Developing an Optimum Material Procurement Schedule

by Integrating Construction Program and

Budget Using NSGA-II
Santu Kar 1; Chirag Kothari 2; and Kumar Neeraj Jha, Ph.D. 3
Downloaded from ascelibrary.org by Georgia Tech Library on 02/19/21. Copyright ASCE. For personal use only; all rights reserved.

Abstract: Procuring the correct materials at the right time and for the lowest cost is essential in construction. This fact underlines the
importance of an optimized material procurement schedule integrated with the construction schedule/program. However, few studies ex-
amine the development of a material procurement schedule by integrating the construction schedule and optimizing material costs as well as
any material shortage impact. In addition, budget constraints and maximum storage capacity are rarely captured in existing models for
material procurement optimization. To address these shortcomings, an optimization model is developed using the nondominated sorting
genetic algorithm II (NSGA-II) and executed in MATLAB R2017a. This model was implemented in a construction project that developed a
material delivery schedule based on the budget constraints and maximum storage capacity for five majorly used building materials, resulting
in a costs saving of 31.17% and a reduction in shortage impact of 83.47% compared with the actual delivery schedule. The developed model
also incorporates the minimum order quantity and standard shipping size as well as avoiding surplus materials, leading to sustainable
procurement of materials. Construction practitioners can use this model for procuring materials when facing budget constraints in a con-
struction project with the lowest cost and least shortage and without excess or insufficient purchasing. The model will aid in completing the
project within the stipulated time and budgeted cost. DOI: 10.1061/(ASCE)CO.1943-7862.0002028. © 2021 American Society of Civil
Engineers.
Author keywords: Material delivery schedule; Criticality of materials; Shortage impact; Optimization; Budget constraint; Storage capacity;
Nondominated sorting genetic algorithm II (NSGA-II).

Introduction 2002). However, the material management process has proven

ineffective in the construction industry (Kim 2014; Jha 2015). In-
Material management is effectiveness is caused by different issues such as overordering or
underordering of materials, forgetting to order, late delivery, short-
an integrated process for planning and controlling all neces-
age of required material, poor communication, poor management
sary efforts to make sure that the quality and quantity of ma-
terials and equipment are appropriately specified in a timely of surplus materials, poor quality of materials, insufficient storage
manner, are obtained at a reasonable cost, and are available facilities, and large inventory holding cost (Caldas et al. 2014;
when needed. (CII 2020) Kulkarni et al. 2017; Kar and Jha 2020b). Ibn-Homaid (2002)
and Doloi et al. (2012) identified a lag in the delivery of materials
It is crucial to project success (Caldas et al. 2014). A systematic as the most critical attribute causing construction delays.
material management process lowers the procurement cost and en- The material management process is well-developed in the
sures timely availability of material; thus, it contributes to better manufacturing industry, unlike in construction (Caldas et al. 2014;
scheduling, reduced costs, and improved productivity in construc- Jha 2015; Jusoh and Kasim 2017). The material management strat-
tion (Jusoh and Kasim 2017; Gurmu 2020). Typically, materials egies that are typical for the manufacturing industry, such as eco-
account for 50%–60% of the total cost of construction and influ- nomic order quantity (EOQ), cannot be directly adopted for the
ence 80% of the construction program or schedule (Ibn-Homaid construction industry due to the fundamental differences between
these industries in uniformity of demand and standardization of
1
Ph.D. Research Scholar, Dept. of Civil Engineering, Indian Institute products (Georgy and Basily 2008; Jha 2015; Kar and Jha 2020b).
of Technology Delhi, New Delhi, Delhi 110016, India (corresponding For instance, the EOQ model assumes a constant delivery of
author). ORCID: https://orcid.org/0000-0002-9618-227X. Email: santu materials, which is not suitable for a construction project because
[email protected] demand may not remain constant in that case. Furthermore, the
2
Ph.D. Candidate, Dept. of Civil, Architectural, and Environmental En-
just-in-time (JIT) system, which is well-implemented in the manu-
gineering, Univ. of Texas at Austin, 301 E. Dean Keeton St., ECJ 5.412,
Austin, TX 78712. ORCID: https://orcid.org/0000-0002-5824-6527. facturing industry, is inadequate in construction, especially for criti-
Email: [email protected] cal building materials like cement, steel, reinforcement bars, and
3
Professor, Dept. of Civil Engineering, Indian Institute of Technology aggregates. These materials must be stored during construction
Delhi, New Delhi, Delhi 110016, India. Email: [email protected] to handle uncertainties due to delays in delivery, material shortages
Note. This manuscript was submitted on May 18, 2020; approved on
in the local market, unfavorable weather conditions, and uncertain-
November 9, 2020; published online on February 8, 2021. Discussion per-
iod open until July 8, 2021; separate discussions must be submitted for ties in the construction schedule (Polat and Arditi 2005). Therefore,
individual papers. This paper is part of the Journal of Construction En- the construction industry requires a specific strategy for material
gineering and Management, © ASCE, ISSN 0733-9364. management.

© ASCE 04021017-1 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2021, 147(4): 04021017

 The preparation of procurement plans and delivery schedules of section also discusses several benefits resulting from the proposed
materials represent the two most important strategies in material optimization model. The final section draws conclusions and also
management (Gurmu 2020; Kar and Jha 2020b). These strategies discusses the contributions and limitations of this research.
assist project team members in knowing when to place an order and
how much to order. However, such a task is not simple. Typically,
in a construction project, managers first prepare a project schedule Literature Review
and then plan the material ordering timetable accordingly. How-
ever, managers neglect the trade-off among different procurement Studies on material management in the construction industry have
costs such as ordering, storage, and capital costs, as well as pen- primarily examined material management practices and issues.
alties due to shortages of materials (Habibi et al. 2019). Ordering Some studies investigated sustainable material management and
materials in large quantities during the beginning stages of con- construction supply chain management. A few other studies have
struction can increase the cost of storing them, whereas ordering included the measurement of material management effectiveness,
too late can result in material shortages and consequent impacts. use of information and communication technology (ICT), and au-
Furthermore, frequent orders of the same materials can increase tomation in material management.
Downloaded from ascelibrary.org by Georgia Tech Library on 02/19/21. Copyright ASCE. For personal use only; all rights reserved.

associated ordering costs, whereas purchasing them in a single The aforementioned research has failed to incorporate the
or few orders can increase overall expenses by locking up capital construction schedule in their studies of material management.
in a large inventory. Therefore, an optimized material procurement Researchers have explained that material management should be
schedule is required to increase the cost efficiency of procurement. integrated with the construction schedule, and a material procure-
Moreover, the procurement schedule should incorporate the avail- ment schedule should be prepared based on the construction
able budget to avoid delay in supplier payment. schedule for timely availability of materials for achieving maximum
Some studies have been conducted to develop an optimized project efficiency (Dodin and Elimam 2001; Dixit et al. 2013; Gurmu
material procurement schedule for a construction project. Re- 2020; Kar and Jha 2020a). As discussed subsequently, some studies
searchers developed some models in this context, considering have attempted to examine this question.
different material procurement costs (Dodin and Elimam 2001;
Georgy and Basily 2008; Said and El-Rayes 2011; Dixit et al. Integration of Material Procurement with Construction
2014; Habibi et al. 2019). However, these models do not emphasize Schedule
the budget constraint situation in a construction project. Typically, Ala-Risku and Kärkkäinen (2006) introduced a shipment tracking–
in a construction project, limited budgets are allotted for the pro- based tool for inventory transparency and a proactive material de-
curement of materials within a specified period, which restricts the livery model based on the construction schedule to ensure the
ability to simultaneously purchase all the materials (Dixit et al. timely availability of materials. To do so, they used a bill of ma-
2013; Kar and Jha 2020a). It may cause a delay in the procurement terials (BOM). However, Ala-Risku and Kärkkäinen (2006) did not
of some materials, resulting in shortage of materials at the neces- illustrate the cost-effectiveness of the material delivery model. An
sary times. automated model for material management and control based on
The shortage of materials may negatively impact the project in construction schedule and planned quantities of materials was in-
terms of time and cost overrun, which should be minimized in a troduced by Navon and Berkovich (2006). However, the automated
construction project. Kar and Jha (2020a) proposed a method to model did not incorporate the changes in the construction schedule,
assess the shortage impact of materials in a construction project. ordering frequency, and storage capacity of materials. Barriga et al.
However, previous studies are limited to incorporating the shortage (2005) proposed a material requirement plan (MRP) for the manu-
impact of materials in the procurement schedule. Previous studies factured housing industry, which incorporated the production
also did not emphasize the storage capacity of materials, which schedule and could act quickly to unexpected changes in demand.
critically constrains some construction projects. Furthermore, the The MRP system was designed based on backward scheduling.
previous literature rarely considered models that took several ma- However, the MRP system is silent on the optimal order quantity
terials into account at once. of materials.
Therefore, this study aims to develop a material procurement Guo et al. (2017) developed a bulk material management system
schedule that is capable of incorporating the construction schedule, to improve business intelligence analysis for procurement planning
optimizing procurement costs and the impact of the shortage of ma- and inventory warning in nuclear power projects. However, this
terials, and adhering to the storage capacity and available budget in system did not consider the storage capacity onsite and budget
material procurement. The material procurement schedule will be availability in procurement of materials. Furthermore, the studies
developed by considering several materials at once. This schedule conducted by Ala-Risku and Kärkkäinen (2006), Navon and
will help procurement managers to procure the appropriate quantity Berkovich (2006), Barriga et al. (2005), and Guo et al. (2017) did
of materials at the best possible time while lowering acquisition not incorporate the procurement costs of materials such as ordering,
costs, which will result in greater success in construction. storage, and capital costs, although they are crucial in an effective
The next section describes the theoretical background on material management system, as discussed in the following section.
material management, studies that consider the context of integrat-
ing material management with the construction schedule, and dif-
ferent models on material procurement. This part is followed by a Material Management Models Incorporating
method for developing an optimum material procurement schedule Procurement Costs
(MPS). This section includes the formulation of the problem for Habibi et al. (2019) stated that usually, construction managers first
the proposed optimization model as well as the optimization pro- prepare a project schedule and then plan the material ordering
cess, which uses the nondominated sorting genetic algorithm II schedule accordingly. However, managers neglect the trade-off be-
(NSGA-II). The proposed model is then illustrated with a case- tween different procurement costs such as ordering or holding costs
study project. Furthermore, the “Discussion” section compares the and penalties. Shtub (1988) developed a mathematical model that
delivery schedule generated from the proposed model and the ac- minimized the penalty and expediting expenses in a large construc-
tual material delivery schedule followed in the case project. This tion project. Although that study considered the procurement costs,

