized
computer-aided design software to achieve higher level of prefabrication through grouping of
individual prefabricated building elements.
An example of the second category that focused on crane selection and operation in
modular construction projects is the 3D crane evaluation system developed by Han et al. (2017).
The system was designed to plan mobile crane operations in modular construction projects by
considering crane lifting capacity, required working radius, and planning the crane module lifting
schedule. Another study by Salama et al. (2017) developed an optimization model for identifying
near optimal module configurations that minimize the cost of modular residential construction.
The model utilized a crane cost penalty index (CCPI) that considered cost of using crane in
Page 4 of 28
Draft
5
modular construction based on the hourly crane renting cost, number of modules in a project, and
hourly module placing rate by the crane. Another study developed a mathematical algorithm for
selecting crane type in modular construction based on required crane lifting capacity and
optimizing crane location and operation within the construction site (Olearczyk et al. 2014).
Moghadam et al. (2012) used simulation and post simulation visualization for scheduling onsite
installation of prefabricated modules and minimizing tower crane idle time in modular
construction projects.
Despite the significant contributions of the aforementioned studies that focused on
assembly of prefabricated modules as well as onsite construction operations in modular
construction, they are all incapable of: (1) quantifying the impact of selecting module finishing
level on the offsite fabrication, transportation and onsite assembly costs of all prefabricated
modules in modular construction projects; (2) selecting an optimal finishing level for each
prefabricated module type in modular construction projects from a set of feasible module finishing
level alternatives; and (3) considering and minimizing the cost of all phases of modular
construction impacted by the selection of module finishing level including module offsite
fabrication, transportation, and onsite assembly costs. Accordingly, there is a pressing need for
additional research to address and overcome these limitations.
Objective
The objective of this research is to develop a novel model for optimizing the finishing level of
prefabricated modules in modular construction projects that is capable of overcoming the
aforementioned limitations of existing models. The model is developed in three main phases: (1)
formulation phase, that identifies the optimization model decision variable, objective function, and
constraints; (2) implementation phase, that performs the optimization computations using integer
Page 5 of 28
Draft
6
programming; and (3) performance evaluation phase, that analyzes a cases study to evaluate the
performance of the optimization model. The following sections describe the three phases of the
model development.
Model formulation
The present model is formulated to support planners in identifying for each module type i an
optimal finishing level z (F*z,i), where z and i are modeled using positive integers, as shown in Fig.
2. The finishing level decision variable (Fz,i) in this model is designed to represent the building
components that are manufactured in the module production factory while the remaining
components for each module type i will be completed onsite after delivering modules to the
construction site. Building components include structural, mechanical, electrical, plumbing,
finishing, and furnishing components. For example, a planner can specify the possibility of three
alternative finishing levels for module type i = 1: (1) module structural frame elements only; (2)
module structural frame elements and walls framing; and (3) all structural, mechanical, electrical,
finishing, and furnishing components, as shown in Fig. 2. The model provides planners with the
flexibility of specifying building components for each module type finishing level based on the
module manufacturing practices and the specific project requirements.
Fig. 2. Model decision variable (Fz,i). [Colour online.]
The objective function in this model is designed to minimize the cost of all phases of
modular construction impacted by module finishing level including offsite fabrication cost (FC),
transportation cost (TC), and onsite assembly cost (AC), as shown in eq. (1).
(1) Minimize FC + TC + AC
Page 6 of 28
Draft
7
The offsite fabrication cost (FC) is calculated in this model by adding up the costs of all
module building components fabricated in the module production factory for every module type,
as shown in eq. (2).
(2) z ∈ Z FC = ∑ = 1(, × , × ) ∀
Where FCz,i is the offsite fabrication cost of module type i with finishing level z in $, Vi is the total
number of modules for each module type i, I is the total number of module types specified by the
planner, and Z is the total number of finishing levels for each module type i specified by the
planner.
The transportation cost (TC) represents the cost of transporting all modules from the
factory to the project site including the cost of wrapping and weather protection for modules during
transportation and it is calculated as shown in eq. (3).
