advantageconsumer.com
Consumer Protection Council, Rourkela 


(As desired by the National Consumer Disputes Redresaal Commission (NCDRC),
as a result of the case which was before its consideration, Consumer Protection
Council, Rourkela vs. Indian Oil Corporation Ltd. & Ors., O.P. No.
224 of 2001, two Professors of Indian Institute of Technology, Kharagpur,
(Dr. R.K.Saha, Professor & Head, Deptt. of Chemical Engineering and
Prof. Subhabrata Ray, of Chemical Engineering Deptt.) undertook the study
of the LPG (Indane) bottling plant, at Balasore, Orissa, in June 2003.
Their Report, with detailed statistical analysis and their inferences were
made available to the NCDRC, in July 2003. Since the Report was under
the consideration of the Commission and the final order had not been passed,
we also delayed publishing this informative report. Readers can now
enjoy reading the full content of the report.  Editor)
Executive Summary: The Plant was noted to normally fill these days around 6,000 cylinders in single shift operation and around 10,000 cylinders in twoshift operation. The capability of the carousel (LPG filling) machine and its accessories for filling correct weight of LPG in cylinders has been evaluated based on sampling of filled cylinders from the carousel machine and study of the operation of the plant, and study of the entire system. Analysis of the filled LPG quantity in these samples, drawn in presence of M/s IOC personnel and Mr. Vaidyanathan showed:
M/s IOC Ltd. Marketing Division. Basis : Target
wt of LPG to be filled in cylinders = 14.2 kg
Filled LPG weight between the lower and upper limit is understood to be
the correct filled weight. This weight is expected in cylinders leaving
the ' carousel and the weight correction system for underfilled/overfilled
cylinders'.
Steps (1)
Sampling
The plant fills around 6000 to 10000 cylinders per day. For this range of production, the sample size code is 'M'. (ref.: IS 2500 code, table 1, inspection level v). from table IIA, M represents sample size of 315, under acceptance quality level of 0.01. the sample size of 305 is close to 315 and is adopted as 'acceptable'. The sampled data is included here in Annexure 1. Two samples were rejected as obvious outliers. The sample computations are therefore based on 303 samples. The sampled data was processed to find the actual weight of LPG filled in the cylinders. Mean and standard deviation of the LPG filled were computed. Frequency plot of the quantity of filled with the corresponding 'normal distribution' was plotted.
It is seen that the distribution follows 'normal distribution' with fair
closeness. This also signifies that the sample is a fair representation
of the production for the day (population) and the mean and standard deviation
of the sample and the population are expected to be close.
(2) Looking for presence of any bias in the filled quantity and the cylinder tare weight A plot
of the tare weight vs. the quantity of LPG filled for the sampled cylinders
was done. The general trend is of filling quantity being above 14.2 kg,
for all cylinder tare weights. No definite trend in the filled quantity
variation with cylinder tare weight is observed.
(3) Computation based on sampled data Sample size = 303 cylinders Filled wt of LPG (mean, xbar) = 14.3206 kg Standard deviation of filled wt of LPG (s ) = 0.2667 kg Mean
overfilling (+)/underfilling () = 0.1206 kg
(4) Preliminary inference from the sampled data analysis The
number of cylinders underfilled and overfilled in the 303 samples observed
correspond to 11.22% (=100 x 34/303 %) and 42.90% (=100 x 130/303 %) respectively.
These are the percentages of the cylinders which should be passing through
the filled LPG bottle correction scales / system. We can say that about
out of ten cylinders filled, one has to be corrected for underfilling,
and four for overfilling. These proportions are 'uncommonly high'. Weight
correction of this large fraction of cylinders from carousel is possible
only at a lower throughput of the filling plant. The carousel performance
is therefore found to be unsatisfactory as it has a large variability in
the quantity of LPG filled in the cylinders.
(5)
Estimation of limits on the cylinders filled for the day @ 95% confidence
(LNTL95, UNTL95)
The range limits within which the average quantity of LPG filled per cylinder (for all cylinders filled in that day) by the carousel is expected to lie with 95% confidence are estimated. LNTL95
= xbar  (Za/2)
x s
/ Ö
n = 14.3206  1.96 x 0.2667/Ö
303
UNTL95
= xbar + (Za/2)
x s
/ Ö
n = 14.3206 + 1.96 x 0.2667/Ö
303
It
is seen that the two limits are close and this signifies that the sample
represents the day's filling quite well. The mean and standard deviation
for the sampled cylinders therefore shall represent the same for all cylinders
filled in that day.
(6)
Process Capability Ratio (PCR)
Process
capability ratio (PCR) aids in evaluation of processes with respect to
their specification limits, and is defined as:
Recommended
minimum values for existing processes are:
In this case, USL = 14.35 kg, and LSL = 14.05 kg, s = 0.2667 kg Hence, PCR (2 sided) = (14.35  14.05)/(6x0.2667) = 0.187 << 1.33 limit PCR (if only LSL exists) = (14.3206  14.05)/(3 x 0.2667) = 0.3382 << 1.25 limit PCR (if only USL exists) = (14.3500  14.3206)/(3x0.2667) = 0.0367<<1.25 limit The
PCR values are substantially away from the corresponding minimum required.
This denotes that a large fraction of the filled cylinders are overfilled
and underfilled from the carousel. Operation of the filled weight correction
system is essential to correct for this deviation. Expectedly about 11%
and 42% of the cylinders have to pass through the filled LPG weight correction
system. Thus 50% of the cylinders filled in the carousel passing through
the weight correction system shall constrain the throughput of the plant.
In other words, the present condition of the carousel and its working system
is not capable of delivering the correct weight of filled LPG (14.2 +/
0.15 kg), at the normal bottling rate of the plant.
Conclusions Based on the typical operation of the plant observed and analysis of the results of the samples drawn the following conclusions are made:


