Transportation Research Part E 74 (2015) 152–168

Contents lists available at ScienceDirect

Transportation Research Part E

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

A strategic approach to improve sustainability in transportation

service procurement
R. Jothi Basu a, Ruibin Bai b,⇑, PL.K. Palaniappan c
a
Department of Mechanical Engineering, Sethu Institute of Technology, Virudhunagar Dt, India
b
Division of Computer Science, The University of Nottingham Ningbo, China
c
Department of Mechanical Engineering, Thiagarajar College of Engineering, Madurai, India

a r t i c l e i n f o a b s t r a c t

Article history: This paper focuses on the procurement of full truck-load transportation service with the
Received 9 October 2013 view to ensuring sustainability in transport logistics. The carrier assignment problem
Received in revised form 21 September (CAP) involved in truck load contract auction is studied using a complete enumeration
2014
method and heuristic approach. Performance of both methods is compared in terms of
Accepted 27 October 2014
Available online 30 December 2014
solution quality and computational time based on the results of sample problems consid-
ered. Also, the impact of combinatorial bidding on reducing empty haul in real world case
is analyzed and environmental sustainability is achieved through reduction in carbon foot
Keywords:
Carrier assignment problem
print.
Combinatorial bidding Ó 2014 Elsevier Ltd. All rights reserved.
Sustainable transportation
Full truck-load transportation

1. Introduction

The importance and the need for sustainable business development in logistics (e.g. green logistics) has been widely rec-
ognized in recent years. Sustainability has become one of the greatest challenges of the 21st century because of the rapid
growth of world population, globalization, outsourcing operations and intensity of customer demands. Sustainable business
development is a holistic strategic management that includes the entire value delivery chain from the origins of raw mate-
rials to production processes, customer applications and the end-of-life solutions (Rainey, 2006). Each and every product
manufactured in the industry reaches the end customer through various stages of a supply chain. The role of logistics in
the distribution of goods along the supply chain is inevitable. In the era of globalization and e-commerce, the role of logistics
has become more and more demanding due to needs for cross continental movement of freights. Efficient logistics service
networks are required so that the goods can be delivered to customers at right time in right quality.
Nowadays, most manufacturing companies have factories and facilities at different parts of the globe and a market that
spans the entire globe. In these circumstances, multi-modal transportation is essential for servicing various transportation
demands in their global supply chains (Barnhart and Laporte, 2007). In multi-modal transportation, a product is transported
from the manufacturing facility to the end users through various channels of supply chain by using different transportation
modes such as air, road, rail and sea routes. It should be noted that most domestic freight is transported by roads and trains.
Trucks are normally used for moving the goods from one part of a country to the other either as a truckload or less-than
truckload (consolidated package). From the sustainable perspective, Life Cycle Assessment (LCA) is a very important tool

⇑ Corresponding author.
E-mail addresses: basurj_be@yahoo.co.in (R. Jothi Basu), ruibin.bai@nottingham.edu.cn (R. Bai), kpal@tce.edu (PL.K. Palaniappan).

http://dx.doi.org/10.1016/j.tre.2014.10.015
1366-5545/Ó 2014 Elsevier Ltd. All rights reserved.
 R. Jothi Basu et al. / Transportation Research Part E 74 (2015) 152–168 153

to identify the scope for improvement in each and every activity of all the LCA phases, including the transportation phase. In
this scenario, it is vital to eliminate unnecessary movement in the distribution channel. According to Blaeser and Whiting
(2012), among the current green technologies used by the logistics industries, route optimization is the most prominent
one especially with the large sized truckload industries. Studying the interaction between sustainability and technological
innovation in the field of logistics is essential (Klumpp et al., 2009). To leap forward in this direction, an attempt is made
in this paper to reduce transportation costs (and emissions) by reducing empty haul of the carriers in the full truckload
industry.
Truckload transportation service can be either full truckload or less-than truckload. In full truckload transport, a dedi-
cated truck is deployed for moving a shipment from its origin to the destination; whereas in the less-than truckload trans-
port, many smaller shipments are consolidated and moved by using a single truck. Full truckload is very common in industry
because of dedicated fleet movements along the supply chain without intermediate stops. This ensures on time delivery of
goods at the destinations which can be difficult to achieve in the case of transfer oriented less-than truckload services.
Manufacturing industries nowadays tend to outsource transport logistics activity to improve efficiency and reduce cost.
For this purpose, a contract will be established by a manufacturer with the freight service companies for the period of two or
three years. Auction procedure is generally followed for allocating the lanes to the different trucking companies. Here a lane
refers to a source–destination pair, between which freight has to be moved. This research work studies the problem of
assigning lanes to various carriers who participate in the auction, formally called as carrier assignment problem (CAP)
(Sheffi, 2004). It is also known as the Winner Determination Problem (WDP). CAP is a complex problem faced in a transpor-
tation service procurement auction for establishing contracts with carriers. So far the research in this area (particularly with
the combinatorial bidding) is limited and hence requires more attention.
In recent times, combinatorial bidding has become very important in transportation service auctions because of the asso-
ciated benefits of utilizing economy of scope. Economy of scope is nothing but an opportunity available for the carrier to
utilize the empty movement in their existing network in the future bidding. If the lane offered in a particular call for bidding
matches with the lanes of empty movement in their current network, then there is an opportunity to aggressively bid for
these lanes since it adds no significant cost to the carrier. Also in combinatorial bidding, carriers can submit bids in the form
of lane bundles that enable the carrier to bid for combinations of lanes of their own choice with less empty movements.
According to Lee et al. (2007), repositioning of empty trucks constitutes a substantial portion of operating expenses. Here,
there is no risk of getting an incomplete set in the bundle and hence carriers can bid without fear of exposure problems. This
way, shippers will benefit as the transportation cost will be reduced, leading to a win–win situation for both carriers and
shippers. This will also lead to logistics sustainability because of the scope for elimination or reduction of empty truck
movement.
From the summary of survey shown in Table 1, it is evident that various authors have addressed the sustainability aspect
in different areas of research in logistics such as 3PL, reverse logistics, logistics operations and customized logistics. Issues
related to social sustainability (Presley et al., 2007; Sarkis et al., 2010) as well as environmental sustainability (Lieb and Lieb,
2010; Dey et al., 2011) are discussed. All these works are confined to the above mentioned topics in logistics and no article is
found in transportation service procurement. In this regard, an attempt is made in this work in such a way that sustainability
is achieved through route optimization. Route optimization achieved by reducing empty movement of trucks, which is the
main advantage of using combinatorial bidding. To solve CAP with combinatorial bids, a heuristic method is proposed and
explained in detail in Section 6.
In this study, combinatorial bids are considered in order to exploit the associated cost related benefits and also to min-
imize the carbon footprint because of the elimination of unwanted empty movement of trucks. This will enhance the chances
of achieving the environmental sustainability in logistics management. Even though various authors (Song and Regan, 2003;
Sheffi, 2004; Ma et al., 2010) have discussed the advantages of combinatorial bidding in truckload procurement auction in
detail, our approach is more holistic, as environmental sustainability benefits derived through combinatorial bidding is also
accounted for along with cost reduction. But in practice, the use of combinatorial bidding in truckload procurement auction
is very limited and in this regard, this research work addresses the real world bidding case. The carrier assignment model is
solved for the problem considered by using both a complete enumeration method and a heuristic approach. Cost savings and
sustainability related benefits are achieved through adoption of combinatorial bidding hence it is considered in this study. In
formulating the carrier assignment model, demands for each lane are taken from the estimated data available in the call-for-
bidding offered by the shipper, but it is assumed as unity in most of the literature (Song and Regan, 2003; Sheffi, 2004; Guo
et al., 2006). In the real world transportation service procurement process, load to be moved in each lane is not unity and it
varies across different lanes. This study considers such a varying demand pattern so that the model is closer to the
reality.
The rest of the paper is organized as follows. Section 2 describes the procedures and terminologies involved in truckload
procurement auction for establishing a contract. In Section 3, literature related to logistics sustainability and carrier assign-
ment problem is discussed separately in detail. Section 4 presents the formulation of the carrier assignment problem. Section
5 describes a complete enumeration method for the proposed carrier assignment problem of different sizes. Section 6
details a heuristic approach for solving the above mentioned problem. In Section 7, results obtained by the proposed meth-
ods are compared and their performance is elaborated. Section 8 summarizes the concluding remarks and scope for future
research.
154 R. Jothi Basu et al. / Transportation Research Part E 74 (2015) 152–168

