An Optimization Approach for Job-shop with Financial Constraints
In the Context of Supply Chain Scheduling Considering Payment Delay between
Members
S. Kemmoe
1
, D. Lamy
2
and N. Tchernev
2
1
CRCGM EA 3849, Université d’Auvergne, Clermont-Ferrand, France
2
LIMOS UMR CNRS 6158, Aubière, France
Keywords: Supply Chain, Payment Delay, Job-Shop, Linear Program, GRASP.
Abstract: In this paper the use of Job-Shop Scheduling Problem (JSSP) is addressed as a support for a supply chain
scheduling considering financial exchange between different supply chain partners. The financial exchange
is considered as the cash flow exchanges between different upstream and downstream partners. Moreover,
several suppliers are involved in operations. The problem under study can be viewed as an extension of the
classical JSSP. Machines are considered as business or logistic units with their own treasury and financial
exchanges happen between the different partners. The goal then is to propose the best schedule considering
initial cash flows in treasuries as given data. The problem is formulated as integer linear programming
model, and then a powerful GRASPxELS algorithm is developed to solve large scale instances of the
problem. The experiments on instances with financial constraints proved the methods addressed the problem
efficiently in a short amount of time, which is less than a second in average.
1 INTRODUCTION
This paper deals with Supply Chain (SC) scheduling
taking into account financial constraints. A SC
composed by individual firms is modeled. In this SC
forward flow of materials and backward flow of
cash appear. Cash flows occur over time in two
forms. Accounts Payable or cash outflows include
expenditures for the logistic activities, or equipment
and materials needed to achieve each operation.
Accounts Receivable or cash inflows are induced by
progressive payment for completed task or product.
The supply chain is modelled as a Job-shop where
each SC member is considered as a machine. The
main goal is to obtain such a schedule which
maintains during the schedule horizon a positive
cash position. Thus, a better synchronization of
material and financial flows avoiding negative cash
position leads to integration of SC performance. An
integer linear programming model is developed
where payment terms and amounts of all suppliers
and distributors are known. A GRASPxELS
algorithm, where the objective is to minimize the
completion time of all activities taking into account
the financial constraints, is proposed to solve large
scale instances.
The next section provides a brief literature review.
The section 3 introduces the assumptions used in
this study. Section 4 presents the integer linear
programming model. In section 5 a customized
GRASPxELS is presented; and the results obtained
thanks to this metaheuristic are compared with the
ones obtained with the CPLEX solver. Finally, a
conclusion and future researches are proposed.
2 RELATED WORK
Inclusion of cash flow in scheduling problem has
been studied with different objective value which
leads to the Resource Investment Problem (RIP)
(Najafi al., 2006) and the Payment Scheduling
Problem (PSP) (Ulusoy G. and Cebelli, 2000).
Depending on the objective, publications encompass
both net present value (Elmaghraby and Herroelen,
1990) and extra restrictions as bonus-penalty
structure (Russell, 1986), or discounted cash-flows
(Najafi al., 2006).
The main objective of cash manager is to have
enough cash to cover day-to-day operating expenses.
Two types of metrics are generally used to optimize
financial flow: during a given period, cash position
190
Kemmoe S., Lamy D. and Tchernev N..
An Optimization Approach for Job-shop with Financial Constraints - In the Context of Supply Chain Scheduling Considering Payment Delay between
Members.
DOI: 10.5220/0005271301900198
In Proceedings of the International Conference on Operations Research and Enterprise Systems (ICORES-2015), pages 190-198
ISBN: 978-989-758-075-8
Copyright
c
2015 SCITEPRESS (Science and Technology Publications, Lda.)
reveals the cash which is available and cash flow,
the cash generated. (Stadtler, 2005) proposes a study
of management on supply chain where the time
horizon relative to the operational schedule
corresponds to the financial schedule. To increase
performance, financial considerations must be done
at every production level, from planning to control,
in order to avoid bank overdraft. (Bertel et al., 2008)
proposed a mixed integer linear program to find an
optimal production plan to maximize average cash
position under a deterministic multi-factory, multi-
stage, and multi-product system, modelled as a flow
shop. A Dynamic Simple Policy (DSP) has been
proposed by (Gormley and Meade, 2007) in order to
minimise transactions costs at short terms periods of
a company and in a national or international context
where financial exchanges are not independently
distributed upon the global costs of the enterprise.
