Sie sind auf Seite 1von 19

ASSIGNMENT 1

AZ Evulukwu
Assignment 1 - Similitude,
Modelling, and Data
Analysis
Ee3580@mun.ca

Memorial
University
Newfoundland

709-691-3790

Dr Leonardo Lye

4/10/2011
TABLE OF CONTENTS
1) ................................................................................................................................................ 4
2) ................................................................................................................................................ 6
a. ............................................................................................................................................. 6
i. .......................................................................................................................................... 6
ii. ......................................................................................................................................... 6
iii. ........................................................................................................................................ 6
b. .......................................................................................... Error! Bookmark not defined.
i. ...................................................................................... Error! Bookmark not defined.
ii. ..................................................................................................................................... 8
c. ............................................................................................................................................. 9
i. .......................................................................................................................................... 9
3)........................................................................................................................................... 11
a. ........................................................................................................................................... 11
i. ........................................................................................................................................ 11
b. ........................................................................................................................................... 12
i. ........................................................................................................................................ 12
ii. ...................................................................................... Error! Bookmark not defined.
c. .......................................................................................... Error! Bookmark not defined.
i. ...................................................................................... Error! Bookmark not defined.
4) .............................................................................................................................................. 16
a. .......................................................................................... Error! Bookmark not defined.
i. ...................................................................................... Error! Bookmark not defined.
ii. ...................................................................................... Error! Bookmark not defined.
b. .......................................................................................... Error! Bookmark not defined.
Summary of Results
Q1)
A brief summary of an experiment I expect to conduct with its
factors and responses is given. This experiment is concerning the
calculation of Mass transfer coefficients in VAPEX systems.

Q2)
Minitab software is used to analyse to sets of data. 2 paired, 2-
sample T and Normal probability test were conducted. It was found
that the two sets of data are much related due to the similarities of
their numerical values

Q3)
Analysis show that we accept our null hypothesis and that all
mean values for chemical types are similar. Also Lab 2 and lab 4 are
the most similar of all.

Q4)
1)
Consider an experiment that you have conducted in the past. What was the response,
the factors and levels considered, and how was the experiment analyzed? What was the
outcome or conclusion of that experiment? If you have never conducted an experiment in the
past, think of an experiment that you may conduct that might be of interest and list the response
(s), factors, possible levels, and what you hope the experiment will answer.

Experiment to be done is the determination of Mass transfer


coefficient.

Bitumen viscosity reduction is required for it to flow properly


and be produced efficiently. The main method for doing this is the
use of steam to directly reduce its viscosity. This is very capital
intensive and waste water.

It is believed that the bitumen viscosity can also be reduced


with the use of light hydrocarbons as a solvent (VAPEX technology).
The main mechanism is diffusion of the solvent into the bitumen to
reduce viscosity. The mass transfer coefficient of this diffusion is
therefore very important.

Responses expected are;

 Variation in the mass transfer coefficient values


 Variation in viscosity reduction of bitumen.

Factors include;

 Type of solvent used


 Temperature of the injected solvent
 Pressure of injected solvent
 Purity of solvent used
Possible levels included;

 Type of solvent – Butane/Propane


 Temperature of solvent- 20-30oc
 Pressure of solvent – 35.177-40psi
 Purity of solvent – 0.85-0.95 mole fraction

I intend to use the experiment to obtain the optimum


conditions that give the best diffusion coefficient (thus viscosity
reduction) and therefore being able to better predict expected flow
rate of bitumen in a VAPEX system.
2)
Air pollution concentrations at 10 locations of a building were measured after
implementing certain control measures. Measurements were taken at two different times at the
same location and the following values in parts per million of SO2 were obtained:
Time 1: 68.3 66.5 62.9 64.8 66.3 69.4 64.2 62.5 70.4 61.2
Time 2: 70.8 60.1 65.1 63.3 68.5 71.8 62.1 68.2 65.1 67.9
It is desired to show that the pollution values obtained at Time 1 are not significantly different
from that at Time 2 at the 5% significance level.

a.
i.
Which statistical test should be carried out?

We can use the Minitab software and do a paired T test on


both sets of values

ii.
Justify your choice

This test is able to analyze 2 sets of data and compare their


means

iii.
What is the conclusion from the test?