Table 1
Summary of sustainability logistics literature.

Research area Author Objective Methodology Key findings

3PL Lieb and Lieb (2010)
Logistic operations Dey et al. (2011)
Kuhl and Zhou (2009)
Reverse logistics Sheu et al. (2005)
Presley et al. (2007)
Sarkis et al. (2010)
Customized logistics Wang and Lalwani
(2007)
 To study the sustainabil-
ity initiatives undertaken
by 3PL companies
 Impact of these initiatives
on the 3PL and their
customers
To examine the current status
of sustainability efforts in
logistic operations
To develop simulation tool for
making transportation
decision
To develop optimization
based model for integrated
logistics operational problems
of Green Supply Chain
Management
To develop a strategic
justification tool for project
evaluation with sustainability
implications
To address social
sustainability aspect in
reverse logistics
Aims to interlink e-business
with different logistics
strategies to achieve
customized logistics
sustainability
Empirical Survey based on data
obtained from 40 large 3PL
companies in North America,
Europe and Asia – Pacific region
Detailed survey of literature on
various logistics functions
General procedure used in
developing simulation tool is
followed. Developed module of
the model is linked with ARENA
for analysis
A multi objective framework is
formulated and it is used to
optimize the operations of
integrated logistics and reverse
logistics
An activity based management
methodology is used to make a
decision. Methodology is tested
in the scenario of selecting the
best between the two reverse
logistics provider
Structure of social sustainability
is developed based on practice
and research output.
Developed a conceptual model
for customized logistics
sustainability and the same is
validated using real world data.
Economic setbacks do not
affect the 3PL companies from
implementing sustainability
related projects
Revealed limited work done
with sustainability concepts
in logistics
With the proposed model the
sustainability and
environmental impact
measures are made as integral
part in decision making
process
Numerical results proves that
by using proposed model,
profit improved by 21.1%
compared to existing scenario
Effect of reverse logistics on
economic, environmental and
social sustainability of the
business is detailed as a
outcome the research
Profile of reverse logistics for
social sustainability is
developed
An effective performance
measurement system to
evaluate customized logistics
sustainability is proposed

2. Transportation service procurement

A transportation service procurement auction starts with a call for bidding by the shipper who wants to move their
freights. The shipper gives information about the lanes to be covered, loads per week (or per month) to be hauled, a contract
period, terms and conditions to be satisfied by carriers etc., Based on this information, willing carriers will go for bidding.
Lanes chosen for bidding should be carefully selected by the carrier after analyzing their existing network in order to deter-
mine the economy of scope by winning the new lanes. In general, a carrier can bid for individual lanes or a set of lanes as a
bundle package, called combinatorial bids. The steps involved in the procurement is shown in Fig. 1.

3. Literature review

In this section, literature related to carrier assignment problem (CAP) is discussed on three aspects, objectives, demand
assumptions and environmental sustainability. CAP is encountered in the bid analysis stage of the transportation procure-
ment process and it is NP-complete (Sandholm, 2002) but small and medium sized instances can be solved by the state-
of-the-art integer programming solvers. A CAP model helps the shipper to minimize the total costs while ensuring that each
lane is served and the capacity required by each lane is satisfied.
There are a reasonable number of studies with cost related objectives in the literature and a few authors solved CAP as a
bi-objective optimization problem by including the service quality as a second objective. Very few researchers have included
other non-price objectives in their study. The general objective of the CAP problem is to minimize the total transportation
cost subject to the requirement that each lane can be assigned to one carrier only in the final allocation of carrier assignment
problem.
A general formulation of CAP (without combinatorial bidding) is given by Song and Regan (2003) as follows:
XX
Minimize C ¼ bij xij
i j
 R. Jothi Basu et al. / Transportation Research Part E 74 (2015) 152–168 155

Call for bidding by shipper

Carrier bids for lanes Bid Generaon

Problem (BGP)

Evaluaon of bids by shipperS

Carrier Assignment
Determinaon of winner
Problem (CAP)
for each lane by shipper

Final Allocaon

Fig. 1. Transportation service procurement process.

Subject to
X
aij xij ¼ 1 8i 2 V
j

xij 2 f0; 1g

where,
C – Total transportation cost spent by the shipper
V – The set of lanes to be serviced
bij – Bid value by carrier j for serving lane i
aij – Binary coefficient that indicates whether carrier j bided lane i or not.
xij – Binary decision variable to indicate whether carrier j is selected for lane i or not.

The objective of minimizing the total cost involved in the transportation is considered by a number of researchers (Caplice
and Sheffi, 2003; Caplice, 2007; Cohn et al., 2008; Srivastava et al., 2008; Yadati et al., 2007). The objective of maximizing the
total saving is considered by Lim et al. (2008). Yadati et al. (2007) considered the case in which the suppliers provide a quan-
tity-discount curve of prices, instead of a single bid. The winner determination problem in this case is still NP-hard, since this
is a generalization of the normal bidding process. They provided mathematical programming formulations for winner deter-
mination applied to hybrid procurement mechanisms. A heuristic solution scheme was proposed and its effectiveness was
demonstrated by the quality of the solutions it obtained. Almost all the literature in full truckload transportation service pro-
curement has cost related objectives.
Some of the researchers (Buer and Pankratz, 2010; Buer and Kopfer, 2014) formulated CAP as a bi-objective model by
including the quality of service offered by the carrier as a second objective. Buer and Pankratz (2010) solved the bi-objective
problem by using a branch and bound method and a multi objective evolutionary algorithm. A hybrid metaheuristic based on
GRASP (Greedy Randomized Adoptive Search Procedure) and ALNS (Adaptive Large Neighborhood Search) was proposed by
Buer and Kopfer (2014) to solve the bi-objective model. Rekik and Mellouli (2012) accounted quality of service as a carrier’s
reputation. Reputation of carriers is computed as unexpected hidden cost and it is added to the total cost of procurement so
that the problem is reduced to a single objective problem.
Sheffi (2004) proposed to include various non-price objectives while solving carrier assignment problem involved in
truckload transportation service procurement. Guo et al. (2006) solved winner determination auctions for transportation
procurement to include shipper non-price objectives and carrier transit point costs. A model was developed by considering
transit point costs and metaheuristic methods were proposed for the new model. Computational experimentation shows
that the algorithms work well and can easily be implemented.
Another important aspect in solving CAP is the assumption of demand of loads in each lane. In some cases, unit demand is
considered for simplification (Guo et al., 2006; Lee et al., 2007; Song and Regan, 2003). Therefore, the total transportation
cost is calculated by assuming a single load in each lane. Moreover, some papers formulate the models based on the known
demand, which is an estimate based on the previous data (Caplice and Sheffi, 2003; Sheffi, 2004; Caplice, 2007; Chang, 2009;
Yadati et al., 2007; Chen et al., 2009; Rekik and Mellouli, 2012). Very few authors (Ma et al., 2010; Remli and Rekik, 2013;
Zhang et al., 2013) have addressed the uncertainty in shipment volumes in each lane. Ma et al. (2010) proposed a two-stage
156 R. Jothi Basu et al. / Transportation Research Part E 74 (2015) 152–168

