Professional Documents
Culture Documents
01-2018-5-13 - P Linear Programming For Aggregate Production Planning in A Textile Company P PDF
01-2018-5-13 - P Linear Programming For Aggregate Production Planning in A Textile Company P PDF
orders that must be met [7], taking into [16], among others. These mathematical
account constraints related to production models provide a solution to the specific
capacity, available materials, equipment, requirements of each problem, offering
labour and other necessary resources for a wide range of possibilities in modeling,
the production system. Once the produc- which allows to pose a great diversi-
tion strategy is obtained, the master pro- ty and quantity of linear and non-linear
duction schedule (MPS) is built, decom- constraints, integer and continuous varia-
posing temporally and spatially the goals bles, and one or multiple objective func-
of the aggregate planning and forecasting tions, as in the case of goal programming
into the manufacturing requirements of [6, 17]. In order to find a solution for
specific products, thus being a guide for mathematical programming models of
the acquisition of resources and materials aggregate planning, several alternatives
[8, 9]. To solve the aggregate production can be found, ranging from spreadsheets
planning problems, level, chase, and hy- and solvers [11, 18, 19] to specialised
Introduction brid strategies are often used. The level mathematical programming software
Nowadays the textile industry faces strategy maintains a steady quantity of such as CPLEX [20] and GAMS [21].
a competitive manufacturing environ- labour and production rate, and for cases
ment that has led manufacturers to imple- in which variations in demand occur, the In literature related to the modelling of
ment planning strategies that guarantee strategy increases or decreases the inven- aggregate production problems, some
operational effectiveness, motivating the tory levels, creates a portfolio of prod- studies propose objective functions relat-
redesign of organisational and productive ucts with complementary demands, and ed to total costs, labour fluctuations, late
structures, to reach satisfactory levels of influences the increase in demand and orders, benefits and utilities, equipment
competitiveness [1]. These conditions pending orders. Chase strategies equal- use, service level, and quality, among
force global textile and clothing sup- ise the production rate with the demand others [6, 13-15, 18, 21]. Likewise these
ply chains to optimise their production rate in each period through hiring and fir- studies take into account parameters,
planning in order to face the challenges ing labour, using overtime and part-time variables and constraints related to the
demanded by market dynamics [2]. For workers and subcontracting production, size of labour, hiring and firing, labour
the long or medium-term, aggregate pro- among others. Hybrid strategies com- learning rates, working hours and legal
duction planning is considered as a tool bine the options proposed by the level restrictions, production capacity, subcon-
that takes into account the constraints of and chase strategies in order to provide tracting of production, multiple produc-
existing capacity [3], providing guidance an optimal production plan that satisfies tion factories, storage capacity, inventory
to the manufacturer regarding efficient demand and other business policies at the levels, machine maintenance, setup and
production and supply strategies [4]. In lowest cost [9-11]. assembly times, customer demand, and
this sense, aggregate production plan- cash management.
ning determines the manufacturing ca- The complexity associated with the for-
pacity of products to satisfy the demand mulation of mixed strategies suggests However, to the best of our knowledge,
in a time horizon, for which it evaluates the use of optimisation models based in the literature, there are currently no
design configurations and manufacturing on mathematical programming, which aggregate planning models for compa-
scenarios that maximise benefits and/or according to the specific conditions of nies in the textile industry that address-
minimise costs through optimal produc- the problem to be solved, can generate es variables related to fabric waste and
tion quantities, capacity, subcontracting, transport models of linear programming shrinkage by line and by process, train-
inventory and shortages [5, 6]. [12], mixed-integer linear programming ing days needed for new employees in
(MILP) [13, 14], multi-criteria mixed-in- each process, and the efficiency of new
For its operation, aggregate planning is teger linear programming (MCMILP) employees in each process. Therefore
based on demand forecasts and customer [15], and non-linear programming (NLP) this article aims to design and implement
Campo EA, Cano JA, Gómez-Montoya RA. Linear Programming for Aggregate Production Planning in a Textile Company. 13
FIBRES & TEXTILES in Eastern Europe 2018; 26, 5(131): 13-19. DOI: 10.5604/01.3001.0012.2525
Line 1 Line 2 Line 3
an aggregate production planning model 3 product lines, which are permanently quired to take into account the efficiency
to provide optimal production strategies in use due to customer demand, called of hired employees because the initial
in the medium term that suit the needs Line 1 – Simple Pocket (one-colour fab- part of the work contract is dedicated to
of companies in the textile industry. For rics weighing less than 130 gr/m2), Line receiving training, and they also have
this, we propose an algebraic formulation 2 – Double Pocket (one-colour fabrics a lower performance in their work due
for the aggregate planning model, which weighing over or equal to 130 gr/m2), to their little experience in the tasks as-
is then validated with an experimental and Line 3 – Staffing (pre-dyed fabrics signed. The waste and contractions of the
development to analyse output variables weighing over or equal to 120 gr/m2). fabric must also be included in the pro-
such as production rates, inventory lev- Figure 1 shows the transformation pro- duction plan.
