You are on page 1of 8

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/303246512

Containerization Method for Logistic Cost Reduction Based on Closed-Loop


Supply Chain Model

Conference Paper · August 2015


DOI: 10.1115/DETC2015-46371

CITATIONS READS
0 72

4 authors, including:

Qing Chang Guoxian Xiao


University of Virginia General Motors Company
126 PUBLICATIONS   1,933 CITATIONS    90 PUBLICATIONS   1,733 CITATIONS   

SEE PROFILE SEE PROFILE

Jorge Arinez
General Motors Company
121 PUBLICATIONS   1,876 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Smart Manufacturing Analysis and Control, CPS in Manufacturing View project

production systems View project

All content following this page was uploaded by Qing Chang on 02 August 2016.

The user has requested enhancement of the downloaded file.


Proceedings of the ASME 2015 International Design Engineering Technical Conferences &
Computers and Information in Engineering Conference
IDETC/CIE 2015
August 2-5, 2015, Boston, Massachusetts, USA

DETC2015-46371

CONTAINERIZATION METHOD FOR LOGISTIC COST REDUCTION BASED ON


CLOSED-LOOP SUPPLY CHAIN MODEL

XinyanOu Qing Chang


Stony Brook University Stony Brook University
Stony Brook, NY, USA Stony Brook, NY, USA

Guoxian Xiao Jorge Arinez


General Motors Corp General Motors Corp
Warren, MI, USA Warren, MI, USA

ABSTRACT decades, efforts to minimize system-wide costs via reduction on


Logistics cost is an important contributor to the overall cost inventory has been increasingly addressed and attempted in
in a supply chain system. By using collapsible containers, the industry. Literatures dealing with this aspect of inventory
frequency of return freight can be reduced and the return of management can be categorized into two major types, the first
containers can be optimized, leading to potential logistic cost being deterministic analytical models [2-4] and the second being
savings. However, the dynamic behavior of container flows due stochastic models [5-7].
to demand, inventory, storage, and repair requirements make it In this paper, we introduce an innovative containerization
difficult to accurately analyze container system performance. An method (CM), which represents an opportunity to reduce
accurate estimation of this collapsible container usage impact is shipping schedules in return freights and thus the amount of
of great importance for decision-making. trucks operated and fuel consumed annually, to mitigate the
This paper describes the development of a mathematical trade-off between holding a low safety inventory level and the
model of the container dynamic flow system by using the requirement for flexible lead time. By using collapsible
collapsible containers. A continuous time, discrete space Markov containers, the return of containers can be optimized, leading to
process is used for stochastic scenario. The model determines the potential logistic cost savings while keeping a low quantity of
total cost savings, based on the collapsible rate, the number of safety inventory, a quantity that can be calculated by container
collapsible containers, the performance of the factory and the number in the system. The dynamic behaviors of container level
supplier and the transportation environment. The presented are investigated based on a closed-lined supply chain model
mathematical formulation enables the evaluation of the system consisting of one supplier and one factory. In order to quantify
performance. A case study of collapsible container supply chain the system-wide cost with or without uncertainties, such as
system demonstrates the advantages of this methodology. In random disruption events in manufacturing, a cost modeling
addition, a simulation model of this stochastic system is approach of the whole system is developed [8, 9], which bases
presented to verify the mathematical model. Simulation tests are on both deterministic and stochastic model. This paper
conducted to demonstrate the potential logistics cost savings in contributes to our understanding about the interactions between
the closed-loop supply chain system. inventory level and lead time adjustment through increasing or
decreasing the delivering frequencies, by which increasing the
deliver frequency would lower inventory level and shorten the
1. INTRODUCTION lead time of demand and vice versa. In addition, the incentives
A supply chain is an integrated manufacturing process to invest on collapsible container is studied based on cost
wherein raw materials are converted into final products [1]. Two analyses. Some key results of this paper demonstrate that there
basic characteristics in a supply chain management system are is a great advantage by using CM and an important improvement
inventory control and logistics management. During the last two of lead time flexibility.