© ASCE 04021017-2 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2021, 147(4): 04021017

 it did not consider the material budget in the model. Dodin and and fuzzy shortage costs. Uncertainty in the activity duration and
Elimam (2001) attempted to explore the trade-off between material lead time were captured using fuzzy numbers. However, this model
ordering and holding costs. They explained that if materials were did not incorporate ordering cost of materials. Moreover, it did not
ordered during the early stage of the project, then ordering costs consider the storage capacity for the materials, which represents
might decrease, but storage costs would increase. Conversely, if a critical constraint in a construction project, especially in urban
materials were ordered just in time, then ordering costs might prove areas.
high, whereas holding costs would be minimal. Hence, managers In a recent study, Kar and Jha (2020a) described how prioriti-
should consider both the holding and ordering costs in material pro- zation of materials was rarely linked with the construction sched-
curement. However, that study did not consider budget availability ule, which created problems, especially when facing budget
and the shortage cost of materials—that is, the cost due to the constraints. To address this, they determined total criticality (TC)
unavailability of materials. of materials by combining material criticality (MC) and activity
An optimized delivery schedule of materials was developed by criticality (AC). MC values of materials were calculated using
Georgy and Basily (2008). A delivery schedule was developed with seven interdependent criteria that relied on analytic network pro-
the MRP by optimizing purchasing, ordering, storage, and capital cess (ANP) and technique for order preference by similarity to an
Downloaded from ascelibrary.org by Georgia Tech Library on 02/19/21. Copyright ASCE. For personal use only; all rights reserved.

costs using a genetic algorithm (GA). However, the optimization ideal solution (TOPSIS). AC values were calculated using the float
model did not consider all the materials at once in the optimization available in associated activities. Kar and Jha (2020a) suggested
process. Moreover, it did not incorporate the shortage cost of ma- using the TC values to quantitatively measure the penalty from the
terials and the budget available for purchase. Said and El-Rayes nonavailability of materials—that is, the shortage impact in the
(2011) developed a construction logistics planning (CLP) model procurement schedule. They stated that procurement based on
that could integrate and optimize material procurement and storage TC values will ensure that the most critical materials for the most
on construction sites. The model used GA to minimize material important activities are procured first.
ordering, financing, stock-out, and layout costs. However, it also However, Kar and Jha’s (2020a) study was limited to assessing
considered a fixed ordering period of materials, which represents the criticality value of materials for prioritizing their procurement
a shortcoming in the model because construction projects involve and did not attempt to develop a procurement schedule of mate-
uncertainties due to delay in delivery, material shortages in the local rials by optimizing their procurement costs and shortage impact.
market, uncertainties in the construction schedule, and unfavorable Moreover, it did not emphasize the storage capacity of materials
weather conditions. in their procurement. However, the present study extends the re-
Tabrizi and Ghaderi (2016a) resolved the issue of simultaneous search conducted by Kar and Jha (2020a) to develop the opti-
planning of the project scheduling and procurement of materials by mized material procurement schedule and used the MC values
developing a mixed-integer programming mathematical model. of materials reported by them to assess the shortage impact of
The model considered the uncertainty issues in activities duration materials.
and execution costs. Furthermore, Tabrizi and Ghaderi (2016b) ex-
tended their study by accounting for the discount on material prices Need for a New Approach for Material Procurement
offered by suppliers in the procurement of materials.
Zoraghi et al. (2017) also examined a problem in the scheduling Georgy and Basily (2008) stated that the EOQ model, which as-
of material ordering that considered the discount on material prices sumes a constant delivery of materials and is typically used in
based on the size of the orders, using renewable as well as non- manufacturing projects, is not suitable for construction projects
renewable resources. They found that NSGA-II performed well because demand may not remain constant. Other material procure-
to solve this problem. Furthermore, Habibi et al. (2019) developed ment approaches such as JIT, fixed order interval system, and fixed
a mathematical model to schedule material ordering in construction order quantity system are inadequate in construction (Polat and
considering these procurement costs and sustainability goals. The Arditi 2005; Georgy and Basily 2008). So, a specific material pro-
model could determine the optimized material ordering time and curement strategy based on the construction schedule is required
quantity. However, the models developed by Tabrizi and Ghaderi for construction projects. However, few studies examined the de-
(2016a, b), Zoraghi et al. (2017), and Habibi et al. (2019) did not velopment of a material procurement schedule that incorporates the
consider the budget constraint situation or maximum material stor- construction schedule into the budget constraints of a construction
age capacity. project to optimize material procurement costs and limit the impact
of unavailability of materials. Such studies that also consider all the
construction materials at once have rarely been performed. Further-
Budget Consideration in Material Procurement more, past studies did not emphasize the storage capacity of ma-
terials in material procurement. Therefore, an optimized material
Kar and Jha (2020a) described the importance of the incorporation
procurement schedule should be developed that considers these as-
of budget in material procurement. They stated that in a construc-
pects to make the most realistic and cost-effective schedule in the
tion project, a limited budget is allotted for material procurement
construction context.
for a certain period, which restricts the procurement of all materials
at the same time. Therefore, a material procurement policy should
consider the available material budget. Budget constraints material Development of an Optimum Material Procurement
management were considered by Dixit et al. (2013, 2014). They Schedule
attempted to integrate material management with project manage-
ment in the manufacturing of complex products, noting that when What, how much, and when to order represents an important
facing budget constraints, managers must prioritize certain materi- decision in construction material management. Thus, the develop-
als in procurement. They should be prioritized based on their criti- ment of a MPS is crucial for a construction project. The method-
cality. This was defined as a relative quantitative measure of the ology used for developing the MPS is shown in Fig. 1. The
time and cost overrun due to that material’s unavailability. Further- methodology shown was implemented using a computer program
more, Dixit et al. (2014) developed a procurement scheduling developed in MATLAB 2017a. Initially, an activity schedule for a
model under budget constraints by considering fuzzy holding costs construction project was prepared based on the activity duration

© ASCE 04021017-3 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2021, 147(4): 04021017

 Activity schedule Furthermore, an activity may require several materials, and one
single material may be required for different activities. Thus, the
Quantity of materials required for required quantity and required date of each material for each ac-
Prepare MRP
the activities tivity should be determined separately in the MRP.

Unit price, ordering cost per order,

storage cost per unit quantity of
Development of the Optimization Model for Generating
Development of the
material, interest rate, criticality optimization model using
the Material Delivery Schedule
values of materials, minimum NSGA-II
order quantity, standard shipping
After preparing the MRP, an optimized material delivery schedule
quantity, budget was developed based on the MRP by optimizing material procure-
ment cost and impact due to the unavailability of materials as dis-
Development of optimized cussed next.
material delivery schedule
Material Procurement Cost
Downloaded from ascelibrary.org by Georgia Tech Library on 02/19/21. Copyright ASCE. For personal use only; all rights reserved.

