Available online at www.sciencedirect.

com
Available online at www.sciencedirect.com
Available
Available online
online at
at www.sciencedirect.com
www.sciencedirect.com
Available
Availableonline
onlineatatwww.sciencedirect.com
www.sciencedirect.com

ScienceDirect
ScienceDirect 
Procedia Manufacturing 00 (2018) 1070–1075
Procedia Manufacturing 00 (2018) 1070–1075 www.elsevier.com/locate/procedia
Procedia
Procedia Manufacturing
Manufacturing 00 (2018) 1070–1075
Procedia Manufacturing 22 00 (2018)
(2018) 1070–1075
1063–1068 www.elsevier.com/locate/procedia
www.elsevier.com/locate/procedia
Procedia Manufacturing 00 (2017) 000–000 www.elsevier.com/locate/procedia
www.elsevier.com/locate/procedia
11th International Conference Interdisciplinarity in Engineering, INTER-ENG 2017, 5-6 October
11th
11th International Conference Interdisciplinarity in
2017, Tirgu-Mures, Engineering,
Romania INTER-ENG 2017, 5-6 October
11th International
International Conference
Conference Interdisciplinarity
Interdisciplinarity
2017,
in
in Engineering,
Tirgu-Mures, Engineering,
Romania
INTER-ENG
INTER-ENG 2017,
2017, 5-6
5-6 October
October
2017,
2017, Tirgu-Mures,
Tirgu-Mures, Romania
Romania
Multiple
Multiple solutions
solutions in
in linear
linear programming
programming problem
problem
Multiple solutions in linear
Multiple solutions Marcel
in linear programming
programming
a,∗
problem
problem
Bogdan
Manufacturing Engineering Society International Conference
a,∗ 2017, MESIC 2017, 28-30 June
2017,
Marcel
Marcel
Vigo
Bogdan
Bogdan
(Pontevedra),
a,∗
a,∗ Spain
a ”Petru Maior”
Marcel
University,
a ”Petru
Bogdan
N. Iorga street, No. 1, Tı̂rgu Mureş, Romania
a ”Petru Maior” University, N. Iorga street, No. 1, Tı̂rgu Mureş, Romania
a ”Petru Maior”
Maior” University,
University, N.
N. Iorga
Iorga street,
street, No.
No. 1,
1, Tı̂rgu
Tı̂rgu Mureş,
Mureş, Romania
Romania
Costing models for capacity optimization in Industry 4.0: Trade-off
Abstract
Abstract
Abstract
Abstract
between used capacity and operational efficiency
By applying the Simplex Algorithm, Matlab, or WolframAlpha one of these two answers is obtained: there is (a) solution or there
By
is noapplying theWe
solution. Simplex
continueAlgorithm, Matlab, started
the investigation or WolframAlpha
in [1], where onetwoof these
more two answers
subcases is attached
were obtained:when
therethe
is (a) solution
solution or there
exists. We
By applying the Simplex Algorithm, Matlab, or WolframAlpha one of
of these two banswers is obtained: there is
is (a) solution or
or there
By
is no
give
is
applying
no also
theWe
solution.
several
solution. We
Simplex
continue
examples,
continue A.
Algorithm,
the
in
the Santana
investigation
which we are
investigation ,
able P.
Matlab, astarted
or
to
started Afonso
WolframAlpha
in
further
in [1],
a,*
[1], where
specify
where ,
one A.
two
the
two Zanin
these
more two
standard
more , R.
answers
subcases Wernke
answers
subcases
is attached
were
weregiven by
attached
b when
obtained: therethe
Simplex
when the
(a) solution
solution
Algorithm
solution
there
exists.
or/and
exists.
We
by
We
is
giveno also
solution.
severalWeexamples,
continue in
thewhich
investigation
we arefor started
able in [1], where
tolinear
further specify two
themore subcases
standard weregiven
answers attached
by when theAlgorithm
Simplex solution exists.
or/and We
by
the computer
give also algebra
several systems
examples, in implemented
which we are abletheto programming
further specify the problems.
standard The problems
answers given are
by solved
Simplex here by
Algorithmour extended
or/and by
give
the also several
computer examples,
algebra systemsin implemented
which wea are able
University
for thetooffurther
Minho,
linear specify
4800-058the
programming standardPortugal
Guimarães,
problems. answers
The given by
problems Simplex
are solved Algorithm
here by our or/and by
extended
algorithm.
the computer In the case systems
algebra of multiple solutions, bwe
implemented for are
the able to programming
linear select the closest one to aThe
problems. previously
problems given
are point.
solved Ahere
model
by of anextended
our electric
the computer
algorithm. algebra
In the systems
case to implemented
of exemplify
multiple solutions, for the linear
Unochapecó, programming
89809-000 problems.
Chapecó, SC, The
Brazil problems are solved here by our extended
power system
algorithm. In is used this goal.we are able to select the closest one to a previously given point. A model of an electric
algorithm.
power In the
system the case of
case to
is used
multiple
multiple solutions,
of exemplify solutions,
this
we are
goal.we are able
able to
to select
select the
the closest
closest one
one to
to aa previously
previously given
given point.
point. A
A model
model ofof an
an electric
electric
c 2018 The
power system Authors.
system isis used
used toPublished
to exemplify by
exemplify thisElsevier
this goal.
goal. B.V.
power
c 2018 The Authors. Published by Elsevier B.V.
Peer-review

