Optimization for Hierarchical Production Planning of...
-
Upload
trinhthien -
Category
Documents
-
view
219 -
download
0
Transcript of Optimization for Hierarchical Production Planning of...
-
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:14 No:02 90
I J E N S IJENSIJENS April 2014 -IJMME-5757-029914
Optimization for Hierarchical Production Planning of
Industrial Processes Asmaa A. Mahmoud
1, Islam H. Afefy
2, M. Abdel-Karim
3 1. Post Graduate student -Industrial Engineering Department, Faculty of Engineering, Fayoum University
2. Assistant Prof. Industrial Engineering Department, Faculty of Engineering, Fayoum University. 3. Professor, College of Engineering & Information Technology, University of Business & Technology, Jeddah, Saudi Arabia
(On leave absence, Dept. of Industrial Engineering. Fayoum University, Fayoum, Egypt).
* Corresponding Author
Islam Helaly Afefy: Industrial Engineering Department, Faculty of Engineering, Fayoum University: e-mail:
Abstract-- In this paper, a generalized mathematical model formulation for cellular manufacturing system (CMS) using
hierarchical production planning (HPP) approach is proposed.
The model is applied to two different real case studies and is
solved by using operation research optimization software (Lingo
(12.0) program). The model is divided into three main steps as
follows: data collection, mathematical model formula, and
results. The proposed mathematical model of the optimization
can solve the problems of the system under utilizing the limited
resources in a production plan. The results show that the
proposed mathematical model can be used to minimize
manufacturing total costs of products for similar cases. To prove
the work, two case studies are introduced; for the first case
(electric water heater with capacity 50 liter (EWH1)), the results
show that the total cost decreases by 8.46 % for the optimum
conditions. For the second case, (electric water heater with
capacity 80 liter (EWH2)), as global results, the total cost
decreases by 3.7% for the optimum conditions.
Index Term-- Hierarchical Production Planning (HPP), Cellular Manufacturing System (CMS), Group Technology (GT) and
General Manufacturing Company (GMC).
1. INTRODUCTION
The hierarchical production planning (HPP) approach is firstly introduced by Hax and Meal (1975). In this approach,
Hax and Meal structured the problem into three levels
hierarchy based on types, families and items. Firstly,
aggregation planning was done on types which include all
production families. Secondly, families are the groups of
item within types classification. These items require a
major equipment setup before they are produced. The HPP
divides the large planning optimization problem known as
monolithic problem into sub- problems, as described in
Figure 1.The components of different levels of aggregation
for products are:
1- Top level result from an aggregate planning model using
linear programming and linear decision rules for variable
according to its types. These types are sets of items that are
similar in term of seasonal demand pattern and production
rates.
2- Middle level considers a heuristic disaggregation of the
type into families. These families are sets of items having
similar setup costs.
3- Low level decision consists of disaggregation families that
sets of items based on equalizing run out times.
The HPP approach is necessary to demonstrate the CMS. The CMS is subset of group technology (GT). The GT is a
new approach for batch production (small lot- production) in
which similarities exist among component part or processes.
While applying the GT, it depends on formation of part
families on basis of design and/or manufacturing.
Furthermore, applying GT leads to a significant improvement
on product design, manufacturing engineering, and the CMS
(Akturk and Wilson, 1998). The group technology family can
have more than one feasible cell for its production. Inside such
feasible cell, a family can be processed entirely. In general,
the feasible cell consists of two sub- cells defined as:
Primary cell, which is capable of producing the family at the lowest possible cost.
Secondary cells, those are ones in which the manufacture of the family is possible at a higher cost
due to the increase of both setup and material handling
costs.
Now, cell loading is a decision activity that determines
the kind of items and quantities to be produced in each cell
within specific time period. The production process is
subject to production capacity and demand forecast. In batch
production (BP) environment, there are two different
approaches for cell loading problem which are monolithic
and hierarchal approaches.
The First approach known as Monolithic Approach, formulates cell loading problem as a large Mixed-
Integer Linear Programming (MILP) problem at an
individual item level. In this case, a heuristic
procedure can be applied to solve cell loading
problem.
The second approach known as Hierarchical Approach, divides the overall problem into a
hierarchy of smaller problems (Akturk and Wilson,
1998).