 We start by stating our null and alternate hypothesis


 H0 : µ1= µ2
 H1 : µ1≠ µ2
 Through the use of mini-tab we make 2 columns in the columns
board
 1 for Time 1 and other Time 2
 We fill in their respective values
 We go Stat/Basic Statistics/2 sample t
 We enter samples in different columns with ‘Time 1’/’Time 2’
 This gives us the plot below. One is the individual value plot
and the other is Box-plot of Time 1, Time 2
 As we can see from the individual value plot and box-plot, the
means of the two sets of values hardly have any difference
between them and as a result the two set of values are very
similar as a result we accept the null hypothesis.

b.
i.
What is the main assumption of the test used?

Our main assumption is that we have a normal distribution.

ii.
Verify the assumption of the test used.

This is verified by the following process:


 Going to Stat/Basic Statistics/Normal.
 We click on the normality test
 We do this test for Time 1 and Time 2
 These give us the probability plot below
 The high P-values for both plots (0.892 & 0.841) indicate the
distribution is a normal distribution
Probability Plot of Time 2
Normal
99
Mean 66.29
StDev 3.793
95 N 10
AD 0.198
90
P-Value 0.841
80
70
Percent

60
50
40
30
20

10

1
55 60 65 70 75
Time 2

c.
i.
What alternative tests to the one you used can also be used?

A 2 sample T-test is also adequate for these set of values. This


follows the same procedure as the paired t – test. The results are
given below. They also show very little variation in the mean and Se
mean values. Also the T-value is close to zero (-0.46) and so the this
favours the null hypothesis (that the mean sets of values are similar)
Paired T for Time 1 - Time 2

N Mean StDev SE Mean


Time 1 10 65.65 3.07 0.97
Time 2 10 66.29 3.79 1.20
Difference 10 -0.64 4.35 1.38

95% CI for mean difference: (-3.76, 2.48)


T-Test of mean difference = 0 (vs not = 0): T-
Value = -0.46 P-Value = 0.653
3)
a.
An article in the journal of Quality Technology (vol 13, No. 2, 1981) describes an
experiment that investigates the effects of four bleaching chemicals on pulp brightness. These
four chemicals were selected at random from a large population of potential bleaching agents.
The data are as follows:

Chemical Pulp Brightness


1 77.2 74.47 92.75 76.21 82.88
2 80.52 79.31 81.91 80.35 73.39
3 79.42 78.02 91.60 80.80 80.63
4 78.00 78.36 77.54 77.36 77.39

i.
Is there a difference in chemical types? Use = 0.05.
 We start by stating our null and alternate hypothesis
 H0 : µ1= µ2= µ3= µ4
 H1 : At least 1 of the 4 means is different
 We start by opening a new design in DOE software
 In the factorial tab we go to general factorial design
 We use categoric factors of 1
 Level 4 with nominal tick and label treatment level 1-4
 We use 5 replicates with a response level of 4
 We then get a table which we right click to put in standard
order
 We enter all our responses into the R1 table accordingly
 The table then looks like below
 We then go to analysis to find out information of all our values
 The ANOVA analysis suggest that there is no difference in the
chemical types as 0.5377<0.75. So we accept the null
hypothesis
 The table of ANOVA values is also shown below.
b.
i.
Analyze the residuals from this experiment.
 We go into the diagnostics tab to take a look at the residuals
 Below are lists of plots we do for the residuals.
Design-Expert® Software Normal Plot of Residuals
R1

Color points by value of


R1: 99
92.75

N o r m a l % P r o b a b ility
95
73.39 90

80
70

50

30
20

10
5

-1.42 -0.38 0.66 1.71 2.75

Internally Studentized Residuals

Design-Expert® Software Residuals vs. Predicted


R1
3.00
In te r n a lly S tu d e n tiz e d R e s id u a ls

Color points by value of


R1:
92.75
1.50
73.39

0.00

-1.50

-3.00

77.73 78.82 79.91 81.00 82.09

Predicted

Design-Expert® Software Externally Studentized Residuals


R1
E x te r n a lly S tu d e n tiz e d R e s id u a ls

3.67
Color points by value of
R1:
92.75
1.84
73.39

0.02

-1.80

-3.62

1 4 7 10 13 16 19

Run Number
 Our normal plot residuals show that the residuals are normally
distributed with a constant variance
 Our studentized residual vs. predicted values shows we quite a
bit of constant error
 Our external studentized plot show we have about 2 influential
values at most

ii.
Comment on the adequacy of the model.