The material management process in a construction project con-

Development of optimized
Delivery lead time material procurement schedule
sists of different functions such as material planning, procurement,
transportation, receiving, warehousing, and issuing. Typically,
these functions involve four categories of costs: (1) purchasing
Fig. 1. Flowchart for development of MPS.
cost, (2) ordering cost, (3) storage cost, and (4) capital cost.
Purchasing Cost. The purchasing cost of a material equals the pur-
chasing price from a supplier. It depends on the unit price of the
material and the order quantity as shown in Eq. (1)
and interrelationships. A MATLAB program was formulated to in-
corporate all four types of relationships between the activities: fin- X
M X
D

ish–start, finish–finish, start–start, and start–finish. Afterward, a Purchasing cost ¼ Qtj × Puj ð1Þ
j¼1 t¼1
MRP was prepared based on the activity schedule and the material
required quantity for each activity. An optimization model mini- where Qtj = quantity delivered on the tth day for the jth material;
mizing the procurement cost and the shortage impact of materials Puj = unit price of the jth material; M = total number of materials;
was then developed to generate an optimum material delivery and D = project duration.
schedule based on the MRP, available budget, and material costs. Ordering Cost. The ordering cost of a material represents the ad-
The optimization was performed with NSGA-II. Finally, the MPS ministrative expense involved in issuing a purchase order (PO) to a
was developed based on the material delivery schedule and lead supplier. It includes requisition, delivery, receiving, and auditing
time of materials. The details of the steps followed for developing costs (Georgy and Basily 2008; Said and El-Rayes 2011). The or-
the MPS are described in the following sections. dering cost of a material increases with the increase in the number
of orders, as shown in Eq. (2). Ordering cost is also an important
cost component of a material management model as emphasized by
Preparation of MRP
Dodin and Elimam (2001), Georgy and Basily (2008), Tabrizi and
The MRP consisted of the required dates and quantities at the site Ghaderi (2016a), Zoraghi et al. (2017), and Habibi et al. (2019)
for the materials for execution of the construction activities. The
authors proposed the following steps for preparing the MRP: X
M
Ordering cost ¼ N j × Coj ð2Þ
• Step 1: Materials were identified for the necessary construction j¼1
activities.
• Step 2: The quantity of materials required for each activity was where N j = number of orders for the jth material; and Coj = order-
estimated according to the drawings and specifications. An ing cost for a single order for the jth material.
acceptable extra percentage of material quantity was added to Storage Cost. The storage cost of a material represents the cost of
this estimated quantity to incorporate material wastage. warehousing and handling the material. It depends on the storage
• Step 3: To start an activity, all the materials required for that quantity of a material and storage cost per unit quantity, as shown in
particular activity were assumed to be made available before Eq. (3). The storage quantity of a material at a particular time can be
the activity is begun. The required date that the materials would calculated by subtracting the total material requirement up to that
arrive at the site was determined as the earliest start date of the time from the total material received until that time. Several re-
associated activity based on the construction schedule, and the searchers emphasized the storage cost as an essential cost compo-
required quantity of the materials was defined as the quantity for nent of a material management model (Dodin and Elimam 2001;
that activity obtained from Step 2. A similar assumption was Georgy and Basily 2008; Dixit et al. 2014; Tabrizi and Ghaderi
also made by Shtub (1988) to determine the required date 2016a; Zoraghi et al. 2017; Habibi et al. 2019)
and quantity of materials when developing the MRP.
Notably, larger activities should be broken down into smaller X
M X
D
Storage cost ¼ SQtj × Csj ð3Þ
sub-activities that are manageable at the lowest level. For example, j¼1 t¼1
the overall concreting work can be divided into subactivities like
foundation concreting, column concreting, and beam and slab where SQtj = storage quantity for the jth material at time t; and
concreting. Similarly, brickwork can be divided into Brickwork Csj = storage cost of the jth material per unit quantity.
phase 1, Brickwork phase 2, Brickwork phase 3, and so on. Thus, Capital Cost. Capital cost of a material equals the cost due to the
for this study, activity was defined as the smallest portion of a funds invested for the storage of the materials that would be later
project for which all the materials must be available before being used. However, these funds could be used for other purposes. Thus,
executed. the capital cost depends on storage quantity, the average unit price

© ASCE 04021017-4 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2021, 147(4): 04021017

 of the material, and the interest rate, as shown in Eq. (4). The im- Kar and Jha (2020a) developed a systematic approach for measur-
portance of capital cost in a material management model was also ing the impact of the unavailability of materials. They suggested
highlighted by Dodin and Elimam (2001), Georgy and Basily using the TC value of material for this purpose. The TC value
(2008), and Zoraghi et al. (2017) of a material was determined by multiplying its MC value with its
AC value. An AC value depends on the float available to the ac-
X
M X
D
tivities. A material can be used for different activities; some are
Capital cost ¼ SQtj × I × Puj ð4Þ
j¼1 t¼1
critical, whereas others are not. Thus, even for a particular material,
AC values may differ from activity to activity, which results in vary-
where SQtj = storage quantity of the jth material at time t; and I = ing TC values for the different activities. Furthermore, an activity
interest rate per unit time. may require several materials with different MC values. Thus, even
This study assumes a fixed unit price of material over the project for a particular activity, the TC values of the materials may not be
duration. Furthermore, the total required quantity of a material is the same.
fixed. Thus, the total purchasing cost does not depend on the de- To obtain the total shortage impact due to a material, TC values
livery times and quantity of materials in different delivery times. for that material must be summed for the duration in which the
Downloaded from ascelibrary.org by Georgia Tech Library on 02/19/21. Copyright ASCE. For personal use only; all rights reserved.

Therefore, purchasing cost was not optimized in the proposed material shortage exists. However, no shortage impact will result
model. if the material is available necessary. As an example, consider a
A material may be required for different activities at different shortage of cement for a total of three days for three different ac-
times. Procuring materials before the start of each activity just as tivities (i.e., shortage of 1 day for each of the activities) in a project
needed (i.e., following the JIT approach) may increase ordering with a duration of D days. Considering that these three activities
cost while also reducing storage and capital costs (Dodin and demonstrated a different total float that resulted in the varying TC
Elimam 2001). Said and El-Rayes (2011) noted that procuring ma- values of cement of 0.4, 0.2, and 0.3, the shortage impact for ce-
terials just as needed for each activity may result in small-order ment would be calculated as 0.9 (¼ 0.4 þ 0.2 þ 0.3).
quantities that lead to high delivery costs because the trucks re- To identify the shortage of a material for a particular activity, the
quired to deliver them are underutilized, with loads below their activitywise delivery quantity of material was deducted from the
maximum capacities. Procurement of the total material quantity activitywise required quantity of material. This was calculated
in a single order or few orders with large quantities would reduce for each period for each material considered in the optimization
the ordering cost; however, it would increase the storage and capital model. Georgy and Basily (2008), Dixit et al. (2013, 2014), and
costs. Conversely, procuring the material with frequent orders and Habibi et al. (2019) also emphasized the importance of including
small quantities would reduce the storage and capital costs; how- the shortage impact of materials in the procurement plan.
ever, it would increase the ordering cost (Georgy and Basily 2008; Thus, the Objective function 2 for the optimization model was
Tabrizi and Ghaderi 2016a; Habibi et al. 2019). Habibi et al. (2019) set as follows.
stated that ordering materials in smaller quantities would cause a Objective 2: Minimize shortage impact
higher probability of delay in individual activities leading to the
postponement of project completion. Therefore, an optimal balance X
M X
N X
D X
t X
t

of ordering, storage, and capital costs are needed for a cost-effective Minimize TCja ; if Qtja < TQja ð6Þ
j¼1 a¼1 t¼1 t¼1 t¼1
material procurement schedule.
Thus, the Objective function 1 for this study was set as follows.
where TCja = total criticality of the jth material for the ath activity
Objective 1: Minimize procurement cost
(i.e., the impact of the unavailability of the jth material for the ath
X M XM X D activity); Qtja = quantity delivered at the tth day for the jth material
Minimize N j × Coj þ SQtj × Csj for the ath activity; TQja = total quantity required for the jth
j¼1 j¼1 t¼1 material for the ath activity; and N = total number of activities.
X
M X
D  The MC, AC, and TC values of the materials were calculated
þ SQtj × I × Puj ð5Þ following the method illustrated by Kar and Jha (2020a). Further,
j¼1 t¼1 constraints were formulated for the optimization model as follows.

