The version of Internet Explorer is outdated.

This website can not be viewed with your browser!

Upgrade your browser to the latest version (Internet Explorer 8) or install another browser, like Firefox or Google Chrome

The Complete Business Leader - Leadership training program - Click to View/Download Brochure
Wednesday, 04 June 2014 00:00

Usage of Statistical Tools in Improving Manufacturing Quality

Written by  Prof.J.Venkatappaiah (Retired), Indian Statistical Institute
Rate this item
(0 votes)

Statistical tools play an important role in improving Manufacturing Quality. The usage of these tools is not very popular in the Indian Manufacturing. The benefits and simplicity of these tools make a strong case for the Manufacturing companies to take a serious look at these tools.

A  Tele- communication plant was experiencing large scale failures, 45 % in 20 pair cable  and still higher % in  larger sizes of cables, in water penetration (WP) Test of PIJF (Polythene Insulated jelly Filled) cables.

 Foreign collaborators concluded, after extensive investigations and trials, the technology was not capable of meeting that rigorous WP Test, which in their view is not required.


However the Customer was insisting conformance to the WP Test. As a consequence the company was losing several Millions of Rupees on account of price cuts, rework and scrap.

World Bank, which financed the project, ordered a Statistical Investigation. Studies were conducted on 20 pair cable, which could easily be extended to larger size cables

Phase 1: Process Control

Large variability, 1.0 mm against a tolerance of 0.4 mm, in cable core diameter was noted, hampering smooth passage of cable core through Jelly filling machine- loose conductors hanging out, insulation getting ruptured and conductors breaking.

Wire drawing and insulation operations were identified as critical stages of manufacturing. Analysis of data showed that the processes were evidencing frequent disturbances (out of control); conductor diameter (CD) was varying by 33 microns and Insulated CD (ICD) was varying by 120 microns as a against tolerances of 10 microns, 20 microns respectively.

 scatter diagram between conductor diameter and resistance, thus, showed large variability

in cable resistance: 86 ± 7 ohms / km as against a specification of Max 91 ohms / km; expected rejection for high resistance was 44 %, and Resistance un –balance was 18 %. These got masked under high WPT failures.

a first step, scientific feed-back through line charts was instituted at both the stages. These helped identify process disturbances, link them with product dimensions and diagnose the problem: goal- post interpretation of specification, Not much effort in controlling process parameters, and not much concern for rework or scrap.

Improvement Actions

1.      Reformulation of specification : Target  ± Tolerance










    510 ± 5


Cond. Elongation


     Min 22

      25 ± 3





   1.05 ± 0.1

Cond. Resistance

Ohms / km

  Max  91

       88 ± 3



2.      Training of operators to aim at Target rather than anywhere within spec. Band

3.      Application of reduced tolerance for test sample, as variation in one km cable length was about twice that of one meter test sample.

4.      Locking line speed and fixing wire Tension at insulation process after studying Regression of these parameters on ICD


5.       Institution of process monitoring  Control charts, and discussions with shop personnel on preventing disturbances. These helped improve ’Process Capability’ of ICD to 16 microns, evaluated using Normal probability plotting method, against a tolerance of 20 microns, and past variability of 120 microns.

6.      Draw-down in CD (wire drawing to final cable stage) after controlling elongation was estimated as 8 microns, revealing scope for reduction in CD

These steps resulted in significant reduction in variability of mechanical and electrical dimensions and consequent large financial benefits.





CD -     Average

-          Variability

-          Elongation










ICD -    Variability




Cable Core dia.

Cond. Resistance

Res. unbalance


Rej.  %

Rej.  %







Phase II: Process Engineering

Failure of cables in WPT continued even though cable core became uniform. A critical examination of graphical representation of WP test results helped diagnose the problem:


                                    “ Plant technology is basically capable of meeting WP test; parameter levels possibly located at unstable values  and highly prone to disturbances”.

Deficient process engineering  and lack of process control of jelly filling operation.

In order to examine the optimality and criticality  of control of parameters  at Jelly filling process a Statistically designed experiment was planned. Two teams- a Top Management Team and an Investigation Team, were setup for smooth conduct of the experiment.

Detailed discussions and elaborate lecture sessions with Managers, Engineers and shop personnel were conducted as a part of planning and organizing the experiment.

‘Experimental factors’ and their levels: 5 factors at two levels and 3 factors at 3 levels, levels for ‘Local control factors’ and ‘ Noise factors’ were identified. 

Out of the 864 (=25x33) factor level Combinations 18 combinations that make up L18 OA (Orthogonal Array) design was selected.

The experimental factors were classified as ‘primary’, ‘secondary’, and ‘tertiary’ depending on the lack of ease in changing the factor levels. Two factors were nested in other factors.

The details of conducting trials, cable samples to be collected and test on samples planned.

The three cardinal principles of experimentation: Randomization, Replication, and Local control were ensured.



One km length of cable was produced for each of the 18 trials, and 3 samples from start, middle end of the cable were tested for WP and in case of failure, the day of failure was recorded.

Tests of Hypotheses on proportions of failure and expected life showed only one factor, Type of core wrap to be having significant contribution to WP failures. A critical examination of the data showed large sampling and experimental errors, in spite of the care taken.

The data were further subjected to ‘Minute Analysis’ of Prof. Taguchi and ANOVA performed. In addition to Type of core wrap, Jelly  Application conditions: Temperature – Pressure, and Jelly Brand were significant.

Cables filled with indigenous Jelly are stable over time compared with imported Jelly, and low temperature, high pressure filling results higher rate of success in WPT compared with high temperature- low pressure filling.

These were contrary to existing belief and practice of the plant, but a vital technological input. Based on the experiment, technological  and economic considerations, optimum levels for the 8 factors determined.

Recommended process parameter levels implemented: higher HP motor for pumping jelly installed, line speed locked, and chilled water plants set right. Process manuals prepared, graphical feed back instituted and shop personnel were trained.

The success rate  of  cables in WPT increased from 55% to 100 % in the 20 pair cable, and 99-9 % in bigger size cables in 2 years. Also, the number of cables marketed increased from 450 to 8050.

The testing of cables reduced to 1/8 th, test duration curtailed to 4 days from the earlier 14 days. The Management proclaimed a saving of 30 Millions of Rupees over an investment of Rs 9 lakhs.

Much more significant aspects are development of  cordial relationships and improvement in work culture with customer and among employees.



About the author

Professor J.Venkatappaiah is a Post-graduate in Mathematics and Statistics (M. Stat., Indian Statistical Institute) with      specialization in Quality Technology and Management. Overfour decades of  rich  professional experience as consultant and faculty on Quality Technology and  Management to many prominent private and public sectorIndustrial and Academic organizations. Worked in association with UNIDO and other International Experts, and  Genichi Taguchi, OA & QE Pioneer. Continuing the professional activity after retirement as Professor, SQC&OR, from Indian Statistical  Institute;

Read 1101 times Last modified on Friday, 20 June 2014 20:44


Leave a comment

Please enter your comments by filling the form below. Your comments are of much value to us. For preventing malicious comments, the comments are published only after moderation. We cannot guarantee a reply to each comment but we assure you that all comments are read and appropriate actions will be taken. Thank you for your understanding.