Professional Documents
Culture Documents
Author(s): Stan van Hoesel, H. Edwin Romeijn, Dolores Romero Morales and Albert P. M.
Wagelmans
Source: Management Science, Vol. 51, No. 11 (Nov., 2005), pp. 1706-1719
Published by: INFORMS
Stable URL: http://www.jstor.org/stable/20110457
Accessed: 04-04-2018 01:07 UTC
REFERENCES
Linked references are available on JSTOR for this article: http://www.jstor.org/stable/20110457?
seq=1&cid=pdf-reference#references_tab_contents You may need to log in to JSTOR to access the
linked references.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of
scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
INFORMS is collaborating with JSTOR to digitize, preserve and extend access to Management
Science
This content downloaded from 190.131.242.26 on Wed, 04 Apr 2018 01:07:49 UTC
All use subject to http://about.jstor.org/terms
MANAGEMENT SCIENCE ?nf?&
Vol. 51, No. 11, November 2005, pp. 1706-1719 DOI i0.1287/mnsc. 1050.0378
issn 0025-19091 eissn 1526-55011051511111706 ? 2qo5 INFORMS
H. Edwin Romeijn
Department of Industrial and Systems Engineering, University of Florida, 303 Weil Hall, P.O. Box 116595,
Gainesville, Florida 32611-6595, romeijn@ise.ufl.edu
Albert R M. Wagelmans
Econometric Institute, Erasmus University Rotterdam, P.O. Box 1738, 3000 DR Rotterdam, The Netherlands,
wagelmans@few.eur.nl
generalizes the uncapacitated serial single-item multilevel economic lot-sizing model by adding stationary pro
duction capacities at the manufacturer level. We present algorithms with a running time that is polynomial in the
planning horizon when all cost functions are concave. In addition, we consider different transportation and
inventory holding cost structures that yield improved running times: inventory holding cost functions that are
linear and transportation cost functions that are either linear, or are concave with a fixed-charge structure. In the
latter case, we make the additional common and reasonable assumption that the variable transportation and
inventory costs are such that holding inventories at higher levels in the supply chain is more attractive from a
variable cost perspective. While the running times of the algorithms are exponential in the number of levels in
the supply chain in the general concave cost case, the running times are remarkably
insensitive to the number of levels for the other two cost structures.
Key words : lot sizing; integration of production planning and transportation; dynamic programming;
polynomial time algorithms
History: Accepted by Thomas M. Liebling, mathematical programming and networks; received June 17, 2002.
This paper was with the authors 11 months for 2 revisions.
This content downloaded from 190.131.242.26 on Wed, 04 Apr 2018 01:07:49 UTC
All use subject to http://about.jstor.org/terms
Hoesel et al.: Integrated Lot Sizing in Serial Supply Chains with Production Capacities
Management Science 51(11), pp. 1706-1719, ?2005 INFORMS 1707
This content downloaded from 190.131.242.26 on Wed, 04 Apr 2018 01:07:49 UTC
All use subject to http://about.jstor.org/terms
Hoesel et al.: Integrated Lot Sizing in Serial Supply Chains with Production Capacities 1708
Management Science 51(11), pp. 1706-1719, ?2005 INFORMS
of solutions to uncapacitated lot-sizing problems as at the next level. From the final warehouse level prod
described in Zangwill (1969) for the multilevel case, and, ucts are then (possibly after having been stored for
for instance, in Wagner and Whitin (1958) for the single- some period) transported to the retailer.
level case. In particular, in our two-level model we work We consider a planning horizon of T periods. In
with the new concept of a subplan, and show that extreme each period t, the retailer faces a nonnegative demand
solutions decompose into a given by dt, while the production capacity of the man
number of consecutive subplans. Our algorithms for ufacturer in period t is equal to bt. We will consider
this model all run in polynomial time in the planning a total of L levels, which includes the manufacturer,
horizon of the problem. The direct generalization of the retailer, and L ? 2 intermediate warehouses. We
this approach to the multilevel case leads to a very say that the manufacturer is at the first level of the
large running time. We achieve substantial savings by chain, and the retailer is at the Lth level. Each of
introducing the concept of a relaxed subplan. In the intermediate levels corresponds to a warehouse.
contrast to existing approaches in the literature, our Let U+ denote the set of nonnegative real numbers.
dynamic program does not necessarily represent all (or For each period t = l,... ,T, the production costs are
even only) extreme point solutions to the MLSP PC. In given by the function pt: U+ -> IR+, the transporta
addition, while the paths in the dynamic pro gram do all tion costs from level ? to level i +1 are given by the
correspond to feasible solutions of the problem, the costs function c\ : U+ -> U+ (i = 1,..., L -1), and the inven
of a path may overestimate the tory holding costs at level ? are given by the function
costs of the corresponding solution to the problem. h\ : R+ -> R+ (i = 1,..., L). Throughout the paper, we
We are nevertheless able to prove (based on the con will assume that all cost functions are concave, non
cavity of the cost functions) that our dynamic pro decreasing, and equal to zero when their argument is
zero.
gram solves the MLSP-PC to optimality. The resulting
algorithm for the case of general concave cost func The MLSP-PC can be formulated as follows:
tions is exponential in the number of levels in the sup ply L-l
minimize
r(Myf)+?cf(*?)+?W))^ (P)
chain. However, it is remarkably insensitive to the
number of levels for the two specific cost structures
t=\ \ e=i ?=i /
mentioned above.