(3) z ∈ Z TC = ∑ = 1(, × , × ) ∀
Where TCz,i is the transportation cost of module type i with finishing level z in $ from factory to
project site. It should be noted that the scope of the developed model focuses on quantifying and
optimizing the impact of module finishing level decisions on the total cost of fabrication,
transportation, and assembly of modular construction. Accordingly, the module transportation cost
(TCz,i) in this model quantifies the impact of module finishing level decisions and does not consider
the impact of other module transportation decisions that are beyond the scope of this paper such
as module delivery time that can have an impact on crane handling and truck idling costs.
The onsite assembly cost (AC) is calculated in this model by summing up: (a) onsite
finishing cost (OC) for all module building components finished on the construction site for every
Page 7 of 28
Draft
8
module type; and (b) crane cost for lifting the modules from trucks and assembling them in their
designated locations in the project, as shown in eqs. (4) and (5).
(4) + k ∈ K AC = ( × CD) ∀
(5) z ∈ Z OC = [∑ = 1(, × , × )] ∀
Where DCk is the daily cost rate of crane type k in $/day, CD is the Crane utilization duration in
days, OCz,i is the onsite finishing cost of module type i with finishing level z in $, and K is the total
number of available crane types specified by the planner.
The model provides planners with the flexibility of selecting the least expensive crane from
a set of feasible alternatives k =1 to K based on the required crane lifting capacity (LC) that is
identified based on the heaviest module weight, as shown in eq. (6). For example, a planner can
specify the availability of two feasible crane alternatives: (1) first crane alternative (k = 1) that has
a lifting capacity less than 2,000 kg and a daily cost (DC1) = $300/day; and (2) second crane
alternative (k = 2) that has a lifting capacity between 2,000 kg and 3,000 kg and DC2 = $600/day.
For this example, the model can utilize eq. (6) to identify the required crane lifting capacity (e.g.
LC = 2500 kg) and accordingly selects the second crane alternative (k = 2) and identifies its daily
cost rate (DC2 = $600/day) to calculate the onsite assembly cost using eq. (4).
(6) Mz,i) SF] z, i = [max(, × × ∀
Where Mz,i is the weight of module type i with finishing level z in kg, and SF is the crane safety
factor specified by the planner that accounts for dynamic forces during lifting of module.
The model is designed to consider all relevant practical optimization constraints to ensure
that it provides feasible solutions, including finishing level assignment constraint, truck capacity
Page 8 of 28
Draft
9
constraint, maximum crane capacity constraint, and decision variable constraint. The finishing
level assignment constraint is designed to ensure the selection of only one alternative from the
feasible set of available finishing levels, as shown in eq. (7). To comply with this constraint, the
model integrates a binary decision variable representing the selection of each feasible alternative
(Fz,i), where Fz,i = 1 if finishing level z of module type i is selected by the model, and Fz,i = 0
otherwise, as shown in eq. (7).
(7) i ∈ I∑ = 1, = 1 ∀
The truck capacity constraint is designed to ensure that the total weight of all modules
transported by each truck does not exceed the truck weight capacity, as shown in eq. (8).
(8) ( Ti) z, i , × , × ≤ ∀
Where Ti is the number of modules of type i that can be transported by one truck, and TW is the
truck weight capacity in kg.
The maximum crane capacity constraint is integrated in the optimization model to ensure
that the required crane lifting capacity (LC) does not exceed the maximum available crane capacity
(maxLC) specified by the planner, as shown in eq. (9).
(9) ≤ maxLC
The decision variable constraint is designed to set the boundary of the decision variable.
For example, if the planner specified that there are four possible finishing levels for module type i
= 1, Fz,1 is set with a lower bound of z = 1 to represent the first finishing level alternative of module
type i =1, and upper bound of z = 4 to represent that there are four finishing level alternatives of
module type i =1. The optimization model can select any of the four alternatives during the
optimization computations and analyze its performance.
Page 9 of 28
Draft
10
Implementation
The present model is implemented using integer programming to perform the optimization
computations. The model is implemented in three main stages: (1) input stage that provides all
data relevant to the optimization model as specified by the planner; (2) optimization stage that
executes the integer programming optimization; and (3) output stage that generates the
optimization results, as shown in Fig. 3.