stochastic integer programming model for the CAP in combinatorial auctions to hedge the shipper’s risk under shipment
uncertainty. In addition, many other important comprehensive business side constraints are included in the model. CPLEX
9.0 solver is used to solve the model developed and the computational results showed that the stochastic model yields very
good solutions compared to the deterministic model with cost minimization objective. A two stage robust optimization
approach is investigated by Remli and Rekik (2013) and the problem is solved by using a constraint generation algorithm.
Zhang et al. (2013) also proposed a two-stage stochastic programming approach to solve CAP under shipment volume uncer-
tainty. A Monte Carlo approximation (MCA) method is used to solve the test instances.
A number of attempts have been made by the researchers to address the needs of green logistics and to study measures to
be taken to achieve environmental sustainability. Dekker et al. (2012) highlights the contribution of operations research to
green logistics. The study found that operational efficiency of the systems is an important factor in the environmental per-
formance. Various allocation methods of Green House Gases (GHG) emissions to shipments in road freight transportation
were studied by Kellner and Otto (2012). They analyzed the performance of various methods with a small numerical prob-
lem illustration. Zhu et al. (2012) analyzed the extent of the adoption of green supply chain management practices among
the manufacturers in China through an empirical study. For this purpose, a pool of manufacturers was clustered into three
major categories namely adopters, followers and laggards based on the time of implementation of Green logistic principles
into their systems. Factorial experiments were conducted by using ANOVA based on the data collected from 245 (sub-)
industries in China to test the significance in performance difference of three categories of industries. The result shows that
there is a very good improvement in performance of adopters compared to followers and laggards.
Also, a literature survey revealed that it is very difficult to find the exclusive case studies with real life data with the
exception of the case Tian et al. (2011), which considered a real life transportation service procurement case in Philips.
The carrier assignment problem was modeled as a linear integer programming problem and an iterative heuristic algorithm
was proposed to solve it.
From the literature survey, it can be seen that there is a very limited research in non-price objectives. Non-price objec-
tives include on time delivery, availability of suitable equipment, area coverage, cargo insurance etc. Also very few research-
ers have analysed the combinatorial auction setting involved in carrier assignment problem with real time data. Another
important equipment observation is that no attempt has been made from a sustainability perspective and it is vital in the
current scenario since most of the industries are moving toward sustainable business development. Hence, environmental
sustainability initiatives like energy efficiency, reduction of carbon emission, electronic documentation etc, taken by the car-
rier need to be considered by the shipper when selecting the carrier for a truckload contract.

4. Carrier assignment problem with combinatorial bids

Carrier assignment problem is the problem of allocating lanes to various carriers who are participating in the full truck-
load transportation service procurement auction after evaluating their submitted bids with the objective of minimizing the
total cost of transportation. In the problem considered, carriers are allowed to submit combinatorial bids. With combinato-
rial bids the problem becomes NP-hard and difficult to solve by existing methodologies. In this research work, demand for
each lane is known and it is obtained from shipper’s call for quotation, but in most of the literature it is assumed as unity.

4.1. Mathematical model

We now present the mathematical model for the carrier assignment problem with combinatorial bidding. The model is
based on the formulation developed by Guo et al. (2006) which is a network flow model with the assumption of unit demand
in each lane. In this study, the model has been extended to accommodate practical issues, where demand in each lane is an
estimated value and lane distance is given in RTKM (Round Trip distance in Kilometer). The estimate of demand is based on
the previous year’s demand pattern in each lane and is provided by the shipper. Demand data and lane distance is revealed to
the bidder through call-for-bidding by the shipper. The objective is to minimize the value of total cost of transportation for
hauling loads in different lanes during the contract period. Bundle package representation is similar to that of Caplice and
Sheffi (2003). Based on their current freight network configuration and existing businesses, carriers choose most appropriate
lane bundles to join the bidding by solving a bid generation problem.

4.1.1. Notations

n the total number of lanes

m the total number of carriers
M a very large constant.
H the length of the contract (in years).
K the set of all possible combinatorial biding bundles for n lanes. denote |K| be the total number of bundles for n lanes.
li the distance of lane i.
 R. Jothi Basu et al. / Transportation Research Part E 74 (2015) 152–168 157

pi monthly load to be moved in lane i.

ckj the cost of serving the bundle k e K.
Bk the list of lanes in bundle k.
k P
dk the distance of bundle k in Round Trip distance in Kilometer (RTKM), and d ¼ i2Bk li :
rkj the biding rate of carrier j for bundle k e K per RTKM. rkj>0 means carrier j bids for bundle k. Otherwise, rkj=M.
dki a binary indicator that takes value of 1 if lane i is included in bundle k and 0 otherwise.
xkj binary decision variables that take 1 if bundle k is assigned to carrier j in the final solution and 0 otherwise.
x the vector of xkj .
s the optimal solution vector.
P
Lk total load to be moved in lanes in bundle k per month, whereLk ¼ i2Bk pi .

Objective function
XX
m X
H
k
minimize T ¼ 12Lk d rkj xkj ð1Þ
k2K j¼1 u¼1

Subjected to:
XX
m
dki xkj ¼ 1 8i ð2Þ
k2K j¼1

xkj ¼ f0; 1g 8j; 8k ð3Þ

The objective is to minimize the total cost incurred for servicing all the demands of all the lanes during the contract period.
P
The term Hu¼1 12Lk in the equation gives the total number of loads to be hauled for the lanes in the particular bundle during
the contract period. During bidding, carriers submit the price rate per RTKM (rkj ) for each bundle k which they bid for. If car-
rier k does not bid for bundle k, rkj is set to a very large value (rkj = M) to prevent bundle k being allocated to carrier k. A binary
decision variable xkj is used in the model to determine whether bundle k submitted by carrier j is allotted or not. Since the
objective is minimizing the total cost, a carrier will not be allocated to a bundle that it did not bid for because otherwise, the
cost of doing this will be significantly more than other alternatives. Constraint (2) ensures that each lane is assigned exactly
to only one carrier in the final allocation. The model is a typical integer (binary) programming problem that can be reduced
to the classic set covering problem which is NP-Hard.