subject to
This paper is organized as follows. In ?2, we
introduce the MLSP with production costs and gen eral x]+l}=yt+lll? t = l.T, (1)
nondecreasing concave production, transporta
xit+lf = xt-1+lf_1, t = l,...,T;l = 2,...,L-l, (2)
tion, and inventory holding cost functions. We char
acterize the extreme points of the feasible region of the df+IfL = xf-1+ifIL1/ t = l,...,T, (3)
problem, and prove a decomposition result that will
yt<bt, t = l,...,T, (4)
form the basis of our algorithms. In ?3, we study the
two-level problem and provide a gen /? = 0, 1 = 1,...,L, (5)
eral dynamic programming framework based on the
decomposition result derived earlier, which yields a
yt>0, t = l,...,T,
polynomial time algorithm in the planning horizon for xf>0, t = l,...,T; i = l,...,L-l,
general concave costs. In ?4, this algorithm is then
generalized to the multilevel lot-sizing prob lem and is
lf>0, t = l,...,T; 1 = 1,...,L,
shown to still be polynomial in the plan where yt denotes the quantity produced in period t,
ning horizon, and better running times are given for x\ is the quantity shipped from level ? to level l + l in
two variants of the model. The paper ends in ?5 period t, and If denotes the inventory quantity at level
with some concluding remarks and issues for further ? at the end of period t. Constraints (l)-(3) model the
research.
balance between inflow, storage, and outflow at the
manufacturer, warehouse, and retailer levels, respec
2. Model Formulation and Analysis tively, in each period. The production quantity in each
period is restricted by constraints (4). Finally, con
2.1. The Model
straints (5) state that all initial inventory levels are
As described in the introduction, we will study a equal to zero. Unlike in the traditional single-level lot-
multilevel lot-sizing problem with a serial structure. In sizing model, this is not an assumption that we can
each period, production may take place at the manu make without loss of generality, due to the non
facturer. The items that are produced may be stored at linearity of the transportation and inventory holding
the manufacturer level or transported to the first ware cost functions. Therefore, we will later discuss how
house level. At each of the warehouse levels, products to deal with problem instances where this constraint
are again either stored or transported to the warehouse is absent, and instead (nonnegative) initial inventory
This content downloaded from 190.131.242.26 on Wed, 04 Apr 2018 01:07:49 UTC
All use subject to http://about.jstor.org/terms
Hoesel et al.: Integrated Lot Sizing in Serial Supply Chains with Production Capacities
Management Science 51(11), pp. 1706-1719, ?2005 INFORMS 1709
quantities at all levels are considered as part of the and T=4. This representation will facilitate the anal
problem data. Then, the developed algorithms can be ysis of the structure of extreme points of the feasi
applied in a rolling horizon scheme, in which new lot ble region of (P) in ?2.4. Before proceeding with this
sizing instances are solved?and their optimal solu analysis, in ?2.2 we will discuss related models and
tions partially implemented?as time proceeds and algorithms from the literature, as well as some special
new demand forecasts become available. cases that reduce to single-level models in ?2.3.
For convenience, we will define dts to be the cumu
2.2. Literature Review
lative demand in periods t,...,s, i.e.,
The single-level variant of the MLSP-PC has received
S (6) a lot of attention in the literature. The uncapacitated
d?=\ J^dT for t = l,...,s; s = l,...,T, problem, the ELSP, is solvable in polynomial time in
the length of the time horizon; see Wagner (1960) for
0 otherwise. this basic result. More efficient algorithms for special
cases have been developed by Aggarwal and Park
To ensure feasibility of (P), we will assume that
(1993), Federgruen and Tzur (1991), and Wagelmans
the cumulative demand in the first t periods cannot
This content downloaded from 190.131.242.26 on Wed, 04 Apr 2018 01:07:49 UTC
All use subject to http://about.jstor.org/terms
Hoesel et al.: Integrated Lot Sizing in Serial Supply Chains with Production Capacities 1710
Management Science 51(11), pp. 1706-1719, ?2005 INFORMS
the problem to a two-level model that inherits its cost allows us to eliminate the production variables as
function structures from the three-level model. well as the inventory variables at the manufacturer,
For their class of fixed-charge transportation costs, they yielding a standard uncapacitated ELSP. The resulting
provide an 0(T*) algorithm to solve the model, problem can be solved in 0(T2) time for general con
even in the case of nonstationary production capac cave transportation costs (see Wagner 1960), and in
ities. For their class of concave transportation costs O(TlogT) time for fixed-charge transportation costs
they provide an 0(T8) algorithm to solve the model in the (see Aggarwal and Park 1993, Federgruen and Tzur
presence of stationary production capacities. They pose the 1991, and Wagelmans et al. 1992).
complexity of their model for more general
2.4. Characterization of Extreme Points
cost structures as an open question. In this paper, we
address this question by deriving an 0(T7) algorithm Problem (P) has a concave objective function, and its
for solving the two-level problem in the presence of feasible region is defined by linear constraints. This
stationary capacities. implies that there exists an extreme point optimal
solution to (P). Consider the flow in the network cor
2.3. Special Cases responding to any extreme point feasible solution. As is
It is common in lot-sizing problems to model the common in network flow problems, we will call the arcs
inventory holding costs as linear functions, i.e., that carry an amount of flow that is both strictly positive
h?(I[) = h?I[ for t = l,...,T; ? = 1,...,L, with h\>0 for all t and strictly less than its capacity free arcs. It
and i. We will therefore consider this class of is well known (see, e.g., Ahuja et al. 1993) that the
problems in ??4.3 and 4.4. In ?4.3, we will in addi subnetwork containing only the free arcs contains no
tion assume that the transportation costs have a fixed cycle.
charge structure without speculative motives. More 2.4.1. Subplans. Note that only arcs that have a
formally, cit(x)=ftel{x>0}+gfx, where 1{X>0} is an indi finite upper bound (which in our case are only the
cator function taking the value 1 if x > 0, and 0 oth erwise. production arcs) may carry flow while they are not
The assumption that there are no speculative motives, which free. Removing all production arcs, the network con
is commonly assumed for the produc taining all remaining free arcs decomposes into a
tion and inventory holding costs in traditional eco number of connected components. Limiting ourselves
nomic lot-sizing models, means in this context that, with for now to connected components that do indeed
respect to variable inventory and transportation carry flow, we identify the first and last nodes in
costs only, it is attractive to transport as late as possi ble. the component at each level. For a given component,
More formally, g? + h?+1 >h\+glM for t = 1,..., T
these nodes can be denoted by (?,sn + l) and
l;i = l,...,L-l. (?,<;&) for 1 = 1,...,L, where
Note that if the transportation cost functions are both
linear and exhibit no speculative motives, it is always ?n<?w,i<?tt^?w,2 for = 1,...,L-1. (8)
optimal to store production at the manufac turer and
transport only when demand needs to be satisfied. (Note that the strict inequality holds due to the defini
Hence, without loss of optimality, we can assume that tion of the subplan: The first period included at level i
J/ = 0 for all t = l,...,T and i = 2,...,L. Similarly, if the is ?n-f-l.) With this approach, some nodes may be
transportation costs are linear and isolated and not included in any component carrying
gf + h?t+1<h*+gf+1 for t = l,...,T-l) ? = 1,...,L-1, it is flow. We assign each of those to the component that
cheaper to transport as soon as we produce and is adjacent to the left of them. The assignment of the
store the production at the retailer level. Then, with isolated nodes is illustrated in Figure 2. After elim
out loss of optimality, we can assume that J/ = 0 inating the production arcs, we obtain two compo
for all t = l,...,T and i = l,...,L ? 1. These two spe nents. The first one is defined by the nodes (1,1) and
cial cases of the MLSP-PC therefore yield a standard (1.4) in Level 1, (2,1) and (2,4) in Level 2, and (3,1)
CLSP. and (3,6) in Level 3, and the second one by nodes
Finally, a variant of the uncapacitated two-level (1.5) and (1,8) in Level 1, (2,6) and (2,8) in Level 2,
MLSP-PC can easily be reduced to an uncapacitated and (3,7) and (3,8) in Level 3. We may observe that
ELSP. When production costs as well as the inven node (2,3) is part of the first component, even though
tory holding costs at both levels are linear, given that no flow passes through this node. As mentioned
we decide to transport in a certain period, we can above, the isolated node (2,5) is assigned to the
easily determine the best production period, i.e., the left component.
period that yields the minimum total unit production Summarizing, we can decompose an extreme point
and manufacturer-level inventory costs for transport solution to (P) into components, each of which con
in period t. Redefining the transportation cost func tains a set of nodes {(?/?^ + l),...,^,^)} (^ = 1, ,?)
tion accordingly, which can be done in 0(LT) time, satisfying (8). We will call the components thus
This content downloaded from 190.131.242.26 on Wed, 04 Apr 2018 01:07:49 UTC
All use subject to http://about.jstor.org/terms
Hoesel et al.: Integrated Lot Sizing in Serial Supply Chains with Production Capacities
Management Science 51(11), pp. 1706-1719, ?2005 INFORMS 1711
Figure 2 The Structure of an Extreme Point Solution to the MLSP-PC, Proposition 2.2. A subplan can contain at most one
/.=3and7" = 8 free production arc.
This content downloaded from 190.131.242.26 on Wed, 04 Apr 2018 01:07:49 UTC
All use subject to http://about.jstor.org/terms
Hoesel et al.: Integrated Lot Sizing in Serial Supply Chains with Production Capacities 1712
Management Science 51(11), pp. 1706-1719, ?2005 INFORMS
This content downloaded from 190.131.242.26 on Wed, 04 Apr 2018 01:07:49 UTC
All use subject to http://about.jstor.org/terms
Hoesel et al.: Integrated Lot Sizing in Serial Supply Chains with Production Capacities
Management Science 51(11), pp. 1706-1719, ?2005 INFORMS 1713
that all demand is produced within the set of produc defined above to each individual subplan would yield
tion periods allowed in the subplan. Because clearly K<T, an algorithm with running time 0(T8) for comput
the number of allowable values for Y is 0(T). ing the costs of all subplans. However, the running
Furthermore, by Proposition 2.3 we know that the time can be reduced by observing that the costs of
cumulative transported quantity up to and including many subplans are related. In particular, observe that
some period is either equal to the total production the dynamic programming network corresponding to
quantity of an initial sequence of production peri any subplan of the form (tX/t2,rx,T2) is actually a
ods or satisfies the demand of an initial sequence of subnetwork of the dynamic programming network
demand periods in the subplan. More formally, this for the subplan (0,t2,Tx,r2). Therefore, using back
means that ward recursion to solve for the shortest path between
nodes (0,0,0) and (t2,dTi+x>T2,dTi+x Ti) in the latter
Xe(\J{kb,kb+s})u( ? {dTl+1/A network yields, as a byproduct, the shortest paths
between nodes (?,0,0) and (t2,dTi+X/t2,dTi+xt2) for each
where, in addition, dTi+xt<X<Y to ensure that t = l,...,Tx. It thus follows that we only need to con
demands are satisfied on time and products are not sider the 0(T3) subplans of the form (0,t2,Tx,r2), the
transported before they are produced, and X = 0 if t< costs of which can be determined in 0(T7) time.
rx and X = Kb + e if t >t2 to ensure that transportation
takes place within the subplan. The number of allow 3.4. Dealing with Initial Inventories
able values for X is thus 0(T) as well, so that the total If the initial inventories at the manufacturer and/or
number of nodes in the network is 0(T3). retailer levels, Iq1 and Iq, are strictly positive, there
Arcs in the network represent production, trans is a slight change in the construction of subplans.
