CHEMICAL ENGINEERING TRANSACTIONS

VOL. 51, 2016

A publication of

The Italian Association

of Chemical Engineering
Online at www.aidic.it/cet

Guest Editors: Tichun Wang, Hongyang Zhang, Lei Tian
Copyright © 2016, AIDIC Servizi S.r.l.,

ISBN 978-88-95608-43-3; ISSN 2283-9216

Lateral Transshipment Model in the Same Echelon Storage
Sites

Yanning Jianga,b, Yue Yang*a, Shuchi Haoc
aSchool of Traffic and Transportation Engineering, Central South University, Changsha 410083,PRC
bSchool of Geographical Sciences, Guangzhou University, Guangzhou, 510006, PRC
cDepartment of Commerce, Guangzhou City Polytechnic, Guangzhou 510405,PRC
yangyue@csu.edu.cn

In case of one-product, one-period, normal distribution demand, emergency lateral transshipment instant
arrival, four transshipment rules in the same echelon storage sites are discussed in this paper. They are one-
time and full-sharing transshipment, one-time and partial-sharing transshipment, multiple-time and full-sharing
transshipment, and multiple-time and partial-sharing transshipment. A total cost model is constituted on the
factors of transshipment cost and shortage cost. Numerical experiments show that multiple-time
transshipment can avoid the shortage risk of supply location; partial-sharing transshipment can drive down the
shortage risk of supply location; decision of transshipment rule mostly depends on unit cost of transshipment,
unit cost of shortage, and distance between the storage sites.

1. Introduction

Starting from the 1970s, researchers did a lot of theoretical study on the lateral transshipment in the same
echelon based on the assumption: there is no safety storage in every storage site, and transshipment will be
carried from the available site to the shortage site in one certain rule. Lateral transshipment is based on the
theory of risk pooling effect or shared inventory in order to realize centralized management and customer
service. Lateral transshipment is an effective tool to adjust the balance between demand and inventory level.
Due to the risk-pooling effect, the total cost will be decreased or the customer satisfaction level will be
increased, when the increased cost of lateral transshipment is less than the cost of safety stock or emergent
replenishment. However, the cost of lateral transshipment is highly dependent on its rules.

2. Research Review

A comprehensive review of current researches is listed as follows.

2.1 Rules and costs of transshipment
(Das, 1975) investigated the rules and costs of transshipment between two locations based on random
customer demand and periodical inventory contral policy, and then discussed the applicable ordering strategy
considering the availability of transshipment. (Karmarkar and Patel, 1977) studied the non-directional
transshipment among several locations. (Kukreja and Schmidt, 2005) considering the change of demand and
lead time, formulated a transshipment model with the assumption of full-sharing transshipment, lateral
transshipment instant arrival, and transshipment cost relevant with transshipment frequency. (Yu and Liu,
2013) studied the transshipment between online stores and brick-and-mortar retailers, and concluded that the
optimum level of inventory increases with the increase of transshipment cost. When the cost of transshipment
is moderate, the inventory of online stores with transshipment is lower than that without transshipment; while
the inventory of brick-and mortor retailers is reverse.

2.2 Dynamic transshipment
(Robinson, 1990) studied the dynamic and random storage problem when transshipment happens among
several storage locations, and constructed a model to optimize the ordering and transhipping strategy in order
to minimize the total expected cost. (Man, 2012) proposed a cause-effect loop diagram for transhipping relief

DOI: 10.3303/CET1651125

Please cite this article as: Jiang Y.N., Yang Y., Hao S.C., 2016, Lateral transshipment model in the same echelon storage sites, Chemical
Engineering Transactions, 51, 745-750 DOI:10.3303/CET1651125

745

material reservation, simulated with system dynamics, and discovered the implementation effect of the same
inventory policy depends on the demand probability, unit transshipment cost and purchasing cost prior to the
disaster. (Donmez and Turkay, 2013) constructed a mixed-integer linear programming model for the design of
reverse logistics network that includes collection, sorting, export, recycling and disposal of waste batteries at a
landfill area.

