Name: Ari Flores
Date Submitted:
4/19/08
Class: IE-673
Serial No: IE673-Spring 2008-61-39
1. Introduction
and Objectives of the Project
2. Descriptions
of the Methodologies Applied
3. Main
Body of the Project
3.1 Steps
3.1.1
Collect the Data
3.1.33 Compute
the Central Line and Control Limits
3.1.2 Plot Data
3.2. Control Chart Spreadsheet
3.3 Analysis
of control charts
4. Summary
5. Further
Work Needed
6. References
1.
Introduction and Objectives of the Project
This
organization is customer focused and understands customer needs and requirements. Embedded in our company’s core values
is to improve continuous by being more effective, efficient, and innovative and ensure customer satisfaction at a lower price.
A management tool that our company uses, is also part of our Total Quality Management framework, and allows for process improvement
is process control analysis. The objective of this project is using control charts
for detecting the variations and minimizing them for the quality improvement. My projects focus is in the Liquefied Natural
Gas process. The control chart variable analysis is on a pressure variable called “Expander/Compressor Discharge”
pressure. Expander/Compressor Discharge” pressure is an important pressure in this LNG process because this pressure
is a primary energy process variable to convert Natural gas to LNG. The control
charts for Attributes is p-chart constant analysis. The p-chart will be on safety
relief valves. These are important safety devices in LNG to safely remove over-pressurized fluid. In the following paragraphs
I will discuss methodology applied, process control charts and their explanation, and summary.
2. Descriptions of the Methodologies Applied
Statistical methods is a management tool that
can be utilized to improve the processes and increase the bottom line, lower cost and become more productive at a higher quality.
Quality control charts are used in quality control. Control charts are the tools for quality control and used for improving
the process and quality systems of the organizations. Quality control charts are important because this allows management
to make informative factual decision on the process. According to E-learning text book, the entire decision making process
in quality control, relies on statistical methods (Ranky). This is important and applies the Total Quality management principles
because as we learned early in the year one of Quality principled defined by ISO is to perform decision on data an factual
information.
My understanding is, in simplest terms, variation
is difference in the design from the actual result and variation can occur in every part of the process. Control charts are
tools to detect and help to make informative decision to reduce the variation in the process or product.
Briefly, to determine the variations samples
are taken at given time intervals or randomly, those samples are tested to determine whether samples are in conformance with
the production specifications. The level of conformance are plotted on the charts and monitored. Depending on the type of
analysis one want to do various control charts can be created. Below are two types of control charts. These two types of control
charts Attribute control charts and variable control charts. Below is a brief description of each control chart. Descriptions
were found in E-learning text book by author Professor Ranky.
Attribute
Control Charts have two types of characteristics that are success (conform) or failure (non-conform) to the specifications.
1. P - Charts deals with the percentage
of the success by dividing the number of non-conforming units to the sample size of the sample.
2. C – Charts observes the number
of nonconformities in the production by dividing the number of nonconformities to the sample size.
3. U – Charts are the number of nonconformities
per unit where the sample size is variable.
Variable
Control Charts maintain the quality control over production by process mean and process variability. Mean is defines
as the average of the values in a sample data set. Purpose of the mean is to measure the central tendency. Standard deviation
is the variability in a sample data set by measuring deviation of the values from the mean.
· X-bar Chart shows
the variations of the mean value of the quality characteristic over several samples.
· R-chart shows
the variability with standard deviation of a sample of n units.
In this
project, a variable control chart was prepared for “Expander/Compressor Discharge” pressure. These sample data
were obtained from Operators which read pressure indicators. The pressure indicators used are calibrated on a yearly basis
by pressure test instrument that can be traced back to NISI standard. NISI standard instruments is quality standard used to
calibrate test instrument to maintain high quality standard.
In addition,
an attribute chart was prepared for safety relief valves. Operators test valves on a bench test named Barbee valve tester.
Barbee valve tester was also calibrated to NISI standard and LNG Operators test valve per Liquefied Natural Gas procedure.
3. Main Body of the Project
In the LNG process,
primary customers, the rate gas payer, wants a process plant to run cost effectively and efficiently. A good indicator that
can provide process indication is “Expander/Compressor Discharge” pressure. By improving “Expander/Compressor
Discharge” pressure performance, better customer satisfaction can be met. A
control chart variable on a pressure variable “Expander/Compressor Discharge” will be prepared to test the mean
and variability.
Another important
feature is safe operation of LNG plant. Therefore we are also testing safety valve conformity or non conformity.
3.1 STEPS
3.1.1 Collect the Data
Control Chart
Variable:
Sample size is
3 and the number of samples is 25 at maximum. Operators took the readings samples three times per day for a total of 25 days.
Samples results were logged into a database.
Control Charts
Attribute:
Sample size is
n is 36. Operators tested 36 Safety Valves per LNG procedure. Sample number m is 12. Valves were grouped three per sample
because in each shift three valves were tested. Samples results were logged into a database.
3.1.2 Plot Data
Control Chart
Variable:
For plotting X-bar chart
we need to find the average of the quality characteristic of 3 units for every sample.
The Formula
is:
_ n
X = Σ Xi/n
i=1
The formula for plotting R-chart
is:
R=MAX (Xi, i=1, n) – Min (Xi, i=1, n)
In this project, R-Chart and
X-bar chart were prepared using Professor Ranky’s excel spreadsheet file ContChartVar_Templ_ver4.xls. The Central Line and Control Limits were also calculated and plotted by above excel
spreadsheet.
The Coefficients A2, D3 and D4 depend on the sample size
which is 3 and was found in above spreadsheet.