(Comelli et al., 2008) used an activity based costing
(ABC) system to link supply chain physical and cash
flows, proposing a tactical production planning
model. (Tsai, 2008) studied the influence of trade
terms, under a stochastic demand process, on cash
flow risks and showed that using trade discounts to
encourage early payment by customers increased
cash inflow risk despite an improved cash cycle.
(Grosse-Ruyken et al., 2011) plotted out that the
Cash Conversion Cycle (CCC) is a good measure of
performance considering upstream and downstream
partners in order to avoid the “domino effect”
resulting in the bankruptcy of a supplier.
The problems studied are usually considering
cash position as variables. Very few works propose
to analyse cash flow and scheduling problem as an
operational problem of cash management (Kemmoe
et al., 2011a). Moreover, (Elazouni and Gab-Allah,
2004) showed that “available scheduling techniques
produce financially non-realistic schedules”.
Recently (Kemmoe et al., 2011a) formulated the
problem so called “Job-shop with financial
constraint” (JSFC) which is defined as a Job-shop
problem with simultaneously consideration of
manufacturing specific resource requirements and
financial constraints. The inclusion of financial
considerations permits to consider the proper
coordination of production units when optimizing
the supply chain. The main goal is to obtain the
smallest duration of a given supply chain operational
planning while respecting the budget limit of each
production unit. Later (Kemmoe et al., 2012)
extended the model of (Kemmoe et al., 2011a) to
take into account the terms of payments and multiple
suppliers per operation.
In this paper the linear model proposed by
(Kemmoe et al., 2012) is improved for small and
medium size instances and a GRASPxELS
algorithm for large size instances for JSFC with
multiple suppliers per operation is developed.
3 SUPPLY CHAIN ASSUMPTIONS
3.1 Physical Flow Assumption
In this study the cash flow of a manufacturer who
acquires materials from suppliers, transforms them
into semi-finished or finished goods and sells them
to distributors, is considered. To better understand
this relationship a model of a given supply chain is
presented on Figure 1, where each product (P
i
) has
its own process plan which defines the product route
through the supply chain. Therefore the product will
be treated successively by a supplier unit (S
i
),
manufacturing units (MU
i
) and distributor (D
i
). This
supply chain can be modeled as a Job-shop
addressing the proper coordination between material
lots (jobs) and financial considerations.
Figure 1: Material and financial flows through the SC.
The Job-shop scheduling problem (JSSP) consists in
scheduling a set of n jobs that have to be sequenced
on m machines. Each job involves a set of machine-
operations, which must be processed in a pre-
determined order. Each operation has to be
processed on a given machine during a processing
time and no pre-emption is allowed. The JSSP
consists in finding a schedule with a minimal global
duration by managing machine disjunctions (see for
review (Jain and Meeran, 1999)). Using the
disjunctive graph (Roy and Sussmann, 1964) the
logistic activities can be modeled by vertices.
Precedence constraints between operations are
represented by an arc. Disjunctive constraints
between two logistic activities which require the
same logistic unit are modeled by an edge. An arc
has a total cost equal to the duration of the logistic
P
1
P
2
P
3
P
3
AnOptimizationApproachforJob-shopwithFinancialConstraints-IntheContextofSupplyChainScheduling
ConsideringPaymentDelaybetweenMembers
191
activity. The corresponding oriented disjunctive
graph of the SC problem of Figure 1is presented on
Figure 2.
Figure 2: Oriented disjunctive graph of the SC of Figure 1.
The physical flow between the manufactures is
presented at the centre of the previous graph (MU
1
and MU
2
). Since physical flows are directly
impacted by cash-flows, the assumptions concerning
cash in- and outflow are presented on the next sub-
section.