1 Copyright © 2015 by ASME and General Motors


The rest of the paper is organized as follows: Section 2
describes the system and models the system as a continuous The structure of the CLSC system can be described as
closed-line supply chain system. The cost saving is formulated shown in Fig. 2. The supplier is modeled as a virtual machine
depending on the logistic saving and the extra containers cost. (denoted as S) producing parts while the factory is modeled as a
Section 3 presents the formulation of the extra container cost virtual machine (denoted as F) consuming parts. The
based on a CLSC with deterministic analysis and stochastic transportations are also modeled as virtual machines (denoted as
analysis. Section 4 gives a complete mathematical formulas of G and R). In addition, four virtual buffers are used to reflect the
the cost saving of the system. It also proposes a simulation uncertainties of the system. A continuous flow model is used to
method of the system and gives a verification for the cost saving analyze this closed-line system with exponential machines. The
formulas based on the simulation result. The paper ends with a definitions of variables and parameters are listed below.
discussion and conclusion remarks in section 5.

2. CLOSED-LOOP SUPPLY CHAIN MODEL

2.1 System Description


In this paper, we investigate a closed-lined supply chain
(CLSC) system with the use of collapsible containers as shown
in Fig. 1, which consists of a supplier and a factory. The supplier
produces parts and loads parts into containers. Then, a trip of
trucks deliver these containers to the factory. The factory
consumes parts, empties the containers, and returns them to the Figure 2. LAYOUT OF CLOSED-LINE SUPPLY CHAIN
supplier. Due to using collapsible containers, the return trips can
accommodate more containers and the frequency of return
freight is reduced so that the frequency of return freight is lower
than the frequency of the freight from the supplier to the factory. 𝑐𝑠 , 𝑐𝑓 : The rated speeds of the supplier and the factory,
This lower return freight frequency could potentially reduce the respectively.
logistic cost, however, it raises the probability of the supplier 𝑣𝑠 (𝑡), 𝑣𝑓 (𝑡): The processing rates of the supplierand the factory
starving the necessary containers, which may lead to the at time t, respectively.
production rate reduction of the factory. To avoid the supplier 𝑣𝑔𝑖𝑛 (𝑡), 𝑣𝑔𝑜 (𝑡), 𝑣𝑟𝑖𝑛 (𝑡), 𝑣𝑟𝑜 (𝑡) : The input rates and output
starving and mitigate the impact on the factory production rate rates of the transportation from S to F and transportation from F
due to using collapsible containers, the supplier uses extra to S at time t, respectively.
containers when there is no available collapsible containers in 𝑣𝑒𝑖𝑛 (𝑡): Input rate of extra containers at time t.
the supplier’s storage. The extra containers are disposable and 𝑣𝑒𝑜𝑢𝑡 (𝑡): The output rate of extra containers at time t.
non-collapsible and the empty extra containers will be discarded 𝐵𝑠𝑖 , 𝐵𝑠𝑜 , 𝐵𝑓𝑖 , 𝐵𝑓𝑜 : The buffer capacity of the upstream and
after use. downstream of the supplier and factory respectively.
𝑏𝑠𝑖 (𝑡), 𝑏𝑠𝑜 (𝑡), 𝑏𝑓𝑖 (𝑡), 𝑏𝑓𝑜 (𝑡) : The buffer level of
𝐵𝑠𝑖 , 𝐵𝑠𝑜 , 𝐵𝑓𝑖 , 𝐵𝑓𝑜 at time t, respectively. The buffer level means
the number of containers in the buffer.
𝑇𝑡𝑟𝑎𝑛𝑠 : The transportation time interval between the supplier
and the factory.
s: The number of trucks in one trip.
m: The maximum number of non-collapsed containers per truck
from the supplier to the factory.
M: The maximum number of collapsed containers per truck
from the factory to the supplier.
𝑇𝑠𝑑,𝑖 : The ith trip departure time from the supplier.
𝑇𝑓𝑑,𝑖 : The ith trip departure time from the factory.
𝜆𝑠 , 𝜇𝑠 : The supplier’s breakdown rate and repair rate.
𝜆𝑓 , 𝜇𝑓 : The factory’s breakdown rate and repair rate

Figure 1. CLOSED-LINED SUPPLY CHAIN (CLSC) We make the following assumptions:

2 Copyright © 2015 by ASME and General Motors


(a) The buffer capacity of 𝐵𝑠𝑖 , 𝐵𝑠𝑜 , 𝐵𝑓𝑖 , 𝐵𝑓𝑜 are big enough such 𝐸𝑥𝑡𝑟𝑎𝐶𝑜𝑛𝑡𝑎𝑖𝑛𝑒𝑟𝐶𝑜𝑠𝑡
that the supplier and the factory are never blocked.
(b)When the buffer 𝐵𝑠𝑖 is empty, extra containers are added and = 𝐶𝑒 ∙ 𝐸𝑥𝑡𝑟𝑎𝐶𝑜𝑛𝑡𝑎𝑖𝑛𝑒𝑟𝐶𝑜𝑠𝑡𝑁𝑢𝑚𝑏𝑒𝑟 (3)
the rated speed of input extra containers is equal to the rated
speed of the supplier 𝑐𝑠 . Where, 𝐶𝑒 is the cost of a single extra container. The extra
(c) At time 𝑡 = 0 , all the collapsible containers are in the container cost is proportional to the extra container number.
buffer𝐵𝑠𝑖 . The buffer 𝐵𝑠𝑜 , 𝐵𝑓𝑖 and 𝐵𝑓𝑜 are empty. However, the extra container number is relative to not only the
(d)The supplier is never starved by using the extra containers. random downtime event of the factory and the supplier but also
(e)The rated speeds of the supplier and the factory keep the same, the collapsed rate M/m.
i.e. 𝑐𝑠 = 𝑐𝑓
(f) The supplier freights the containers at predetermined
departure time 𝑇𝑠𝑑,𝑖 with a fixed interval sm/𝑐𝑠 . Namely, 3. EXTRA CONTAINER ANALYSIS
𝑠𝑚
𝑇𝑠𝑑,𝑖 = 𝑖 . As we describe before, the supplier uses the extra container
𝑐𝑠
(g) The factory freights the containers at predetermined when the buffer Bsi is empty. Two factors lead to the result that
the supplier runs out of the containers before the factory return
departure time 𝑇𝑓𝑑,𝑖 with a fixed interval sM/𝑐𝑓 . Namely,
𝑠𝑚 𝑠𝑀 the containers: first, the collapsible containers are not sufficient;
𝑇𝑓𝑑,𝑖 = + 𝑇𝑡𝑟𝑎𝑛𝑠 + 𝑖 . second, disruption events prevent the factory returning enough
𝑐𝑠 𝑐𝑓
(h) The parts in collapsible containers have higher priority to be containers in time. Firstly, we discuss the effect of the initial
consumed than those in extra containers in the factory. In other number of collapsible container on the production system with
words, if there are both collapsible and non-collapsible deterministic downtime event. Then, we consider the effect of
containers (extra containers) in buffer 𝐵𝑓𝑖 , the factory consumes random disruption events.
the parts in collapsible containers first.
3.1 Deterministic and Transient Analysis
Based on the conservation of flow, the buffer levels during
2.2 Cost Classification and Cost Allocation (0, T] satisfies the following equations:
A general supply chain is a complex network system which 𝑇
encompasses a set of organizations and activities facilitating 𝑏𝑠𝑜 (𝑇) − 𝑏𝑠𝑜 (0) = ∫0 [𝑣𝑠 (𝑡) − 𝑣𝑔𝑖𝑛 (𝑡)]𝑑𝑡 (4)
economical and efficient flow of raw material to final products.
Cost is one of the most effective indexes to evaluate the 𝑇
𝑏𝑓𝑖 (𝑇) − 𝑏𝑓𝑖 (0) = ∫0 [𝑣𝑔𝑜 (𝑡) − 𝑣𝑓 (𝑡)]𝑑𝑡 (5)
performance of a supply chain system, for its clarity and ability
to provide guidance for system wide cost saving. However, an 𝑇
accurate cost estimation of all activities involved in raw material 𝑏𝑓𝑜 (𝑇) − 𝑏𝑓𝑜 (0) = ∫0 [𝑣𝑓 (𝑡) − 𝑣𝑟𝑖𝑛 (𝑡) − 𝑣𝑒𝑜𝑢𝑡 (𝑡)]𝑑𝑡 (6)
sourcing, procurement, conversion and logistics within a supply
𝑇
chain is important but difficult for any single entity get involved 𝑏𝑠𝑜 (𝑇) − 𝑏𝑠𝑜 (0) = ∫0 [𝑣𝑟𝑜 (𝑡) + 𝑣𝑒𝑖𝑛 (𝑡) − 𝑣𝑠 (𝑡)]𝑑𝑡 (7)
in the system. The primary goal in this paper is to present a
quantitative method for cost saving of using collapsible 𝑇
∫0 𝑣𝑠 (𝑡)𝑑𝑡 is the container count of S, which is the number
containers. The total cost saving problem includes: the cost of
extra containers and logistic saving. Other operational impacts, containers used by S during (0, 𝑇].
𝑇
including extra labor cost for container handling and ∫0 𝑣𝑓 (𝑡)𝑑𝑡 is the container count of F, which is the number
maintenance, inventory control and warehouse management can containers consumed by F during (0, 𝑇].
all be translated into calculating container cost in the system. 𝑇
∫0 𝑣𝑒𝑖𝑛 (𝑡)𝑑𝑡 is the container count of total number of extra
Thus, the total cost saving per unit time is formulated as
containers used by the supplier during (0, 𝑇].
𝑇
𝐶𝑜𝑠𝑡𝑆𝑎𝑣𝑖𝑛𝑔𝑡𝑜𝑡𝑎𝑙 ∫0 𝑣𝑔𝑖𝑛 (𝑡)𝑑𝑡 is the integration of the input rate of
transportation, showing the container count departing from the
= 𝐿𝑜𝑔𝑖𝑠𝑡𝑖𝑐𝑆𝑎𝑣𝑖𝑛𝑔 − 𝐸𝑥𝑡𝑟𝑎𝐶𝑜𝑛𝑡𝑎𝑖𝑛𝑒𝑟𝐶𝑜𝑠𝑡 (1) supplier during (0, 𝑇].
𝑇
∫0 𝑣𝑔𝑜 (𝑡)𝑑𝑡 is the container count received by the factory
The logistic saving per unit time can be obtained as during (0, 𝑇].
𝑇 𝑇
m 𝑠𝑚 Similarly, ∫0 𝑣𝑟𝑖𝑛 (𝑡)𝑑𝑡 and ∫0 𝑣𝑟𝑜 (𝑡)𝑑𝑡 indicate the
𝐿𝑜𝑔𝑖𝑠𝑡𝑖𝑐𝑆𝑎𝑣𝑖𝑛𝑔 = (1 − ) C 𝑇 / (2) container counts departing from the factory and arriving at the
M 𝑐𝑠
suppliers, respectively.
Where, C 𝑇 is logistic cost per trip. Fig. 3a and 3b illustrate the container count trajectories with
The extra container cost per unit time is different initial number of collapsible containers without any
random disruption events.