Hierarchical approach that is used directly toward
enhancements in several ways utilizing knowledge of group
technology based on manufacturing system. These
enhancements can be summarized as:
Formulation of production planning problem with adding capacity constraints of the systems. The
-
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:14 No:02 91
I J E N S IJENSIJENS April 2014 -IJMME-5757-029914
results of the production rates are determined
within capacity of the system.
At the HPP approach the decision making is top-down, being from highest level to lower level.
At cell loading level, there are three basic ways of responding
to change in demand:
1. Holding alternatively constant production rate and using inventory to satisfy demand peaks.
2. Using changes in level of production which include overtime and assigning some of items to their
secondary cells with an additional production cost.
3. Combining these two strategies to meet demand modified by Akturk and Wilson (1998).
This paper is divided into five sections. In section 1, the
HPP approach is described shortly. In section 2, the
background of our research (literature reviews) is discussed.
In section 3 a methodology for solving the CMS is given. In
section 4 a description of case study and application
methodology for HPP are introduced. Finally; conclusions and
recommendations in section 5.
2. LITERATURE REVIEWS
The main objective of many researchers was to obtain an optimization method to solve the problems of the system
under utilizing the limited resources in a production plan to
meet variability in products demand. So a classification for the
HPP approach at many cases can be summarized as following:
Goloven (1975); based on Hax and Meal was first
introduced the HPP approach for single-stage production
planning and the concept of hierarchal planning through the
difference between planning and operational. He also
modified multi-stage production planning as an extension to
the HPP approach for single stage models to systems with
multi-stages. A pioneer at the HPP approach at the case of
seasonal demand was introduced by Haas (1979), as proposed
a heuristic to perform the levels of hierarchical planning.
Gfrerer and Zipfel (1995) were proposed the HPP approach at
the case of uncertain demand for multi-period model .It
consist of an aggregate planning level and a detailed planning
level. Then, this approach modified by Wu and Ierapetritou
(2007). Carravilla and Sousa (1995) discussed a complex
analysis production planning problem in a make- to- order
company using hierarchical production planning approach.
Katayama (1996) proposed a relevant production planning
procedure for multi-item continuous production in term of
two- stage automated hierarchical planning procedure. The
HPP approach for complex manufacturing systems is studied
in details by Mehraz et al., (1996). Akturk and Wilson (1998)
discussed the CMS by proposing a hierarchical cell loading
approach to solve the production planning problem. Ozdamar
and Birbil (1999) described the HPP approach at case of
energy intensive production environment for energy intensive
industries (e.g. steel, tile and glass ware industries).This
search was to minimize the number of active kilns throughout
the year besides optimizing the process design in the curing
department. El-Houssaine et al., (2011) investigated a robust
HPP approach for a two-stage real world capacitated
production system operating in an uncertain environment.
Habibie et al (2012), developed multi objectives model using
Lingo program (software).
3. METHODOLOGY
In this research, the CMS problem is solved using Lingo
program, which is a comprehensive tool designed to make
building and solving mathematical optimization models easier
and more efficient. Lingo (12.0) provides a completely
integrated package that includes a powerful language for
expressing optimization models and a full-featured
environment for building and editing problems. In addition, it
has a set of fast built-in solvers capable of efficiently solving
most classes of optimization models. Lingos results of the
CMS are reducing costs in order to be competitive, saving
time, increasing achieving market demand and achieving goal
of the system under certain criteria and constraints design.
3.1 Mathematical Modeling
This section will develop three parts for solving the CMS
problem. The first part illustrates objective function; the
second part defines the mathematical formulation while the
third part is devoted to illustrate index and notation. These
parts are described briefly in the following subsections.
A. Objective Function
The objective function is to minimize variable production
cost. Such as (production cost, cells setup cost, inventory
holding cost and regular capacity cost). These variables can be
simplified as:
4
1
4
1
2
1
7
1 )(
.)/()(t t i
itit
j
jtjtijt
Pi
ijtij
jFSi
ijtijt IFhRrXQBPXCCostMinj
(1)
B. Mathematical Formulation
There are three types of constraints at hierarchical model
for cell loading problem in the CMS. These constraints can be
defined as:
Production Constraints for families and items described by Eqs.(2) and (3)
Inventory Constraints for families and items described by Eqs.(2), (3), and (4)
Production and resources capacity constraints over the planning horizon described by Eqs (5), (6), (7), (8),
and (9)), respectively.