3.2 Cash Flow Assumption
The cash outflow assumption supposes that the
manufacturing units and distributors units always
pay its suppliers (suppliers or manufacturing units)
at the maturity of its accounts payable, which has a
given credit term. The cash inflow assumptions
suggest that sales/shipments occur at the end of each
processing time and that there is a given credit term
offered to customers (manufacturing units or
distributors). Using these assumptions and the
hypothesis that each activity has a known duration
and two suppliers paid with different delays, the
different events occurring during an activity can be
represented, using the following notations:
t
α
i
, t
β
i
: delays, respectively for the first (
) and the
second (
) suppliers, to receive financial
amount for delivering the resource
necessary to execute the corresponding
activity i.
r
i
: account receivable (financial resource or
inflow) generated by the operation i.
s
i
: starting time of activity i.
δ
i
: delay between the starting time of activity i
and the time of account receivable r
i
.
p
i
: duration of material flow on activity i.
c
α
i
, c
β
i
: financial resource required to pay the first
and the second suppliers of the activity i.
Using these notations a set of basic examples of
events occurring during an activity is proposed in
the Figure 3. In the first case presented in the Figure
3, the supplier represented by α is paid after t
α
while
the second supplier represented by β is paid after t
β
,
both inside the activity. When processing time is
over, an inflow occur with amount r
i
after a delay δ
i
.
Figure 3: Different cash-flows events.
In the second case, two suppliers are paid at the
same time while the operation is processed. The
inflow occurs at a moment after the end of the
activity. In the third case the supplier
is paid
during the activity, while the other (
) is paid after
its end. Since the inflow occurs after a given delay,
it may happen after or before the payment of the
second supplier as presented on the fifth case. The
fourth case shows two suppliers paid at the same
time after the end of the operation. In the fifth case,
the suppliers are all paid after the end of the
operation but at different times. Finally, the sixth
case is a special one where the first supplier is paid
inside the activity but the inflow happen before the
payment of the second supplier who is paid after the
end of the activity. This case can be encountered
when an enterprise has negotiated with its supplier a
larger delay. Thanks to the income of money, the
enterprise may perhaps use this inflow for some
financial optimization involving bank interests.
The main objective during the SC scheduling is
to find a schedule which minimizes the lead time
while respecting the budget limit of each SC
member avoiding negative cash position as shown in
Figure 4.
Figure 4: Schedule of operation i with and without
negative cash position.
1,0
2,0
3,0
p
11
1,1 1,2 1,3
2,1 2,2 2,3
3,1 3,2 3,3
p
12
p
22
p
21
p
31
p
32
p
1
1
*O
p
10
p
20
p
30
p
23
p
3
3
S
1
S
2
D
1
D
2
MU
1
MU
2
S
1
D
1
MU
2
MU
1
MU
1
MU
2
...
...
...
0
0
time
time
time
time
MU
Cash
Position
Supplier payment conducting to a
negative cash position for operation i
Cash
Position
MU
c
α
i
c
β
i
t
β
i
-t
α
i
t
α
i
-pi
pi
ri
δi
...
c
α
i
c
β
i
t
β
i
-t
α
i
t
α
i
-pi
pi
ri
δi
Starting time of operation i is delayed to avoid
negative cash position
ICORES2015-InternationalConferenceonOperationsResearchandEnterpriseSystems
192
In the first part of Figure 4 the supplier payment
leads to a negative cash position on the treasury
associated to the MU. In the example proposed, an
inflow occur after a given period and is important
enough to have a positive cash position. Thus, in
order to keep positive cash position when the second
supplier is paid, the starting time of activity i on MU
must be increased. This is shown on second part of
Figure 4. Concerning the cash position, it can be
seen that the shift did not affect the previous cash
flows (the first part are identical), they just happen
later in the Gantt diagram. We end up having the
following problem. We are given a set of jobs,
machines and precedence constraints between the
job operations and then we want to find a scheduling
such that at each step of the time the cash flow is
positive and the finish time is minimized.
4 LINEAR PROGRAMMING
The model presented in this section has been built to
obtain exacts solutions avoiding bank overdraft by
repartition of financial resources among the different
stakeholders. It relies on a flow added to the
incumbent Job-shop.