4.2. Assumptions

i. Demand is known at the time of bidding.

ii. Auction is a single round sealed bid.
iii. Bidding rate for each lane is same for the whole contract period.
iv. Combinatorial bids are allowed and carriers can submit bundle packages.
v. Distance of each lane is measured in RTKM.
vi. No restriction on maximum and minimum number of lanes won by the particular carrier.
vii. No upper limit for the maximum number of carriers in the final allocation.

5. Complete enumeration method

For the purpose of evaluating the performance of the heuristic method for this problem (see Section 6), a complete enu-
meration method is first developed. In addition, the complete enumeration method can be used to solve small problem
instances. This method searches entire search space for the optimal solution. The method is used as a benchmark for com-
paring performance between approximation methods in terms of solution quality. A flow chart for this method is shown in
Fig. 2.

Step 1: Get information regarding Number of lanes n, source–destination pair, lane distance li.
Step 2: Generate possible number of bundle packages |K| for the given number of lanes by using the term 2n  1.
Step 3: Get the details of number of bidders or carriers m, biding bundle package K submitted by carriers, quoting price ckj
for serving a particular bundle k by carrier j.
Step 4: For each bundle package find out the minimum total cost and the corresponding carrier.
Step 5: Generate various combinations of bundle packages with the already obtained minimum values so that obtained
combination of bundles is equal to the number of lanes n.
Step 6: Find out the least cost combination of bundles from Step 5. The corresponding bundles served by each carrier are
then recorded, which will be the final allocation.
158 R. Jothi Basu et al. / Transportation Research Part E 74 (2015) 152–168

Fig. 2. A flowchart of the complete enumeration method.

Table 2
Information in call for quotation issued by the shipper.

Lane number Source Destination RTKM Load in 2010–2011 per month Load in 2011–2012 per month
1 Jaipur Reengus 110 53 55
2 Jaipur Chaksu 102 24 26
3 Jaipur Alwar 286 41 44
4 Jaipur Dausa 165 57 60
5 Jaipur Naraina 160 32 34
6 Jaipur Sikar 226 52 55

The main difficulty with this method is its long computational time and therefore it is impossible to solve the large sized
problem by using this method. So, we can use this approach only for small and medium sized problems.
The complete enumeration procedure is applied to the problem of carrier assignment problem involved in full truckload
transportation service procurement. For this purpose, real data was obtained from the call for bidding proposal reported by
one of the leading petroleum company in India for moving LPG cylinders from a plant to different distributors located at var-
ious places. The company considered for the study has a worth of USD 6.97 billion and it has 13,213 permanent employees.
Annual revenue generated by the company is USD 41.42 billion with a local market share of 23.1%. The freight cost amounts
for USD 178.59 million and customer for LPG products alone accounts for USD 37.4 million. Information given by the shipper
while calling for bidding is given in Table 2.
 R. Jothi Basu et al. / Transportation Research Part E 74 (2015) 152–168 159

Table 3
Bidding details of various carriers.

Bidder Number of bundle Bundle Bidding rate per Bidder Number of bundle Bundle Bidding rate per
number packages package RTKM in US$ number packages package RTKM in US$
1 12 1 0.0993 4 11 1 0.0988
2 0.0962 3 0.0973
4 0.0982 5 0.0955
6 0.0945 1,3 0.0933
1,4 0.0922 5,6 0.089
4,6 0.0947 2,3,4 0.0872
3,5 0.0923 1,3,5 0.0913
1,3,6 0.0887 2,4,6 0.0888
1,4,5 0.0917 1,2,3,5 0.091
2,3,5,6 0.0915 1,3,4,5 0.087
1,3,4,5 0.0888 1,4,5,6 0.0893
2,3,4,6 0.0913
2 12 2 0.0943 5 12 1 0.0912
3 0.0915 4 0.0972
5 0.0932 2,3 0.0943
6 0.0908 3,5 0.095
1,3 0.0905 4,6 0.0953
3,5 0.0897 1,3,6 0.0923
4,6 0.0893 1,5,6 0.0913
2,3 0.0905 2,3,6 0.0893
1,3,5 0.089 2,4,6 0.0908
2,4,6 0.0882 3,4,5 0.0873
3,4,5 0.0885 1,2,3,5 0.0888
3,4,5,6 0.0888 2,3,5,6 0.0918
3 10 2 0.098 6 12 3 0.0908
4 0.0963 6 0.0898
5 0.0972 1,3 0.0887
1,2 0.0898 3,4 0.0927
4,5 0.0903 5,6 0.0945
1,3,5 0.0917 2,4 0.0937
2,4,6 0.0913 3,6 0.0928
3,4,6 0.0918 1,3,5 0.09
1,3,4,5 0.091 4,5,6 0.0888
2,3,5,6 0.0895 3,5,6 0.0898
1,3,5,6 0.0903
2,3,5,6 0.089

For this particular example, the number of lanes for bidding is 6. The distance between the source and destination is given
in terms of RTKM and also the load to be moved in each lane for the period of two years is also given in request for quotation.
Based on the above information supplied by the shipper, carriers bid for the lanes in the form of combinatorial bids. The
number of carriers, bundles offered by each carrier and bid rate quoted by each carrier for the particular bundle are assumed
as shown in Table 3. Here, bidding rate is assumed based on the current market value.
The total number of possible bundle packages with various lane combinations is given by the term 2n  1, where n is the
total number of lanes for which bidding is called for. Therefore for n = 6, the total number of possible lane combinations is
26  1 = 63.
The next step is to find out the minimum cost for each bundle and which can be obtained by the following equation.

X
H
k
C kmin ¼ min 12Lk d r kj ð4Þ
j
u¼1

where C kmin denotes the minimum cost for serving the bundle k.
Once the values of minimum cost for each bundle package and the corresponding carrier are obtained (if a particular bun-
dle is not submitted by any of the carrier, then minimum value is set to infinity), the next step is to explore various sets of
combinations of bundles in such a way that each feasible set covers all the lanes exactly once. For the above example where
n = 6, various sets of bundle combinations include {1,2,3,4,5,6}, {12,345,6}, {145,23,6}, {13,52,4,6} and so on. For this illus-
tration problem of n = 6, the maximum number of such combination sets is 203 and the number will increase exponentially
with the number of lanes n. The total number of bundle combinations for n lanes can be computed by the following Bell
Number Bn

Bn ¼ Sðn; 0Þ þ Sðn; 1Þ þ Sðn; 2Þ þ . . . þ Sðn; nÞ

160 R. Jothi Basu et al. / Transportation Research Part E 74 (2015) 152–168

Table 4
Solution to the problem of n = 6 in combinatorial bidding setup.