3 Copyright © 2015 by ASME and General Motors


The container count of S and F is denoted as two slope lines
in Fig. 3a. The container count of S increases as S rated speed When the container count trajectory of S coincides with the
c𝑠 . Due to the transportation delay, the container count of F is dash line, it means all the input containers and initial containers
right shifted and parallel to the trajectory of container count of has been used up by S, so the buffer 𝐵𝑠𝑖 is empty. Thus, the
S. coincided part of lines indicates the time period of using extra
The step lines between the container count lines of S and F container.
show the container counts of the transportation from S to F. And The initial number of collapsible containers 𝑏𝑠𝑖 (0) is
𝑇
the two step lines below the trajectory line ∫0 𝑣𝑓 (𝑡)𝑑𝑡 shows crucial to the utilization of extra containers. Fig. 3b indicates that
the container counts of the transportation from the F to S. when the initial collapsible containers are not sufficient, the extra
𝑇 containers are incurred periodically (shown as the trajectory line
𝑏𝑠𝑖 (0) + ∫0 [𝑣𝑒𝑖𝑛 (𝑡) + 𝑣𝑟𝑜 (𝑡)]𝑑𝑡 is the container count of 𝑇
of ∫0 𝑣𝑒𝑖𝑛 (𝑡)𝑑𝑡 ) to keep the factory normal operation. .
total input containers and initial container number of buffer 𝐵𝑠𝑖 .
It is denoted as the dash step line in Fig. 3a and 3b. The minimum number of initial collapsible containers for
the system without a need of using extra containers periodically
is formulated as following,

𝑏𝑛𝑒𝑒𝑑 = 𝑠𝑚 + 𝑠𝑀 + 2𝑇𝑡𝑟𝑎𝑛𝑠 ∙ 𝑐𝑠 (8)

𝑏𝑛𝑒𝑒𝑑 is a threshold to indicate whether the collapsible


containers are sufficient.
Thus, the extra container number per unit time without
downtime event is

ExtraContainerNumber

0 𝑏𝑠𝑖 (0) ≥ 𝑏𝑛𝑒𝑒𝑑


={ 𝑏𝑛𝑒𝑒𝑑 −𝑏𝑠𝑖 (0)
(9)
⌈𝑏𝑛𝑒𝑒𝑑 /𝑠𝑀⌉∙(𝑠𝑀 ⁄𝑐𝑓 )
𝑏𝑠𝑖 (0) < 𝑏𝑛𝑒𝑒𝑑

Then, we denote a disruption event 𝑒⃗ as a 3-tuple (s, 𝑇𝑑 , 𝑑)


or (F, 𝑇𝑑 , 𝑑) to indicate that the supplier or the factory is
(a) COLLAPSIBLE CONTAINERS ARE SUFFICIENT disrupted at time 𝑇𝑑 for a duration of d. We use 𝑣𝑠 (𝑡; 𝑒⃗) and
𝑣𝑓 (𝑡; 𝑒⃗) to denote the processing rate of the supplier and the
factory at time t when the system is disturbed by event 𝑒⃗ ,
respectively.
Fig.4 illustrates the container count trajectories when the
supplier is subject to a disruption event (s, 𝑇𝑑 , 𝑑). As shown in
Fig. 4, when the system has sufficient collapsible containers, a
single disruption event on the supplier will not lead to using extra
T containers, and it has effect on the container count of the factory
0
vgo (t )dt
and the supplier.
T Fig.5 illustrates the container count trajectories when the
 0
vgin (t )dt
T factory is subject to a disruption event (F, 𝑇𝑑 , 𝑑). Only when the
T  vro (t )dt
downtime lasts long enough for the buffer 𝐵𝑠𝑖 to become empty,

0
vs (t )dt
0
T the extra containers will have to be introduced into the system.
T  vein (t )dt 𝑇
The trajectory of ∫0 𝑣𝑒𝑖𝑛 (𝑡; 𝑒⃗)𝑑𝑡 in Fig. 5 indicates that extra

0
vrin (t )dt
T 0

0
v f (t )dt containers are not needed after the factory back up. On one hand,
more initial collapsible containers brings longer maximum
downtime duration for the factory without introducing extra
containers. On the other hand, more initial collapsible containers
(b) COLLAPSIBLE CONTAINERS ARE INSUFFICIENT can reduce the extra containers usage when the downtime last
long enough to require extra containers.
Figure 3. CONTAINER COUNTS WHEN COLLAPSIBLE
CONTAINERS ARE SUFFICIENT AND INSUFFICIENT