els, labour level, hiring, and firing. Final- cesses for each product line.
ly we present the main conclusions and Due to the policies established by the
contributions of this study. The company needs to determine its an- company, the use of overtime or produc-
nual aggregate production plan to mini- tion subcontracting is not considered,
mise the total costs represented by labor because these decisions can significantly
Formulation of the aggregate
costs and inventory management,, and affect the quality of the product and put
production planning model
ensure compliance of constraints related industrial secrets at risk. In the same way,
The company selected for this study to each product line demand, production since it is a medium-term production plan
manufactures pre-dyed and one-colour capacity expressed in meters (m), work based on 12 month’s horizon planning, op-
fabrics suitable for dyeing which have in process storage, and operational effi- erational details related to the setup time,
high turnover rates. The company has ciency. In addition, the company is re- assembly and maintenance of machines
are not considered directly; however,
Table 1. Indices, parameters, coefficients and variables for the LIPROTEX model. these factors are included proportionally
to the operating times of each product line
Indices Definition
based on time and motion studies in each
I Number of processes (i = 1, … , I)
process mentioned in Figure 1. There-
T Number of months (t = 1, … , T)
fore we propose the use of mathematical
L Number of product lines (l = 1, … , L)
programming to create a linear program-
Parameters and coefficients Definition
ming model, which will be referred to as
CEi Monthly cost per employee in process i, $
the LIPROTEX model. The use of a line-
CAi,t Cost of inventory management in process i in month t, $/m
Ai Storage capacity in process i
ar programming model is justified due to
PMi Maximum production capacity for process i, m
the complexity of the model related to the
DTl,i Waste and shrinkage of product line l in process i
number of parameters and variables taken
Dl,t Expected fabric demand of line l in month t, m
into account in a medium-term planning
DEi Required training days for new employee in process i
horizon, and to the constraints of resourc-
Ei New employee efficiency in process i
es, capacities, inventories, demand, labour
MHl,i Production of meters per hour of product line l in process i
and product lines, which guarantee the
HDt Real available hours per employee in month t feasibility and optimisation of the aggre-
CAC Administrative cost of hiring employee, $ gate production plan.
CAD Administrative cost of firing employee, $
Output variables Definition Variables for the LIPROTEX model
Hl,i,t Production hours dedicated to line l in process i in month t This study identifies variables that have
Sl,i,t Final inventory of product line l in process i in month t, m not been taken into account in previous
Xi,t Number of employees in process i in month t
models of aggregate programming which
XDi,t Number of employees fired in process i in month t
influence the total cost of a production
XNi,t Number of employees hired in process i in month t
plan and generate a great impact on de-
Decision variables Definition
cision-making processes in the textile
Pl,i,t Fabric production of product line l in process i in month t, m
industry. Among these variables, fabric
This study identifies variables that have not been taken into account in previous models of aggregate programming
which influence the total cost of a production plan and generate a great impact on decision-making processes in the
textile industry. Among these variables, fabric waste by product line and by process is highlighted due to the nature
of fabric processing, which commonly generates contractions and losses; the training days needed for new
waste by product
employees in line
eachand by process
process Table 2.
is alsoisincluded, sinceMonthly
a new demand
employeefor one mustyearbe fortrained
the product
before lines (thousands
starting of and
a job, meters).
a salary
highlighted
must be duepaidtoequally
the nature whileofthisfabric
training is being provided. Proportionally the efficiency of new employees in each
Month, t Jan Feb Mar Apr May Jun Jul Ago Sep Oct Nov Dec
processing,
process which
is taken commonly
into account, generates
since this varies according to the employees' expertise and adaptation to the job.70
Line 1 60 70 70 85 80 60 70 90 110 80 75
contractions and losses;
Table 1 presents thethe training
indices, days
parameters, Demand
coefficients, and variables of the LIPROTEX model.
forecast Line 2 64 71 139 115 133 80 123 99 163 135 104 84
needed for new employees in each process (Dl,t) Line 3 100 105 166 123 177 124 172 194 206 219 140 149
is also
Tableincluded, sinceparameters,
1. Indices, a new employee coefficients and variables for the LIPROTEX model
must be trained beforeIndices starting a job, and Definition
Table 3. Production capacity and production per process.