7
1
1,
j
itittiijt dIFIFX (2) , for j=1, 2,7,
t =1,2, 4
-
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:14 No:02 92
I J E N S IJENSIJENS April 2014 -IJMME-5757-029914
)(
0iTIk
n
Iw
itktw IFI
(3)
jtPi
ijtijtij
jFSi
ijtij RXQbpXaj
)./()(
(4)
)lim(0 itupperR jt j (5)
)( )(
)(jFSi iTIk
LT
n
Iw
kjtwkl RRZPR LR(j),j,t (6)
)1).(lim(0 LtLt itupperRR L,t (7)
0)(
jt
jLRl
Lt RRR (8)
0,,,,,,, LtLtktwkjtwjtjtitijt RRORIZROIFX j,t (9)
C. Index and Notation
In this section, the notations in the formulation will be
described. It should be noted that, all cost parameters are
measured in Egyptian pound (L.E). The notations are
classified into parameters and decision variables as shown in
details in the following:
Parameters:
ija
: Average total time required to produce one unit of
family i at cell j, (h/unit).
ijBP
: Setup cost for family i in primary cell j, (L.E).
ijbp
: Setup time for family i in primary cell j, (h/unit).
ijtC
: Average unit cost for producing one unit of
family i by cell j in period t, (L.E).
itd
: demand for family i in period t, (Unit).
ktwd
: Demand for item k in sub period w of period t,
(Unit).
FS (j): A feasible set of families assignable to cell j.
ith
: Average holding cost for family j in period t,
(L.E).
J : number of cells
K : number of items
L : number of resources
jP
: Set of families which their primary cell is j, (Unit)
klPR
: Average total time required to produce one unit
of item k using resource l, (h).
ijtQ
: Initial estimation for the lot size of family i in cell
j in period t, (Unit)
jtr
: Average cost of one regular time unit for cell j
during period t, (L.E).
T : planning horizon.
TI (i): A set of items belonging to family i
t : n*w, where n is an integer multiple , and w is sub
period.
lt
: Average proportion of time resource l is down in
period t.
Decision variables:
kitwI
: Inventory of item k at the end sub period w of
period t, (Unit).
itIF
: Inventory of family i at the end of period t, (Unit).
jtR
: Regular time used by cell j in period t, (h/month).
ltRR
: Regular time used by total resources l in period t,
(h/month).
ijtX
: Number of units of family i produced by cell j in
period t, (Unit).
kitwZ
: Number of units of item k produced by cell j in sub
period w of period t, (Unit).
The most important part in this study is conceptual
framework and mathematical model which were proposed for
CMS. The study shows the basic principles of calculations of
the method for CMS model. Through; define Lagrange theory,
apply decomposition procedures, update the values of
Lagrange method using gradient method and use optimization
software program Lingo (12.0) to solve mathematical model.
Moreover, the flowchart of algorithms framework is carried
out. This chart concluded in details all the steps of solving the
problem, and the final part is devoted to illustrate the proposed
software methodology in solving the mathematical
optimization models in an easier and more efficient way. In
the previous study M. Akturk, and G. Wilson, were stopped at
using Lagrange theory but our solution complete the previous
study as it consists of formulating a Lagrange relaxation of the
initial model, solving this dual problem by efficient and
iterative solution procedures, applying decomposition
procedures, updating the values of Lagrange method using
gradient method and using optimization software program
Lingo (12.0) to solve mathematical model.
4. CASE STUDY This case study analyzes the impact of the changes in
parameters of several goals that must be achieved. First, lets
define a real case study applied at General Manufacturing
Company (GMC) for manufacturing engineering in Egypt.
The company manufactures three products: full automatic
washing machine, gas cooker and electrical water heaters
(EWH) with capacity of (30, 40, 50, 80,100 Liter). Its main
With respect to
-
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:14 No:02 93
I J E N S IJENSIJENS April 2014 -IJMME-5757-029914
interest is studying the last product i.e. EWH especially with
capacity of 50 L and 80 L in details. Hence, the proposed
mathematical model is applied through data collection as a
first step then through mathematical formula as a second step
and finally through obtaining results in the last step.