Constraints
Impact of the Shortage of Materials The availability of budgets for materials at different periods was
The budget constraints of a particular construction project may re- considered the constraint in the optimization model. Thus, Eq. (7)
strict the ability to procure all the materials when required, which should be satisfied in the development of the material delivery
may cause a material shortage. This shortage may result in time and schedule
cost overruns as well as loss of productivity. However, each mate-
rial’s shortage may not result in the same impact. This depends on X
M X
t X
t
many factors, including the criticality level of the activity for which Qtj × Puj ≤ Bt ð7Þ
the material is needed, the percentage contribution of the material j¼1 t¼1 t¼1
to the activity, volatility in the price of the material, and so on, as
explained by Kar and Jha (2020a). A procurement or project man- where Bt = available budget for materials at time t.
ager intends to reduce this shortage impact. Therefore, the impact Furthermore, the total required quantities for the materials
of the shortage of materials should be minimized when preparing should be procured to ensure the execution of the activities for
the MPS. which the materials are required. Moreover, the total quantities pro-
The impact of the shortage of materials is difficult to measure in cured for the materials should not exceed their required total quan-
a construction project due to the complexities and interdependen- tities for the project to avoid a surplus of materials. This will also
cies of the activities. Furthermore, an activity may require several reduce the material wastages in the project and will help achieve the
materials, which increases the difficulty of the assessment of global aim of sustainability. Thus, the material procurement
the impact of unavailability for an individual material. However, schedule should satisfy Eq. (8)

© ASCE 04021017-5 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2021, 147(4): 04021017

 T=1 T=2 T=3 T=4 T=5 T=6 .. T=D T=1 T=2 T=3 .. T=D .. T=1 T=2 .. .. .. T=D

Q 11 Q 21 Q 31 Q 41 Q 51 Q 61 .. Q D1 Q 12 Q 22 Q 32 .. Q D2 .. Q 1M Q 2M .. .. .. Q DM

Genes

Fig. 2. Chromosome representing materials deliveries.

X
D Optimization Process Using NSGA-II for Developing the
Qtj ¼ TQj ; j ¼ 1; 2; : : : ; M ð8Þ Material Delivery Schedule
t¼1 In this study, NSGA-II, which was developed by Deb et al. (2002),
was used to generate the optimum material delivery schedule by
where TQj = total quantity required for the jth material in the
minimizing the procurement cost and the impact of material short-
Downloaded from ascelibrary.org by Georgia Tech Library on 02/19/21. Copyright ASCE. For personal use only; all rights reserved.

project.
The order quantity at a time for the material should at least equal age. This material scheduling problem involved many possible so-
or exceed its minimum order quantity. Furthermore, quantity lutions because the delivery date and order quantity for a particular
should be ordered in the multiple of the standard shipping size material might result in many combinations (i.e., delivery of the
of the material to make the material schedule cost-efficient. This material is possible on any day within the project duration with
will also reduce carbon emissions due to the transportation of any quantity from the total required quantity). Moreover, several
materials and will improve the sustainability of material procure- materials are required to be procured for a construction project.
ment. Thus, order quantity at a time for the material should fol- Thus, the solution space increases further. Moreover, no specific
low Eqs. (9)–(11): order quantity or number of orders were known beforehand. There-
fore, using a classical optimization technique would prove
Qtj ≥ MQj ð9Þ challenging.
This material scheduling problem qualifies as a combinatorial
If (TQj =SSj ) is an integer optimization problem (Paquete and Stützle 2018). Thus NSGA-II,
an evolutionary algorithm, was used in this study to perform the
Qtj ¼ R × SSj ð10Þ optimization. This optimization model involved two objective func-
tions. The presence of multiple objectives resulted in a set of opti-
If (TQj =SSj ) is not an integer mal solutions (known as Pareto-optimal solutions), instead of a
 single optimal solution.
R × SSj for upto ðN j − 1Þth order
Qtj ¼ ð11Þ In the NSGA-II, each solution was represented as a chromo-
Qb for N j th order some of size equal to the total number of decision variables. In this
optimization problem, the total number of decision variables is
N
X j −1 equal to the product of the project duration and the total number
Qb ¼ TQj − ðR × SSj Þ; j ¼ 1; 2; : : : ; M of materials considered in the optimization problem. A chromo-
1 some consisted of a certain number of genes, where genes con-
tained information on decision variables (i.e., delivery quantity
where MQj and SSj = minimum order quantity and standard ship-
at a specific period for specific material). A group of chromosomes
ping quantity, respectively, for the jth material; R = any random
integer number; and Qb = balanced quantity after (N j − 1)th or- represented the population. An example depicting chromosome and
ders. Standard shipping quantity equals the full truckload capacity genes are shown in Fig. 2. In this figure, the gene’s Qtj value rep-
for a material. R was represented by an integer because the order resents the quantity on the tth time for the jth material. A zero value
quantity was considered a multiple of the standard shipping quan- for a gene indicates that no delivery is made at that time.
tity. For example, if one truckload capacity of reinforcement is 4 t, The steps followed in the NSGA-II are briefly as follows (Deb
then it will be delivered as 4, 8, 12, and 16 t, and so on. If the et al. 2002; Zahraie and Tavakolan 2009):
reinforcement requirements for two activities are 3.5 and 4.5 t, re- 1. Initially, a random parent population (P0 ) with size N was
spectively, then as per the model, reinforcement will either be pro- created.
cured in a single trip with 8 t or in two trips with 4 t each time, not in 2. Based on nondomination, the population was sorted.
a partially loaded truck. This assumption is also realistic as suppli- 3. Each nondominated solution was assigned a fitness (or rank)
ers prefer to supply materials in a full truckload. Furthermore, the equal to its nondomination level (one is the best level, two is
delivery of materials in a partially loaded truck will increase the the next best level, and so on).
transportation charges of the materials. 4. A child population (Q0 ) with size N was created using binary
Typically, in a construction project, the storage area is limited. tournament selection, recombination, and mutation.
Therefore, maximum storage capacity should be considered when 5. After creating the initial generation, further generations were
procuring materials. The storage quantity of materials at any time created using the following steps:
should not exceed their maximum storage capacity. The procure- a. The parent population (Pt ) and the child population (Qt ) were
ment schedule should follow Eq. (12): combined to form a combined population (Rt ) of
size 2N.
SQtj ≤ SCj ; j ¼ 1; 2; : : : ; M; t ¼ 1; 2; : : : ; D ð12Þ b. The combined population (Rt ) was sorted based on fast
nondominated sorting to identify the nondominated fronts
where SQtj = storage quantity for the jth material at time t; and (F1 ; F2 ; : : : ).
SCj = maximum storage capacity for the jth material. c. A new parent generation (Ptþ1 ) of size N was created by add-
After setting the objective functions and constraints, the optimi- ing nondominated solutions. For this, the solutions in the first
zation was performed using NSGA-II as discussed next. ranked front (F1 ) were selected first followed by the

© ASCE 04021017-6 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2021, 147(4): 04021017

 Nondominated Crowding
sorting distance sorting
Selection
F1 +
P0 /Pt Crossover
+
F2 Mutation

F3
Pt + 1 Qt + 1
P0 Q0
F4
Q0 /Qt Rejected
Downloaded from ascelibrary.org by Georgia Tech Library on 02/19/21. Copyright ASCE. For personal use only; all rights reserved.

R 0 /Rt

Fig. 3. NSGA-II algorithm.

subsequently ranked fronts, that is, F2 , F3 , and so on. To se- obtained from the past record of the material delivery or based on
lect exactly N solutions, the solutions in the last accommo- the experience of the procurement manager.
dated front were sorted based on the crowding distance
assigned to each solution of the last front. Some lower-ranked
nondominated solutions may have to be rejected from the last Illustration of the Optimization Model through a
front for selecting exactly N solutions (Fig. 3). Case-Study Project
d. A new child population (Qtþ1 ) of size N was created from the
parent solution (Ptþ1 ) by performing selection, crossover, and Description of Case-Study Project and Specific
mutation. Materials
6. Step 5 was repeated until the maximum number of generations
The construction of a 4-story hostel building project located at
was reached.
Delhi, India, was selected as the case-study project to illustrate
By performing the optimization process using NSGA-II, the op-
the results of the MPS. The building contained 670 m2 of total
timized delivery schedules of materials were produced. These
built-up area. The construction lasted 36 weeks. The major con-
schedules maintain procurement cost and shortage impact at the
struction works of this project consisted of (1) reinforced cement
lowest possible levels.
concrete (RCC), (2) brickwork, (3) plastering, and (4) flooring. The
most commonly used materials for these activities included (1) ce-
Development of MPS ment, (2) sand, (3) aggregates, (4) reinforcement, and (5) tiles.
These works were executed in 22 weeks, whereas other construc-
After obtaining the material delivery schedule, the MPS was devel- tion activities such as surveying, excavation, painting, electrical fit-
oped by assuming that materials are delivered to the site according tings, and plumbing took the remaining 14 weeks.
to the PO. In the MPS, the procurement dates (i.e., PO dates) were
determined by deducting the lead time of the respective materials
from their delivery dates and considering procurement quantities Applying the Optimization Model
(i.e., PO quantities) as the delivery quantities. A material’s lead The first step in the MPS development consisted of the preparation
time indicates the amount of time required to deliver the material of the MRP. The MRP was prepared based on the construction
to the site after the PO is placed with the supplier. Lead time can be schedule of the project, which is shown in Fig. 4. The earliest start

0 RCC 4 FS0 4 RCC 8 FS0 8 Brickwork 11 15 Flooring 18 FS0 19 Flooring 22

0 Ph I 0 0 Ph II 0 0 Ph II 0 1 Ph I 1 0 Ph II 0
0 4 4 4 4 8 8 3 11 16 3 19 19 3 22
FS0 FS0 FS0
FS0
4 Brickwork 7 FS0 11 Plastering 15 FS0 15 Plastering 19
4 Ph I 4 0 Ph I 0 0 Ph II 0
8 3 11 11 4 15 15 4 19

EST EFT
Activity
Legend TF TF
LST Duration LFT

Fig. 4. Schedule of activities.

© ASCE 04021017-7 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2021, 147(4): 04021017

 © ASCE

Table 1. Material input data for the case-study project

Material Material Unit Minimum Standard Maximum Ordering Storage
Material Material Associated required required price order shipping storage cost (Coj ) cost (Csj )
code description activity date (week) quantity MC AC TC (USD) quantity quantity capacity (USD) (USD)
1 Cement (bags) RCC Phase I 1 975 0.457 1 0.457 4.3 200 200 1,200 26.0 0.2
Downloaded from ascelibrary.org by Georgia Tech Library on 02/19/21. Copyright ASCE. For personal use only; all rights reserved.

2 Sand (tons) RCC Phase I 1 55 0.238 1 0.238 12.7 50 50 400 21.3 0.6
3 Aggregate (tons) RCC Phase I 1 105 0.323 1 0.323 11.2 25 25 150 12.0 0.6
4 Reinforcement (tons) RCC Phase I 1 19 0.671 1 0.671 600.0 4 4 24 56.0 30.0
1 Cement (bags) RCC Phase II 5 975 0.457 1 0.457 4.3 200 200 1,200 26.0 0.2
2 Sand (tons) RCC Phase II 5 55 0.238 1 0.238 12.7 50 50 400 21.3 0.6
3 Aggregate (tons) RCC Phase II 5 105 0.323 1 0.323 11.2 25 25 150 12.0 0.6
4 Reinforcement (tons) RCC Phase II 5 19 0.671 1 0.671 600.0 4 4 24 56.0 30.0
1 Cement (bags) Brickwork Phase I 5 325 0.457 0.2 0.091 4.3 200 200 1,200 26.0 0.2
2 Sand (tons) Brickwork Phase I 5 98 0.238 0.2 0.048 12.7 50 50 400 21.3 0.6
1 Cement (bags) Brickwork Phase II 9 325 0.457 1 0.457 4.3 200 200 1,200 26.0 0.2
2 Sand (tons) Brickwork Phase II 9 98 0.238 1 0.238 12.7 50 50 400 21.3 0.6
1 Cement (bags) Plastering Phase I 12 580 0.457 1 0.457 4.3 200 200 1,200 26.0 0.2
2 Sand (tons) Plastering Phase I 12 120 0.238 1 0.238 12.7 50 50 400 21.3 0.6
1 Cement (bags) Plastering Phase II 16 580 0.457 1 0.457 4.3 200 200 1,200 26.0 0.2
2 Sand (tons) Plastering Phase II 16 120 0.238 1 0.238 12.7 50 50 400 21.3 0.6
1 Cement (bags) Flooring Phase I 16 210 0.457 0.8 0.366 4.3 200 200 1,200 26.0 0.2
2 Sand (tons) Flooring Phase I 16 310 0.238 0.8 0.190 12.7 50 50 400 21.3 0.6
5 Tiles (m2 ) Flooring Phase I 16 290 0.478 0.8 0.382 14.7 50 50 400 42.7 0.7
1 Cement (bags) Flooring Phase II 20 210 0.457 1 0.457 4.3 200 200 1,200 26.0 0.2
04021017-8

2 Sand (tons) Flooring Phase II 20 310 0.238 1 0.238 12.7 50 50 400 21.3 0.6
5 Tiles (m2 ) Flooring Phase II 20 290 0.478 1 0.478 14.7 50 50 400 42.7 0.7
J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2021, 147(4): 04021017

 66.6
66.3
22
time (EST), latest start time (LST), earliest finish time (EFT), latest

0
0
0
0
0
finish time (LFT), total float (TF), and duration in weeks of each

66.6
66.3
activity are shown within the box for the activity (Fig. 4). Critical

21
0
0
0
0
0
activities are highlighted in gray (Fig. 4).

66.6
66.3
A MATLAB program was formulated to develop the construc-

20
0
166
0
0
80
tion schedule incorporating all four types of relationships. How-
ever, all the relationships in the case-study project consisted of

66.6
63.0
19
finish–start (FS) relationship with zero lead time. Each of the ac-

0
0
0
0
200
tivities was performed in two phases: Phase I and Phase II. Each

62.6
60.1
phase represents the work of two floors. Materials like cement and

18
0
100
0
0
0
sand were required for several activities such as RCC, brickwork,
plastering, and flooring. Moreover, each of the activities, including

62.6
58.8
17
RCC, brickwork, plastering, and flooring, required more than one

380
100
0
0
0
material. Thus, the required quantity and required date at the site of
Downloaded from ascelibrary.org by Georgia Tech Library on 02/19/21. Copyright ASCE. For personal use only; all rights reserved.

a specific material was estimated for each activity separately in the

62.6
55.9
16
600
50
0
0
150
preparation of the MRP, as presented in Table 1. Each material was
labeled with a specific material code.

56.0
50.5
The inputs for developing the optimized material delivery

15
0
0
0
0
100
schedule for this case-study project such as unit price (Puj ), order-

Material delivery quantity in different periods

ing cost (Coj ), storage cost (Csj ), minimum order quantity (MQj ),

56.0
49.0
14
0
50
0
0
0
standard shipping quantity (SSj ), and maximum storage capacity
(SCj ) of materials were obtained from the planning manager

56.0
48.4
13
0
100
0
0
50
and procurement manager involved in this project, as indicated
in Table 1. The weekly interest rate (I) for this optimization prob-

50.6
46.4
lem was set at 0.3%. To determine the TC values of materials, their

12
0
100
0
0
0
MC values were taken from Kar and Jha (2020a), and AC values
were calculated using the formula suggested by Kar and Jha

50.6
45.1
11
200
150
0
0
0
(2020a) based on the float of the relevant activities. The TC values
of the materials were calculated by multiplying their MC values

50.6
42.4
and AC values. The MC, AC, and TC values of the selected ma-
10
0
0
0
0
0
terials for different activities are provided in Table 1. The TC values
were used in the optimization model to determine the impact of

44.0
42.4
600
50
0
0
0
9

material shortages as previously described. The amount of money

available for procuring all five materials for the case project was
44.0
39.2
USD 26,666.70, USD 10,666.70, USD 6,666.70, USD 6,666.70,
0
150
0
0
0
8

USD 5,333.30, USD 6,666.70, and USD 4,000 in Weeks 1, 4,

7, 10, 13, 16, and 19, respectively, as obtained from the planning
44.0
37.3
7
0
0
0
0
0
Table 2. Material delivery schedule generated from the proposed optimization model

manager.
Further, the MRP aided in the development of the delivery sched-
37.3
37.3
6
0
0
0
0
0

ule for the selected materials for the case-study project by optimizing
procurement costs and minimizing the impact of material shortage.
37.3
37.3
0
60
18
0
1,000

The proposed optimization model was implemented using MATLAB

5

R2017a. The available budget, total required quantity, minimum

order quantity, standard shipping quantity, and maximum storage
37.3
21.6

capacity were implemented as the constraints for the model. The

400
0
0
0
0
4

parameters for the NSGA-II algorithm, such as population size and

maximum number of generations, were varied in several trials to ob-
26.6
19.8
3
0
0
0
0
0

tain the best possible combination. After several trials, the population
size and maximum number of generations were set to 1,000 and 250,
26.6
19.8
2
0
0
50
0
0

respectively.
NSGA-II generated a set of Pareto-optimal solutions. The pres-
26.6
19.3
1,000
150
100
20
0

ence of multiple objectives in an optimization problem provides a

1

set of optimal solutions instead of a single optimal solution. These

solutions are known as Pareto-optimal solutions. Because the opti-
Cumulative budget (USD × 103 )

mization problem was a minimization problem involving two ob-

jectives, a solution on the Pareto front that was closest to the origin
was selected to compare with the actual delivery schedule followed
Cumulative purchasing

in the case project. The optimized material delivery schedule for the
Reinforcement (tons)
Material description

materials considered for the case-study project is given in Table 2.

cost (USD × 103 )
Aggregate (tons)

The optimized delivery schedule satisfied the minimum order

Cement (bags)

quantity and standard shipping quantity constraints (Table 2).