2.3 Transshipment in different inventory policies
(Xu et al., 2003) studied the transhipment problem assuming fixed ordering cost, independent (Q, R) inventory
strategy and partial-sharing inventory, and subsequently analyzed the impact of transshipment on probability
of non-shortage and demand satisfaction level. (Hu et al., 2005) by constructing the dynamic programming
algorithm, minimized the cost of inventory and transshipment based on (s, S) inventory strategy, and studied
how the transhipment policy is influenced by transhipment cost, holding cost and shortage cost. (Huo and Li,
2007) established a batch ordering model for transhipping spare parts among multiple locations in the same
echelon, and calculated the probabilities of demand satisfaction, transshipment and shortage by configuring
three factors: inventory level, required lead time and net inventory.
In summary, the current studies, first of all, defined the amount of transshipment relying either on empirical
data or on one-time transshipment rule, which may lead to the shortage of supply location after transshipment,
therefore this is not an optimized strategy in the long run. Secondly, to calculate the transshipment cost, only
the quantity was taken into consideration, neglecting the impact of transshipment distance. Lastly, the risk of
shortage after transshipment at the supply location was ignored in the current studies. Therefore it is practical
to investigate the realistic transshipment rules and their applicability, and calculate the costs of various
transshipment rules.

3. Model description and assumption

Two factors are considered when categorizing the transshipment rules:

3.1 Times of transshipment: one-time or multiple-time transshipment
One-time transshipment means the required quantity (reorder point minus inventory) of demand location
(whose inventory is less than reorder point) is transported in one-time from the supply location (whose
inventory is more than reorder point), which later on may lead to the shortage of the supply location. Multiple-
time transshipment means only the surplus (inventory minus reorder point) is transported from the supply
location to the demand location, which may require transhipments from multiple supply locations. An example
is illustrated in Figure 1, where Vj0 is the initial inventory; Rj is the reorder point; and djj is the distance between
two locations.

Figure 1: One-time and multiple-time transshipment Figure 2: Full sharing and partial sharing transshipment

As shown in Figure 1, for V10<R1, location 1 runs into an inventory shortage of 20; the surplus in location 2 is 8;
and the surplus in location 3 is 7. The replenishment will always be transported from the nearest location.
Let’s firstly apply one-time transshipment principle. The shortage of 20 in location 1 will be transported from
location 3, which will lead to a shortage of 13 in location 3. Accordingly, the shortage in location 3 will be
replenished from location 2, and eventually this will result in a shortage of 5 in location 2.
By following the multiple-time transshipment principle, the surplus of 7 in location 3 will firstly be transported to
location 1, and another 8 from location 2 to location 1, and finally there will be a shortage of 5 in location 1.

3.2 Degree of sharing: full-sharing and partial-sharing.

As illustrated in Figure 2, full-sharing means transshipment occurs when Vj0>Rj, partial-sharing occurs only
when Vj0>Rj+Hj, where Hj is the reserved safety inventory.

d12=50

03
V =75，

R3=68，
H3=9

01
V =30，

R1=50，
H1=8

02
V =87，

R2=70，
H2=9

d13=40

d23=28 1

3
03

V =77，

R3=70

d23=28

02
V =58，

R2=50

d13=40

d12=50
01

V =30 ，

R1=50

746

As shown in Figure 2, a shortage of 20 occurs in location 1. By the partial-sharing principle, only location 2 is
qualified as the supply source. By the full-sharing principle, both location 2 and 3 are qualified as the supply
source.
Two problems will be discussed in this paper. They are which is the most optimal transshipment rule in
general and how the factors of unit cost of shortage, unit cost of transshipment and distance between storage
sites impact on the transshipment rules. We assume family of the storage sites is J, and the demand of every
storage site follows normal distribution.

3.3 Assumptions
We make the following assumptions:
a) transshipment is taken place among the same echelon storage sites, the coordinates of each storage
locations are known;
b) single product, single period, the independent requirement at each storage site complies to normal
distribution and (Q, R) inventory strategy;
c) assuming instant arrival of transshipment, distance of transshipment is calculated by coordinates of supply
and demand sites;
d) holding cost during transshipment is included into transshipment cost;
e) total cost includes transshipment cost and shortage cost; and
f) the nearest supply location is the preference select.

4. Modeling the transshipment among locations at the same echelon

We begin with a few basic definitions. The notation for a series of parameters that are used in the model is
presented in Table 1.