a salary must be paid Iequally while this Number of processes (i = 1, … , I)
training is being provided. Proportionally
T Number of monthsProduction (t = 1, … , T)Production, m/hour (MHl,i) Shrinkage percentage (DTl,i)
the efficiency of new employees in each Process (i) capacity, Product Product Product Product Product Product
L Number of product (l =i) 1, …
lines (PM
m/month , L) line 2
process is taken into account, since this line 1 line 3 line 1 line 2 line 3
Parameters
varies according and
to the Coefficients
employees’ exper- Definition
Warping 425.000 120 120 115 0% 0% 0%
tise and adaptation toCE theijob. Table 1 pre- Monthly Warpingcost checkper employee
420.000 in process 390 i ($)390 350 0% 0% 0%
CA i,t coefficients, Cost
sents the indices, parameters, of inventory management
Gumming 570.000 in219 process 219 i in month 240 t ($/m)3% 3% 2%
Ai
and variables of the LIPROTEX model. Storage
Weaving 490.000 i
capacity in process 157 120 125 10% 12% 7%
PM i Maximum
Weaving check production capacity for
410.000 318 process318 i (m) 280 2% 2% 3%
Objective function DT andl,iconstraints Waste and shrinkage of product line l in process i
The objective function D l,t
and constraints of Expected straints for fabric
the demand of lineby
final inventory l inperiod,
month t (m)tion hours, final inventory, hiring, firing,
the mathematical model DE i for the produc- Required process and product line. The storage ca- inand
training days for new employee process i
production level.
tion programming ofEcompanies i in the New employee
pacity constraints efficiency
in in
Equation process
(4) i
ensure
textile industry (LIPROTEX)
MH l,i are pre- Production
that the final inventory
of meters perofhour
a process
of product does lineOnce l in process i
the aggregate production planning
sented in Equations HD (1)-(9).t not exceed the maximum storage
Real available hours per employee in month t capaci- problem has been modelled, it is identified
CAC ty of the process. Equation
Administrative cost of hiring employee ($) (5) represents as a linear programming model that does
The objective function represented in labour balance equations for hiring and not require entire variables, due to the
CAD Administrative cost of firing employee ($)
Equation (1) minimises the total costs of firing. The production of meters of fab- nature of the production process, which
Output variables Definition
the production plan in terms of the labour ric in a period, process, and product line produces fabric continuously and not in
and inventory management costs. Labor Production
H l,i,t hours dedicated to line l in process i in month t
is represented in Equation (6). Equa- discrete units. Likewise, for labour-relat-
costs are represented by S l,i,tthe monthly cost Final inventory
tion (7) presents of the
product
number lineofl in process i ed
monthly month t (m)
in variables, the company’s management
of employees per process X i,t Number of employees in process i in month t
in a month, hours available in the production plant, decides that it does not require entire var-
the training time ofXD employees,
i,t and the Numberwhich resultsof employees fired in process
from multiplying the avail- i in month
iablest of workers, and according to the
administrative cost XN of hiring
i,t and firing Number able hours of employees
per employee hiredinina month
processby i in month
results,t it decides to hire part-time staff,
employees. Decision
The inventory variables management Definitionthe number of employees available per redistribute operators with idle capacity,
costs are proportional P l,i,tto the monthly Fabric process and month.
production The constraints
of product of take
line l in process i in advantage
month t (m) of high-skilled employ-
inventory level in each process. Equa- the maximum production capacity are ees, and assign employees with idle time
tion (2) guarantees
Objective fulfillment
function of the fin- found in Equation (8). Equation (9) rep- to other processes, among others. In ad-
and constraints
ished product demand for each product resents non-negativity restrictions for the dition, the mathematical programming
Equation (3)
line.The represents
objective function andthe con- variables
constraints of the model model
of the mathematical related for
to produc-
the productionmodelprogramming
proposed handles deterministic
of companies in
the textile industry (LIPROTEX) are presented in Eq. (1-9).