4 Copyright © 2015 by ASME and General Motors


𝑏𝑠𝑖 (0) and collapsed rate M/m . We use the exponential
reliability model to represent the supplier and the factory. The
uptime and downtime probability density function of the supplier
and the factory are given by the exponential distributions given
by parameters λ𝑠 , 𝜇𝑠 , λ𝑓 , 𝜇𝑓 , respectively. The states of the
T factory and the supplier are independent. Hence, the dynamic
0
vgo (t )dt system can be described by a continuous time, discrete space
T Markov process.
 0
vgin (t )dt
According to the assumptions, the supplier is never starved
T (by using extra container) and the factory is never blocked. Thus
0
vro (t )dt the entire closed line system can be formulated in two cases:
T
 vs (t )dt
(a) 𝑏𝑠𝑖 (0) ≥ 𝑏𝑛𝑒𝑒𝑑
T

0
vrin (t )dt
T 0


0
v f (t )dt
𝑒𝑠 ∙ 𝑄(λ𝑓 , 𝜇𝑓 , λ𝑠 , 𝜇𝑠 , 𝑁𝑒𝑞 ) 𝑒𝑠 ≤ 𝑒𝑓
𝑃𝑅𝑒 = { (10)
𝑒𝑓 ∙ 𝑄(λ𝑠 , 𝜇𝑠 , λ𝑓 , 𝜇𝑓 , 𝑁𝑒𝑞 ) 𝑒𝑠 > 𝑒𝑓
Figure 4. CONTAINER COUNT WITH A SUPPLIER
DOWNTIME EVENT
Where,
𝑣 ∙ 𝑇𝑡𝑟𝑎𝑛𝑠
𝑁𝑒𝑞 = [𝑏𝑠𝑖 (0) − 𝑏𝑛𝑒𝑒𝑑 ]
𝑠𝑀
1
𝜆𝑓 𝑠𝑀−𝑠𝑚−𝑣∙𝑇𝑡𝑟𝑎𝑛𝑠
+ (𝜇𝑠 − + 21 ) 𝑐𝑠 (11)
𝜆𝑓 +𝜇𝑓 𝑠𝑀+𝑠𝑚+𝑣∙𝑇𝑡𝑟𝑎𝑛𝑠
2

T
𝑄(λ1 , 𝜇1 , λ2 , 𝜇2 , 𝑁)
 0
vgo (t )dt
T (1−𝑒1 )(1−Φ) λ1 λ2
 0
vgin (t )dt
1−Φ𝑒 −𝛽𝑁 𝜇1
𝑖𝑓 ≠
𝜇2
T 𝜆1 (𝜆1 +𝜆2 )(𝜇1 +𝜇2 )

T
vs (t )dt  0
vro (t )dt = (𝜆1 +𝜇1 )[(𝜆1 +𝜆2 )(𝜇1 +𝜇2 )+λ2 𝜇1 (𝜆1 +𝜆2 +𝜇1 +𝜇2 )𝑁]
(12)
0
T λ1 λ2
0
vrin (t )dt
T { 𝑖𝑓
𝜇1
=
𝜇2
T  vein (t )dt
 0
v f (t )dt 0
𝜇𝑖
𝑒𝑖 = 𝑖 = 1,2 (13)
λ𝑖 +𝜇𝑖