Table 1: Notations for key parameters and variables in the model

Notation Definition Notation Definition

K Safety factor of inventory, as the index of
inventory availability, which is related to
the service level

J Family of the storage sites,
where j is a certain site

C1 Unit lateral transshipment cost
(RMB/t.km)

C2 Unit shortage cost (RMB/t)

uj Mean value of demand at location of j
j Standard deviation of the

demand at location of j
Lj Mean value of lead time at location of j

(day)
σLj Standard deviation of lead time

at location of j (day)
djj Distance between locations Rj Reorder point at location of j

Vjn Storage amount at location of jafter
taking n iterations

Vj0 Initial storage amount at
locaiton of j

Zjn Required transshipment amount at
location of j after taking n iterations, is
equal to the shortage at location of j after
taking n iterations

Hj Reserved storage for avoiding
the risk of stockout at location
of j

n Iterative times

We consider that shortage occurs in location of j, that is, Vjn<Rj, in the following four kinds of models.
In case of one-time and full-sharing transshipment model, we consider location of j’(j’ ∈j) satisfied by the
conditions of Vjn>Rj and Rj=Lj×uj is exist. Thus supply location is selected by djj’=min{djk,k=1,2,…j’}, and then
the shortage amount in demand location of j, that is Zjn can be presented as Zjn=Rj-Vjn.
In case of one-time and partial-sharing transshipment model, we assume there exist location of j’(j’ ∈j)
satisfied by the conditions of Vj’n>Rj’+Hj’ and Rj’=Lj’ ×uj’, select the supply location by djj’=min{djk, k=1,2,…j’},
and conclude the shortage amount in demand location of j, that is, Zjn=Rj-Vjn.
In case of multiple-time and full-sharing transshipment model, assuming shortage is not allowed in supply
location, that is, Zjn≤(vjn-Rj’), a demand location may be served by several supply locations, or a supply

747

location may serve several demand locations. We consider that location of j’ (j’ ∈J) satisfied by conditions of

Vj’n>Rj’ and Rj’=Lj’ ×uj’ is exist, supply location is selected by djj’=min{djk, k=1,2,…j’}, and the formula of Zjn, the
shortage amount in demand location of j , is Zjn =min[(Rj- Vjn),(Vjn-Rj’)].

In case of multiple-time and partial-sharing transshipment model, location of j’ (j’ ∈J) can be satisfied by
conditions of Vj’n>Rj’+Hj and Rj’=Lj’ ×uj’. We can select supply location by djj’=min {djk, k=1, 2, …j’} and conclude
the shortage amount in demand location of j , that is, Zjn= min [(Rj- Vjn), (Vjn-Rj’-Hj’)].

5. Model Solution

The four steps in the model solution are as follows.
Step 1, we calculate and sequence the initial shortage amount in location of j, that is, (Rj-Vj0); the location of
maximum initial shortage amount is supplied proper transshipment amount from selected supply location
according to the above transshipment rules. Thus, the first transshipment is taken place. And then we
calculate cost of transshipment, that is, CLT1=C1×djj×zj0, and the inventory amount of supply location and
demand location after transshipment, that is, Vj1;
Step 2, we calculate the shortage amount in location of j once again after the first transshipment, that is,
CLT2…CLTn, repeat step 1 until there is not demand location or supply location any more;
Step 3, we calculate the shortage cost in location of j after end of iterations, that is, Cq1…Cqj, the formula is
presented as follows:

)()()(2 xdxfVxCC
nj

nj
V

jq 


 (1)

where ）（ 222
2

)(

222
2

1
)(

jLjjj

uLx

Ljjj

xf 









 ;

Step4, we obtain the total cost, that is, TC, where

jn qqqLTLTLT
CCCCCCTC  

2121
(2)

6. Numerical examples

In our examples, for K=1.28, C1=0.3/(t.km), C2=15/t, and other data is as shown in Table 2.

Table 2 Coordinates of each location and variables of storage

locations Coordinates(km) Vj0 (t) Hj (t) uj (t) σj (t) Lj (day) σLj (day)

1 (-40, 10) 73 11 45 5 2 0.1

2 (-20, 70) 125 17 37 4 3 0.3

3 (-42, -32) 70 19 29 6 2 0.4

4 (-70, -43) 95 12 38 3 3 0.2

5 (-10, -60) 107 20 44 6 2 0.3

6 (-40, 32) 165 20 57 6 3 0.2

7 (-68, 10) 190 68 52 4 4 1

8 (12, 15) 50 9 35 5 1 0.1

9 (0, -40) 79 17 28 7 2 0.3

10 (-10, -10) 145 24 56 4 3 0.3

Programming and calculating in Matlab, we get the following results as shown in Table 3 and Table 4.

