Scribd
Upload
661 views

Design and Analysis of Experiments 3

The document discusses the results of an experiment measuring conductivity at four locations in an aerated lagoon. ANOVA and multiple comparison tests found significant differences between locations, indicating non-homogeneous mixing. A second experiment examined factors affecting reaeration rate in a model stream. ANOVA showed velocity, depth, and roughness had significant effects and interactions on reaeration rate. Residual analysis found one outlier but otherwise acceptable residuals.

Uploaded by

api-3723257
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
661 views

Design and Analysis of Experiments 3

The document discusses the results of an experiment measuring conductivity at four locations in an aerated lagoon. ANOVA and multiple comparison tests found significant differences between locations, indicating non-homogeneous mixing. A second experiment examined factors affecting reaeration rate in a model stream. ANOVA showed velocity, depth, and roughness had significant effects and interactions on reaeration rate. Residual analysis found one outlier but otherwise acceptable residuals.

Uploaded by

api-3723257
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
You are on page 1/ 20

Design and Analysis of Experiments

Take Home Work Exam 1


Student: Nguyen Thi Hang Mark:
ID: R9605005
Comment:

Problem.1. Conductivity measurements (µmho/cm) were taken at four different


locations in the aerated lagoon of a pulp and paper mill. The lagoon is supposed to
be mixed by aerators so the contents are homogeneous. The conductivity results
were shown in the following table:
(1) Is the lagoon homogeneously mixed? Conduct an ANOVA for the data and
draw your conclusions.
(2) Please check if the assumption of equal variances is satisfied.
(3) Test all pairs of means using both the Turkey’s test and Fisher LSD test.
Please also compare the results from these two multiple comparison
methods. What are your conclusions?
(4) Use the Kruskal-Wallis test for the experiment. And also compare the
conclusions obtained with those from the usual ANOVA.

Table1 Conductivity measurements (µmho/cm) at the different locations on the


aerated lagoon
--------------------------------------------------------------------------------------
Location A Location B Location C Location D
------------------------------------------------------------------------------------
620 630 680 560
600 670 660 620
630 710 710 600
590 640 670 610
650 680 630
660 680 640
630
590
Solution.
Conducting an ANOVA for the data we get below procedure
With hypothesis: H0 : µ1 = µ2 = µ3 = µ4
H1 : µi ≠ µ j
Using Minitab software (chose α = 0.05) with the following steps:

1. Input and prepare the data for the test


2. Select One-Way of ANOVA method

3. Set up the suitable parameters


we can use some options such as Graphs, Compatisons
Then, we can see the results from using the Minitab

Results and Discussion


(1) Is the lagoon homogeneously mixed? Conduct an ANOVA for the data
and draw your conclusions.
F0 = 13.57 > F0.05, 3, 20 = 3.10 or P = 0.000 < 0.05 thus we can conclude that H0
can be rejected or the lagoon is not homogeneously mixed.

(2) Please check if the assumption of equal variances is satisfied.

H0 : δ 1 = δ 2 = δ 3 = δ 4
H1: δι ≠ δj

Use test for equal variances we get these results


To fix suitable parameters:
Both of Pvalue obtained (Pvalue = 0.644 and Pvalue = 0.593) are larger than 0.05 thus
H0 can not rejected. It means all variances are the same.

(3) Test all pairs of means using both the Turkey’s test and Fisher LSD
test. Please also compare the results from these two multiple
comparison methods. What are your conclusions?
From

Above results indicate pairs of means that are significantly different for:
A vs B, A vs C, B vs D, C vs D
And there are not significantly different for: A vs D, B vs C.
Above results indicate pairs of means that are significantly different for:
A vs B, A vs C, B vs D, C vs D
And there are not significantly different for: A vs D, B vs C.

To compare the results from these two multipule comparison methods:


These two method give the same results when we compare all pairs of means
using both the Tukey’s test and Fisher ‘s test.

(4) Use the Kruskal-Wallis test for the experiment. And also compare the
conclusions obtained with those from the usual ANOVA.
We have hypotheses: H0 : µ1 = µ2 = µ3 = µ4
H1 : µi ≠ µ j
Above results indicate both of H = 16. 86 and H = 17.02 larger than χ0.05, 3 = 7.8
1 or P = 0.001 < 0.05 we conclude to reject H0 or the lagoon is homogeneously
mixed

Problem 2. The data below are from an experiment that attempted to examine
the factors affecting the reaeration rate (y) in a laboratory model stream
channel. The three experimental factors are stream velocity (V, in m/sec),
stream depth (D, in cm), and channel roughness (R).
Table 2 Results of the reaeration rate
Run V D R y
1 0.25 10 smooth 107 117 117
2 0.5 10 Smooth 190 178 179
3 0.25 15 smooth 119 116 133
4 0.5 15 smooth 188 191 195
5 0.25 10 smooth 119 132 126
6 0.5 10 smooth 187 173 166
7 0.25 15 smooth 140 133 132
8 0.5 15 smooth 164 145 144

(1) Analyze the data and draw conclusions.


(2) Analyze the residuals.

Solution
There are three factors V, D and R, each at two levels, are of interest. The
design is called a 23 factorial design.

Using DOE method for 23 design to analyze the data:


1. Input and prepare the data on Minitab
2. Set up suitable parameters.
3. Output the data
3. Results and Discussion
(1) Analyze the data and draw conclusions.
The ANOVA in the above Table is used to confirm the magnitude of these
effects. From these Table we note that the main effect of V is highly significant
(because has very small P-value). The V*D, V*R, D*R, and V*D*R are also
highly significant; thus there are strong interaction between V and D, V and R;
V, D, and R on the reaeration rate (y).

(2) Analyze the residuals.


From the normal probability plot of these residuals show that doesn’t
reveal anything particularly troublesome.

Although the largest positive residual (13 at V = 0.05, D = 15, R =


coarse) does stand out some what from others.The standardized value
of this residual is 13 / 59.4 = 1.69, and this is only residual whose
absolute value is smaller than 2.
Residuals Versus the Order of the Data
(response is y)
15

10

5
Residual

-5

-10
2 4 6 8 10 12 14 16 18 20 22 24
Observation Order

Residuals Versus the Fitted Values


(response is y)
15

10

5
Residual

-5

-10
110 120 130 140 150 160 170 180 190 200
Fitted Value

From plots the residuals versus the fitted values. This plots indicates that a
mild tendency for the variance of the residuals to increase.

You might also like