Fig. 3. Model implementation. [Colour online.]
Performance evaluation
This phase evaluates the performance of the developed optimization model by analyzing a case
study to demonstrate the use of the model and illustrate its contributions and original capabilities
in optimizing the finishing level of all building prefabricated modules in modular construction
projects. The case study involves the offsite fabrication, transportation, and onsite assembly of
steel modules for a 6,000 m2 healthcare facility. The modular construction of this facility consists
of (a) 4,530 m2 area of offsite prefabricated modules that require the fabrication, transportation,
and onsite assembly of five different types of modules, as shown in Table 1 and (b) 1,470 m2 area
of onsite construction of the healthcare facility structural frame and other building components.
The input data for this case study includes (1) module data, (2) finishing level data, (3)
transportation data, and (4) crane data that are summarized in Tables 1, 2, 3, and 4, respectively.
First, the module input data in this case study includes module types, numbers, dimensions,
feasible finishing levels, and weights that address the specific requirements of healthcare facilities,
as shown in Table 1 (ASCE 2017; CFM 2014; Jellen and Memari 2014; Lawson et al. 2014;
SteelConstruction.info 2018).
Page 10 of 28
Draft
11
Second, the finishing level input data in this case study lists all the building components
included in every feasible finishing level in each module as well as their offsite and onsite unit
costs, as shown in the sample data in Table 2. These building components can be identified using
historical data from similar projects and/or data from the Uniformat II classification system (The
Construction Specifications Institute 2010). For example, the planner in this case study specified
that there are three feasible finishing levels for the exam room module (i = 2): (1) first finishing
level (F1,2) that requires offsite fabrication of the module structural frame only represented by
building components B1010 and B1020 in Table 2; (2) second finishing level (F2,2) that requires
offsite fabrication of the module structural, mechanical, electrical, plumbing, and finishing
components that are represented by all the listed building components shown in Table 2 except for
the equipment and furnishings components (E1020 and E2020); and (3) third finishing level F3,2
that requires offsite fabrication of a completely finished module including exam room equipment
and furnishings (E1020 and E2020), as shown in Table 2. The onsite unit costs of all building
components in Table 2 were estimated using the square foot costs provided by RSMeans Data
(RSMeans 2018a). These onsite unit costs of all building components were then adjusted to
estimate their offsite unit costs that considers the impact of productivity and cost improvements
resulting from prefabrication and modularization. Accordingly, the offsite prefabricated unit costs
in this case study (see Table 2) were estimated to be 10% less than their onsite unit costs based on
the reported savings of 7% to 20% in the literature (Lawson et al. 2014; McGraw-Hill Construction
2011). It should be noted that the offsite and onsite unit costs vary from one project to another and
therefore need to be specified as input by the planner in this model and the assumed 10% savings
in this case study is used only for illustrative purposes.
Page 11 of 28
Draft
12
Third, the transportation input data in this case study includes number of modules that can
be transported by one truck for every module type as well as transportation cost rates that are
summarized in Table 3. These transportation cost rates were estimated based on a transportation
distance of 500 km from the factory to the construction site and complying with Illinois State
regulations and transportation fees (ATRI 2018; Battelle 1995; C.H.Robinson 2015; Garrison and
Tweedie 2008; IDOT 2015a; b; c, 2018; Salama et al. 2017; Seaker and Lee 2006). Furthermore,
The truck weight capacity (TW) in this case study was assumed to be 20,000 kg which is the weight
capacity of a standard flatbed truck (Smith 2010).
Fourth, the crane input data includes available crane types that can be used in this case
study as well as their lifting capacities and daily costs, as shown in Table 4. These crane lifting
capacities and costs were estimated using RSMeans cost data (RSMeans 2018b). The crane
utilization duration (CD) in this case study was estimated to be 60 days based on the planner-
specified total number of modules in this case study and the reported crane daily lifting rate in the
literature (Azari et al. 2013; Lawson et al. 2014; Olearczyk et al. 2014; Salama 2018; Salama et
al. 2017). It should be noted that the crane utilization duration varies from one project to another…