748

Table 3 Results of various transshipment rules

Transshipment
rules

One-time and
full-sharing
transshipment

One-time and
partial-sharing
transshipment

Multiple-time
and full-sharing
transshipment

Multiple-time and
partial-sharing
transshipment

Cost(RMB) TC1=2,531 TC2=2,026 TC3=2,353 TC4=1,949

Table 4 Sensitivity analysis of parameters

Parameters
Cost(RMB)

Ranking of rules
TC1 TC2 TC3 TC4

djj’
50%×djj’ 1,762 1,630 1,673 1,890 TC2, TC3, TC1, TC4

150%×djj’ 3,300 2,421 3,033 2,007 TC4, TC2, TC3, TC1

C1
C1=0.15 1,762 1,630 1,673 1,890 TC2, TC3, TC1, TC4

C1=0.45 3,300 2,421 3,033 2,007 TC4, TC2, TC3, TC1

C2
C2=7.5 2,034 1,408 1,856 1,032 TC4, TC2, TC3, TC1

C2=22.5 3,028 2,643 2,850 2,864 TC2, TC3, TC4, TC1

Table 3 shows the ranking of lateral transshipment rules is TC4, TC2, TC3, TC1. In the case of one-time
transshipment, the supply location may be new demand location after transshipping, which may cause
circuitous transshipping and increasing transshipment cost. While full-sharing transshipment may give rise to
the increasing shortage cost at supply location. Therefore, the transshipment rule integrated multiple-time
transshipment with partial-sharing is preferable choice.
Table 4 presents in the case of one-time transshipment, supply location may be new demand location, which
may lead to the increase of total transshipment distance and amount, therefore, multiple-time transshipment
rule is better, especially when the distance between storage sites is longer or unit cost of transshipment is
higher. Bigger unit cost of shortage may decrease the ratio of transshipment cost to total cost, so one-time
and partial-sharing transshipment is superior.

7. Conclusion

Lateral transshipment is a new and effective way to regulate the balance between demand and supply. The
following conclusions are reached after modeling and experimenting:
a) One-time transshipment may lead to the shortage of the supply location itself when its inventory surplus is
less than the replenishing amount required at the demand location, which will cause subsequent secondary or
multiple cross-haul transshipment and greatly increase the total transshipment cost. However, multiple-time
transshipment will avoid this puzzlement;
b) Full-sharing transshipment will increase the risk of subsequent shortage of the supply location and hence
the total cost of shortage, but the partial-sharing transshipment will effectively tackle this problem; and
c) Other factors, such as distance and cost, have impact on the ranking of various transshipment rules, so
proper rules are highly dependent on the real scenario of each case.

Acknowledgments

This work was supported by The General Program of National Natural Science Foundation of China (No.
41271175): The spatial structure and dynamic mechanism of innovation in knowledge-intensive business
services in China’s metropolitan.

749

Reference

Das C., 1975, Supply and redistribution rules for two-location inventory systems: One-period analysis[J].
Management Science, 21(7): 765-776.

Donmez I., Turkay M., 2013, Design of reverse logistics network for waste batteries with an application
in Turkey[J]. Chemical Engineering Transactions, 35:1393-1398.

Hu J., Watson E., Schneider H., 2005, Approximate solutions for multi-location inventory systems with
transshipments[J]. International Journal of Production Economics, 97(1): 31-43.

Huo J.Z., Li H., 2007, Batch Ordering Policy of Multi location Spare Parts Inventory System with Emergency
Lateral Transshipments[J]. Systems Engineering-Theory & Practice, 12(12): 62-67.

Karmarkar U.S., Patel N.R., 1977, The one‐ period, N‐ location distribution problem[J]. Naval Research
Logistics Quarterly, 24(4): 559-575.

Kukreja A., Schmidt C.P., 2005, A model for lumpy demand parts in a multi-location inventory system with
transshipments[J]. Computers & Operations Research, 32(8): 2059-2075.

Man J.H., 2012, Inventory Policy and Simulation for Relief Supply Based on Transshipment[J]. Soft Science ,
26(8): 49-54.

Robinson L.W., 1990, Optimal and approximate policies in multiperiod, multilocation inventory models with
transshipments[J]. Operations Research, 38(2): 278-295.

Xu K., Evers P.T., Fu M C., 2003, Estimating customer service in a two-location continuous review inventory
model with emergency transshipments[J]. European Journal of Operational Research, 145(3): 569-584.

Yu A.M., Liu L.W., 2013, The Transshipment between Two Types of Retailers in the Downstream Supply
Chain [J]. Journal of Systems & Management, 22(1): 1-9.

750