Sand (tons)

Tiles (m2 )

Moreover, if the balance required quantity after an order was less

than the minimum order quantity, then the balance quantity was
accumulated in that order. The actual material delivery schedule

© ASCE 04021017-9 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2021, 147(4): 04021017

 Table 3. Actual material delivery schedule
Material delivery quantity in different periods
Material description 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Cement (bags) 800 0 400 0 400 0 400 400 0 400 400 0 400 200 0 200 0 0 0 180 0 0
Sand (tons) 150 0 100 0 100 0 0 100 0 100 50 0 0 100 0 150 0 0 150 0 166 0
Aggregate (tons) 100 50 0 60 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Reinforcement (tons) 8 12 0 8 4 0 6 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Tiles (m2 ) 0 0 0 0 0 0 0 0 0 0 0 100 0 100 0 100 100 100 0 0 80 0

Table 4. Material storage quantity based on the optimized delivery schedule

Material storage quantity in different periods
Downloaded from ascelibrary.org by Georgia Tech Library on 02/19/21. Copyright ASCE. For personal use only; all rights reserved.

Material description 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Cement (bags) 25 25 25 425 125 125 125 125 400 400 600 20 20 20 20 0 210 210 210 0 0 0
Sand (tons) 95 95 95 95 0 0 0 92 44 44 194 174 274 324 324 0 44 144 144 0 0 0
Aggregate (tons) 0 45 45 45 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Reinforcement (tons) 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
Tiles (m2 ) 0 0 0 0 0 0 0 0 0 0 0 0 50 50 150 10 10 10 210 0 0 0

obtained from the procurement manager is provided in Table 3. The reduction in shortage impact. These results indicate that the adoption
storage quantity of the materials in different periods based on the of the optimized schedule in the case project could improve avail-
optimized delivery schedule is provided in Table 4. ability of required materials.
After developing the material delivery schedule, the material Furthermore, comparison of the material demand (i.e., MRP
procurement schedule was prepared by considering delivery lead schedule), optimized delivery quantity, and actual delivery quantity
time of materials. The lead time of materials was obtained from the are shown in Fig. 5. The comparison between the proposed model
procurement manager in the case-study project. Cement, reinforce- and actual schedule revealed that material demands were less often
ment, sand, and aggregate required a 1-week lead time, whereas met in the latter than in the former (Fig. 5). This result produced a
tiles needed 2 weeks. The PO dates were determined by deducting higher shortage impact in the actual delivery schedule than the pro-
lead time from the delivery dates. The delivery quantities were set posed one. In the optimized delivery schedule, the total amounts of
as the PO quantities. For example, procurement schedule for tiles cement, sand, aggregate, reinforcement, and tiles required were
included order quantity of 50, 100, 150, 200, and 80 m2 for Weeks procured in 7, 11, 3, 2, and 5 orders, respectively. Conversely,
11, 13, 14, 17, and 18, respectively. in the actual delivery schedule, these five materials were procured
in 11, 10, 3, 5, and 6 orders, respectively. This difference resulted in
Comparison with Actual Material Delivery Schedule lower ordering costs in the optimized model. Furthermore, the
optimized schedule resulted in smaller stock quantities. Over the
The cost data and shortage impact for the optimized schedule gen- 22-week period, the stored quantity (SQtj ) of cement, sand, aggre-
erated from the proposed model as well as the actual delivery gate, reinforcement, and tiles totaled 3,100 bags, 2,182 t, 135 t, 4 t,
schedule followed in the case project are provided in Table 5. The and 490 m2 , respectively, in the optimized model, which compared
ordering cost, storage cost, and capital cost fell by 27.38%, 32.17%, favorably to the 3,175 bags, 2,970 t, 195 t, 11 t, and 1,140 m2 that
and 32.17%, respectively, in the optimized schedule compared with resulted from the actual delivery schedule. This resulted in lower
those in the actual schedule. The proposed model resulted in an over-
storage and capital costs in the optimized delivery schedule.
all procurement cost of USD 3,534.30, which favorably compared
with the procurement cost of USD 5,134.80 incurred from the actual
schedule. These differences resulted in a cost saving of about 31.17%
using the proposed model. Further, the proposed model produced a Discussion
shortage impact of 1.022, whereas the actual schedule generated an
In comparison with the actual material delivery schedule, the opti-
impact of 6.182, meaning the proposed model resulted in an 83.47%
mized material delivery schedule produced a significant cost saving
and reduction in the shortage impact (Table 5). The amount of total
orders fulfilled but did not exceed the total requirement, thus avoid-
Table 5. Comparison between optimized schedule and actual schedule ing a surplus of materials. Avoiding material surplus reduces cost of
Obtained from Obtained from transporting said excess materials to other projects. Furthermore,
Description actual delivery optimized delivery surplus materials often become deteriorated or damaged due to a
costs/shortage impact schedule schedule long storage at the project site, which ultimately results in wastage.
Ordering cost (USD) 1,071.3 778.0 Thus, avoiding such a surplus would help to improve an organiza-
Storage cost (USD) 3,833.5 2,600.3 tion’s approach to sustainability.
Capital cost (USD) 230.0 156.0 Furthermore, the optimized schedule adhered to the cumulative
Total procurement cost 5,134.8 3,534.3 budget availability because cumulative material purchasing cost did
(Objective function 1) (USD) not exceed the cumulative budget at any time in the project dura-
Shortage impact (Objective 6.182 1.022 tion, as can be observed from Table 2. Adhering to the budget re-
function 2)
duces the delay in payment to the suppliers, a circumstance often

© ASCE 04021017-10 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2021, 147(4): 04021017

 Cumulative quantity of cement (bags)

Cumulative quantity of aggregate (tons)

Cumulative quantity of sand (tons)
(a) Duration (weeks) (b) Duration (weeks) (c) Duration (weeks)
Downloaded from ascelibrary.org by Georgia Tech Library on 02/19/21. Copyright ASCE. For personal use only; all rights reserved.

Cumulative quantity of tiles (sqm)

Cumulative quantity of
reinforcement (tons)

(d) Duration (weeks) (e) Duration (weeks)

Legend Material demand Optimized delivery quantity Actual delivery quantity

Fig. 5. Comparison among material demand, optimized delivery quantity, and actual delivery quantity for (a) cement; (b) sand; (c) aggregate;
(d) reinforcement; and (e) tiles.