©cc 2018
2018 The under responsibility
Authors. Published of
by the scientific
Elsevier B.V. committee of the 11th International Conference Interdisciplinarity in Engineering.
 2018 The
Peer-review Authors.
The under
Authors. Published
Published by
responsibility Elsevier
bythe
of Elsevier B.V.
B.V.committee of the 11th International Conference Interdisciplinarity in Engineering.
scientific
Peer-review
Peer-review
Abstract
Peer-review under
under responsibility
responsibility
underprogramming;
responsibility ofof the
the scientific
scientific
of thealgorithm; committee
committee
scientific multiple
committee ofof the
ofthe 11th
the11th International
11thInternational
International Conference
Conference
Conference Interdisciplinarity
Interdisciplinarity in
in Engineering.
in Engineering.
Interdisciplinarity Engineering.
Keywords: linear simplex solutions; electric power system.
Keywords: linear programming; simplex algorithm; multiple solutions; electric power system.
Keywords:
Keywords: linear programming; simplex algorithm; multiple solutions; electric power system.
Under the linear programming;
concept simplex 4.0",
of "Industry algorithm; multiple solutions;
production processeselectric power
will besystem.
pushed to be increasingly interconnected,
information based on a real time basis and, necessarily, much more efficient. In this context, capacity optimization
goes beyond the traditional aim of capacity maximization, contributing also for organization’s profitability and value.
1. Introduction
1. Introduction
Indeed, lean management and continuous improvement approaches suggest capacity optimization instead of
1. Introduction
Introduction
1.
maximization.
The standard The study
form of aoflinear
capacity optimization
programming and costing
problem with data models
c ∈ Risn an important research topic m that deserves
n , A ∈ Mm,n (R), and b ∈ Rm is considered to
The
be The standard
contributions
The standard from
standard form form
both
form ofof a
the linear
of aa linear programming
practical and
linear programming problem
theoretical
programming problem problem with
perspectives. data
with data c ∈
This
data cc ∈ ∈RR n , A ∈presents
paperA∈
Rn ,, A ∈M M m,n (R),
Mm,n (R), and
and
(R), and b ∈ R
discusses
and bb ∈ ∈R m is
Rm is a considered
considered to
mathematical
is considered to
be  with m,n to
model
be for capacity management based on different 


 c,costing
x −→models min (ABC and TDABC). A generic model has been
be 

 c, x −→ min
developed and it was used to analyze idle capacity  and



 c,
A · xto
x =design
x −→ min
b min strategies towards the maximization of organization’s (1)


 c, −→
value. The trade-off capacity maximization vs operational 


 A · x = b efficiency is highlighted and it is shown that capacity (1)
(1)