4.1 Parameters
Some extra parameters are taken into account and
added to those presented on the Cash flow
assumption sub-section:
M : set of logistic units;
J
i
: Job of the activity i;
V : set of all activities;
i,j : indexes representing the different activities
to schedule, i=1,..,|V|, j=1,..,|V| ;
μ
i
: logistic unit required to process activity i,
μ
i
M;
T
m
: initial cash flow for the logistic unit m
M
F : set of suppliers;
f : index representing the different suppliers of
an activity, f=1,..,|F| (precedently α β);
H : a large positive number.
4.2 Variables
C
max
: completion date of all activities;
s
i
: starting time of activity i;
x
i,j
: binary variable equal to 1 if activity i is
realized before activity j and equal to 0
otherwise;
y
i,j,f
: binary variable equal to 1 if there is a non-
null cashflow from activity i to defray the f
th
supplier of activity j and equal to 0
otherwise;
φ
i,j,f
: denotes, when two activities i and j are
performed by the same logistic unit, the
number of financial units directly
transferred from supply chain activity i to
supply chain activity j (φ
ijf
0 if k=μ
i
=μ
j
and
φ
i,j,f
=0 if k≠μ
i
);
4.3 Linear Formulation
max
CMin
(0)
ma
x
Cps
ii
,
Vi
(1)
1
,,
ijji
xx
,
ji
jiVji
,/,
(2)
iij
pss
,
jiJJVji
ji
,/),(
(3)
FfjiVji
ytH
pyHxss
ji
fjifji
ifjijiij
,,,,
)(
)1(
,,,
,,,
(4)
jm
VjFf
fj
mMmT


/,
,,0
(5)
fjjji
fj
Vi
fji
tjiFfVj
c
,
,,,
,,/,
,
(6)
fj
Vi
fji
c
,,,
,
fjjji
tFfVj
,
,/,
(7)
fiiji
i
Vj
fji
tjiFfVi
r
,
,,
,,/,
(8)
fiiji
i
Vj
fji
tFfVi
r
,
,,
,/,
,
(9)
jijifji
FfVjixHy
/,,,.
,,,
(10)
jifjifji
FfVjiyH
/,,,.
,,,,
(11)
jifjifji
FfVjiy
/,,,
,,,,
(12)
jifijji
jiFfVjiyx
,/,,,1
,,,
(13)
jifji
FfVji
/,,,0
,,
(14)
The first line (equation 0) refers to the objective of
the problem: minimizing the completion time of all
operations. Constraint (1) gives the expression of the
makespan. Constraint (2) defines the precedence
between activities occurring on the same logistic
unit. Constraint (3) ensures that precedence
constraints are respected between activities of a job.
Constraint (4) adjusts starting dates of activities
when an inflow is needed. If no inflow is needed
(y
i,j,f
= 0), the activity j starts after the end of
operation i, if i is processed before j on the logistic
unit. If y
i,j,f
= 1
,
then, the solver refers to (δ
i
-t
j,f
) as the
AnOptimizationApproachforJob-shopwithFinancialConstraints-IntheContextofSupplyChainScheduling
ConsideringPaymentDelaybetweenMembers
193
time needed between operation i and j. Constraint
(5) avoids to exceed the initial treasury available
when allocating resources to logistic unit. Constraint
(6) ensures that the sum of cash flows from logistic
units and initial treasury is equal to the cash outflow
needed for the supplier f of activity j. Constraints (7)
is identical but take into account the case where the
logistic unit receive an inflow from itself before the
payment of a supplier. Constraint (8-9) ensures that
the sum of cash flows from the considered logistic
unit to the next ones never exceeds the inflow
resulting from its activity. Constraint (10) stipulates
that if logistic activity i occurs before activity j
(x
ij
=1) then a cash flow is possible from i to j. If
activity i does not come before activity j (x
ij
=0) then
no flow is allowed between i and j. Constraints (11-
12) ensure that if there is a cash flow from i to j for
the supplier f then y
i,j,f
=1. If y
i,j,f
=0 then no flow is
possible from i to j. Constraint (13) stipulates that if
activity i occurs before activity j then no cash flows
are possible from j to any supplier of i. Constraint
(14) ensures that no flow is possible between
different logistic units, overall suppliers.