experienced in construction projects. This would therefore result in Phase I had a float of 4 weeks. Therefore, the sand required for
a better buyer–supplier relationship in the project. brickwork Phase I displayed the lowest AC and TC values
Cement, sand, aggregate, reinforcement, and tiles had a maxi- (AC ¼ 0.2 and TC ¼ 0.048), so its delivery was delayed by
mum storage capacity of 1,200 bags, 400 t, 150 t, 24 t, and 400 m2 , 3 weeks, as can be seen in Table 6. The complete delivery of sand
respectively. Table 4 demonstrates that the amount of materials for brickwork Phase I, as suggested by the proposed model, oc-
stored in different periods based on the optimized delivery schedule curred in the eighth week (Table 6).
never exceeded maximum storage capacity. Thus, the optimized Furthermore, the TC values were determined based on the MC
schedule conformed to the storage area at the construction site, values of materials. The MC values were assessed based on the
which increased the practicality of the schedule. criteria including percentage contribution, lead time, and volatil-
The proposed model generated an activity-specific material de- ity in price of materials. A material with higher volatility in price
livery schedule to determine which materials would be required for has a higher MC value (Kar and Jha 2020a). The proposed model
each of the associated activities in different time periods, as pro- prioritizes the procurement of more critical materials because this
vided in Table 6. This result aids in the tracking of materials for minimizes the shortage impact, i.e., TC values of materials.
a particular activity and improvement of material control in con- Therefore, the proposed model not only incorporates the procure-
struction. The proposed model considered the penalty cost that ment costs of materials but also prioritizes the procurement of
would result from material unavailability due to construction materials that have a higher volatility in price, such as reinforce-
budget constraints by determining and minimizing the impacts ment bars.
of material shortage. This reduces construction delays and aids The optimized material delivery schedule was generated based
in the project’s successful completion. on the construction schedule and materials required for the activ-
When facing budget constraints, the model resulted in the delay ities by considering the budget available at any given time. This
of the delivery of those materials that lowered the impact of a improves the efficiency of the material management system, which
material shortage (i.e., those with a lower TC value). For example, can in turn aid in completing the project on time and within budget,
materials for brickwork Phase I and RCC Phase II were required in thus achieving project success. The optimized delivery schedule
the fifth week, as can be seen from Table 1. The cumulative was generated using NSGA-II, a novel approach in this area. It ef-
material budget for the fifth week was USD 37,333.30, which fell ficiently handled the complexities of the proposed model, including
USD 160 below the amount of money required to purchase the many decision variables, multiple objectives, and several con-
MRP quantity. Therefore, procuring all the materials required by straints. The time needed to calculate the optimal solutions for
the MRP in the fifth week was infeasible. In this situation, the the case-study project was about 10 min, which seems reasonable
model delayed the materials which demonstrated the lowest TC considering all its advantages. Further, the proposed model consid-
value—that is, the material whose shortage resulted in the lowest ered all materials at once in the development of the optimized
possible overall impact. As can be seen, between these two activ- material delivery schedule, which had rarely occurred in past
ities, RCC Phase II demonstrated no float, whereas brickwork studies.

© ASCE 04021017-11 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2021, 147(4): 04021017

 © ASCE

Table 6. Activitywise material delivery schedule generated from the optimization model
Material delivery quantity in different periods
Material Associated
description activity 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Cement (bags) RCC Phase I 975 975 975 975 975 975 975 975 975 975 975 975 975 975 975 975 975 975 975 975 975 975
Downloaded from ascelibrary.org by Georgia Tech Library on 02/19/21. Copyright ASCE. For personal use only; all rights reserved.

Sand (tons) RCC Phase I 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55

Aggregate (tons) RCC Phase I 100 105 105 105 105 105 105 105 105 105 105 105 105 105 105 105 105 105 105 105 105 105
Reinforcement (tons) RCC Phase I 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19
Cement (bags) RCC Phase II 25 25 25 425 975 975 975 975 975 975 975 975 975 975 975 975 975 975 975 975 975 975
Sand (tons) RCC Phase II 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55
Aggregate (tons) RCC Phase II 0 45 45 45 105 105 105 105 105 105 105 105 105 105 105 105 105 105 105 105 105 105
Reinforcement (tons) RCC Phase II 1 1 1 1 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19
Cement (bags) Brickwork Phase I 0 0 0 0 325 325 325 325 325 325 325 325 325 325 325 325 325 325 325 325 325 325
Sand (tons) Brickwork Phase I 40 40 40 40 40 40 40 98 98 98 98 98 98 98 98 98 98 98 98 98 98 98
Cement (bags) Brickwork Phase II 0 0 0 0 125 125 125 125 325 325 325 325 325 325 325 325 325 325 325 325 325 325
Sand (tons) Brickwork Phase II 0 0 0 0 0 0 0 92 98 98 98 98 98 98 98 98 98 98 98 98 98 98
Cement (bags) Plastering Phase I 0 0 0 0 0 0 0 0 400 400 580 580 580 580 580 580 580 580 580 580 580 580
Sand (tons) Plastering Phase I 0 0 0 0 0 0 0 0 44 44 120 120 120 120 120 120 120 120 120 120 120 120
Cement (bags) Plastering Phase II 0 0 0 0 0 0 0 0 0 0 20 20 20 20 20 580 580 580 580 580 580 580
Sand (tons) Plastering Phase II 0 0 0 0 0 0 0 0 0 0 74 120 120 120 120 120 120 120 120 120 120 120
Cement (bags) Flooring Phase I 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 40 210 210 210 210 210 210
Sand (tons) Flooring Phase I 0 0 0 0 0 0 0 0 0 0 0 54 154 204 204 254 310 310 310 310 310 310
Tiles (m2 ) Flooring Phase I 0 0 0 0 0 0 0 0 0 0 0 0 50 50 150 290 290 290 290 290 290 290
Cement (bags) Flooring Phase II 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 210 210 210 210 210 210
04021017-12

Sand (tons) Flooring Phase II 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 44 144 144 310 310 310

Tiles (m2 ) Flooring Phase II 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 10 10 10 210 290 290 290
J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2021, 147(4): 04021017

 Conclusion schedule is more practically oriented, cost-effective, and specifi-
cally tailored to construction projects. The model considers several
Materials represent the critical resources in a construction project aspects including material budget, shortage impact, and storage
because they account for a major portion of the project cost and capacity, which are not typically incorporated in the EOQ and
contribute significantly to the amount of time required in the project JIT systems of material procurement. Such aspects of material pro-
schedule. Thus, effective material management is very crucial in curement were rarely investigated in the previous literature.
achieving project success. Such management has focused on pro- Moreover, incorporating these aspects along with the optimiza-
curing the correct materials at the best time for the lowest possible tion of procurement costs when developing the material procure-
cost. This fact demonstrates the importance of integrating material ment schedule differentiates this study from the typical time-cost
management with the construction schedule and optimizing the trade-off analysis problems and thus contributes to the body of
procurement cost of materials. knowledge. This study also contributes to the theory by demon-
An optimization model has been developed herein that mini- strating a method for the assessment of the impact of material
mizes the procurement cost as well as the shortage impact of shortage on a construction project, which is less emphasized in lit-
materials and generates the material procurement schedule. An il- erature. The proposed model for material procurement schedule
Downloaded from ascelibrary.org by Georgia Tech Library on 02/19/21. Copyright ASCE. For personal use only; all rights reserved.

lustration of the model by using a case-study project shows that the benefits from the broad capabilities of NSGA-II as an optimization
model can save on procurement costs of materials and reduce the technique. A construction practitioner can use the proposed model
shortage impact of materials significantly, even when the project is as a decision support system for procuring materials when facing
small. Because the model minimizes the shortage impact of mate- budget constraints in a construction project with the lowest cost and
rials, it will reduce delay in construction activity due to the late least shortage, adequate site storage capacity, and without excess or
delivery of materials, leading to a reduction in time overrun in insufficient purchasing. Such a tool could aid in the completion of
the project. The proposed model considers the budget constraints the project on time and within the budget.
in the procurement of materials. Thus, it can avoid the issue of too The time required to reach an optimal solution for the case-study
little money to pay the suppliers. The order quantity of a material project is quite minimal. Although the processing time in a NSGA-II
optimized using the model also satisfies the conditions of minimum algorithm depends on several parameters including number of var-
order quantity and standard shipping size. The model also considers iables (i.e., number of materials and project duration), population
the physical conditions of the site such as storage capacity, thus size, and maximum number of generations, it is largely influenced
improving its practicality. by the latter two aspects. Therefore, it is believed that an optimal
Moreover, the proposed model avoids obtaining an excess of solution could be reached in a reasonable processing time even for a
materials in a construction project, thus improving the sustainabil- more complex project with longer project duration that consisted of
ity of the procurement process. It assumed that to start an activity, several activities and materials.
the materials required for the activity should be available at the be- Although the proposed model can be generalized for any num-
ginning. To achieve the maximum model efficiency, larger activ- ber of materials and activities and any project duration, this study
ities should be broken down into smaller subactivities that are considers only five important and majorly used materials in a small
manageable at the lowest level. Furthermore, the model helps to case project to test the model. Further studies can be conducted
avoid storing large quantities of materials for a long time because with more materials and activities to develop an optimal material
it incorporates the maximum storage capacity and storage cost of procurement schedule. In a future study, the model can consider a
materials. Therefore, the model is applicable in any size project. larger project to assess time and costs savings that would result.
Furthermore, the procurement schedule is developed in this Moreover, this study assumed a fixed unit price of materials. In
study based on the construction schedule. However, the construc- the future, a discount on the unit price for the bulk procurement
tion schedule is changeable due to many different reasons during of materials and distribution of material usage over the activity du-
construction. Thus, if any substantive changes are made in the ration could be incorporated in the proposed model to advance the
schedule during the construction, the model must be redone for model. Furthermore, applicability of the proposed model for the
the remaining activities. In such situation, there are two kinds of procurement of equipment may be investigated in the future.
work: one is works being under construction, and the other is works Finally, delivery lead time is implemented in the model based on
that have not been started. At the same time, within the procure- the experience of the procurement manager; however, it may be
ment schedule, there will be two types of activities: one for which analyzed based on some construction project data, which the au-
materials are already being procured and the other for which ma- thors plan to examine in the future.
terials need to be procured. The revised MRP schedule will only
consist of the balance quantity (i.e., quantity of total materials re-
quired quantity minus the quantity of total materials received up to Data Availability Statement
that date) of materials. The materials’ requirement date in the re-
vised MRP will depend on the revised construction schedule—that Some or all data, models, or codes that support the findings of this
is the date of requirement of the materials for the ongoing activities study are available from the corresponding author upon reasonable
will remain unchanged, and the date of requirement of materials for request.
the activities that have not started will be changed in the MRP if
their start date is changed. Accordingly, based on the revised MRP,
the optimization model will generate a revised optimized procure- References
ment schedule for the materials required for ongoing activities as
Ala-Risku, T., and M. Kärkkäinen. 2006. “Material delivery problems
well as the activities that have not started.
in construction projects: A possible solution.” Int. J. Prod. Econ.
This study demonstrates several practical implications and theo- 104 (1): 19–29. https://doi.org/10.1016/j.ijpe.2004.12.027.
retical contributions. This study addresses an important material Barriga, E. M., J. G. Jeong, M. Hastak, and M. Syal. 2005. “Material re-
strategy that is being less investigated in the literature, which is quirements planning for a manufactured housing facility.” J. Archit.
developing an optimized material procurement schedule based Eng. 11 (3): 91–98. https://doi.org/10.1061/(ASCE)1076-0431(2005)
on the construction schedule. The developed material procurement 11:3(91).