 A
x≥
A · x b
· xRn=0bn .
= (1)
optimization might hide operational inefficiency.  



 x ≥ R n 0 n .
© 2017

 x ≥ n 0n .
x ≥R 0n . [2–4] and the interior point algorithm [5], the last one
R
TwoThe Authors. Published
algorithms are usually by used:
ElseviertheB.V. simplex algorithm n

Two
Peer-reviewalgorithms
implemented
Two under
in
algorithms are
Matlab
are usually
responsibility
[7] and
usually of used: the
the scientific
WolframAlpha
used: the simplex algorithm
committee
[8].
simplex [2–4] and
of the Manufacturing
algorithm theEngineering
interior point algorithm
Society [5], Conference
International the last one
Two algorithms
implemented
2017. in are usually
Matlab [7] and used: the simplex
WolframAlpha [8]. algorithm [2–4] [2–4] andand the
the interior
interior point
point algorithm
algorithm [5],
[5], the
the last
last one
one
The commonly
implemented
implemented in
in used[7]
Matlab
Matlab simplex
[7] and
and algorithm has[8].
WolframAlpha
WolframAlpha two exits: either provides a solution when the problem has one or
[8].
The commonly
establishes
The that thereused
is nosimplex
solution algorithm has
for the problem, two theexits: either
objective provides
function aa solution
being when
unbounded. the
Veryproblem
recently,has in one or
[1] we
The commonly
Keywords: commonly
establishesCostthat
Models;
used
used
thereABC;
is no
simplex
simplex
TDABC;
solution
algorithm
algorithm
Capacity
for
has
has two
Management;
the twoIdle
problem,
exits:
exits:
the
either
either
Capacity;
objective
provides
Operational
function a solution
providesEfficiency
beingsolution when
when the
unbounded. the
Very
problem
problem
recently,
has
hasin
one
one
[1]
or
or
we
proposed
establishes
establishes a continuation
that
that there
there is
is no
noof the algorithm
solution
solution for
for the
the when at
problem,
problem, least
the
the a solution
objective
objective exists
function
function and to
being
being decide whether
unbounded.
unbounded. or
Very
Very not the
recently,
recently, solution
in
in [1]
[1] is
we
we
proposed
unique.
proposed Ouraa continuation
comments
continuation of
were
of the
the algorithm
related to
algorithm the when at least
procedures
when at least to aaobtain
solution
solution exists
multiple
exists and to
solutions
and to decide
if
decide whether
there is the
whether or not
case.
or not the solution is
proposed
unique. Our a continuation
comments of the
were algorithm
related to the when at leasttoaobtain
procedures solution exists solutions
multiple and to decideif whether
there is the or not the solution is
case.
the solution is
1.We
unique. give
Our
Introduction
unique.
We Our
give
examples
comments
comments
examples
when
were
were the solution
related
related to
to the
theisprocedures
not a unique
procedures to
to point tomultiple
obtain
obtain illustrate
multiple the newif
solutions
solutions if cases
there
there in
is
is the case.
the
the proposed algorithm,
case.
whenWethe
We givebounded
give examples
examples setwhen
when
when
the
of solutions
the solution
is as is
the solution
solution is not
not a unique
isaa convex combination
not aa unique
unique
point
point
point to
toofillustrate
solutions
toofillustrate
illustrate
the
the new
theor/and
new cases
new cases
cases
in
in the proposed
the unbounded
in the set of algorithm,
the proposed
proposed solutions,
algorithm,
algorithm,
when
when the
respectively.
the bounded
These
bounded set
were
set of
of solutions
formulated
solutions is
isin as
[1].
as a convex
convex combination
combination of solutions
solutions or/and
or/and the
the unbounded
unbounded set
set of
of solutions,
solutions,
Thethe
when costbounded
respectively. ofThese
idle capacity
set
were is a fundamental
offormulated
solutions isinas information
a convex
[1]. combinationfor companies and their
of solutions management
or/and of extreme
the unbounded set ofimportance
solutions,
respectively.
in These
modern production
respectively. were
These were formulated
systems. in
In general,
formulated [1].
in [1].it is defined as unused capacity or production potential and can be measured
in∗ several ways: tons of production, available hours of manufacturing, etc. The management of the idle capacity
Corresponding author. Tel.: +40-265-233-112
*∗∗ Paulo Afonso.
Corresponding Tel.: +351
author. 253+40-265-233-112
Tel.: 510 761; fax: +351 253 604 741
∗ E-mail address:
Corresponding
Corresponding
marcel.bogdan@science.upm.ro
author. Tel.: +40-265-233-112
+40-265-233-112
E-mail address:author.
E-mailaddress: Tel.:
psafonso@dps.uminho.pt
marcel.bogdan@science.upm.ro
E-mail
E-mail address:
address: marcel.bogdan@science.upm.ro
marcel.bogdan@science.upm.ro
2351-9789 c 2018 The Authors. Published by Elsevier B.V.
2351-9789
2351-9789
Peer-review 
©cunder
2018responsibility
2017 TheAuthors.
The Authors.Published
Published
of byElsevier
by Elsevier
the scientific B.V.of the 11th International Conference Interdisciplinarity in Engineering.
B.V.
committee
2351-9789
2351-9789 c 2018
cunder The Authors.
2018responsibility
The Authors.of Published
Published by
by Elsevier
Elsevier B.V.
B.V.of the 11th International Conference Interdisciplinarity in Engineering.
Peer-review
Peer-review under responsibility of the scientific
the scientific committee
committee of the Manufacturing Engineering Society International Conference 2017.
Peer-review
Peer-review under responsibility of the scientific committee of
under responsibility of the scientific committee of the
the 11th
11th International
International Conference
Conference Interdisciplinarity
Interdisciplinarity in
in Engineering.
Engineering.
2351-9789 © 2018 The Authors. Published by Elsevier B.V.
Peer-review under responsibility of the scientific committee of the 11th International Conference Interdisciplinarity in Engineering.
10.1016/j.promfg.2018.03.151
1064 Marcel Bogdan / Procedia Manufacturing 22 (2018) 1063–1068
Marcel Bogdan / Procedia Manufacturing 00 (2018) 1070–1075 1071