portation, and inventory decisions. Arcs are present Recall that we construct subplans corresponding to
between pairs of nodes__in_ the network of the a given extreme point solution by considering all arcs
form (t,Y,X) and (f + l,Y,X), with Y-Ye{0,s,b} (except production arcs) that carry positive flow. The
(where the value ? is only allowed if Y = kb for subplans are then formed by the resulting connected
some k = 0,l,...,K), and X>X (where Xg{X,Y}U components together with some isolated nodes. When
(UjLTl+i{^t1+i,s}))- I* is easy to see that there are 0(T) there are initial inventories, however, there may be
arcs emanating from each node in the network, so that one or more components that carry flow but do
the entire network has 0(T4) arcs. not contain a production period. In these compo
From the information contained in the nodes defin nents, demand is satisfied using initial inventories
ing an arc, we can easily_compute the production at warehouse and retailer levels only, and they can
quantity in period f+ 1 (Y ? Y), the transportation be assigned to the component containing produc
quantity in period t + 1 (X ?X), the inventory held tion Period 1 (i.e., the component containing node
at^ the^ manufacturer level at the end of period t + 1 (1,1)). The results in ??2.4.2 and 2.4.3 are clearly still
(Y ? X), and the inventory held at the retailer level at valid for subplans in which tx > 0. However, for sub
the end of period t + 1 (X ? d +1 t+x). The costs of an plans with tx = Tx= 0, the results continue to hold pro
arc are thus given by vided we view the total initial inventories Iq+Iq as a
cumulative production quantity up to and including
pt+1(Y-Y)+c/+1(X-X)
Period 0, and the initial inventory Iq at Level 2 as the
+ h]+1(Y-X) + hl1(X-dTi+ht+1). cumulative transportation quantity up to and includ
ing Period 0. Unless r2 = T, these subplans can only
If all cost functions can be evaluated in constant time,
have a feasible solution if the total initial inventories do
the costs of a given arc can be computed in con not exceed the total demand that needs to be satis
stant time provided that we determine all cumulative
demands dtt, (in 0(T2) time) in a preprocessing step. fied in the subplan. For subplans with dXj2 > Iq+Iq, we
obtain K=l(dXT2-I0l-lQ2)/b\ and e = dXT2-IG1-I02-Kb.
Any path in the network from the source (tx, 0,0) to
the sink (t2,dTi+1/T2,dTi+1/T2) represents a feasible flow As already mentioned, dXri <Iq+Iq can only occur if t2
in the subplan (tx,t2, tx,t2) with just one free produc = T. If indeed dXT < Iq+Iq, an extreme point solu
tion arc. Moreover, it is easy to see that the reverse tion will contain only a single subplan: (0,T,0, T),
is also true. Therefore, the subplan costs are given by and no production will take place in any period in
the minimal cost path in this network from the source that subplan, i.e., K = ? = 0. The only remaining diffi
node to the sink node. The time required for finding culty in this case is that we do not want to specify in
this minimal cost path is proportional to the number advance in which level the excess inventory will end
of arcs in the network, so that the cost of a single up as ending inventory. This can easily be dealt with
subplan can be determined in 0(T4) time. by extending the planning horizon by one period,
Because there are 0(T4) subplans, a straightforward say T + 1. Then, define the production cost function
application of the dynamic programming algorithm for that period as pT+1(0) = 0 and PT+i(yT+i) = 0C f?r
This content downloaded from 190.131.242.26 on Wed, 04 Apr 2018 01:07:49 UTC
All use subject to http://about.jstor.org/terms
Hoesel et al.: Integrated Lot Sizing in Serial Supply Chains with Production Capacities 1714
Management Science 51(11), pp. 1706-1719, ?2005 INFORMS
all O < yT+x < b and the transportation cost function as based on the idea that the most important informa
Cj+X (xT+x ) = 0 for all xT+x > 0. Finally, set dT+x = ?? + Iq ? tion present in the definition of a subplan is the set of
dXT. The costs of the single subplan (0,T,0,T) in the production periods tx + l,...,t2 and the set of demand
original problem can then be found by finding the periods t1 + 1,...,t2. The basis of our improved algo
costs of the subplan (0, T + 1,0, T + 1) in the modified rithm is then to allow transportation in the periods
problem. tx + l,...,T2 (while of course retaining the given pro
Now consider the dynamic programming network duction and demand periods). We can then use the
used to compute the costs of a subplan. For subplans same dynamic programming approach as in the two
that contain initial inventories, we let the source node level case, where we replace the two-level subplan
be (0,I?+Iq2,Iq) and the sink node be (t2,lQ1+I02 + Kb + costs 4>(tx,t2,Tx,T2) by the minimum costs of satisfy
s,Il+Il + Kb + s). For a state (t, Y,X), this also means
ing demand in periods rx +1,..., r2 using production
that K in periods tx + l,...,t2, where at most one of the pro
Ye\J{lQ1+lQ2+kb,lQ1+I02+kb + s} duction quantities may be different from both 0 and b,
k=0 and where transportation at all levels is allowed in
and periods tx + l,...,r2. We will denote the latter costs
by \j)(tx,t2,Tx,T2), and refer to vectors (tX/t2,rx,T2)
Xe(\J[%+I*+kb,I?+I*+kb+B})u( ? {du}\ as relaxed subplans.
\jfc=0 / Vs^Ti+1 / To illustrate the concept of relaxed subplans, con
sider the following problem instance of the 2LSP-PC.
