Вы находитесь на странице: 1из 10

Lead Time Prediction Using Simulation in Leather

Shoe Manufacturing
Hermela Solomon, Kassu Jilcha, and Eshetie Berhan
Addis Ababa University, Addis Ababa Institute of Technology, Addis Ababa, Ethiopia
{hermu.y,jkassu}@gmail.com, eshetie_ethio@yahoo.com

Abstract. The purpose of this study is to acquire lead time prediction on the basis of actual data to optimize it. Most Manufacturing industries experiences
problems of lead time prediction. The determination of planning values for
manufacturing lead times has been viewed both as a problem of estimating independent, uncontrollable variables, and as a control problem, where emphasis
is placed on managing the average lead times to match predetermined norms.
Once a new order with specific and known processing requirements enters to
manufacturing system, an exact lead time estimate is assigned to it that is based
on the manufacturing current status. Thus, the customers of the case company
was being influenced by unpredictable lead time. As a result of this, the customers were unable to get the product of shoe on time due to indeterminate and
fluctuating lead time. To come +up with acceptable lead time, Monte Carlo and
Arena simulation tools were used to analyze optimal lead time. The finding of
the tools simulation resulted in fixed lead time and order cycle numbers for the
products order over specified periods (what is the final result reported by simulation?)
Keywords: Production, Lead time, Leather shoe, Distribution, Monte Carlo.

Introduction

Many companies fail to meet customer requirements in their product on time delivery
and companies fail to predict their lead time as many studies finding results proved.
Therefore, this research paper focuses on lead time prediction for the leather shoe
company using simulation techniques optimize lead time so that customers get satisfaction. To support these problems, it is important to summarize the previous results
and conclusion.
More and more companies on the global market are today capable of manufacturing individual or small series orders at comparable prices and quality. The main difference between these companies is the expected production order development time
and the observance of delivery deadlines [21].
Since planned lead times represent the amount of time allowed for orders to flow
through the production facility, they have a very significant impact on the system
performance: tight lead times may lead to past orders and expediting, while loose
lead times aggravate the overall material requirement planning (MRP) performance
significantly [7].
Springer International Publishing Switzerland 2015
A. Abraham, P. Krmer, & V. Snel (eds.), Afro-European Conf. for Ind. Advancement,
Advances in Intelligent Systems and Computing 334, DOI: 10.1007/978-3-319-13572-4_23

283

284

H. Solomon, K. Jilcha, and E. Berhan

Lead Time is also regarded an essential element in negotiating with the customer
[9]. Lead Time estimation is critical as it exclusively affects customer relations and
shop floor management practices. Due date quoting, which means commitment to
meeting customer orders on time, is a direct outcome of lead time estimation. Short
lead times improve a manufacturer's image and future sales potential. However, not
only short but also accurate and precise lead time estimates are desirable.
Leather product manufacturing industries are one of the major industries in Ethiopia. Competitive advantage of the selected case is providing a large variety of products with reasonable price in a short time. In order to be more responsiveness, meet
customer needs, reduces response time considering the companys ability.
This study was conducted in one of the oldest leather shoe company which is
Anbessa Leather Shoe manufacturer company launched in 1927 E.C and producing
Men's , women's and Children's shoe. The company has a capacity to produce 4000
pairs per day and operates 6 days a week and 70% of their products are sold in local
market and the rest are exported to different countries.
In this study the problem of indeterminate and fluctuate production lead time was
found a big problem in the company. To solve this problem of the company, this
study was conducted in selecting the best method of simulation which are Monte Carlo and ARENA Simulation. These tools helped to optimize lead time of Anbessa
leather Shoe Company to meet customer orders to make delivery on time and to have
fixed production lead time period.
From the late 1960s, the lead time management problem has been consistently addressed in the literature. In an early works by [20] the determination of planning values for manufacturing lead times has been viewed both as a problem of estimating
independent, uncontrollable variables, and controllable problems. Monte Carlo Simulation is one of techniques using random variables to estimate possible activity duration in a probability distribution [1].
There are published studies aimed at solving lead time by [18, 24] were modeled
three lead time estimation models. Each model requires the estimation of two or more
parameters. The parameters in each model were estimated via linear regression from
steady state simulation data. These were Jobs in queue (JIQ), Jobs in bottleneck
queue (JIBQ) and Combined model (COMB) and the experience obtained during
many tests of practical implementation of these procedures, led us to the conclusion
that it would be possible to predict lead times of planned orders on the basis of actual
operational and assembly order lead times achieved in the past. These predictions (on
the basis of enterprise requirement planning (ERP)-system data or on the basis of
manually acquired past data) are accurate enough for individual production. Using
these data, it is possible to predict lead times even for fairly complex products with
several machining operations and individual order features. Based on [21]. There are
six procedure for predicting lead times for future production orders.
Lead time estimation are closely related to job due dates. In practice, job due dates
are determined either internally by production personnel or externally as a result of
negotiation between marketing personnel and the customer. Internally set due dates
are often used in make to stock operations, assemble-to-order operations may set due
dates internally for the production of components while setting the due dates for finished products externally, while externally set due dates are more prevalent in make
to order operations. Whether due dates are set externally or internally, the degree to

Lead Time Prediction Using Simulation in Leather Shoe Manufacturing

285

which they are met is largely dependent on the ability to accurately predict the time
necessary to complete product processing requirements, i.e. lead time[15].
Job and shop characteristics are the two factors affecting flow times. Earlier studies
have taken job related parameters (such as total work content and number of operations) into account in rules like CON, TWK, SLK, and NOP [10]. Shop related measures such as total shop load at the time of order arrival or shop congestion on an
orders prospective route have been considered in later rules JIQ and JIS are two wellknown examples [5]. Using job and shop status data in combination has proven to be
more effective [25,4]. Experiments have shown that shop related measures specific to
an order are more effective than aggregate shop congestion indicators in estimating
flow times [14]. This work makes use of queuing theoretic findings and assumes that
utilization rate and process time are factors defining the flow time. The proposed
model is quadratic in both factors. Higher order terms also appear in another approach
[16]. Studies bases their OFS (operation flow time sampling) and COPS (congestion
and operation flow time sampling) rules on the fact that flow times are correlated
[23]. A sample of recently completed orders is used to infer ongoing average flow
time per operation to reflect near term shop congestion. Their results support the previous findings about superiority of using shop conditions. However; significant
differences between the rules are found depending on shop balance. Probability distribution based approaches explicitly consider the impact of variance either of the
workload (e.g. JIQ) or of the error in estimation [6] refers to an earlier work making
use of the normal distribution in quoting LT estimation subject to uncertainty. Threat
flow time as a normal distributed random variable and base LT estimation on a
service level requirement [8].
In general, the study focused on lead time prediction and optimized leather shoes
company products delivery time using simulation techniques.

Material and Methods

The study was conducted considering different materials and methods to achieve the
goal of this study. The literature reviews and software applications discussed in this
paper were utilized to come up with good optimized lead time prediction solution.
Before starting with the procedure for predicting production order, the right data
were collected from the case company. At the beginning of lead time prediction, it is
necessary to define the interval for data acquisition of past operational and assembly
orders and so that it was done. This interval can be a month, a quarter, a year or several years. As mentioned, a company wishing to predict lead times must have an
ERP/PPC system as a basis for all further steps, because this is the database of orders
processed in the past. Operational or assembly order codes, type and sequence of
operations in manufacturing and assembly orders, actual execution times of operational or assembly orders, date of completing a particular operational or assembly
order in the previous workplace, and date of completing a particular operational or
assembly order in the observed workplace.
The input data for this predicting method were the actual lead times of operational
and assembly orders processed in the past in the case companys workplaces.

286

H. Solomon, K. Jilcha, and E. Berhan

The principles of modeling methods such as simulation, mathematical optimization


and heuristic models have been described in detail by reference [17]. Some customer
service criteria which influenced by delivery time and configuration were: customer
experience, response time, return ability, order visibility, product variety and product
availability. Hence, simulation has appeared as a powerful method to investigate and
design method to predict lead time of the product for the case company.
Different authors finding showed that simulation arena is the best tool to solve
such a stochastic system of the companies. This model also customized with Monte
Carlo simulation in order to compromise random customer order because today's customer order is not the same with tomorrow order's by simulating those random customer order distribution move towards up with good result.
2.1

Data Analysis

Data was collected from the case company and identified to which tool it fits. Normally Chi Square test is used to identify fitted statistical distribution. Process of leather shoe product including loading and unloading time, transportation time from one
station to another station and demand probability distribution with in past two years
has been collected using primary and secondary data collection techniques. Chi
Square test was done in this study to find fitted distribution for data by using easy fit
software. One statistical test that addresses this issue is the chi-square goodness of fit
test. This test is commonly used to test association of variables in two-way tables,
where the assumed model of independence is evaluated against the observed data.