𝑇𝑇 𝐼𝐼 𝑇𝑇 𝐼𝐼 𝐿𝐿
𝐷𝐷𝐷𝐷𝑖𝑖
𝑀𝑀𝑀𝑀𝑀𝑀 𝑍𝑍 = 𝐶𝐶𝐶𝐶𝑖𝑖 × 𝑋𝑋𝑖𝑖,𝑡𝑡 + 𝑋𝑋𝑋𝑋𝑖𝑖,𝑡𝑡 × + 𝐶𝐶𝐶𝐶𝐶𝐶 × 𝑋𝑋𝑋𝑋𝑖𝑖,𝑡𝑡 + 𝐶𝐶𝐶𝐶𝐶𝐶 × 𝑋𝑋𝑋𝑋𝑖𝑖,𝑡𝑡 + 𝐶𝐶𝐶𝐶𝑖𝑖,𝑡𝑡 × 𝑆𝑆𝑙𝑙,𝑖𝑖,𝑡𝑡 (1)
(1)
30
𝑡𝑡=1 𝑖𝑖=1 𝑡𝑡=1 𝑖𝑖=1 𝑙𝑙=1
𝐿𝐿
𝑙𝑙=1 𝑆𝑆𝑙𝑙,𝑖𝑖,𝑡𝑡 ≤ 𝐴𝐴𝑖𝑖 for i = 1, … , I t = 1, … , T (4)
(4)
𝐿𝐿
𝑙𝑙=1 𝐻𝐻𝑙𝑙,𝑖𝑖,𝑡𝑡 = 𝐻𝐻𝐻𝐻𝑡𝑡 × 𝑋𝑋𝑖𝑖,𝑡𝑡−1 + 𝑋𝑋𝑋𝑋𝑖𝑖,𝑡𝑡 × 𝐸𝐸𝑖𝑖 − 𝑋𝑋𝑋𝑋𝑖𝑖,𝑡𝑡 for i = 1, … , I t = 1, … , T (7)
(7)
𝐿𝐿
𝑙𝑙=1 𝑃𝑃𝑙𝑙,𝑖𝑖,𝑡𝑡 ≤ 𝑃𝑃𝑃𝑃𝑖𝑖 for i = 1, … , I t = 1, … , T (8)
(8)
𝐻𝐻𝑙𝑙,𝑖𝑖,𝑡𝑡 ≥ 0, 𝑆𝑆𝑙𝑙,𝑖𝑖,𝑡𝑡 ≥ 0, 𝑋𝑋𝑙𝑙,𝑖𝑖,𝑡𝑡 ≥ 0, 𝑋𝑋𝑋𝑋𝑙𝑙,𝑖𝑖,𝑡𝑡 ≥ 0, 𝑋𝑋𝑋𝑋𝑙𝑙,𝑖𝑖,𝑡𝑡 ≥ 0, 𝑃𝑃𝑙𝑙,𝑖𝑖,𝑡𝑡 ≥ 0 for l = 1,…, L i = 1,…, I-1 t = 1,…, T (9)
(9)
The objective function represented in Equation 1 minimises the total costs of the production plan in terms of the
Equations (1), (2), (3), (4), (5), (6), (7), (8) and (9).
labour and inventory management costs. Labor costs are represented by the monthly cost of employees per process
in& aTEXTILES
FIBRES month, the Europe
in Eastern training
2018, time
Vol. 26, of employees, and the administrative cost of hiring and firing employees. The inventory 15
5(131)
management costs are proportional to the monthly inventory level in each process. Equation 2 guarantees fulfillment
of the finished product demand for each product line. Equation 3 represents the constraints for the final inventory by
period, process and product line. The storage capacity constraints in Equation 4 ensure that the final inventory of a
Table 4. Estimated waste by process, costs and inventory levels. variables, and in the case of integrating
uncertainty with some parameters and/or
Monthly inventory Initial inventory (m) (S l.i.0)
Process (i)
Storage capacity.
cost per. variables, the model must be adjusted by
m/month (Ai) Product Product Product
$USD/m (CAi.t) line 1 line 2 line 3 implementing techniques such as fuzzy
Warping 170.000 $ 0.013 20.000 10.000 15.000 logic [22, 23].
Warping check 120.000 $ 0.013 10.000 5.000 7.500
Gumming 30.000 $ 0.021 30.000 15.000 22.500
Experimental development
Weaving 350.000 $ 0.025 40.000 20.000 30.000
Weaving check 150.000 $ 0.025 20.000 10.000 15.000 According to the demand forecasts and
negotiations carried out with the com-
Table 5. Labour costs, efficiencies and training requirements. pany’s customers, Table 2 presents the
Cost of employee Initial Training requirements
monthly demands for each of the product
New employee
Process (i) per process, employees
efficiency (Ei)
for new employees, lines for a planning horizon equivalent to
$USD/month (CEi) (Xi.0) days (DEi)
one year.
Warping $ 696.7 13 70% 5
Warping check $ 443.3 3 80% 12
Table 3 presents the production capacity
Gumming $ 696.7 4 70% 5
by process per month, a breakdown of
Weaving $ 823.3 11 60% 5
the production capacity per hour for each
Weaving check $ 443.3 3 80% 12
of the product lines, and estimated fabric
waste due to contractions between pro-
Table 6. Available hours per employee per month.
cesses in each product line.