Finally, note that for subplans with tx > 0, we should All demands are equal to 1; the production and trans
have no positive inventory inflow. Therefore, in case portation costs are given by
there are nonzero initial inventory levels, we actually
need to compute the costs of all subplans (0,t2,0,r2)
while taking into account the initial inventory levels, p?y) = 100.1{y>0}+y iit^2,
as well as the costs of all subplans (0, t2,rx, r2) for rx > 0 1 l|y>0} +y otherwise,
without taking into account the initial inventory level and
at the manufacturer. This clearly does not influence the
overall running time of the algorithm. 50-1
c)(x) = {*>o} +x iit^2,
{*>o}"
4. The Multilevel Case
4.1. Introduction Finally, let all inventory holding cost functions be
equal to zero. The optimal flows in 0(1,4,2,4) and
We may extend the dynamic programming approach
(^(1,4,2,4) are given in Figure 3. When calculating
developed in ?3.1 for the two-level case to the multi the costs </>(l,4,2,4), transportation is only allowed
level case, where again a Phase 2 dynamic program
in periods in which both production may take place
ming network represents all extreme point solutions and demand is satisfied (i.e., in Periods 3 and 4 in
to the MLSP-PC. To this end, we should define the example), while in the relaxed version of the same
F((?n)^=1) to be the minimum cost associated with sat
subplan transportation is permitted in any period
isfying the retailer demands in periods {?L1 + l,...,T}
where either production may take place or demand is
using production in periods {?n + l,...,T}, and ware satisfied (i.e., in Periods 2, 3, and 4 in the exam
house I in periods {?m/1 + 1,.../T} for* each 1 = ple). Therefore, the costs 1^(1,4,2,4) are lower than
1,..., L ? 2. We would then be interested in computing
F((0)?=1). It is easy to see that the running time of the
Figure 3 The Optimal Flows in 0(1,4,2,4) and <//(1,4,2,4)
corresponding generalization of the Phase 2 dynamic
2 2
program would be 0(T2L). In this section, we will
derive a modification of the Phase 2 dynamic program
that runs in 0(T4) time. This modification does not
\ \
make Phase 1 computationally more expensive,
and may even make it less expensive.
In particular, we will develop a more efficient
approach in which the Phase 2 dynamic program does
not necessarily represent all (or even only) extreme
point solutions to the MLSP-PC, and in addition over
estimates the costs of many of the nonextreme point
solutions that it represents. However, as we will show,
it does contain an optimal extreme point solution and
is guaranteed to find this solution. This approach is 1 1
This content downloaded from 190.131.242.26 on Wed, 04 Apr 2018 01:07:49 UTC
All use subject to http://about.jstor.org/terms
Hoesel et al.: Integrated Lot Sizing in Serial Supply Chains with Production Capacities
Management Science 51(11), pp. 1706-1719, ?2005 INFORMS 1715
0(1,4,2,4) because in the relaxed subplan we can The next theorem shows that there exists an optimal
transport in the second period. solution to the lot-sizing problem that is represented
These changes have two major consequences. Con by a path in the Phase 2 dynamic programming net
sider a path from the source to a sink in the Phase 2 work whose length is equal to the optimal costs.
network. First, while it is easy to see that the corre Theorem 4.3. The Phase 2 dynamic programming net
sponding solution of the MLSP-PC is indeed feasible, it is work contains a path that corresponds to an optimal solu
not necessarily an extreme point solution because tion to our lot-sizing problem, and the length of the path is
production and demand nodes in two relaxed sub equal to the cost of this solution.
plans contained in the solution may be connected by
arcs containing positive flow. Second, it is possible that Proof. Consider an extreme point optimal solution
certain arcs are used in more than one relaxed to the lot-sizing problem, say with cost O*. As dis
subplan. This means that the length of the path in the cussed in ?2.4, this optimal solution decomposes into
network may not be the same as the costs of the corre a sequence of consecutive subplans. It is easy to see
sponding solution to the MLSP-PC. Dealing with this that the Phase 2 dynamic programming network con
latter issue first, the following theorem shows that the tains a path for which the production and demand
path length is never smaller than the actual costs of the periods of each of the arcs correspond to this sequence
solution, and is equal to the costs of the solution of subplans. Lemma 4.2 now says that the length of
if all transportation and inventory cost functions are linear. the path in the dynamic programming network, say
^, will not exceed O*. However, by Theorem 4.1 we
know that ty is an overestimation of the costs of a cor
Theorem 4.1. Each path from the source to a sink in responding feasible solution. Optimality of O* now
the Phase 2 dynamic programming network corresponds to implies that in fact W = <?>*, which proves the desired
a feasible solution to the MLSP-PC. The length of this path result. D
cannot be smaller than the cost of the corresponding solu Theorems 4.1 and 4.3 clearly imply that our two
tion, and is equal to the solution cost if all transportation and phase algorithm solves the MLSP-PC. We can now
inventory cost functions are linear. conclude that Phase 2 of the algorithm runs in 0(TA)
Proof. The fact that a path from the source to time, given all values il/(tx,t2,rx,r2). The remaining
a sink in the Phase 2 dynamic programming net challenge is thus to provide efficient algorithms for
computing these values.
work corresponds to a feasible solution to the lot
sizing problem follows immediately from the fact 4.2. Concave Costs
that all production capacity constraints, as well as all
demands, are satisfied. However, certain transporta 4.2.1. The Costs of the Relaxed Subplans. In this
tion and inventory arcs may carry positive flow in the section, we will formulate the problem of determin
partial solutions corresponding to more than one arc ing the costs if/(tx,t2,Tx,T2) as a dynamic program
in the path, and each of the partial flows is charged ming problem. Put differently, we define, for each
separately according to the corresponding cost func (tx,t2,rx, t2), a network with the property that \?j(tx, t2,
tion. Due to the concavity of all cost functions, it fol tx,t2) is equal to the shortest path between a pair
lows that the cost of the total flow will not exceed the of source and sink nodes in this network. The nodes
sum of the costs of the individual flows on any partic in this network are of the form (t,Y, X1 ,...,XL~l),
ular arc, and therefore the length of a path will never where t indicates a period, Y is equal to the cumula
be less than the costs of the corresponding solution. In tive production quantity up to and including period
addition, when all transportation and inventory cost t, and Xe is equal to the cumulative transportation
functions are linear, the path length and solution costs quantity from level I to level I +1 up to and including
are clearly equal. D period t. Note that feasibility dictates that we should
The next lemma gives a relationship between the restrict ourselves to values dTi+xt < XL~l < < X1 < Y.