Std. Dev. 4.90


Mean 79.91
C.V. % 6.13
PRESS 599.98

R-Squared 0.1234
Adj R-Squared -0.0409
Pred R-Squared -0.3697
Adeq Precision 1.992

 We have a negative predicted r-square value of –0.37


 This value shows the use of overall mean is a better predictor
of responses than our current model
 Our adequacy precision which is the signal to noise ratio shows
an inadequate signal of 1.99
 This model is not sufficient to navigate the design space

i.
If there is a statistically significant difference, which labs are different and which are
similar? Ignore this part if there is no significant difference.

 Prob > t value gives a feel for the difference between two
means.
 A value less than 0.05 makes there a great probability that the
means of the two groups are significantly different
 A value of larger than 0.1 means the difference in the two
means is not considered significant.
 All the Prob > t values surpass 0.1 mean they are not very
statistically different.
 Lab 2 and 4 having the highest Prob > t value shows they are
the most similar
4)
a.
The data below are from an experiment conducted to investigate the warping of copper
plates. The two factors studies were temperature and the copper content of the plates. The
response variable is the amount of warping. The data are as follows:
Copper Content %)
Temperature (C˚) 40 60 80 100
50 17, 20 16, 21 24, 22 28, 27
75 12, 9 18, 13 17, 12 27, 31
100 16, 12 18, 21 25, 23 30, 23
125 21, 17 23, 21 23, 22 29, 31

i.
Analyze the data and draw the appropriate conclusions, checking all necessary
assumptions

 First we state our null hypothesis and alternate hypothesis


 H0 : µ1= µ2= µ3= µ4
 H1 : At least 1 of the 4 means is different
 We click on general 2 factorial design
 We choose 4 levels for A and name A Temp (temperature) with
units of degrees
 We also put all the temperatures for since this is a quantitative
parameter
 We choose 4 levels for B and name B Wt (weight) with units of
%
 We also put all the weight % since this is also a quantitative
parameter
 We go for 2 replicates a piece as we have with 1 response
 We now input all the response values from our experiment
 The given table is shown below
Std Run Temp Wt% R1
1 28 50 40 17
2 29 50 40 20
3 10 75 40 12
4 16 75 40 9
5 24 100 40 16
6 7 100 40 12
7 17 50 40 21
8 9 125 40 17
9 13 50 60 16
10 21 50 60 21
11 2 75 40 18
12 31 75 60 13
13 3 100 60 18
14 8 100 40 21
15 20 125 60 23
16 11 125 60 21
17 23 50 80 24
18 26 50 80 22
19 5 75 80 17
20 22 75 80 12
21 32 100 80 25
22 4 100 80 23
23 19 125 80 23
24 30 125 80 22
25 1 50 100 28
26 6 50 100 27
27 25 75 100 27
28 12 75 100 31
29 18 100 100 30
30 14 100 100 23
31 27 125 100 29
32 15 125 100 31

 We now make a list of plots to analyse the data


 First we have the normality plots from the residuals
 This shows the data follows a normal distribution and therefore
has a fairly constant variance
 Our predicted vs actual graph has a very poor goodness fit
showing the values are independent of each other
 Our residuals vs predicted value plot shows a lot of scattered
and values and a fair amount of independence
 All these plots are shown below
Design-Expert® Software Normal Plot of Residuals
R1

Color points by value of


R1: 99
31

N o r m a l % P r o b a b ility
95
9 90

80
70

50

30
20

10
5

-1.69 -0.78 0.13 1.04 1.95

Internally Studentized Residuals

Design-Expert® Software Predicted vs. Actual


R1
31.00
Color points by value of
R1:
31
25.50
9
P r e d ic te d

20.00

14.50

9.00

9.00 14.50 20.00 25.50 31.00

Actual

Design-Expert® Software Residuals vs. Predicted


R1
3.00
In te r n a lly S tu d e n tiz e d R e s id u a ls

Color points by value of


R1:
31
1.50
9

0.00

-1.50

-3.00

13.00 17.25 21.50 25.75 30.00

Predicted

Das könnte Ihnen auch gefallen