Control Charts
Attribute:
“In p-charts, we
focus on the number of nonconforming units in a population (Ranky).”
The formula for fraction
“p” is
where : fraction of nonconforming
D : number of nonconforming units in
the ith sample
n
: sample size of the ith sample
|
Pi=Di/ni
where
: fraction of nonconforming
D : number of nonconforming
units in the ith sample
n : sample size of the ith sample
3.2 Control Chart Spreadsheet
3.2.1
Control_Chart_Variable
See Control Chart below.
3.2.2
p-chart_constant_size
See P-chart below
3.3 Analysis of control charts
For both chart the analysis criteria is obtained from Professor Ranky’s website cimwareukandusa.com. Criteria
for analyzing each chart can be seen below.
First Step is to look for if the points that are out of control limits.
Second
Step is to look for if there are no points out of control limits. In this step we should look if there are 7 consecutive
points above or below the central line and If there are 7 consecutive points increasing or decreasing which shows that the
process is out of control.
If above steps do not apply to each chart, we the process is in statistical control.
P-Chart:
According to Control_Chart_Attributes_Description
word document “attributes concern quality characteristics which are able to be classified in two types, conform and
not conform to specifications (Ranky)”. P-charts are used to determine
“the number of nonconforming units in a population (Ranky).” As indicated above these
36 Safety Valves must be tested per federal code. Valves are separated into three valves per sample batch because this is
how work was performed. Separating into group batches as such will allow for management to determine who performed testing
on that particular day. In this project the p-chart sample size 36 and sample
number is 12. The bottom line is the p-chart is statistically in control showing there is no non conformity (all 36 safety
valves are conforming). The p varied, where 6-point are above the p-average 0.016 and 6- below the p-average. The p-chart
shows no safety valves were out of the chart limits. The chart limits are upper limit UCL is 0.079 and lower limit LCL is
0.00. In addition the p-bar is 0.016. Management now knows this batch tested conforms. What if the valves do not conform?
The p-chart would show points out of limit. These point out of limits would mean valves are non-conforming. The valves that
are non-conforming may be caused by the following. Valve itself damaged, Operator tested wrong, procedure is incorrect, Barbee
tester miss calibrated. Because we know the other valves are correct, Barbee was calibrated, procedure was good enough to
calibrate the other valves, and then mostly likely valve damaged do to wear, tear or debris. The other possibility can be
the operator tested incorrectly even with the correct procedure. One can possibly go to another P-chart, check for nonconformity
and determine if those inconformity match same operator. By doing this check one can rule out the Operator.
According to word document control chart variables,
the goal of variables control charts is to control mean and variability of a process. The sample size for our analysis is
3 and the number of samples is 25. The R-chart and the X chart are both statistically in process control after several iterations were performed with the data. Several data points were removed
from the analysis because they were not within limits. The r-chart and x-bar chart were both showed points above and below
outer and lower limits respectively after iteration one. The UCLR and LCLR limits in the first R-chart were 40.00 and 0.00
respectively. The R-bar average is 15.76 and 8-points were found above the R-bar, but below the UCLR. Point 13 was significantly
over UCLR causing us to remove the point and once again chart the data points.
In the second R-chart The
UCLR and LCLR limits in the first R-chart were 32.34 and 0.00 respectively. The new R-bar average is 12.57 and 8-noncontinous
points were found above the R-bar. After the first iteration, all points were within UCLR and LCLR and no points were non-continuous.
Therefore, we can go ahead and create the R-chart based on the first iteration.
The first new X chart was
created based on the new R-chart data. The new X-chart had several points out of UCLR limit and was removed to plot a new
X-chart. The second revision new X-chart was created by removing the point that were out of UCLR limit. Once these points
were removed the third and final X-chart had all points within UCLR and LCLR limits 565 and 547. The new X-Dbar is now 556.7.
No points are outer limits. Two points appear to be outer limit but because they
are out of limits by a tenth we do not worry about it. This third and final X-chart is acceptable because they are no seven
points continuously connected above or below the X-Dbar average. Furthermore, the estimated process deviation is 3.564. The
process is stable at steady state than production is more effective.
The value of this X-chart
analysis is that with the real time technology the Operator will be able to determine within guidelines he/she should operate
within. In addition, management will have the data to determine areas to improve. For example, we can look at the samples
that were out of control limits and determine if the Operator needs more training or determine if training was not sufficient.
. In addition, we know equipment may show signs of wear and tear, and then immediate intervention to reduce downtime may be
possible.
For this situation we have
found area for improvement. Within two years a new plant will be constructed. Therefore, in the engineering design specification
better reading and measurements will be provided to provide more real time readings and allow for data statistical analysis.
The Operator will be able to have readily available X-chart analysis.
4. Summary
Control charts are the tools for quality control and used for improving the process
and quality systems of the organizations. Control charts are utilized to monitor and control the process. In this project
X-chart and R- charts were plotted. The sample size for our analysis is 3 and the number of samples is 25. The R-chart and
the X chart are both statistically in process control after iteration each. The p-chart sample size 36 and sample number 12
showed p-chart is statistically in control.
5. Further Work Needed
For continuous improvement, continuous
monitoring is required and smart initiations are required. Identify the areas of improvement and put together a quality plant
to resolve any issues. These charts should be documented and placed as part of record for future use. Also, the charts prepared
above can be compared to last year chart. Findings should be communicated by management to stakeholders.
6. References
Ranky, IE673, Spring 2008, e-Learning
Pack ID IE673_Spring 2008-61-39