costs associated with a subplan and the correspond The source is the node (tx,0,...,0), while the sink is
ing relaxed subplan.
the node (t2,Kb + s,...,Kb + s). As in ?3.3, we have
that K
Lemma 4.2. For any subplan ((?n/^Li)' we ^ave
Ye[j{kb,kb+8},
f/zfl?0((?n,?^=1)>^(?n,?12,?L1,?L2). k=0
Proof. This result follows immediately by noting and the number of allowable values for Y is
that both 0((?*i/?*2)Li) and ^n^w ^li^li) are the Moreover, in a similar fashion to that of the two
optimal value of an optimization problem with iden case, we have that
tical cost functions, but where the feasible region of
the former is a subset of the feasible region of the X<e(\J{kb,kb + e})u( ? {dTl+i,s}Y ? = 1,...,L-1,
latter. D \k=0 / \s=T!+l /
This content downloaded from 190.131.242.26 on Wed, 04 Apr 2018 01:07:49 UTC
All use subject to http://about.jstor.org/terms
Hoesel et al.: Integrated Lot Sizing in Serial Supply Chains with Production Capacities 1716
Management Science 51(11), pp. 1706-1719, ?2005 INFORMS
so that the number of allowable values for X? is 0(T) as general concave production costs, and linear inven
well. This means that the total number of nodes in tory holding costs at all levels, as well as one of the
the network is 0(TL+1). following two transportation cost structures: (i) fixed
Arcs in the network represent production, trans charge without speculative motives; or (ii) linear.
portation, and inventory decisions, and are present
between pairs of nodes in the network of the form 4.3. Fixed-Charge Transportation Costs Without
(?,Y,Xl,...,XL-1) and (t + 1,Y,X>,..., X1'1), where Y- Speculative Motives
Ye{0,s,b} and X?>Xl (l = l,...,L-l). It is easy
4.3.1. Introduction. In this section, we consider
to see that there are (^(T1'1) arcs emanating from each
node, for a total of 0(T2L) arcs in the network. the case of fixed-charge transportation costs with
Similar to ?3.3, we can easily_compute the produc tion out speculative motives and linear inventory hold
quantity in period t + 1 (Y ? Y), the transporta ing costs. As before, we will determine the costs of
tion quantity between levels I and l + l in period t + 1 each relaxed subplan using dynamic programming.
(Xl ? Xl), the inventory held_at the manufacturer level After a preprocessing step that runs in 0(LTA), this
at the end of period t + 1 (Y ?X1), and the inventory dynamic program runs in 0(T4) time for each indi vidual
held at the retailer level at the end of period t + 1 relaxed subplan. By using the reduction tech nique at
(XL~1 ? dTi+xt+x). The costs of an arc are thus given by the end of ?3.3, the cost of all 0(T4) relaxed subplans
can be computed simultaneously in 0(T7)
time. Therefore, the running time of the dynamic pro
pt+1(Y-Y) + ?c?+1(X'-X') + A}+1(Y-X) gramming approach for this special case of the MLSP
?=l
PC is 0(T7 + LT*). When L = 2, we can reduce this
running time to 0(T6).
+E^+i(x<-1-xO+^+1(xL-1-dTi+u+1).
4.3.2. Zero-Inventory-Ordering Property at the
Retailer. We will show that, under fixed-charge trans
If all cost functions can be evaluated in constant time,
portation costs without speculative motives, solutions
the costs of a given arc can be computed in 0(L) time
satisfying the zero-inventory-ordering (ZIO) property
in the same way as in the 2LSP-PC after a prepro
at all levels in \2,...,L\, i.e., I[xf~?=0 for t = l,...,T
cessing step taking 0(T2) time. We conclude that the
1; 1 = 2,...,L, are dominant. That is, given any feasi
cost of a single relaxed subplan can be determined in
0(LT2L) time. ble solution to the relaxed subplan (tx,t2,Tx,r2), there
always exists another solution that is at least as good
Finally, noting that there are 0(T4) relaxed sub
and satisfies the ZIO property at all levels in {2,..., L}.
plans and applying the same technique for reduc
ing the running time as used at the end of ?3.3, Theorem 4.4. Given a relaxed subplan (tx,t2,rx,r2),
we obtain an algorithm for the MLSP-PC with arbi the set of solutions with the ZIO property at all levels in
trary concave production, transportation, and inven {2,...,L} is dominant.
tory holding costs and stationary capacities that runs Proof. Let (y,x,I) be a feasible solution to the
in 0(LT2L+3) time. Although this time is exponential
in the number of levels, the order of the running time relaxed subplan (tx,t2,Tx,r2) that does not satisfy
will be limited by the fact that the number of levels the ZIO property at some level. Let I be the last
will typically be relatively small. level, such that the ZIO property holds for all I e {I +
This approach can easily be extended to deal 1,...,L\, but is not true for level I. We can construct a
with initial inventories. Recall from ?3.4 that only new solution at least as good as (y,x,I), such that the
relaxed subplans with ^=0 need to be considered. ZIO property holds for all I e {I,..., L). If 1 = 2, then
For such a relaxed subplan, we should view the total we have obtained the desired result. Otherwise, we
initial inventories Y?i=\Iq as a cumulative production repeat the procedure with the new solution. Observe
quantity up to and including Period 0, and the initial that this procedure converges because the new I has
inventory Y%=s+1Iq as the cumulative transportation decreased by at least one unit.