4.1. Problem Statement
In this section, we describe the problem of planning
production in this application. We first discuss the problem
statement, and then we introduce the solution method. In
general, systems play an important role in many industrial
processes. Particularly, the main objective of the CMS is to
obtain the optimum method to solve the problem of
minimizing the variable production costs (production cost,
cells setup cost, inventory holding cost and regular capacity
cost). The problem of our case study is to minimize the total
product cost, and increase profitability ; therefore, we use the
HPP approach for solving this problem .To achieve this target,
it is necessary to manage and control the relation between the
limited resources in the plan of the production and variability
in products demand.
4.2. Solution Method
A schematic flow chart for framework of algorithm used
to the CMS is illustrated in details in figure 2. Using this
algorithm provides optimal solution for the system. It should
be noted that, the algorithm consists of three main steps
named as: data collection, mathematical model and results.
These steps are discussed too. This schematic is introduced in
details at Appendix. A and B and its steps of optimal solution
are shown.
4.3 Data Collection In order to illustrate the capabilities of the proposed-model,
actual data from the GMC (with emphasis on the case of EWH
with only two capacities i.e. 50 Liter and 80 Liter) have been
collected. A comprehensive numerical example is presented to
illustrate the applicability of the proposed model in
environment. This example is solved with the Lingo (12.0)
software.
In this case, all different types of cost parameters that are
used for the computational study are carried out. The data
collections concerning of studying EWH which is presented as
following: 2 families of EWH with capacity of 50L (EWH1)
and EWH with capacity of 80L (EWH2). In addition to 7
manufacturing cells, 4 planning horizon period (4 weeks), 5
resources of production (machine, manpower, material ,
maintenance (spare parts)and energy), 65 (parts) items, no
secondary cells in production , no overall overtime, no
backorder, P1,P2,,P7 ={family 1.2}, where P1 is a set
of families at primary
cell(1),FS(1),.FS(7)={family1.2},where FS(1) is a
feasible set of families assignable to cell (1), TI (1) =TI (2)
=65 items, where TI (1) is a set of items to family(1),and
LR(1)=LR(2),LR(7)={resource1,2,3,4,5},where
LR(1) is a set of resources belonging to cell (1). Moreover,
variable parameters in the system are summarized in table (1),
where UN~ [a, b] represents a uniformly distributed random
variable in interval [a, b]. It should be noted that, most of the
value of these parameters are constant throughout the planning
horizon .In addition, the processing time of each item at each
feasible resource klPR , and the number of operations for each
item are fixed.
4.4 Results In this section, the results for different cases which are
under consideration, will be presented and discussed in details.
Two comprehensive examples with full input data and results
analysis are presented. Also, an explanatory scheme shows the
results of our system for the planning period. First, let us
consider the first product i.e., the EWH1 product. The
comparison between the initial and optimum total product cost
for EWH1 is presented in figure 3. A significant difference is
founded in the total product cost. It can be observed that the
optimum total product cost (Z) of EWH1 decreases with
respect to the initial values from 4.48*108 L.E in the initial
value conditions to 4.1*108 L.E in the optimum.
There are different types of variables which are shown in
table 2. We have 15 variables, but the most effective three
were discussed. These three effective variables cause
decreasing in the total product cost as plotted in figure 4 and
6. The optimization problem has been solved for the variable
parameter values given in table 2. The optimum value of each
variable was calculated, in which minimizing the total product
cost for EWH1 is represented by that variable. This table
shows the average cost of one regular time unit of cell
decreases from 21164.73 L.E in the initial conditions to
20251.47 L.E in the optimum conditions. Moreover,
approximately 8.43% of the total product cost for EWH1 was
saved according to the optimization results in the initial
conditions.
A Comparison between Initial and Optimal Variables
Values for EWH1 Product are plotted in figure 4. It is found
that there are improvements of these variables which reflect
the reducing total product cost of EWH1. For example,
inventory of the product at the end of planning period
decreases from 1295 unit/ month in the initial
conditions to 1275 unit/ month in the optimum conditions.