Month Jan Feb Mar Apr May Jun Jul Ago Sep Oct Nov Dec
Working hours
208 184 200 208 200 200 216 200 208 208 192 208
Regarding inventories, Table 4 shows the
per employee (HDt)
storage capacity per process per month,
Table 7. Fabric production per product line, process and month Pl,i,t (m x 1000).
Month, m
Product line, l Process, i
1 2 3 4 5 6 7 8 9 10 11 12
1 – 24.8 125.4 52.2 91.6 148.9 0.0 182.2 79.1 59.5 85.9 80.2
2 – 44.8 125.4 52.2 91.6 148.9 0.0 182.2 79.1 59.5 85.9 80.2
1 3 – 54.8 86.3 91.2 91.6 72.3 76.6 103.1 126.0 91.6 85.9 80.2
4 22.8 59.4 77.8 94.4 88.9 70.1 74.3 100.0 122.2 88.9 83.3 77.8
5 40.0 70.0 70.0 85.0 80.0 63.1 66.9 90.0 110.0 80.0 75.0 70.0
1 130.6 53.2 74.9 158.4 195.8 147.8 70.1 186.1 111.4 123.8 121.8 98.4
2 140.6 53.2 74.9 158.4 195.8 147.8 70.1 186.1 111.4 123.8 121.8 98.4
2 3 145.6 53.2 74.9 158.4 132.1 186.6 95.0 186.1 111.4 123.8 121.8 98.4
4 155.8 51.6 72.7 153.7 128.1 181.0 92.2 180.5 108.0 120.1 118.2 95.5
5 134.0 66.0 63.9 135.3 112.7 159.3 81.1 158.8 89.7 111.0 104.0 84.0
1 71.8 99.5 143.7 212.5 129.4 123.3 354.9 46.7 229.7 236.6 153.6 163.5
2 86.2 100.0 143.7 209.4 132.5 123.3 349.9 51.7 229.6 236.8 153.6 163.5
3 3 40.2 115.2 182.1 154.5 174.6 136.1 265.3 136.3 226.0 240.3 153.6 163.5
4 61.4 112.9 178.5 151.4 171.2 133.3 258.0 135.5 221.5 235.5 150.5 160.2
5 85.0 105.0 166.0 140.8 159.2 124.0 238.4 127.6 206.0 219.0 140.0 149.0
Table 8. Final inventory per product line, process, and month S l,i,t (m x 1000).
Month, m
Product line, l Process, i
1 2 3 4 5 6 7 8 9 10 11 12
1 20.0 – – – – – – – – – – –
2 10.0 – 39.0 – – 76.6 – 79.1 32.2 – – –
1 3 6.5 – 6.2 – – – – – – – – –
4 18.4 – – – – – – – – – – –
5 – – – – – 3.1 – – – – – –
1 – – – – – – – – – – – –
2 – – – – 63.8 24.9 – – – – – –
2 3 23.5 – – – – – – – 6.1 – – –
4 – – – – – – – – – – – –
5 80.0 75.1 – 20.3 – 79.3 37.4 97.3 24.0 – – –
1 0.5 – – 3.1 – – 5.0 – 0.1 – – –
2 53.6 38.4 – 54.9 12.7 – 84.6 – 3.5 – – –
3 3 – – – – – – 2.0 – – – – –
4 – – – – – – 1.7 – – – – –
5 – – – 17.8 – – 66.4 – – – – –
Table 9. Level of labour, hiring and firing by product line and month X i,t, XN i,t, XD i,t
Month, m
Process, i
1 2 3 4 5 6 7 8 9 10 11 12
1 8 8 17 17 18 18 17 18 17 17 16 14
2 3 3 5 6 6 6 5 5 6 6 5 4
Employees
3 4 6 8 9 9 9 9 10 10 10 8 7
(Xi.t)
4 9 9 15 15 15 15 15 17 17 17 14 12
5 4 4 5 6 6 6 6 6 7 7 6 5
1 – – 9 – – – – 1 – – – –
2 – – 2 1 – – – – – – – –
Hiring
3 – 2 2 1 – – – 1 – – – –
(XNi.t)
4 – – 6 – – – – 1 – – – –
5 1 – 1 1 – – – – – – – –
1 5 – – – – – 1 – – – 1 2
2 – – – – – – – – – – – 1
Firing
3 – – – – – – – – – – 1 1
(XDi.t)
4 2 – – – – – – – – – 2 2
5 – – – – – – – – – – 1 1