Figure 5.CONTAINER COUNT WITH A FACTORY Φ=


𝑒1 (1−𝑒2 )
(14)
𝑒2 (1−𝑒1 )
DOWNTIME EVENT
(𝜆1 +𝜆2 +𝜇1 +𝜇2 )(𝜆1 𝜇2 −𝜆2 𝜇1 )
β= (𝜆1 +𝜆2 )(𝜇1 +𝜇2 )
(15)
From the deterministic analysis, we can conclude that more
initial collapsible containers can reduce the utilization of extra (b) 𝑏𝑠𝑖 (0) < 𝑏𝑛𝑒𝑒𝑑
containers. When the collapsible containers are less than a
threshold value 𝑏𝑛𝑒𝑒𝑑 , using more collapsible containers can When 𝑏𝑠𝑖 (0) < 𝑏𝑛𝑒𝑒𝑑 , extra containers are added
reduce the periodic utilization of extra containers. When the periodically into the system according to deterministic analysis.
initial number of collapsible containers is beyond 𝑏𝑛𝑒𝑒𝑑 , more The production rate of extra containers can be obtained as
initial collapsible containers can better protect the factory following:
disruption events.
[𝑏𝑛𝑒𝑒𝑑 −𝑏𝑠𝑖 (0)]/𝑐𝑠
𝑃𝑅𝑒 ≈ 𝑒𝑠 ∙ ⌈𝑏 (16)
𝑛𝑒𝑒𝑑 /𝑠𝑀⌉∙(𝑠𝑀 ⁄𝑐𝑓 )

3.2 Stochastic Analysis


According to the previous assumptions, the rated speed of
In this section, we consider the system encountering input extra containers is equal to the supplier rated speed. Thus,
multiple random disruption events on both the factory and the extra container number and cost per unit time are obtained,
supplier. Then average extra container per unit time depends on

5 Copyright © 2015 by ASME and General Motors


𝐸𝑥𝑡𝑟𝑎𝐶𝑜𝑛𝑡𝑎𝑖𝑛𝑒𝑟𝑁𝑢𝑚𝑏𝑒𝑟 = 𝑃𝑅𝑒 ∙ 𝑐𝑠 (17)
Table 1. PARAMETERS FOR SIMULATION
𝐸𝑥𝑡𝑟𝑎𝐶𝑜𝑛𝑡𝑎𝑖𝑛𝑒𝑟𝐶𝑜𝑠𝑡 = 𝑃𝑅𝑒 ∙ 𝑐𝑠 ∙ 𝐶𝑒 (18)

𝐶𝑇 281 dollars per trip


4. COST SAVING AND SIMULATION VERIFACATION
From the stochastic analysis and the total cost saving 𝐶𝑒 100 dollars per extra container
formulas, we obtain the total cost saving. And the results are
shown in Fig. 6. 𝑇𝑡𝑟𝑎𝑛𝑠 1 day per trip

𝐶𝑜𝑠𝑡𝑆𝑎𝑣𝑖𝑛𝑔𝑡𝑜𝑡𝑎𝑙 𝑣𝑠 10 containers per day


20 containers per trip
= 𝐿𝑜𝑔𝑖𝑠𝑡𝑖𝑐𝑆𝑎𝑐𝑖𝑛𝑔 − 𝐸𝑥𝑡𝑟𝑎𝐶𝑜𝑛𝑡𝑎𝑖𝑛𝑒𝑟𝐶𝑜𝑠𝑡 𝑠𝑚
from the supplier to the factory
= (1 − 𝑚/𝑀)𝐶𝑇 /(𝑠𝑚/𝑐𝑠 )−𝑃𝑅𝑒 ∙ 𝑐𝑠 ∙ 𝐶𝑒 20 ∙ 𝑀/𝑚 containers per trip
(19) 𝑠𝑀
from the factory to the supplier
Unit time 1 day