Also, the total product cost decreased with the decreasing of
the inventory cost.
As shown in figure 5, is presented the comparison between
initial and optimum total product cost of EWH2 as obtained
from Lingo program. A significant difference in the total
product cost of EWH2 can be observed. This means reduction
in the total product cost of EWH2 can be obtain by using a
suitable optimum technique. This figure shows approximately
5.4*106 L.E was saved according to the optimization results in
the case study.
To discuss these results in details, the comparison between
initial and optimum variable values is presented in table 3.
The critical securitizations for this table indicate that. There is
notable improvement at specific effective variables
parameters. It found that the optimum total product cost
decreased with existence of the initial values. This table shows
that the average cost of one regular time unit of cell decreases
-
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:14 No:02 94
I J E N S IJENSIJENS April 2014 -IJMME-5757-029914
from 36751.25 L.E in the initial conditions to 36001.29 L.E in
the optimum conditions. Approximately 4.04 % of the total
product cost was saved according to the optimization results in
the case study of (EWH2). These results agree with the results
of those previous investigations for Akturk and Wilson, 1998.
For more clear and discussed data in table 3, where there
are effective variable parameters. These effective variable
parameters are presented and analyzed in graph as shown in
figure 6. From figure 6, it is observed that there are
improvements at these variables that reflect the reducing of
total product cost. For example, initial estimation for the lot
size of the product in cell at planning period decreases from
2055 unit/ month in the initial conditions to 2047 unit/ month
in the optimum conditions. Also, the total product cost is
decreased with respect to the decrease of the production cost.
4.5 Sensitivity Analysis on the Optimization Model for CMS Results
In order to carry out an analysis to determine the sensitivity
of the problem developed for CMS using the HPP approach, a
different parameters utilized in the case study for (EWH1,
EWH2) with emphasis on the effective variable parameters:
ijtX , ijtQ , itIF , itd ,and kitwI . Those values, of all these
parameters, that resulted in minimum total product cost are
chosen to be used for sensitivity analysis. The sensitivity of
the approach against variations in the numbers of ijtX , ijtQ ,
itIF , itd , and kitwI is presented in figures 7, 8, 9and 10.We
observe the effective variable parameters for (EWH1, EWH2)
against the variance and the variance percentage (%)
respectively. This sensitivity analysis agrees with Adnan.
Tariq (2010) and Tarik. Aljuneidi (2013). Figure 7 is plotted that the most effective variable on the
variance is jtR = 913.26 h/ month, which indicate that it is
effective variable in minimization of total product cost for the
case study. Further, as far as analysis is presented in figures 8,
9and 10 respectively. The analysis on whole is showing the
most effective parameter variance, and variance (%) for
(EWH1, EWH2).
Figure 8 discusses the most effective variable on the
variance (%) which is ijtQ = 21.45 %, and ijtX = itd = 4.46
%, that indicats these effective variables in the same sound at
minimization of total product cost for the case study.
As shown in figure 9, which presents an analysis regarding
that there are two equal effective variables on the minimizing
of total product cost which are ijtX and itd .therefore; these
two variables play an important role in the case study. The
same as for figure 10, which introduces an analysis of the
effective variables which cause minimizing the total product
cost for the case study .These variables are ijtX and itd .
Also there is a relation between total product costs saved
for (EWH1, EWH2) against the variance (%) that is shown in
figure 11. It found that there are big variances (%) through
using the proposed mathematical model. For more clear data
figure 12 presents the total product costs saved for (EWH1,
EWH2).
It can be concluded from the analysis of the effective
variable parameters which is presented above that serves the
satisfactory performance of the proposed algorithm. The total
product costs are saved for two products and the aim achieved
by using the HPP approach for CMS, which saved 37.85*106
L.E (34.8%), 54.57*106 L.E (4.04 %) respectively. These
results are obtained from the improvement.