Final minimum combination Bundle packages in the Corresponding Bundle cost in Total cost of transportation service procurement
set set carrier US$ in US$
{1345,2,6} (1,3,4,5) 4 68,961.42 100,802.82
(2) 2 5773.2
(6) 6 26,068.2

where S(n, k) can be obtained from the recursive equation of

Sðn; kÞ ¼ k:Sðn  1; kÞ þ Sðn  1; k  1Þ if n  1 and k  1
with the initial conditions of
Sð0; 0Þ ¼ 1
Sðn; 0Þ ¼ 0 if n > 0
Sð0; kÞ ¼ 0 if k > 0
For example, let us consider n = 4, the total number of combination sets are calculated as
B4 ¼ Sð4; 0Þ þ Sð4; 1Þ þ Sð4; 2Þ þ Sð4; 3Þ þ Sð4; 4Þ
By using the recursive equation, the values obtained are detailed below:
Sð4; 0Þ ¼ 0; Sð4; 1Þ ¼ 1; Sð4; 2Þ ¼ 7; Sð4; 3Þ ¼ 6; Sð4; 4Þ ¼ 1;
This results in a total number of combination sets of B4 = 15.
For all the generated combination sets, the total cost is calculated for each set by summing the minimum cost of individ-
ual bundle package (i.e. cmink) in the set. For example, consider the combination set {12,345,6} and in this set there are three
bundle packages (1,2), (3,4,5) and (6). It can be found that the minimum bundle cost (in USD) for these bundles are 18304.44,
56775.4 and 26068.2 respectively. Hence the total cost for the combination set {12,345,6} is the sum of the individual bundle
costs above. Similarly this total cost is calculated for the all combination sets and the minimum among all the 147 sets is the
optimal objective, which is US$ 100802.82 as per the details given in the Table 4.
Hence, the final optimal assignment is that lanes 1, 3, 4 and 5 are assigned to carrier 4; lane 2 to carrier 2 and lane 6 to
carrier 6. The method was programmed in C language.
If combinational bids are not permitted (i.e. carriers can only bid for individual lanes), the total cost of procurement
becomes USD105458.63. Therefore, by allowing the carriers to submit combinatorial bids, the shipper can save as much
as USD 4655.81 for a 6-lane allocation problem. This cost saving is only from shipper perspective and from carrier point
of view. There is also reduction in empty haul movement in their network which should also be taken into consideration.
Table 5 shows the result of various problems solved by using this method.
The complete enumeration method gives optimal solutions for smaller problems. Large problems will be more challeng-
ing to solve since the problem considered is NP-hard. The computational time for the complete enumeration method
increases exponentially as the problem size increases. Alternative methods are needed in these cases.

6. A heuristic approach

This method is developed in order to overcome the excessive computational time by the complete search method which
becomes unsuitable for large sized problems. The proposed heuristic approach selects suitable combinations and gives near
optimal solutions. With minor compromise in solution quality, the heuristic method is more efficient in terms of computa-
tional time compared with the complete search method. Fig. 3 shows the flow chart for the heuristic approach.
The proposed heuristic uses bundle packages submitted by the carrier as input and it is based on a linear search method.
We now explain it in detail below.

Step 1: Read problem data, including lanes, carriers and bundle bids.
Step: 2Initialize x kj :=0;
Step 3: Find out the minimum cost ckmin for each submitted lane bundle and also find out the corresponding bidder i.
Step 4: Store the lane bundles submitted by the bidders in an array and denote the total number of elements of the array
as Np. For j = 0 to Np, read the lane package k.
Step 5: Compare bundle k with all packages from index 0 to Np. The packages which are not matched (or different) with
the package k are combined into one combination till all lanes are included and the total cost is calculated. If one com-
bination is formed, do the same comparison from the next index to Np for other possible combinations.
Step 6: Increment j and repeat step 5 for the remaining lane packages up to Np.
 R. Jothi Basu et al. / Transportation Research Part E 74 (2015) 152–168 161

Table 5
Results for complete enumeration method.

Number of Number of Carrier Bundles submitted by each Final assignment Total cost of transportation
lanes carriers number carrier in US$
Bundles Cost in US$
allotted
2 3 1 3 Nil Nil 18,100.68
2 3 (1,2) 18,100.68
3 3 Nil Nil
3 4 1 4 Nil Nil 43,310.52
2 4 Nil Nil
3 4 (1,2) 17,930.88
4 4 (3) 25,379.64
4 5 1 3 Nil Nil 64,016.72
2 4 (2,3) 30,586.4
3 4 Nil Nil
4 5 (1,4) 33,430.32
5 5 Nil Nil
5 5 1 10 Nil Nil 74,965.02
2 10 Nil Nil
3 10 (1,2) 18,406.32
4 10 (3,4,5) 56,558.7
5 10 Nil Nil
6 6 1 12 Nil Nil 100,802.82
2 12 (2) 5773.2
3 10 Nil Nil
4 11 (1,3,4,5) 68,961.42
5 12 Nil Nil
6 12 (6) 26,068.2
7 7 1 15 Nil Nil 115,108.53
2 14 (2,4,6) 51,405.05
3 13 Nil Nil
4 12 (1,3,5,7) 63,703.49
5 15 Nil Nil
6 14 Nil Nil
7 14 Nil Nil
8 7 1 15 Nil Nil 142,889.7
2 14 Nil Nil
3 13 (2,7) 20,792.16
4 12 {(5,6), {37,104.46,
(1,3,4,8)} 84,993.08}
5 15 Nil Nil
6 14 Nil Nil
7 14 Nil Nil
9 8 1 17 Nil Nil 154,804.13
2 15 (2) 5803.8
3 13 (7,9) 26,902.8
4 14 {(5,6), {37,104.46,
(1,3,4,8)} 84,993.08}
5 15 Nil Nil
6 14 Nil Nil
7 15 Nil Nil
8 15 Nil Nil

Step 7: Finally the combination which has minimum cost is found and the corresponding bidders are obtained.

The proposed heuristic approach was coded in C language and run on a PC with Intel Pentium Core 2 Duo 1.19 GHz pro-
cessor and 2 GB. The proposed heuristics is applied to the same set of problem instances and the results are tabulated in
Table 6 which also includes the final allocation of bundles to various carriers.
It can be seen from Table 6 that the heuristics approach gives exact solution to small size problems and the quality of
solution declines as the size of the problem increases, but by very small margin.

7. Results and discussions

Since the number of possible bundle packages and their combinations increases exponentially with the increase in the
number of lanes, the computational time rise steeply and the complete enumeration method becomes ineffective. The con-
162 R. Jothi Basu et al. / Transportation Research Part E 74 (2015) 152–168

Fig. 3. A flow chart for the heuristic approach.

dition worsens further with the increase in the number of carriers and the number of bundle packages submitted by the
carrier.
Barring the above mentioned problem of prolonged time consumption for computation, there is an uncompromised solu-
tion quality with complete enumeration method. A heuristics is also developed to solve this problem in reasonable time with
little difference in solution quality. Fig. 4 shows the comparison of results obtained by using both the methodologies.
The amount of reduction in computational time is understood from the fact that the time taken for getting the output for
the 9 lanes problem is almost 28 h for the complete enumeration method, whereas it is only 10 s for the heuristics to produce
the result. At the same time, the solution objective value obtained by the heuristic deviates away from the optimum by
 R. Jothi Basu et al. / Transportation Research Part E 74 (2015) 152–168 163

Table 6
Solution by heuristic approach.