quantity up to and including Period 0 from level s to Let te{tx + l,...,T2 ? 1} be a period so that
level s + 1, for all se{l,...,L-l}. As in ?3.4, without lfxj~l>0. The positive inventory If has been trans
increasing the running time, these initial inventories ported to level ? in some earlier period. However, due
can be incorporated into the dynamic programming to the absence of speculative motives, we can resched
approach to calculating \\f(tx,t2,Tx,72) by appropriate ule the transportation of the If units to period t + 1
redefinitions of the possible values of Y and X. without increasing the costs. Repeating this argument
In the next sections, we will show how the run for each period t violating the ZIO property at level ?,
ning time can be dramatically reduced for problem we obtain a solution where the ZIO property is true
instances that have stationary production capacities, for each level le{l,...,L). D
This content downloaded from 190.131.242.26 on Wed, 04 Apr 2018 01:07:49 UTC
All use subject to http://about.jstor.org/terms
Hoesel et al.: Integrated Lot Sizing in Serial Supply Chains with Production Capacities
Management Science 51(11), pp. 1706-1719, ?2005 INFORMS 1717
We may recall that i?j(tx,t2,Tx,T2) is equal to the sx = t,...,T', and s2 = sx,...,T in 0(LT4) time; see the
minimal costs among the solutions of the relaxed sub online appendix. It is important to note that although
plan (tx,t2,rx,T2) with at most one free production arc. Zangwill's model allows for general concave trans
The following corollary to Theorem 4.4 states that for portation and inventory holding cost functions, we
finding this constrained minimum we can cannot use the same approach as described above in
again restrict our search to solutions satisfying the the presence of production capacities. The reason is
ZIO property at the retailer. that in the uncapacitated case, the ZIO property holds
Corollary 4.5. The cost associated with the relaxed for arbitrary concave arc cost functions, while this is
not the case in the capacitated case. However, as we
subplan (tx,t2, tx,t2) can be found among all feasible solu
have shown, in the case of fixed-charge transporta
tions satisfying the ZIO property at all levels in {2,...,L}.
tion costs that exhibit no speculative motives, we also
Proof. This follows immediately from the proof of Theorem obtain the ZIO property, enabling the use of Zangwill's
4.4 by observing that the modification of the algorithm to determine inputs to our algorithm.
solution to obtain a solution satisfying the ZIO prop The problem of determining \?f(tx,t2,rx,T2) reduces
erty does not alter the production flows. D to finding the length of the shortest path in the net
This corollary implies that when searching for ip(tx, work from the source to the sink, which can be done
t2, tx, r2), we can assume that any amount shipped is in linear time in the number of arcs. It is easy to see
equal to the demand of a set of consecutive periods. This that the number of nodes in the network is 0(T3) and
will help to reduce the information maintained the number of arcs 0(T4). Using the same approach to
in the dynamic programming approach described in ? computing multiple values of the function if/ at once as
4.2.1. we have discussed for the function \p at the end of
?3.3, this yields an 0(T7 + LT4) algorithm for solving
4.3.3. The Costs of a Relaxed Subplan. In this
the multilevel variant of this problem.
section, we will formulate the problem of determin
When L = 2, this running time can be reduced
ing the costs ^(tx,t2,Tx,r2) as a simplification of the
to 0(T6). Recall that the number of nodes in the
dynamic programming problem defined in ?4.2.1. All
dynamic programming approach above is 0(T3).
nodes in the dynamic programming network are of
We will show that the number of arcs is also 0(T3).
the form (t,Y,s), where t indicates a period, Y is
For each t there are 0(T) nodes of the form (t,-,t), and
equal to the cumulative production quantity up to 0(T2) nodes of the form (t,-,s) with s>t. Each node
and including period t, and s represents the last of the form (t,-,t) has 0(T) successors, and each
period whose demand is satisfied using transporta node of the form (t,-,s) with s>t has O(l) succes
tion from Level 1 to Level 2 up to and including sors, which makes for a total of 0(T)-(0(T)-0(T) +
period t, where dTi+X/S< Y and t <s. We may observe 0(T2)-0(1)) = 0(T3) arcs in the network. This yields
that from Theorem 4.4, we have that X1=dTi+Xs in the an 0(T6) algorithm for solving the two-level variant
dynamic program of ?4.2.1. The source is the node of this problem.
(tx,0,rx), while the sink is the node (t2,Kb + s,r2). As
Unfortunately, in the presence of nonzero initial
before, we know that Y can only assume the values inventory levels the ZIO property is not necessarily
K dominant anymore. However, in these cases the more
Ye\J{kb,kb + s}. general procedure developed for the case of arbitrary
k=0
concave cost functions of course still applies.
Arcs are present between pairs of nodes4.4in.Linearthe Transportationnet Costs work of the form (t,Y,s)
and (t,Y,s), where Y ? Ye
4.4.1. Introduction. In this section, we will con
{0,s,b} and s>s. It is easy to see that there are 0(T)
sider the case where the transportation costs and
arcs emanating from each node in the network, so that
the entire network has 0(T*) arcs. inventory holding costs are linear. We will develop a
The costs of an arc between nodes (t,Y,s)dynamicandprogramming(t+ approach that finds the opti
1, Y, s) are now given by mal costs of each relaxed subplan. After a prepro cessing step that runs
in 0(LT2) time, this algorithm
pM(Y-Y) + h]+x(Y-dTi+X/?) runs in 0(T2) time for a single relaxed subplan, but
the costs of all 0(T4) relaxed subplans can be com
+ c]+x(ds+X/s) + Ct+Xf2(s + l,s), puted simultaneously in 0(T5) time. This results in
an 0(T5+LT2) algorithm for solving this class of
where Cu(sx,s2) are defined as the optimal costs of instances of MLSP-PC.