λ𝑠 , λ𝑓 0.1

𝜇𝑠 , 𝜇𝑓 0.5

Figure 6. COST SAVING VS. bsi(0) WITH DIFFERENT


COLLAPSED RATES

From Fig. 6, higher collapsed rate leads to more cost saving


only when the initial number of collapsible containers is large
enough. The lower collapsed rate is more economic when the
number of collapsible containers limited to a small value. Figure 7. COMPARISON BETWEEN SIMULATION
A simulation model is built by using MATLAB code. In the RESULT AND MATHEMATICAL RESULT
simulation, the exponential random function is used to generate
the downtime events to satisfy the stochastic property of the
system. The simulation can calculate the average cost saving per It can be observed that the analytic derivation of cost savings
unit time during the simulation time. matches the simulation results very well. The cost saving will
In the simulation study, the simulation time is set as 10000 saturate when 𝑏𝑠𝑖 (0) → ∞ and the maximum cost saving is
unit time with step interval 0.01 repeated for 25 iterations. The equal to the logistic saving due to extra container cost is equal to
parameters are listed in Tab. 1, which is based on a real factory zero. Since the cost saving increases slightly when 𝑏𝑠𝑖 (0) is
scenario and the data is mockup for confidential reason. large enough, there is no reason to have an extremely large
Fig. 7 shows the comparison between the simulation result 𝑏𝑠𝑖 (0).
and the mathematical result from the previous formulas. In addition, the simulation takes a long time to calculate the
average cost saving. Normally, each scenario takes 20 minutes
and it needs hours to finish a rough curve showing different
scenarios of 𝑏𝑠𝑖 (0). However, the analytic method can estimate

6 Copyright © 2015 by ASME and General Motors


the curve in seconds with an acceptable precision. It substantially of the 2008 Industrial Engineering Research Conference,
reduces the calculating cycle of researching the tradeoff under J.Fowler and S.Mason, Eds.
different parameters or variables. [9] Chang, Q., Liu, J., Biller, S., and Xiao, G.,“The Costs of
Downtime Incidents in Serial Multistage Manufacturing
Systems”, ASME Transaction, Journal of Manufacturing
5. CONCLUSION Science and Engineering, Vol. 134, 2012.
This paper investigates the potential cost saving of using [10] Chang, Q., Ni, J., Bandyopadhyay, P., Biller, S., and Xiao,
collapsible containers in a supply chain system. Through the G., “Supervisory Factory Control Based on Real-Time
analysis of cost classification and allocation, the calculation of Production Feedback”, ASME Transaction, Journal of
the cost in the system can be divided into two parts, the first Manufacturing Science and Engineering, Vol. 129, pp653-
being the extra containers cost and the second being the logistic 660, 2007.
saving. The deterministic analysis and stochastic analysis are [11] Li, Jingshan, and Semyon M. Meerkov. “Production
used to formulate the cost saving function depending on the systems engineering”. Springer, 2008.
initial number of collapsible container and the collapsed rate.
From our simulation tests and the formulas, this paper
demonstrates the cost savings by using the Containerization
Method and illustrates the tradeoff between the collapsible
containers and the cost saving. The formulas can rapidly estimate
the system cost saving with a high precision.

ACKNOWLEDGMENTS
This work is supported in part by the General Motors R&D
Center.
The authors would like to thank Ms. Jingfang Tang for her
help in this work.

REFERENCES
[1] Benita M. Beamon, “Supply chain design and analysis
Models and methods”, Int.J.Production Economics 55 281-
294, 1998.
[2] K. Ishii, K. Takahashi, R. Muramatsu, “Integrated
Production, inventory and distribution systems”,
International Journal of Production Research, 26 (3) 81-
104, 1988.
[3] M.A. Cohen, H.L. Lee, “Resource deployment analysis of
global manufacturing and distribution networks”, Journal
of Manufacturing and Operations Management, 2 81-104,
1989.
[4] M.A. Cohen, H.L. Lee, “Strategic analysis of integrated
production distribution systems: Models and methods”,
Operations Research, 36 (2) 216-228, 1988.
[5] H.L. Lee, C. Billington, “Material management in
decentralized supply chains”, Operations Research, 41 (5)
835-847 1993.
[6] H.L. Lee, C. Billington, B. Carter, “Hewlett-Packard gains
control of inventory and service through design for
localization”, Interfaces, 23 (4) 1-11, 1993.
[7] D.F. Pyke, M.A. Cohen, “Performance characteristics of
stochastic integrated production-distribution systems”,
European Journal of Operational Research, 68 (1) 23-48,
1993.
[8] Chang, Q., Liu, J., Biller, S., and Xiao, G., “Cost Modeling
of Quality Loops in Manufacturing Systems”, Proceedings

7 Copyright © 2015 by ASME and General Motors

View publication stats

You might also like