A simple application for an electric power system is given to find the closest solution of (1) to a given point.

2. The Simplex Algorithm 2.0

In Step 4 we add the situations that provide the multiple solution and we are going to use a2α for the application
in section 2.1.

Algorithm

Step 1. Chose primal admissible base B generated by matrix A = (A j ) j∈{1,2,...,n} ;


Determine α0 j ( j ∈ B := { j ∈ {1, 2, ..., n} | A j ∈ B}) such that b = j∈B α0 j A j ;

Step 2. Determine αi j (i ∈ B := {1, 2, ..., n} \ B, j ∈ B) such that Ai = j∈B αi j A j ;
Step 3. Compute  
αi0 = αi j c j − ci , ∀i ∈ B; α00 = α0 j c j ;
j∈B j∈B

Step 4. Look for the signs of the numbers αi0 (i ∈ B)

a) ∀i ∈ B, αi0 ≤ 0 then declare: ”point x0 ∈ Rn , with the components x0j = α0 j , if j ∈ B, x0j = 0, if j ∈ B; is a

solution for problem (1) and α00 is the minimum value of the objective function”;

a1) max αi0 < 0, then declare: ”x0 is the unique solution generated by B”.
i∈B
s := 1;
a2) max αi0 = 0, then declare: ”x0 is NOT the unique solution”;
i∈B
form the set
B0 = {i ∈ B | αi0 = 0} = {ī1 , ī2 , ..., ī p };

a2α) ∃ī ∈ B0 , ∃k ∈ B such that αīk > 0;