5 GRASPXELS APPROACH
5.1 GRASPxELS Principles
The GRASPxELS is a multi-start metaheuristic
based on a GRASP (Greedy Randomized Adaptive
Search Procedure (Feo et al., 1994)) extended with
an ELS (Evolutionary Local Search (Wolf and
Mertz, 2007)). The GRASPxELS, first proposed by
(Prins, 2009), helped to bring very good results in
term of quality and speed to several problems. The
association of both, GRASP and ELS, aims to
propose a better metaheuristic which will explore a
wider range of solutions. A template algorithm of
the GRASPxELS is proposed below:
Algorithm 1: GRASPxELS.
Procedure name GRASP
ELS
Begin
1. S* Ø
2. for p := 1 to np do
3. S Construction_Phase
4. S Local_Seach_Phase
5. if (f(S) < f(S*)) then
6. S* S
7. endif
8. SEvolutionaryLocalSearch_Phase
9. if (f(S) < f(S*)) then
10. S* S
11. endif
12. endfor
13. return S*
end
As stressed in the Algorithm 1, a GRASPxELS is
divided into three phases: the construction phase, the
local search phase and the ELS phase. The different
specificities corresponding to those different phases
are presented in the next sub-section.
5.2 Specificities
Construction phase: As the main objective is to
propose a solution with minimal makespan, a
construction rule based on the duration of the
activities is chosen. At each construction step an
activity is randomly chosen from a list of activities
with small durations.
Local search phase: We chose to use a local
search relying on the neighborhood from (Van
Laarhoven et al., 1992). The algorithm of the local
search procedure can be found in (Kemmoe et al.,
2011b).
ELS phase: In the ELS phase, neighborhood of
local optimum solutions is explored through
mutations and then ameliorated thanks to the local
search. The mutation consists in permuting elements
in the repetition vector used by (Bierwirth, 1995) if
they belong to different jobs. Principles of the ELS
are shown in the Figure 5.As it can be seen in this
Figure, a neighbor could have its makespan
ameliorated or not depending on its initial quality –
this is represented with arrows between the
generation of neighbors and the local search.
...
...
For j from 1 to nb_ELS
Input solution
Generated
Neighbours
Neighbours
after LS
Best found
neighbour
Best solution
found
Higher
quality
Lower
quality
Figure 5: ELS principals.
Finally, in the next sub-section the evaluation
function for the Job-shop problem with financial
constraints is presented, as it is the most important
algorithm of this study.
5.3 Evaluation of a Bierwirth Vector
As mentioned before, a sequence of the operations
relies on a Bierwirth’s vector. The evaluation
ICORES2015-InternationalConferenceonOperationsResearchandEnterpriseSystems
194
function has been split into two parts. The first part
concerning the evaluation of the vector without cash
flows is presented in the Algorithm 2. The
Algorithm 2 returns the makespan and the starting
dates of the activities. However, no information are
given about the cash position of the treasuries.
Algorithm 2: Evaluation.
Procedure name Evaluation
Input/output
: sequence to evaluate
Input
n: number of operations
j: number of jobs
m: number of machines
Variables
i: loop index
op_M[]: last operation on machine
t_Job[]: time we have treated job
job: job treated
vertex: vertex of the job treated
machine: machine for the operation
d: end date of conjunctive predecessor
dPD: end date of disjunctive predecessor
father: predecessor
fatherD: disjunctive predecessor
Begin
1. FOR i :=0 to m DO op_M[i] = -1; END
2. FOR i :=0 to j DO t_job[i] = 0; END
3. FOR i :=0 to n DO
4. job :=
.sequence[i] ;
5. vertex := Vertex of job operation;
6. machine := machine of vertex;
7. d := 0; dPD := 0;
8. father := -1; fatherD := -1;
9. IF t_Job[job] <> 0 THEN
10. //Conjunctive father
11. father := vertex – 1 ;
12. d:= End[father];
13. END IF
14. IF (op_M[machine] <> -1) THEN
15. //Disjunctive father
16. fatherD := op_M[machine] ;
17. dPD:= End[fatherD] ;
18. END IF
19. IF (dPD > d) THEN
20. //father is the disjunctive one
21. father := fatherD ;
22. d :=dPD ;
23. END IF
24. save d and father into
;
25. Increment t_Job[job];
26. op_M[machine] vertex;
27. END
28.