Number of Number of Carrier Bundles submitted by Final assignment Total cost of

lanes carriers number each carrier transportation in US$
Bundles allotted Cost in US$
2 3 1 3 Nil Nil 18,100.68
2 3 (1,2) 18,100.68
3 3 Nil Nil
3 4 1 4 Nil Nil 43,310.52
2 4 Nil Nil
3 4 (1,2) 17,930.88
4 4 (3) 25,379.64
4 5 1 3 Nil Nil 64,016.72
2 4 (2,3) 30,586.4
3 4 Nil Nil
4 5 (1,4) 33,430.32
5 5 Nil Nil
5 5 1 10 Nil Nil 74,965.02
2 10 Nil Nil
3 10 (1,2) 18,406.32
4 10 (3,4,5) 56,558.7
5 10 Nil Nil
6 6 1 12 Nil Nil 100,802.82
2 12 (2) 5773.2
3 10 Nil Nil
4 11 (1,3,4,5) 68,961.42
5 12 Nil Nil
6 12 (6) 26,068.2
7 7 1 15 Nil Nil 116,272.58
2 14 (1,4,7) 47,875.88
3 13 Nil Nil
4 12 (5,6) 37,104.46
5 15 Nil Nil
6 14 Nil Nil
7 14 (2,3) 31,292.24
8 7 1 15 Nil Nil 142,889.7
2 14 Nil Nil
3 13 (2,7) 20,792.16
4 12 {(5,6), (1,3,4,8)} {37,104.46,
84,993.08}
5 15 Nil Nil
6 14 Nil Nil
7 14 Nil Nil
9 8 1 17 Nil Nil 156,110.33
2 15 Nil Nil
3 13 (7,9) 26,902.8
4 14 (2,4,6) 51,890.92
5 15 (1,3,5,8) 77,316.63
6 14 Nil Nil
7 15 Nil Nil
8 15 Nil Nil
10 8 1 17 {(7), (9)} {16,133.28, 167,088.3
12,436.25}
2 15 Nil Nil
3 15 {(2), (5)} {5803.8,
11,784.96}
4 13 Nil Nil
5 15 (1,3,4,8) 84,993.08
6 15 Nil Nil
7 14 (6) 26,019.83
8 15 (10) 9917.11
15 12 1 15 Nil Nil 246,666.53
2 15 Nil Nil
3 15 (15) 11,486.02
4 15 Nil Nil
5 15 {(7), (1,4,9,14)} {16,274.8,
77,370.13}
6 15 Nil Nil
7 15 (13) 13,524.72
8 15 {(6), 10)} {26,068.2,
9917.11}

(continued on next page)

164 R. Jothi Basu et al. / Transportation Research Part E 74 (2015) 152–168

Table 6 (continued)

Number of Number of Carrier Bundles submitted by Final assignment Total cost of

lanes carriers number each carrier transportation in US$
Bundles allotted Cost in US$
9 15 (2,3,5,8,11) 83,216.77
10 15 Nil Nil
11 15 Nil Nil
12 15 (12) 8808.8
20 12 1 12 {(18), (1,6,17,19)} 100,215.28 331,030.97
2 12 {(2,8,13),(7,9,10,12,14,16)} 144,073.68
3 12 Nil Nil
4 15 Nil Nil
5 15 Nil Nil
6 15 (3,4,5,11) 68,013.75
7 15 Nil Nil
8 15 Nil Nil
9 15 Nil Nil
10 15 Nil Nil
11 15 (15,20) 18,728.26
12 15 Nil Nil
Procurement Cost in US $

350000
300000
250000
200000
150000
100000
50000
0
2 3 4 5 6 7 8 9 10 15 20
Number of lanes
Complete Enumeraon Method
Heurisc Approach

Fig. 4. Comparison of results.

Table 7
Comparison of results obtained by both methodologies.

No. of lanes Total Cost of transportation in US$ Cost difference in US$ Deviation in %
Complete search method Heuristics approach
2 18,100.68 18,100.68 0 0
3 43,310.52 43,310.52 0 0
4 64,016.72 64,016.72 0 0
5 74,965.02 74,965.02 0 0
6 100,802.82 100,802.82 0 0
7 115,108.53 116,272.58 1164.04 1.011
8 142,889.7 142,889.7 0 0
9 154,804.13 156,110.33 1306.2 0.843
10 —— 167,088.3 —— ———
15 —— 246,666.53 —— ——
20 —— 331,030.97 —— ——

merely 0.84% but the time saving is of the order of 10,080 times. Table 7 compares the two methodologies results for various
lanes.
Few studies have discussed the importance of restricting the number of lanes allocated to a particular carrier (Sheffi,
2004; Guo et al., 2006). This restriction ensures better spread and the participation of a huge number of carriers with more
choices available for shippers. Hence we analyzed the sensitivity of solution for the real data with the introduction of lane
restriction. The results are shown in Table 8.
It can be seen from the result that there is no change in the total transportation cost for small sized problems. But for
larger problems, cost increases with restriction on the maximum number of lanes won by each carrier. Even though
 R. Jothi Basu et al. / Transportation Research Part E 74 (2015) 152–168 165

Table 8
Sensitivity analysis.

Number of lanes Complete enumeration method Heuristic approach

Total cost with Total cost with Total cost with Total cost with
no constraint constraint of Lmax = 3 no constraint constraint of Lmax = 3
2 18,100.68 —— 18,100.68 ——
3 43,310.52 43,310.52 43,310.52 43,310.52
4 64,016.72 64,016.72 64,016.72 64,016.72
5 74,965.02 74,965.02 74,965.02 74,965.02
6 100,802.82 103,360.12 100,802.82 106,089.97
7 115,108.53 116,272.58 116,272.58 116,272.58
8 142,889.7 144,028.55 142,889.7 146,370.15
9 154,804.13 158,948.65 156,110.33 158,948.65
10 —— —— 167,088.3 169,077.07

158000
Minimum total cost in US$

138000

118000

98000

78000

58000

38000

18000
2 3 4 5 6 7 8 9 10
No. of Lanes
CEM with No Constraint
CEM with constraint
HA with no constraint
HA with constraint

Fig. 5. Effect of constraint on solution quality.

transportation service procurement cost is increased, the shipper can have more carriers in their network and thereby the
risk of dependence on a few carriers in specific area of service is avoided. Table 8 shows the results of both the methods with
and without restriction. The maximum number of lanes that can be assigned to any particular carrier is limited to three
(Lmax = 3).
It is noticed from Fig. 5 that there is no significant change in quality up to five lanes with the restricted search space.
Hence both methodologies give exact minimum value with the condition of no constraint imposed. Similarly, for small prob-
lem size, there is not much difference in the total cost with or without restriction. The reason for this is that for problems of
this size lane restriction may be either 2 or 3. As far as the limitation of this work is concerned, small instances were tested
by the proposed methodologies. But in practice, even thousands of lanes may be allotted to the carrier and in such situations
more efficient techniques are needed in order to minimize the computational time.
This research work emphasizes inclusion of sustainability aspects while establishing truckload contracts. Even though a
number of researchers have discussed green logistics, this work addresses how combinatorial bidding will be useful in
attaining the goal of environmental sustainability in addition to the cost saving. Further, it is evident from the real life truck-
load procurement process that there is a considerable gap between literature and actual practice. In practice, one of the
important conditions imposed by the shipper on carriers is that the trucks awarded contract are exclusively for moving their
freight only and the particular truck should not be used for moving the loads of other shippers. In such cases, it is difficult for
the carrier to minimize the empty haul.
Table 9 shows the reduction in CO2 emissions by following combinatorial bidding in real world truckload procurement
auction. Carbon emission factor values at different levels of empty haul is based on the study of Alan McKinnon and
Piecyk, 2010 for 40 ton capacity truck with the payload of 25 tons. By adopting a flexible system of combinatorial bidding,
carbon emission can be reduced up to 66% when compared to conventional method of individual lane bidding with dedicated
trucks for each shipper. CO2 emissions can be reduced to the extent of 32.4 tons per year for the problem with 6 numbers of
lanes with already discussed data set. In this work, a real world truckload procurement auction is considered and it is solved
166 R. Jothi Basu et al. / Transportation Research Part E 74 (2015) 152–168

Table 9
Reduction of CO2 emissions by reducing empty haul.