As a result of the HPP approach for the CMS, the
following recommendations (critical issues for future study)
could be outlined as: Development of the model under more
variables parameters such as costs,
processing routes and machine availability, Aggregating
proposed model with other assumptions like layout problem
considerations of CMS, as the layout design of CMS is more
important, Lingo (12.0) can be used rather than Lingo (8.0)
because it is the most effective optimization software program
for solving the problem of CMS . The variables sound great
effects as they are significant for current study, but they could
be variable for another. Finally, verification should be applied
to check the validity of the proposed model in the study
5. CONCLUSIONS AND RECOMMENDATIONS
In this paper, the proposed mathematical model is
developed and applied to two different real case studies. The
proposed mathematical model is solved using the Lingo
program and good results. By analyzing and comparing theses
results with the current value, many improvement at total cost
result for similar cases .consequently; applying such technique
can be used to minimum manufacturing total costs of products
that is required at any manufacturing system to:
Reduce costs in order to be competitive.
Achieve goal of the system under certain criteria and constraints design.
Save time and increase achieving market demand.
Acknowledgement
The authors wish to express their thanks to the staff
members of the (GMC) for manufacturing engineering
Company Egypt, for their support during carrying out this
work.
REFERENCES
[1] Achmad Habibie, Muh. Hisjam, Wahyudi Sutopo,and Kuncoro H. Widodo , A Relationship Model between Supplier and
Manufacturer for Securing Availability of Teak Log in Export
Oriented Furniture Industry with Sustainability Considerations, In
Proceeding of the International MultiConference of Engineers
and Computer Scientists, Hong Kong , March 2012, Vol .
[2] Ahelmut Gfrerer, Ginther Zipfel, Hierarchical model for production planning in the case of uncertain demand, European
Journal of Operational Research, 86, pp. 42-161, 1995.
[3] Arnoldo C. Hax, Dan Candea, Production and Inventory Management, prentice- Hall, INC, Englewood Cliffs , New Jersey,
1984.
[4] Adnan. Tariq, Operational Design of a Cellular manufacturing System, PH.D. Thesis, National University of Science and
Technology (NUST), Rawlpindi, Pakistan, 2010.
-
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:14 No:02 95
I J E N S IJENSIJENS April 2014 -IJMME-5757-029914
[5] Bowman.F, Cerard.F.A, Design of Thermal system, Cambridge University press, London, 1990.Dan Wu, Marianthi. Ierapetritou, Hierarchical approach for production planning and Scheduling
under uncertainty, Chemical Engineering and Processing, 46, pp.
11291140, 2007 [6] El-Houssaine Aghezzaf, Carles Sitompul, Frank Van den Broecke,
A robust hierarchical production planning for a capacitated two-
stage production system, Computers and Industrial Engineering, 60, pp. 361372, 2011.
[7] Elizabeth Ann Haas,Hierarchical production planning, PH.D thesis, Massachusetts Institute of Technology, August 1979.
[8] Hiroshi Katayama,On a two-stage hierarchical production planning system for process industries, International Journal of
Production Economics, 44, pp. 63-72, 1996. [9] Hong Sen Yan, Hierarchical stochastic production planning for
the highest business benefit, Robotics and Computer Integrated
Manufacturing, 17, pp. 405419, 2001. [10] Jonathan. J .Goloven,Hierarchical Integration of Planning and
Control, PH.D thesis, Cornell University, Augest 1975.
[11] Linet Ozdamar, Sevket Ilker Birbil, A hierarchical planning system for energy intensive production environments,
International Journal of Production Economics, 58, pp. 115-129,
1999. [12] Maria Antonia Carravilla, Jorge Pinho de Sousa,Hierarchical
production planning Make-To-Order company: case study,
European Journal of Operational Research, 86, pp. 43-56, 1995. [13] Mehraz. A, Minis. I, Proth. J. M, Hierarchical production planning
for complex manufacturing systems, Advances in Engineering Software, 26, pp. 209-218, 1996.
[14] Saad. G. H, Hierarchical production planning system: extension and modifications, the Journal of the operational Research Society, 41, pp. 609-624, 1990.
[15] Selim. M. Akturk, George R.Wilson,A hierarchical model for the cell loading problem of cellular manufacturing systems International Journal of production Research, vol. 36, No.7, pp.
2005-2023, 1998.
[16] Tarik. Aljuneidi, A Mathematical Approach to the Design of Cellular Manufacturing System Considering Dynamic Production
Planning and Worker Assignments, MS.C. Thesis, Concordia
University, Montreal, Quebec, Canada, 2013.