Replace in the optimal base B, vector Aī with Ak , and go to Step 6 to find solution x s ;
and declare: ”each element of the convex hull of {x0 , x s } is solution for problem (1)”;
s := s + 1; go to case a2;
and declare: ”each element of the convex hull of {x0 , ..., x s−1 } s≥1 is solution for problem (1)”;
a2β) ∀ī ∈ B0 , ∀k ∈ B, αīk ≤ 0 then declare: ”the set of solutions is unbounded”;

b) ∃i ∈ B, αi0 > 0 then form the set

B+ = {i ∈ B | αi0 > 0};

Step 5. ∀i ∈ B+ , Look for the signs of the numbers αi j ( j ∈ B)

a) ∃i ∈ B+ such that ∀ j ∈ B, αi j ≤ 0 then declare: ”the objective function for (1) is unbounded from below
over S , therefore (1) has no solution”.
b) ∀i ∈ B+ , ∃ j ∈ B such that αi j > 0, chose an index h ∈ B+ ;

Step 6. Chose an index k ∈ B such that αhk > 0 and

α0k α 
0j
= min | j ∈ B, αh j > 0 ;
αhk αh j

Step 7. Replace in B, vector Ak with Ah then go to Step 2.

We gave in [1] the following commonly used example that exemplifies the uses of its extension.
 Marcel Bogdan / Procedia Manufacturing 22 (2018) 1063–1068 1065
1072 Marcel Bogdan / Procedia Manufacturing 00 (2018) 1070–1075

Example 2.1. Let us consider the linear programming problem




 x1 − x2 − 3x3 + 5x4 −→ min





 −x1 + x2 + x3 + x4 =1




 x1 − 2x 2 + 3x 3 + x 5 =2 (2)





 x1 − x2 + x6 = 3



 x1 , ..., x6 ≥ 0,
         
 −1   1   1   1   0 
         
with data c = (1, −1, −3, 5, 0, 0), A = (A j ) j∈{1,...,6} , where A1 =  1  , A2 =  −2  , A3 =  3  , A4 =  0  , A5 =  1  ,
         
1 −1 0 0 0
   
 0   1 
   
A6 =  0  , and b =  2  .
   
1 3
The canonical basis {A4 , A5 , A6 } is primal admissible. We have B = {4, 5, 6} (as input in step 1) the set of indexes

for which b = j∈B α0 j A j . By the simplex algorithm, one gets {A1 , A2 , A3 } as optimal basis, so B = {1, 2, 3} (as output
in step 4).
The solution xB has the components xBj = α0 j , if j ∈ B, else xBj = 0, and it is

xB = (16, 13, 4, 0, 0, 0) ∈ R6 ,
the same with the one generated by WolframAlpha
 
LinearProgramming[{1, −1, −3, 5, 0, 0}, {−1, 1, 1, 1, 0, 0}, {1, −2, 3, 0, 1, 0}, {1, −1, 0, 0, 0, 1} , {{1, 0}, {2, 0}, {3, 0}}].
Given the data c = [1 −1 −3 5 0 0], Aeq = [−1 1 1 1 0 0 ; 1 −2 3 0 1 0; 1 −1 0 0 0 1], beq = [1; 2; 3], lb = zeros(1, 6),
x = linprog(c, [ ], [ ], Aeq, beq, lb, [ ]), the solution obtained by Matlab is
x = (102.8876, 99.8876, 4.0000, 0, 86.8876, 0).
Note that the set of solutions for (2) can be written as
{xB + α · (x − xB ) | α ≥ 0}.
Remark 2.1. Related to the case a2α from Step 4 of the Simplex Algorithm 2.0 we cannot conclude on the bounded-
ness of the solution set of (1). Indeed, the set of solutions could be bounded or unbounded (see [1]), which can be
seen by the following two problems:

1. 


 x1 + x2 + x4 −→ min




 x1
 + x3 − x4 = 2

 (3)


 x2 − x3 + 2x4 = 3



 x1 , x2 , x3 , x4 ≥ 0,



 x3 − x4 ≤ 2



with the complete solution (convex) bounded. Problem (3) reduces to the system of inequalities 
−x3 + 2x4 ≤ 3



 x3 , x4 ≥ 0,
that represents co{(0, 0), (2, 0), (7/2, 5), (0, 3)}.
2. 


 x1 + x2 −→ min




 x1
 + x 3 − x4 = 2

 (4)


 x 2 − x 3 + x4 = 3



 x1 , x2 , x3 , x4 ≥ 0,
1066 Marcel Bogdan / Procedia Manufacturing 22 (2018) 1063–1068
Marcel Bogdan / Procedia Manufacturing 00 (2018) 1070–1075 1073
     
1 0 1
with the complete solution (convex) unbounded. From data A = (A j ) j∈{1,2,3,4} , A1 = , A2 = , A3 = ,
0 1 −1
 
−1
A4 = , B = {A1 , A2 } is an optimal basis with solution x0 = (2, 3, 0, 0) and B0 = {3, 4}. Following the steps
1
of the algorithm we obtain
 the solution x1 = (0, 5, 2, 0), and the case a2β with ī = 4. Problem (4) reduces to the


 x3 − x4 ≤ 2



system of inequalities −x 3 + x4 ≤ 3 The complete solution contains co{x0 , x1 }.



 x3 , x4 ≥ 0.

In both cases WolframAlpha returns one solution.

2.1. Examples

We use Simplex Algorithm 2.0 and WolframAlpha (WA) for the following problem and compare the results.

Example 2.2. The usual Simplex Algorithm stops after one obtains a solution for



 −x1 − x2 −→ min




 x1 + x2 ≤ 1 (5)



 x1 , x2 ≥ 0.



 −x1 − x2 −→ min



From its standard form   x1 + x2 + x3 = 1 we have data c = (−1, −1, 0), A = (A j ) j∈{1,2,3} , A1 = 1, A2 = 1,



 x1 , x2 , x3 ≥ 0,
A3 = 1, and b = 1. {A1 } can be considered as basis, that is optimal. We are in the case a2α, B = {2, 3} = B0+ .
Using the usual simplex algorithm we get the solution x1 = (1, 0, 0). Replacing A2 with A1 , we obtain another solution
x2 = (0, 1, 0). Therefore, the solution for (5) is the convex combination of (1, 0) and (0, 1), that is

S γ = {(1 − γ, γ) | γ ∈ [0, 1]}.

We take into account the general form of a linear programming problem implemented in [8], adapting data, the
command LinearProgramming [ ] for WA, in our case is

LinearProgramming[{−1, −1}, {{1, 1}}, {{1, −1}}].

The (unique) result generated is {1, 0} meaning the vector (x1 , x2 ) = (1, 0).
Let x0 = (1; 0, 1) ∈ R2 be given. The optimization problem

min xγ − x0 
xγ ∈S γ

gives the solution (0, 95; 0, 05) ∈ S γ , for γ = 0, 05.

In contrast to this example, in Example 2.1 we cannot continue with the classic simplex algorithm (A5 cannot be
introduced in the optimal base).
Similarly to Example 2.1, we consider the following trivial one.