Individual Lane CO2 emissions in ‘g/tonne’ for single haul Reduction in Total reduction Total reduction in CO2
Lane distance CO2 in CO2 emissions emissions per year
number in RTKM emissions for single haul in (considering number of
for single ‘g’ (for 25 tonne loads in each lane per
haul in ‘g/ load) year) in ‘g’
tonne’
At the rate of 70.3 g/tonne- At the rate of 42.3 g/tonne-
km for empty return (50% km for no empty return (0%
empty haul) in the case of empty haul) in the case of
individual dedicated freight flexible combinatorial freight
movement movement
1 110 7733 4653 3080 734,300 32,396,700
2 102 7170.6 4314.6 2856
3 286 20,105.8 12,097.8 8008
4 165 11,599.5 6979.5 4620
5 160 11,248 6768 4480
6 226 15,887.8 9559.8 6328

with setting of combinatorial bidding by the proposed methodologies and it proves that the above mentioned condition by
the shipper not only adds to the cost of procurement but is also a hindrance to achieving sustainability. Sample problems
discussed in this research wok can be used as benchmarks by future researchers.
As an important theoretical implication of the study, it is recommended to consider the CO2 emission associated with
freight movement while modeling CAP. Also, various sustainability factor and non-price factors (Sheffi, 2004) are studied
for their importance in the process of truckload service procurement. For managers, it is suggested to fix norms for CO2 emis-
sions to carriers and penalties may be imposed if the emission level exceeds the allowable limit. Also, the managers may
relax certain conditions in practice so that empty haul movement can be reduced which in turn leads to reduction in CO2
emissions.

8. Conclusion and future scope

Assignment of lanes to various carriers is one of the important tasks to be completed in the process of transportation ser-
vice procurement using auction mechanism, popularly known as Carrier Assignment Problem. This problem deals with
determining the final allocation of lanes to various carriers who are involved in the auction procedure, so that the total cost
of transportation is minimized. In this paper, carrier assignment problem, a NP-hard problem, is considered for full truckload
transportation service procurement and it is solved by using the complete enumeration method as well as a heuristic
method. Their performances are compared.
Most of the problem data for experiments was obtained from the call for bidding proposals reported by a gas industry for
moving gas cylinders in full truckloads to its distributors. Some of the data are estimated artificially by the authors. As per
the actual procedure followed by the particular gas industry, carriers are allowed to bid for only individual lanes not as com-
binatorial bids. But in this paper, combinatorial bids are considered because of their potential advantage and wide spread
acceptance across industries. Also carriers can submit any number of bundles while bidding and it is not necessary to be
similar.
But in practice it is found that some of the shippers impose the condition of allotting dedicated trucks for moving their
freights alone on the carrier when establishing contracts. On the other hand combinatorial bidding allows the carriers to uti-
lize the economy of scope and by this way it minimizes the empty movement in the existing network. Minimization of
empty movement is not only associated with cost saving, but also with reduction of carbon emission due to empty move-
ment of trucks. In today’s context, it is important for the industries to implement the strategies to achieve sustainable busi-
ness development and in this setting the usage of combinatorial bids along with flexibility of truck usage by the carrier ie.,
use of truck for moving goods of different shippers becomes critical.
In this research work, procurement of transportation service is viewed from a sustainability perspective. The extent to
which reduction in CO2 emissions can be achieved by using combinatorial bidding is illustrated with a numerical example.
For the small instance considered, reduction in CO2 emissions amounts to 32.4 tons and hence implementation of combina-
torial bidding without any restriction is proved to be an important step toward achieving environmental sustainability.
As far as the scalability is concerned, the proposed heuristic method is capable of handling large instances efficiently.
Coming to the methods, complete enumeration can produce very good results consistently when problem size increases,
but time complexity increases exponentially. Hence, large instances are too difficult to solve. At the same time, the heuristic
method can solve large size problems but the solution quality decreases slightly when compared with exact solution
obtained by the complete enumeration method. The contributions of the paper to the literature are (i) the study of real world
truckload service procurement problem to draw insights from actual practice (ii) the development of Complete Enumeration
 R. Jothi Basu et al. / Transportation Research Part E 74 (2015) 152–168 167

Method, an exact solution methodology, can be used as a reference to compare the solution quality of other methodologies
and (iii) more importantly, this research work emphasizes the consideration of the sustainability aspect while procuring
truckload transportation service. As an extension of this work, non price objectives can also be included along with the cost
objective and can be modeled as multi objective models. The proposed algorithm can be modified to solve the multi objective
problem. Potential meta-heuristics can be utilized to solve this multi objective problem. Furthermore future researchers can
explore the possibility to solve CAP with stochastic nature of demand.

Appendix

Terminologies involved

Shippers: Shippers may be manufacturers, distributors, retailers, and other organizations that need to move freight.
Carriers: Carriers are trucking companies that have trucks and other transportation facilities.
Lane: It is the basic unit of interest and is defined as a unidirectional movement from an origin to a destination for a cer-
tain time period.
Cycle: A cycle is a set of lanes originating and terminating at the same physical location. In general, cycle generation is
achieved by shipper or carrier collaboration.
Individual Bids: Carriers submit bid for each lane separately and the shipper will decide the winner based on the bid price
for the particular lane submitted by the carriers.
Combinatorial Bids: In Combinatorial bidding, carriers bid for set of lanes and each carrier can submit any number of sets.
If a particular set is won by the carrier then they have to haul all the lanes in the set.
Economy of Scale: Cost of serving a lane depends upon the amount of loads to be moved. For example let us assume a
situation in which the amount of loads to be moved in a particular lane is less than the full capacity of dedicated truck.
In this case, increase in volume up to full capacity of truck won’t affect the operating cost of the carrier very much. Thus
the cost per unit load can be reduced further with increase in amount of load to be moved for specific lane. This is called
as economy of scale in transportation.
Empty haul: It is the empty movement of the vehicle from delivery point to the home. This may be because of reposition-
ing of truck for further freight loading or change of driver or for servicing the vehicle etc.
Economy of Scope: The cost of serving a lane not only depends upon the amount of loads to be moved on that lane but also
on the number of loads carried on other related lanes. Suppose a carrier is moving 10 loads from city A to city B and return
to the city A without load (empty haul) for further loading. In this scenario, load from city B to city A will reduce the repo-
sitioning cost of the carrier and it is referred as economy of scope.
Winner’s curse: It is the phenomenon in which carriers that are most unfamiliar with a shipper’s operations will make
very optimistic assumptions and end up winning a lane, but incurring a loss.
Exposure problem: Sometimes a shipper allocates lanes individually to the carriers. This often poses risks for carriers since
there might be several lanes that might be of interest and every lane in this set must be won separately. Obtaining an
incomplete set of items may render the set much less valuable or even worthless to them.
NP Complete Problem: A decision problem is NP-complete if any problem in NP can be polynomially reduced to it.
NP Hard Problem: An optimization problem (or a computational problem) is NP-hard if some NP-complete problem can be
polynomially reduced to it.