-
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:14 No:02 96
I J E N S IJENSIJENS April 2014 -IJMME-5757-029914
Appendix A
Family /Cell Aggregation Sub-
problem
Item /Resources Disaggregation
Sub-problem
Solution procedures of mathematical model using:
Lagrange multipliers
Decomposition procedures
Update the values of Lagrange method using gradient method
1. Solution procedure for cell loading consists of formulating a Lagrange relaxation
jtjLRl
Lt
J
j
T
t
jtjt
jLRl
Lt
J
j
T
t
jt
iTIk
n
Iw
itktw
n
i
T
t
it
T
t
T
t
N
i
itit
J
j
jtjtjtjt
RRROOR
IFIIFhRrOoAMinL
)(1 1
2
)(1 1
1
)(1 11 1 11
21 .),,(
Part (2): According to Item / Resource Disaggregation sub-
problem for each t (IRDt):
)(
2
)(
1
)(
)(jLRl
Ltjt
jLRl
Ltjt
iTIk
n
Iw
ktwit RRORIZMin
Part (1): According to Family / Cell Aggregation Sub-
problem (FCA):
T
t
T
t
N
i
ititit
J
j
jtjtjtjtjtjt IFhRrOoAYMin1 1 11
21 ).()()()(
Outline for the Proposed Mathematical Model
1 1
-
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:14 No:02 97
I J E N S IJENSIJENS April 2014 -IJMME-5757-029914
3. Update the values of Lagrange method using gradient Method A gradient optimization method is used to update the values of Lagrange
multipliers .for example the it values at step c are updated by the following
formula:
itc 1 it
c + c
)i(TIk
n
Iw
itktw IFI
Where:
Optimal solution c =
2)(
21 )),,((
iTIk
n
Iw
itktw
c
IFI
LZ
Cell loading decision
(Detailed reports)
citjtjtjtjtijijtjijtijtLtLtktwkjtwjtjtitijt hRrOoBPQBSCQRRORIZROIFX ,,,,,,,,,,,,,,,,,,
To shop floor execution
End
Solving last equation to find the values of it , 1
jt , 2
jt
Solving last equation to find the values of it , 1
jt , 2
jt
Obtain the value of it
, 1
jt , 2
jt Obtain the value of it
, 1
jt , 2
jt
1 1
For Variable :
2
it
1
it
IT
0t0i
it
iZ
iY
0
it
it
For Variable 2
2212
00
22
2
2
2 0
iZ
iY
jtjt
jt
jt
jt
JT
tj
For Variable 1
2111
JT
0t0j
11
1
iZ
iY
0
jtjt
1
jt
jt
1
jt
-
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:14 No:02 98
I J E N S IJENSIJENS April 2014 -IJMME-5757-029914
Fig. 1. Components of levels of aggregation for products
Fi. 2. Schematic flow chart of algorithm used to the CMS
390
400
410
420
430
440
450
460
Z ,
(10
6L
.E/
Mon
th)
Eletrical Water Heater (50L)
Intial
Optimal
Fig. 3. Comparison between Initial and Optimum Total Product Cost for EWH1.
Equipment type
and amount
Production
allocation
among plants
Plant capacity
planning
Performance characteristics,
operating results
Constraints
Item production
schedules
Plant sizing and
location
Start
Data Collection
Mathematical Model
Results
End
See Appendix. A
GT families and items
GT cells and resources
Demand information on each item
Process plans (processing time routs, set of feasible cells)
Cost figures
See Appendix. A
Cell Loading (production
Planning) problem
-
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:14 No:02 99
I J E N S IJENSIJENS April 2014 -IJMME-5757-029914
Fig. 4. Comparison between Initial and Optimal Variables Values for EWH1.
1.27
1.28
1.29
1.3
1.31
1.32
1.33
1.34
1.35
1.36
Z ,
(10
9L
.E/
Mon
th)
Electrical Water Heater (80L)
Intial
Optimal
Fig. 5. Comparison between Initial and Optimum Total Product Cost for EWH2.