Example 2.3. Given the linear programming problem




 x1 − x2 −→ min




 −x1 + x2 ≤ 0 (6)



 x1 , x2 ≥ 0,
 Marcel Bogdan / Procedia Manufacturing 22 (2018) 1063–1068 1067
1074 Marcel Bogdan / Procedia Manufacturing 00 (2018) 1070–1075



 x1 − x2 −→ min



from its standard form 
 −x1 + x2 + x3 = 0 the simplex algorithm provides the solution (0, 0, 0), thus for (6) we



 x1 , x2 , x3 ≥ 0,
take x0 = (0, 0). We are in the case a2β. Another solution for (6) is x1 = (1, 1), (or the one provided by Matlab
(84.7173, 84.7173) can be used) therefore the set of solutions is

{(α, α) | α ≥ 0}.

The (unique) solution returned by WA, using the instruction LinearProgramming[{1, −1}, {{−1, 1}}, {{0, −1}}] is
{0, 0}, i.e. (x1 , x2 ) = (0, 0).

Suppose that from an estimation we have an approximate feasible point or a simple x0 given. We consider the
following situation to be implemented. If the linear programming problem has multiple solution the next step is to
refine it and to find the closest one to x0 .
We take over from [6] the model of an electric power system.

Example 2.4. Suppose there are two consumers C1 , C2 and two sources R1 , R2 . The necessary powers to be covered
for the consumers are P1 and P2 , respectively. The distances are li j , i ∈ {1, 2}, j ∈ {1, 2}, from the source S i to the
consumer C j . The unknown power to be transported are x1 = P11 , x2 = P21 , x3 = P12 , x4 = P22 . The function to be
optimized is
l11 x1 + l21 x2 + l12 x3 + l22 x4 .
With the nominal values of the data, the canonical form is



 3x1 + 3x2 + 2x3 + 2x4 −→ min




−x1 − x2
 ≤ −10




 −x 3 − x4 ≤ −20



 x1 , x2 , x3 , x4 ≥ 0.

By the Simplex Algorithm 2.0 we get the solutions x1 = (10, 0, 20, 0), x2 = (0, 10, 20, 0), x3 = (0, 10, 0, 20),
x4 = (10, 0, 0, 20) and the set of solutions is
4 4
  
S = co{x1 , x2 , x3 , x4 } = αi · xi | αi ≥ 0, αi = 1 .
i=1 i=1

By using data c = [3 3 2 2], A = [−1 −1 0 0; 0 0 −1 −1], b = [−10; −20], lb = zeros(1, 4), and linprog(c, A, b, [ ], [ ], lb, [ ]),
Matlab generates the solution x M = (5, 5, 10, 10) ∈ S . WA returns x1 .
Let x0 = (10, 0, 19, 0) ∈ R4 be given. We have to solve

min xα − x0 .
xα ∈S

 1
The solution is the projection of x0 onto S and it is x̄α = 10, 0, 39
2 , 2 .

Acknowledgement

The author modified and improved a previous abstract thanks to the reviewer remark.

References
[1] M. Bogdan, Some comments on a linear programming problem, 2017 (submitted).
[2] E.B. Breckner, De la poliedre la jocuri matriceale, Editura Fundaţiei pentru Studii Europene, Cluj-Napoca, 2007 (in Romanian).
[3] E.B. Breckner, N. Popovici, Probleme de cercetare operaţională, Editura Fundaţiei pentru Studii Europene, Cluj-Napoca, 2006 (in Romanian).
[4] G. Dantzig, Linear Programming and Extensions, Princeton University Press, 1963.
1068 Marcel Bogdan / Procedia Manufacturing 22 (2018) 1063–1068
Marcel Bogdan / Procedia Manufacturing 00 (2018) 1070–1075 1075

[5] R. Robere, Interior Point Methods and Linear Programming, http://www.cs.toronto.edu/ robere/paper/interiorpoint.pdf
[6] D. Sarchiz, Optimizări ı̂n electroenergetică, Multimedia System, Tı̂rgu Mureş, 1993 (in Romanian).
[7] Matlab, http://www.mathworks.com/help/optim/ug/linprog.html
[8] WolframAlpha, https://www.wolframalpha.com/