.makespan:=0
29. FOR i:=0 to m DO
30. IF End[op_M[i]]> makespan THEN
31.
.makespan:= End[op_M[i]]
32. END IF
33. END
End
Consequently, another algorithm must deal with the
cash flows. The inclusion of cash flows can be done
in several ways. First the Algorithm 1 could
compute the makespan and a reparation procedure
would modify the starting dates of activities to
respect cash position of the treasuries. However this
is a bad solution because it will imply changing in
cascade in order to keep the solution consistent, thus
increasing the computation time uselessly. Hence
treasury handling must be done inside the evaluation
function with a call to the Algorithm 3 presented
below.
Algorithm 3: cashFlow.
Procedure name cashFlow
Input/Output
d: theoretic start date of operation
tr: treasury of the current machine
iT: index for moves in tables
pTr: number of negative cash position
Input
mac: current machine
del[]: suppliers delays for operation
cost[]: suppliers cost for operation
waitingP[]: awaiting inflows for tr
dInPay[]: dates of inflows for tr
Variables
dPayF: theoretic outflow date for a
supplier
f: loop index for suppliers
Begin
1. FOR f 1 to nbF DO
2. IF tr cost[op][f] THEN
3. disburse tr[mac] ;
4. ELSE
5. dPayF≔ d + del[op][f] ;
6. WHILE (dPayF > dInPay[mac][iT]
DO
7. Disburse tr ;
8. Increment i_T ;
9. END
10. IF tr cost[op][f] THEN
11. Disburse tr;
12. ELSE
13. IF mac receives payment before
payment of suppliers THEN
14. collect tr;
15. END IF
16. WHILE tr < cost[op][f] AND
17. i_T < size(dInP) DO
18. collect tr;
19. d := dP[mac][i_T] – del[op][f] ;
20. I_T +=1 ;
21. END
22. Disburse tr;
23. IF tr < 0 THEN
24. Increment pTr;
25. exit(FOR) ;
26. END IF
27. END IF
28. END IF
29. END
End
The call to the Algorithm 3 in the evaluation
function is done between lines 23 and 24 of the
Algorithm 2. The important part in this algorithm is
the variable pTr as it stores the number of operations
that conduce to a negative cash position on the
AnOptimizationApproachforJob-shopwithFinancialConstraints-IntheContextofSupplyChainScheduling
ConsideringPaymentDelaybetweenMembers
195
treasury. This variable is used in a Lagrangian
relaxation-like way, keeping the solutions even if
they violate the constraint. It allows to explore non-
suited solutions that can lead to better ones while
exploring their neighborhood as it is not certain that
a direct path exists between two good solutions
without considering bad ones. Thus, if a power of
ten (PT
a
) directly superior to the worst possible
solution is considered, a sequence’s (seq) cost will
be formed as follows:
Seq.cost = (seq.pTr)PT
a
+seq.makespan
Finally, sequences are compared on their
respective costs and not on their makespan anymore.
It can be deduced from the previous formula that if
there is no problem encountered on the treasuries,
then seq.pTr = 0, and consequently seq.cost =
seq.makespan which is the wanted value. Results
obtained are presented in the next sub-section.
5.4 Computational Evaluation
The experiment is performed on twenty instances
built upon the Lawrence’s instances for the Job-shop
problem. The algorithms have been implemented in
C++ and have been executed on a 2523.09 MFLOPS
computer (Linpack Benchmark). The parameters
used in the GRASPxELS for the number of restart,
the number of ELS and the number of neighbours
are respectively 100, 50, 10. For each instance ten
replications have been made. The results (Table 1)
are compared with the ones obtained thanks to the
linear model. On Table 1 the columns S and TT(s) of
the linear model part refer to the solutions obtain
with the CPLEX 12 solver. Concerning the
GRASPxELS part, the column S corresponds to the
average makespan, TT to the total average execution
time, TTB to the average time to the best solution,
DEV to the deviation to the best know solution
(BKS). The three other columns refer to the best
found solution (BFS), the time to found BFS and the
deviation from the BKS.