-
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:14 No:02 100
I J E N S IJENSIJENS April 2014 -IJMME-5757-029914
Fig. 6. Comparison between Initial and Optimal Variables Values for EWH2
Fig. 7. Effective Variable Parameters for EWH1on the Variance.
Fig. 8. Effective Variable Parameters for EWH1on the Variance (%)
-
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:14 No:02 101
I J E N S IJENSIJENS April 2014 -IJMME-5757-029914
Fig. 9. Effective Variable Parameters for EWH2on the Variance.
Fig. 10. Effective Variable Parameters for EWH2 on the Variance (%).
Fig. 11. The Total Product Costs Saved for (EWH1, EWH2) Against the Variance (%).
-
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:14 No:02 102
I J E N S IJENSIJENS April 2014 -IJMME-5757-029914
Fig. 12. The Total Product Costs Saved for (EWH1, EWH2).
Table I Different parameters used through data collection (Fixed parameters).
Parameters Set of Values
Total number of items , k 65
Total number of resources ,L 5
No of periods, T 1
No of sub periods, w 4
Cost of production, ijtC UN~[500,750]
Cost of regular time, jtr UN~[80,100]
Inventory Holding Cost, ith UN~[10,12]
Setup cost for the families, ijBP UN~[50,65]
Processing time, klPR UN~[0.07,0.075]
No of operation per part 130
Setup time, ijbp UN~[0.361,0.331]
Total time required to produced one unit of family i at cell j, ija UN~[3.37,4.5]
Table II
Comparison between Initial and Optimal Variables Values for EWH1 Product.
Variables IV OP V V %
Cijt ( L.E/Unit) 500 510 -10 -2
Xijt (Unit/ Month) 1570 1500 70 4.46
BPij ( L.E/ Month) 233 320 -87 -37.34
Qijt (Unit/ Month) 2000 1571 429 21.45
Rjt ( h/ Month) 21164.73 20251.47 913.26 4.32
hit ( L.E) 40 45 -5 -12.5
IFit (Unit/ Month) 1295 1275 20 1.54
Dit (Unit/ Month) 1569 1499 70 4.46
Aij (h/unit) 13.48 13.5 -0.02 -0.15
BOij (h/unit) 1.44 1.54 -0.1 -6.94
Dktw (Unit/ Month) 68595 68589 6 0.01
Zkitw (Unit/ Month) 68596 68590 6 0.01
Iktw (Unit/ Month) 323.75 318.75 5 1.54
RRlt (h/ Month) 21165.02 22634.78 -1469.76 -6.94
PRkl (h/ Month) 0.3085 0.33 -0.0215 -6.97
TotalProductCostZ(L.E/ Month) 4.49E+08 4.11E+08 37850400 8.43
V = Initial value - Optimum value
V % = ( V/ Initial value) *100
IV : Initial value
OP : Optimum value
-
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:14 No:02 103
I J E N S IJENSIJENS April 2014 -IJMME-5757-029914
Table III
Comparison between Initial and Optimal Variables Values for EWH2 Product.
Variables IV OP V V %
Cijt ( L.E/Unit) 750 780 -30 -4.00
Xijt (Unit/ Month) 2100 2000 100 4.76
BPij ( L.E/ Month) 210 260 -50 -23.81
Qijt (Unit/ Month) 2055 2047 8 0.39
R jt (h/ Month) 36751.25 36001.29 749.96 2.04
Hit ( L.E) 47 48 -1 -2.13
IFit (Unit/ Month) 1495 1480 15 1.00
Dit (Unit/ Month) 2099 1999 100 4.76
Aij (h/unit) 17 18 -1 -5.88
BOij (h/unit) 1.22 1.32 -0.1 -8.20
Dktw (Unit/ Month) 74425 74416 9 0.01
Zkitw (Unit/ Month) 104778.7 102710.3 2068.4 1.97
Iktw (Unit/ Month) 373.75 370 3.75 1.00
RRlt (h/ Month) 36751.5 36001.54 749.96 2.04
PRkl (h/ Month) 0.35 0.35 0 0.00
TotalProductCostZ (L.E/Month) 1.35E+09 1.30E+09 54576000 4.04
V = Initial value - Optimum value
V % = ( V/ Initial value) *100
IV : Initial value
OP : Optimum value