Discussion and Results

3.1

Production Process Flow

In Anbessa shoe factory, there are 3 main sections (Fig 1). Raw materials in the form
of leather are issued from store. They will be sent to cutting section and wait for cutting processes. After cutting processes, numbering, skiving the bundle, it will goes

Fig. 1. Product Process flow chart

Lead Time Prediction Using Simulation in Leather Shoe Manufacturing

287

to stitching section to assemble all components of the leather shoe. After stitching
complete luggage products are then going for finishing to lasting section. Finally,
inspected and packaging will be done. The product will be kept in finishing storage
before delivery to customer.
3.2

Conceptual Model

Simulation modeling is all about building the conceptual model of the case. In fact,
modeler should identify and gather all the details and formulations that are necessary
for the model in mind map then convert all of them into simulation software. Conceptual model includes inputs .Inputs to Arena models were order per arrival, time between arrivals, assembly time in each section, waiting jobs in queue in each section
and time for inspection until deliver to customers.
Customer
Demand

Arrivals

Update
Inventory

True
Inventory
sufficient
False

True
Stitching
section

Quality
check 1

Cutting
section

True
Lasting
section

Quality
check 2

Packing

Order
delivered

False
False

Fig. 2. Conceptual model using Arena Software

3.3

Model Assumption

This study represents continuous event modeling with below assumptions


The manufacturer working time is eight hours per day
There is no maintenance performed during the production period.
Machines are available all the time and no breakdowns
There is always sufficient raw materials in storage, so the production process
never starves.
All Workers are assumed to produce equal product with in equivalent period
of time.
Transportation time is included in processing time.
No power interruption
We Are Interested in
Simulating the system for 10,000 hours
estimating process utilizations, average job waiting times and average job
flow times (the lead time for a job from start to finish)

288

3.4

H. Solomon, K. Jilch
ha, and E. Berhan

Model Formulation

nitions and assumptions in the prior section, the schem


matic
Considering the given defin
model of the system presen
nted in equation (1) helps in calculating delivery time ussing
Monte Carlo tool as:

(1)
Where: tij = delivery tim
me from station i to station j, T= total delivery time of the
whole processes,
A set of uniformly distrib
buted random numbers is needed to generate the arrivalls at
the checkout counter. Random numbers have the following properties:
1. The set of random tim
me is uniformly distributed between 5 and15 days.
2. The set of random num
mbers of orders are uniformly distributed between 500 and
20,000.
3. Successive random nu
umbers are independent.

Reality

Monte Carlo Method

Input

System

Output

F 3. Monte Carlo process Analysis


Fig.

Anbesa Shoe Factory deemand or order arrival is classified into four intervalss to
make the study easy in oveer the two years as shown Table 1. For each order arrival
random numbers also generrated in line with the probability distribution of orders.

Lead Time Prediction Using Simulation in Leather Shoe Manufacturing

289

Table 1. Possibility of Order Arrival

Order Arrive

Probability

Random number assigned


Less than 0.1
Between 0.1 and 0.15
Between 0.15 and 0.40
Between 0.35 and 0.40

0.1
0.15
0.40
0.35

0 - 500
600 - 5000
5001 - 10000
10001 - 20000

The key to this simulation is to use a random number to key a lookup from the table range 500 up to 20,000 orders (named lookup). Random numbers greater than or
equal to 0 and less than 0.10 yielded a demand of 500 and random numbers greater
than or equal to 0.35 yielded a demand of 20,000. The study generated 200 random
numbers using Monte Carlo simulation tool and would got final optimal result.
Table 2. Final average results with Monte Carlo Tool

Order Arrive

Processing time
in all section(hr)

Jobs waiting in the


queue(hr)

500

26.2

9.01

5000

246.5

90.1

10000

491

180.2

20000
Sum

980

380.4

1743.7

659.71

435.925

164.9275

A Monte Carlo based simulation method is designed for running the model to estimate the quantity(order) arrival and the lead time of product in order to help managers to make decisions regarding product lead time and order arrival cycle to customers with alternative scenario that they can see results for many possible variants.
This approach offers not just one outcome, but a distribution of possible outcomes.

Fig. 4. Beta distribution of demand

Since order arrival distribution was between 500 and 20,000 the study test their
distribution using input analyzer and select minimum corresponding p-value which is
Beta distribution with the values of BETA (0.882, 0.831)( Fig 4.).

290

H. Solomon, K. Jilcha, and E. Berhan

Model Validation

Several techniques for validation of model have been offered [13]. In this study, validation of the model was done by comparing the actual data with the results that were
generated by simulation software. In addition, for validating of the model, estimating
lead time with in company when producing for one day was simulated and it was
compared with actual lead time. Table 3 shows the validation data of Monte Carlo and
Arena Simulation tools.
Table 3. Validation of Simulation Tools
Order Arrive
0 - 500
501 - 5000
5001- 10000
10000 -20000

Actual Lead Time of


the Company Min(hr)
48
352
704
1056

Lead Time using Monte


Carlo and Arena(hr)
35.21
336.6
671.2
980

The Monte Carlo and arena software simulation results shown is less than the actual lead time of the company (table 3). This implies that the simulation result is the
predicted lead time for the company.

Conclusion

Due to ever-fiercer market competition, companies must predict lead times and delivery times with ever greater accuracy to handle its customers and satisfy them. If they
give incorrect deadlines, they may not get a request from a particular company next
time, which can lead the company into crisis. This article proposes a technique for
predicting production order lead times on the basis of actual lead times of past operational or assembly orders. Using the proposed procedure, the company can predict the
lead time required for delivery of any new order to any customer and make variations
of delivery lead time calculations on the basis of an acceptable risk level by selecting
the confidence interval with respect to the size and complexity of an order, and taking
into account the companys policy towards its customers.
The study purpose alternatives simulation tools were suggested by the experts to
predict lead time of products by using MONTE CARLO and ARENA software. Finally results are suggested to the company for further improvement in its production lead
time network within different customers orders. For example, when the customer
orders is (501-5000) pieces, the predicted lead time has been reduced to 336.6 hrs
from the current companys lead time of 352hrs. On this basis of the tests, it was
found that the procedure for predicting lead times of production orders was well designed and provided very useful data for sales, as well as for production planning and
control. Signing a supply contract on the basis of reliable statistical data is completely
different from signing a contract on the basis of uncertain, experience-based guesswork.

Lead Time Prediction Using Simulation in Leather Shoe Manufacturing

291

For future study in this regard, it is recommended to improve lead time by taking
into account the sequence of operations required to complete an order, the influence
of the number of operations per order, and the influence of the processing time of
operations and machine breakdown.

References
1. Yahia, A.A.E.F.: Time Schedule Preparation By Predicting Production Rate Using Simulation
2. Arena Professional Reference Guide. Rockwell Software Inc. (2000)
3. Benk, J.: Modeling Kanban Systemic Production with Arena Simulator. GpGyrts
XLIX(4) (2009)
4. Bertrand, J.W.M.: The use of workload information to control job lateness in controlled
and uncontrolled release production systems. Journal of Operations Management 3(2),
7992 (1983)
5. Eilon, S., Chowdhury, I.G.: Due dates in job shop scheduling. International Journal of
Production Research 14(2), 223237 (1976)
6. Enns, S.T.: A dynamic forecasting model for job shop flow time prediction and tardiness
control. International Journal of Production Research 33(5), 12951312 (1995)
7. Ho, J., Chang, Y.: An integrated MRP and JIT framework. Computers and Industrial Engineering 41, 173185 (2001)
8. Hopp, W.J., Sturgis, M.L.R.: Quoting manufacturing due dates subject to a service level
constraint. IIE Transactions 32(9), 771784 (2000)
9. Moodie, D.: Demand management: The evaluation of price and due date negotiation strategies using simulation. Journal of Production Operations Management Society 8(2),
151162 (1999)
10. Nyhuis, P., Wiendahl, H.P.: Logistische Kennlinien, pp. 8194. Springer, Heidelberg
(1999)
11. Percentile (2006), http://en.wikipedia.org/wiki/Percentile
12. Harrell, R.: Simulation using ProModel. McGraw-Hill, New York (2003)
13. Sargent, R.: Validation and Verification of Simulation Models. In: Winter Simulation Conference (1999)
14. Ragatz, G., Mabert, V.: A framework for the study of due date management in job shops.
International Journal of Production Research 22(4), 685695 (1984)
15. Ruben, R.A., Mahmoodi, F.: Lead time prediction in unbalanced production systems (November 14, 2010)
16. Ruben, R.A., Mahmoodi, F.: Lead time prediction in unbalanced production systems. International Journal of Production Systems 38(7), 17111729 (2000)
17. Chopra, S.: Designing the distribution network in a supply chain, pp. 123140. Elsevier
Science Ltd. (2003)
18. Smith, M.L., Siedman, A.: Due date selection procedures for job-shop simulation. Computers and Industrial Engineering 7(3), 199207 (1983)
19. Smith, M.L., Siedman, A.: Due date selection procedures for job-shop simulation. Computers and Industrial Engineering 7(3), 199207 (1983)
20. Tatsiopoulos, I.P., Kingsman, B.G.: Lead time management. European Journal of Operational Research 14(4), 351358 (1983)
21. Berlec, T., Starbek, M.: Predicting Order Due Date, doi:10.1007/s13369-012-02791

292

H. Solomon, K. Jilcha, and E. Berhan

22. Veral, E.: Computer simulation of due-date setting in multi-machine job shops. Computers
and Industrial Engineering 41(1), 7794 (2001)
23. Vig, M.M., Dooley, J.K.: Dynamic rules for due-date assignment. International Journal of
Production Research 29(7), 13611377 (1991)
24. Vig, M.M., Dooley, K.J.: Mixing static and dynamic flow time estimates for due date assignment. Journal of Operations Management 11, 6779 (1993)
25. Weeks, J.K.: A simulation study of predictable due-dates. Management Science 25(4),
363373 (1979)
26. Wiendahl, H.P.: Load-oriented Manufacturing Control, pp. 37199. Springer, Berlin
(1995)

Вам также может понравиться