© ASCE 04021017-13 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2021, 147(4): 04021017

 Caldas, C. H., C. L. Menches, P. M. Reyes, L. Navarro, and D. M. Vargas. Kar, S., and K. N. Jha. 2020a. “Assessing criticality of construction
2014. “Materials management practices in the construction industry.” materials for prioritizing their procurement using ANP-TOPSIS.” Int.
Pract. Period. Struct. Des. Constr. 20 (3): 1–8. https://doi.org/10.1061 J. Constr. Manage. 1–11. https://doi.org/10.1080/15623599.2020.17
/(ASCE)SC.1943-5576.0000238. 42637.
CII (Construction Industry Institute). 2020. “Material management: Best Kar, S., and K. N. Jha. 2020b. “Examining the effect of material manage-
practice definition.” Accessed March 5, 2020. https://www.construction ment issues on the schedule and cost performance of construction
-institute.org/resources/knowledgebase/best-practices/materials-manage projects based on a structural equation model: A survey of Indian ex-
ment. periences.” J. Constr. Eng. Manage. 146 (9): 05020011. https://doi.org
Deb, K., A. Pratap, S. Agarwal, and T. Meyarivan. 2002. “A fast and /10.1061/(ASCE)CO.1943-7862.0001906.
elitist multiobjective genetic algorithm: NSGA-II.” IEEE Trans. Evol. Kim, S. B. 2014. “Assessment of CII best practices usage in the construc-
Comput. 6 (2): 182–197. https://doi.org/10.1109/4235.996017.
tion industry.” KSCE J. Civ. Eng. 18 (5): 1228–1238. https://doi.org/10
Dixit, V., R. K. Srivastava, and A. Chaudhuri. 2013. “Integrating materials
.1007/s12205-014-0221-z.
management with project management of complex projects.” J. Adv.
Kulkarni, V., R. Sharma, and M. Hote. 2017. “Factors affecting material
Manage. Res. 10 (2): 230–278. https://doi.org/10.1108/JAMR-05
management on construction site.” Int. Res. J. Eng. Technol. 4 (1):
-2013-0031.
474–478.
Downloaded from ascelibrary.org by Georgia Tech Library on 02/19/21. Copyright ASCE. For personal use only; all rights reserved.

Dixit, V., R. K. Srivastava, and A. Chaudhuri. 2014. “Procurement sched-

uling for complex projects with fuzzy activity durations and lead times.” Navon, R., and O. Berkovich. 2006. “An automated model for materials
Comput. Ind. Eng. 76 (Oct): 401–414. https://doi.org/10.1016/j.cie.2013 management and control.” Constr. Manage. Econ. 24 (6): 635–646.
.12.009. https://doi.org/10.1080/01446190500435671.
Dodin, B., and A. A. Elimam. 2001. “Integrated project scheduling and Paquete, L., and T. Stützle. 2018. “Stochastic local search algorithms for mul-
material planning with variable activity duration and rewards.” IIE Trans. tiobjective combinatorial optimization: A.” In Handbook of approxima-
33 (11): 1005–1018. https://doi.org/10.1023/A:1010994519405. tion algorithms and metaheuristics: Methodologies and traditional
Doloi, H., A. Sawhney, K. C. Iyer, and S. Rentala. 2012. “Analysing factors applications, 1. Boca Raton, FL: CRC Press.
affecting delays in Indian construction projects.” Int. J. Project Manage. Polat, G., and D. Arditi. 2005. “The JIT materials management system in
30 (4): 479–489. https://doi.org/10.1016/j.ijproman.2011.10.004. developing countries.” Constr. Manage. Econ. 23 (7): 697–712. https://
Georgy, M., and S. Y. Basily. 2008. “Using genetic algorithms in optimiz- doi.org/10.1080/01446190500041388.
ing construction material delivery schedules.” Constr. Innovation 8 (1): Said, H., and K. El-Rayes. 2011. “Optimizing material procurement and
23–45. https://doi.org/10.1108/14714170810846503. storage on construction sites.” J. Constr. Eng. Manage. 137 (6):
Guo, C., J. Guo, L. Ji, M. Chen, and L. Tian. 2017. “Development of 421–431. https://doi.org/10.1061/(ASCE)CO.1943-7862.0000307.
bulk material management system and research on material balance Shtub, A. 1988. “The integration of CPM and material management in
applications based on business intelligence.” In Proc., 2017 IEEE 14th project management.” Constr. Manage. Econ. 6 (4): 261–272. https://
Int. Conf. on e-Business Engineering, 31–37. Beijing: China Nuclear doi.org/10.1080/01446198800000023.
Power Engineering Company. https://doi.org/10.1109/ICEBE.2017.15. Tabrizi, B. H., and S. F. Ghaderi. 2016a. “A robust bi-objective model for
Gurmu, A. T. 2020. “Construction materials management practices concurrent planning of project scheduling and material procurement.”
enhancing labour productivity in multi-storey building projects.”
Comput. Ind. Eng. 98 (Aug): 11–29. https://doi.org/10.1016/j.cie.2016
Int. J. Constr. Manage. 20 (1): 77–86. https://doi.org/10.1080/15623599
.05.017.
.2018.1462447.
Tabrizi, B. H., and S. F. Ghaderi. 2016b. “Simultaneous planning of the
Habibi, F., F. Barzinpour, and S. J. Sadjadi. 2019. “A mathematical model
project scheduling and material procurement problem under the pres-
for project scheduling and material ordering problem with sustainability
considerations: A case study in Iran.” Comput. Ind. Eng. 128 (Feb): ence of multiple suppliers.” Eng. Optim. 48 (9): 1474–1490. https://doi
690–710. https://doi.org/10.1016/j.cie.2019.01.007. .org/10.1080/0305215X.2015.1114772.
Ibn-Homaid, N. T. 2002. “A comparative evaluation of construction and Zahraie, B., and M. Tavakolan. 2009. “Stochastic time-cost resource uti-
manufacturing material management.” Int. J. Project Manage. 20 (4): lization optimization using nondominated sorting genetic algorithm and
263–270. https://doi.org/10.1016/S0263-7863(01)00013-8. discrete fuzzy sets.” J. Constr. Eng. Manage. 135 (11): 1162–1171.
Jha, K. N. 2015. “Construction material management.” In Construction https://doi.org/10.1061/(ASCE)CO.1943-7862.0000092.
project management: Theory and practices. 2nd ed. London: Pearson. Zoraghi, N., A. Shahsavar, and S. T. A. Niaki. 2017. “A hybrid project
Jusoh, Z. M., and N. Kasim. 2017. “Influential factors affecting materials scheduling and material ordering problem: Modeling and solution
management in construction projects.” Manage. Prod. Eng. Rev. 8 (4): algorithms.” Appl. Soft Comput. 58 (Sep): 700–713. https://doi.org/10
82–90. https://doi.org/10.1515/mper-2017-0039. .1016/j.asoc.2017.05.030.

© ASCE 04021017-14 J. Constr. Eng. Manage.

J. Constr. Eng. Manage., 2021, 147(4): 04021017