The results show the strength of the
GRASPxELS. Best solutions have been found in
less than two tenth of a second. The makespan of the
solutions are at less than 0.33 percent from the LP
BFS, and the algorithm found the optimal solution
sixteen times on the twenty instances. The presented
results show that the use of a metaheuristic is really
helpful when searching for good solutions rapidly.
Even if the results are not always the best ones, their
quality and their low deviation to the best known
solution enlighten their unavoidability when
studying large size instances.
Table 1: Results obtained with the GRASPxELS and the CPLEX solver.
Linear programming GRASPxELS
INSTANCE S TT(s) S TT TTB DEV BFS T_BFS DEV_BFS
la01
Financial
666* 176.80
666* 0.04 0.04 0 666* 0 0
la02
Financial
655* 342.26
655* 0.04 0.04 0 655* 0.01 0
la03
Financial
639* 1837.00
650.1 1.35 0.5 1.74 646 0.28 1.1
la04
Financial
615* 280.38
616.7 0.66 0.56 0.28 615* 0.35 0
la05
Financial
593* 417.07
593* 0.02 0.02 0 593* 0 0
la06
Financial
926 86400.00
926 0.01 0.01 0 926 0 0
la07
Financial
890 86400.00
890 0.02 0.02 0 890 0.01 0
la08
Financial
888 86400.00
888 2.51 0.13 0 888 0.02 0
la09
Financial
951 86400.00
951 0.07 0.07 0 951 0.01 0
la10
Financial
958 86400.00
958 0.01 0.01 0 958 0 0
la11
Financial
1222 86400.00
1222 0.03 0.03 0 1222 0.02 0
la12
Financial
1039 86400.00
1039 0.03 0.03 0 1039 0.01 0
la13
Financial
1150 86400.00
1150 0.02 0.02 0 1150 0.01 0
la14
Financial
1292 86400.00
1292 0.01 0.01 0 1292 0.01 0
la15
Financial
1216 86400.00
1207 0.09 0.09 0.74 1207 0.03 0.74
la16
Financial
979* 13172.13
979* 0.04 0.04 0 979* 0.01 0
la17
Financial
784* 274.01
784.6 1.85 0.82 0.08 784* 0.38 0
la18
Financial
853* 198.35
871.9 2.27 0.75 2.22 857 0.06 0.47
la19
Financial
842* 280.42
846.9 1.77 1.03 0.58 842* 0.32 0
la20
Financial
913* 301.01
934.6 2.3 0.66 2.37 927 1.11 1.53
Average :
0.66 0.24 0.33 0.13 0.12
*Asterisks denote proven optima using the LP
ICORES2015-InternationalConferenceonOperationsResearchandEnterpriseSystems
196
6 CONCLUSIONS
This study aims to present the relationship between
physical flows and cash flows through a supply
chain. The different actors of a supply chain should
carefully understand the relationship between supply
chain material activities and cash flows in order to
make operational decisions which will not
jeopardize the whole supply chain. While taking
such decisions, the goal still is to propose the highest
productivity among the supply chain. The problem is
modeled as a Job-shop scheduling problem with
financial consideration as an additional constraint. In
this study it is proposed to schedule operations or
activities while handling cash flows on treasuries in
order to always have a positive cash position. As a
consequence, the results of our study could also
affect the costs of bank overdraft that could be
negotiated. Our case study shows the relevance of
the proposed approach for a “company supply
chain”, since cash flow constraint is addressed
simultaneously with operational planning and
scheduling. Even if a mixed integer linear program
is proposed, it is difficult to solve the problem
exactly since it considers both operation scheduling
and cash-flow resolution simultaneously.
Furthermore, our instances were not representative
of the size of the problems that could be encountered
in the industry. Therefore a strong metaheuristic has
been implemented, the GRASPxELS, in order to
obtain faster results. The provided results are of
good quality, closed to the best solutions
encountered thanks to the solver which validate our
work. This study comes in addition of the past ones
on the subject of Job-shop’s like scheduling
problems with extra cash-flow constraints. A
dynamic Job-shop with random payment delays for
suppliers could be mentioned as a future study, or
the use of a flexible Job-shop model with different
payment costs depending on the chosen logistic units
for the activities.
