You are on page 1of 8

POLAROID CASE STUDY

• CTOE ASSIGNMENTS
GUIDED BY - Prof. Siddharth Joshi

SUBMITTED BY -

Sakshi Sharda (20192341) | Subhasish Das (20191108) | K. Partha Saradhi (20192234) | Dharmik Tandel
(M20191561) | Shubham Bopche (20191099) | Baisakhi Singha Roy (B20195205) | K Md Riaz (20191047)
INTRODUCTIO
N

The quality control processes at the R2 assembly plant of Polaroid were inefficient and traditionally
based on double sampling.
Sampling destroyed a perfectly good product and created scrap.
With Project Greenlight, a six sigma perspective known as Statistical Process Control was brought in.
However, the defects detected by the quality auditors rose to 10x, while defects reported in production
processes fell to half.
For Bud Rolfs, the manager heading the SPC, the perception of SPC is an area of concern.
At the same time, he has to build up confidence in the management that his project didn’t sacrifice
quality. Also, he has to convince the staff that quality comes from controlled and capable processes and
not from ‘tweaking’ the machine.
He also has to make a decision to be able to send Microline (current measurements tool) data to the
computer directly. Or to purchase more complex data capturing device, which will cost $100k more,
and require a full-time operator.
The case analysis is thus based on observing the data from two items (finger height, pod weight) after
the project was implemented
Using the data you we can make a preliminary analysis of whether project greenlight has achieved its
aim of ensuring that the processes that produce the pod weight, finger height measurements are under
control or not.
The control limit within which the mean values should lie is calculated is shown in excel.
Observation from EXCEl

The process is under control with respect to pod weight mean & range , but in finger height mean was
quite out of range.
As seen, by exhibit 6, the only out of range is the mean of finger height, so analysing it more carefully
and seeing where the defect coming from we run an Anova on the finger height mean.

Please prefer excel attach at mail for reference.


OUTCOMES from R
Observation from R

As seen the p-value of this is <0.05, the null hypothesis is rejected; so the three shifts doesn’t belong
in the same population. That’s why there is so much randomness.
To find which one doesn’t belong to same population we run TukeyHSD test.
By analysing the output, we can see that v3 which denotes the C shift doesn’t belong to the same
population as V1 and V2. so we can work on that shift and make the process In control
CTOE ASSIGNMENT

THANKYOU

You might also like