shipping dSiS2 units from node (t,l) to their desti
nations, i.e., demand nodes (L,sx),...,(L,s2). We can 4.4.2. Preprocessing. In terms of the underlying
use Zangwill's algorithm, in a preprocessing stage, network (as described in ?2.1), one unit produced in
to determine the values Q^^) for all t = 2,...,T; period t for satisfying demand in period r>t will,
This content downloaded from 190.131.242.26 on Wed, 04 Apr 2018 01:07:49 UTC
All use subject to http://about.jstor.org/terms
Hoesel et al.: Integrated Lot Sizing in Serial Supply Chains with Production Capacities 1718
Management Science 51(11), pp. 1706-1719, ?2005 INFORMS
in the optimal solution, flow along the minimum-cost quantity produced in period t + 1 are equal to Gt+
path from (l,t) to (L, t). In a preprocessing stage, we can When Y ? Y>8, the transportation and inven
determine the minimal unit transportation costs holding costs for this arc consist of up to_ three ponents: G?+lfS5+GD?
associated with producing one unit in period t for +1/S/S-_1 + G?+M(ds--5), an thus indeed be computed in constant time.
consumption in period r, which we will call GtT. All these The problem of determining ifj(tx,t2,Tx,T2) red to finding the length of
values can be computed in 0(LT2) time by solv ing T shortest the shortest path in the
path problems in acyclic graphs with work from the source to the sink, which can b
0(TL) arcs using backward recursion. Using these in linear time in the number of arcs. It is easy to that the number of nodes
values, we can then again determine the total trans in the network is 0(T2) the number of arcs 0(T2). Using the same approac
portation costs associated with producing, in period t, computing multiple values of the function ifj at
the entire demand of the consecutive periods tx +
1,..., t2, assuming that transportation is allowed in all
periods t,...,r2, i.e., GDfTiT2=ErLr1+idrGfr. In 0(T3) time, as in ?4.3, this yields an 0(T5+LT2) algorith
these costs can be calculated for all t = l,...,T solving this variant of the MLSP-PC.
and t < rx +1 < r2 < T. This information will enable us As in the CLSP, initial inventories can be i
to compute the total transportation costs associated porated when all transportation and inventory
with production in period t in constant time. ing cost functions are linear. In particular, the in
4.4.3. The Costs of a Relaxed Subplan. In this inventories are used to satisfy the earliest dem via the appropriate
section, we will formulate the problem of determining shortest paths in the netw
the costs if/(tx,t2,rx,T2) as a further simplification of after which the demands are updated and the rem ing problem without
the dynamic programming problem defined in ?4.2.1. initial inventories is solved.
All nodes in the dynamic programming network are of the
form (t, Y), where t indicates a period, and Y is equal to 5. Concluding Remarks and Future
the cumulative production quantity up to
Research
and including period t, where dTi+xt<Y and
In this paper, we have considered a generalizati
K
the classical ELSP with stationary production c
Ye\J{kb,kb+e}. ties that allows for multiple levels of storage, as
k=0
as corresponding transportation decisions for
The source is the node (tx,0), while the portingsinkis thebetween the different levels. We have i
node (t2,Kb + e). fied two important special cases of this problem
Arcs are present between pairs of nodesareinsolvable_the in polynomial time. The running tim
network of the form (t,Y) and (f + l,Y)thewhencorrespondingY? algorithms are remarkably ins Y e{0,s,b}. Each arc of the network
describedtivetoabovethenumber of levels in the supply chain.
represents a possible production decision. WeOpenletissuesthe for future research in this area c costs of the arcs be equal to the
total costsdividedassociatedinto three general directions. First, the
with the production amount. It remains toplexities,showthatalthough polynomial in the planning the transportation and inventory holding
costszon, canare beof relatively high order: 0(T5) to 0(T7
computed in constant time. In addition tothethetwoinfor-level cases. It would be interesting if the mation gathered in the preprocessing
phaseof describedtherunning time could be reduced, for inst
in ?4.4.2, we also will find, for each nodeby (t,Y)investigating whether more time can be sav the network, the first period whose
demanddeterminingisnot the costs of many or all subplans
fully satisfied by the cumulative productiontaneouslyY(say.In addition, although the number of le
s) as well as the part of the demand of willthatgenerallyperiod be relatively small, it would nev that remains to be satisfied (say 8).
with general concave
Usinglessthebecumuinteresting to determine if the multilevel lative demands dTi+X/t, (V = tx + 1,...,t2) as well asthe
cost functions can be solv
fact that the value of Y can only be equalpolynomialtokbor time in both the time horizon and kb + s for
k = 0,...,K, this additional informationnumber canof levels. A second direction is the be obtained in 0(T) time. As we will see
oflater,serialthissupply chains in the presence of capa does not increase the running time of findingat othercostsor additional levels in the
chain. Final
of a single relaxed subplan.would be interesting to consider more comple
Now consider an arc connecting the plytwochainnodesstructures, including, for example, pro
(t, Y) (with first remaining demand periodassemblywith structures at the producer level, or mul
retailers.
This content downloaded from 190.131.242.26 on Wed, 04 Apr 2018 01:07:49 UTC
All use subject to http://about.jstor.org/terms
Hoesel et al.: Integrated Lot Sizing in Serial Supply Chains with Production Capacities
Management Science 51(11), pp. 1706-1719, ?2005 INFORMS 1719
This content downloaded from 190.131.242.26 on Wed, 04 Apr 2018 01:07:49 UTC
All use subject to http://about.jstor.org/terms