REFERENCES
Bertel, S., Féniès, P., and Roux, O., 2008. Optimal cash
flow and operational planning in a company supply
chain. International Journal of Computer Integrated
Manufacturing, 21(4), 440-454.
Bierwirth, C. 1995. A generalized permutation approach to
Job-shop scheduling with genetic algorithms. OR
spektrum, 17, 87-92.
Comelli M., Féniès, P., and Tchernev, N., 2008.
Combined financial and physical flows evaluation for
logistic process and tactical production planning:
Application in a company supply chain. International
Journal of Production Economics, 112(1), 77-95.
Elazouni, A. and Gab-Allah, A., 2004. Finance-based
scheduling of construction projects using integer
programming, Journal of Construction Engineering
and Management ASCE 130 (1), pp. 15–24.
Elmaghraby, A. E., and Herroelen, W. S., 1990. The
scheduling of activities to maximize the net present
value of project. European Journal of Operational
Research, 49, 35-49.
Feo T. A., Resende M. G. C., Smith S. H., A greedy
randomized adaptive search procedure for maximum
independent set. Operations Research 1994;42: 860–
878.
Gormley, F. M., and Meade, N., 2007. The utility of cash
flow forecasts in the management of corporate cash
balances. European Journal of Operational Research
182, 923–935.
Grosse-Ruyken, P. T., Wagner, S. M., and Jönke, R.,
2011. What is the right cash conversion cycle for your
supply chain. Int. J. Services and Operations
Management 10, 13-29.
Jain, S., Meeran, S., 1999: Deterministic job-shop
scheduling: Past, present and future. European
Journal of Operational Research 113, 390-434.
Kemmoe, S., Lacomme, P., Tchernev, N., and Quilliot, A.,
2011a. Mathematical formulation of the Supply Chain
with cash flow per production unit modelled as job-
shop. IESM, May 25-27, Metz, France, 93-102, ISBN.
978-2-29600532-3-4.
Kemmoe, S., Lacomme, P., Tchernev, N., and Quilliot, A.,
2011b. A GRASP for supply chain optimization with
financial constraints per production unit, MIC, July
25-28, Udine, Italy.
Kemmoe, S., Lacomme, P., Tchernev, N., 2012.
Application of job-shop model for supply chain
optimization considering payment delay between
members. ILS, August 26-29, Quebec (Canada).
Najafi, A. A., Taghi, S., and Niaki, A., 2006. A genetic
algorithm for resource investment problem with
discounted cash flow. Applied Mathematics and
Computation, 183:1057-1070.
Prins, C., 2009. A GRASP x evolutionary local search
hybrid for the vehicle routing problem. In: Pereira FB,
Tavares J, editors. Bio-inspired algorithms for the
vehicle routing problem, Studies in computational
intelligence, 161, 35–53.
Roy, B., Sussmann, B., 1964. Les problèmes
d’ordonnancement avec contraintes disjonctives. Note
DS no. 9 bis, SEMA, Paris.
Russell, A. R., 1986. A comparison of heuristics for
scheduling projects with cash flows and resource
restriction. Management Science, 32(10):1291-1300.
Stadtler, H., 2005. Supply chain management and
advanced planning––basics, overview and challenges,
European Journal of Operational Research, 163(3),
575-588.
Tsai, C. Y., 2008. On supply chain cash flow risks.
Decision Support Systems 44, 1031–1042.
AnOptimizationApproachforJob-shopwithFinancialConstraints-IntheContextofSupplyChainScheduling
ConsideringPaymentDelaybetweenMembers
197
Ulusoy, G., and Cebelli, S., 2000. An equitable approach
to the payment scheduling problem in project
management. European Journal of Operational
Research, 127,262-278.
Van Laarhoven, P. J. M., Aarts, E. H. L. and Lenstra, J.
K., 1992: Job-shop scheduling by simulated annealing;
Operations Research; 40(1), 113-125.
Wolf, S., Merz, P., 2007. Evolutionary local search for the
super-peer selection problem and the p-hub median
problem. In: Lecture notes in computer science.
Berlin: Springer, 4771, 1–15.
ICORES2015-InternationalConferenceonOperationsResearchandEnterpriseSystems
198