References

Barnhart, C., Laporte, G. (Eds.), 2007. Handbooks in Operations Research and Management Science: Transportation, vol. 14, North-Holland.
Blaeser, J., Whiting, G., 2012. Environmental Sustainability Benchmark study: Leaders find green ROI. Research Report by American Shipper Society.
Buer, T., Kopfer, H., 2014. A Pareto-metaheuristic for a bi-objective winner determination problem in a combinatorial reverse auction. Comput. Oper. Res.
41, 208–220.
Buer, T., Pankratz, G., 2010. Solving a bi-objective winner determination problem in a transportation procurement auction. Logist. Res. 2 (2), 65–78.
Caplice, C., 2007. Electronic markets for truckload transportation. Prod. Oper. Manage. 16 (4), 423–436.
Caplice, C., Sheffi, Y., 2003. Optimization-based procurement for transportation services. J. Bus. Logist. 24 (2), 109–128.
Chang, T.S., 2009. Decision support for truckload carriers in one-shot combinatorial auctions. Transp. Res. Part B 43 (5), 522–541.
Chen, R.L., AhmadBeygi, S., Cohn, A., Beil, R.D., Sinha, A., 2009. Solving truckload procurement auctions over an exponential number of bundles. Transp. Sci.
43 (4), 493–510.
Cohn, A., Beil, D., Sinha, A., 2008. Using implicit bidding to solve truckload procurement auctions. In: Proceedings of NSF Engineering Research and
Innovation Conference, 7–10 January, Knoxville, Tennessee, USA.
Dekker, R., Bloemhof, J., Mallidis, I., 2012. Operations research for green logistics – an overview of aspects, issues, contributions and challenges. Eur. J. Oper.
Res. 219 (3), 671–679.
Dey, A., LaGuardia, P., Srinivasan, M., 2011. Building sustainability in logistics operations: a research agenda. Manage. Res. Rev. 34 (11), 1237–1259.
Guo, Y., Lim, A., Rodrigues, B., Zhu, Y., 2006. Carrier assignment models in transportation procurement. J. Oper. Res. Soc. 57, 1472–1481.
Kellner, F., Otto, A., 2012. Allocating CO2 emissions to shipments in road freight transportation. J. Manage. Control 22 (4), 451–479.
Klumpp, M., Bioly, D.S., Zelewski, S., 2009. Sustainability and technology innovation in logistics – friends or foes? In: Proceedings of Second International
Conference on Multinational Enterprises and Sustainable Development, 4–6 November, Nancy-Mets, France.
Kuhl, M.E., Zhou, X., 2009. Sustainability tool kit for simulation based logistics decisions. In: Proceedings of winter Simulation conference, pp. 1466–1473
Lee, C.G., Kwon, R.H., Ma, Z., 2007. A carrier’s optimal bid generation problem in combinatorial auctions for transportation procurement. Transp. Res. Part E
43 (2), 173–191.
168 R. Jothi Basu et al. / Transportation Research Part E 74 (2015) 152–168

Lieb, K.J., Lieb, R.C., 2010. Environmental sustainability in the third-party logistics (3PL) industry. Int. J. Phys. Distrib. Logist. Manage. 40 (7), 524–533.
Lim, A., Rodrigues, B., Xu, Z., 2008. Transportation procurement with seasonally varying shipper demand and volume guarantees. Oper. Res. 56 (3), 758–771.
Ma, Z., Kwon, R.H., Lee, C.G., 2010. A stochastic programming winner determination model for truckload procurement under shipper uncertainty. Transp.
Res. Part E 46 (1), 49–60.
McKinnon, A.C., Piecyk, M., 2010. Measuring and Managing CO2 Emissions of European Chemical Transport. Cefic, Brussels.
Presley, A., Meadeb, L., Sarkis, J., 2007. A strategic sustainability justification methodology for organizational decisions: a reverse logistics illustration. Int. J.
Prod. Res. 45 (18–19), 4595–4620.
Rainey, D.L., 2006. Sustainable Business Development: Inventing the Future through Strategy, Innovation, and Leadership. Cambridge University Press,
Cambridge, UK.
Rekik, M., Mellouli, S., 2012. Reputation-based winner determination problem for combinatorial transportation procurement auctions. J. Oper. Res. Soc. 63,
1400–1409.
Remli, N., Rekik, M., 2013. A robust winner determination problem for combinatorial transportation auctions under uncertain shipment volumes. Transp.
Res. Part C: Emerg. Technol. 35, 204–217.
Sandholm, T., 2002. Algorithm for optimal winner determination in combinatorial auctions. Artif. Intell. 135 (1–2), 1–54.
Sarkis, J., Helms, M.M., Hervani, A.A., 2010. Reverse logistics and social sustainability. Corporate Soc. Responsibility Environ. Manage. 17 (6), 337–354.
Sheffi, Y., 2004. Combinatorial auctions in the procurement of transportation services. Interfaces 34 (4), 245–252.
Sheu, J.B., Chou, Y.H., Hu, C.C., 2005. An integrated logistics operational model for green-supply chain management. Transp. Res. Part E 41, 287–313.
Song, J., Regan, A., 2003. Combinatorial auctions for transportation service procurement: the carrier perspective. Transp. Res. Rec. 1833, 40–46.
Srivastava, N.K., Viswanadham, N., Kameshwaran, S., 2008. Procurement of global logistics services using combinatorial auctions. In: Proceedings of fourth
IEEE Conference on Automation Science and Engineering, August 23–26, Key Bridge Marriott, Washington DC, USA, 297–302.
Tian, T., Wang, N., Ma, H., Lim, A., 2011. A transportation service procurement problem with combinatorial auction. In: Proceeding of 8th International
Conference on Service Systems and Service Management (ICSSSM), 25–27 June. Nankai University, Tianjin, China, pp. 1–6.
Wang, Y., Lalwani, C.S., 2007. Using e-business to enable customised logistics sustainability. Int. J. Logist. Manage. 18 (3), 402–419.
Yadati, C., Oliveira, C.A.S., Pardalos, P.M., 2007. An approximate winner determination algorithm for hybrid procurement mechanisms logistics. Adv.
Comput. Manage. Sci. 9 (1), 51–66.
Zhang, B., Ding, H., Li, H., Wang, W., Yao, T., 2013. A Sampling-Based Stochastic Winner Determination Model for Truckload Service Procurement. Networks
and Spatial Economics, Accepted in October 2013.
Zhu, Q., Sarkis, J., Lai, K., 2012. Green supply chain management innovation diffusion and its relationship to organizational improvement: an ecological
modernization perspective. J. Eng. Technol. Manage. 29 (1), 168–185.