ChE 452 Lecture 06
-
Upload
cody-wilkinson -
Category
Documents
-
view
37 -
download
3
description
Transcript of ChE 452 Lecture 06
ChE 452 Lecture 06
Analysis of Direct Rate Data
1
Objective
How do you fit data Least squares vs lowest variance
Strengths, weaknesses Problem with r2
2
Analysis Of Direct Rate Data
3
0.020
0.100
10 100
Oxygen Pressure (Torr)
Etc
h R
ate
(mic
rons
/min
)
Metallic Color
Oxide Color
Slope = 0.5
0.030
0.040
0.050
0.060
0.080
0.010
General method – least squares with rate vs time data
Figure 3.10 The rate of copper etching as a function of the oxygen concentration. Data of Steger and Masel [1998].
Usually Not So Easy
Results vary with how fitting is done Cannot tell how well it works by
looking at r2
4
Example: Fitting Data To Monod’s Law
Table 3.A.1 shows some data for the growth rate of paramecium as a function of the paramecium concentration. Fit the data to Monod’s Law:
where [par] is the paramecium concentration, and k1 and K2 are constants.
5
]par[K1
]par[Kkr
2
21p
(3.A.1)
There are two methods that people use to solve problems like this: Rearranging the equations to get a
linear fit and using least squares Doing non-linear least squares to
minimize variance
I prefer the latter, but I wanted to give a picture of the former.
6
Methodology
There are two versions of the linear plots:• Lineweaver-Burk Plots• Eadie-Hofstee Plots
7
(3.A.2)
Methodology
In the Lineweaver-Burk method, one plots 1/rate
vs. 1/concentration.
8
12121
2
21p k
1
]par[Kk
1
]par[Kk
]par[K
]par[Kk
1
r
1
(3.A.2)
Methodology
]par[K1
]par[Kkr
2
21p
Rearranging
A B C D E F
01 k_1 139.400 0.194 2.000
02 K_2 0.037 0.007 9642.8
03 r2z 0.900 Calculated
04 conc rate 1/conc 1/rate rate error
05 0.000 0.000 0.000 0.000
06 2.000 10.400 0.500 0.096 9.602 0.636
07 3.600 12.800 0.278 0.078 16.382 12.829
08 4.000 23.200 0.250 0.043 17.967 27.379
09 5.200 17.600 0.192 0.057 22.488 23.897
10 7.800 46.400 0.128 0.022 31.216 230.566
11 8.000 23.200 0.125 0.043 31.833 74.534
12 8.000 46.400 0.125 0.022 31.833 212.189
13 11.000 32.000 0.091 0.031 40.319 69.210
Table 3.A.3 The numerical values in the spreadsheet for the Lineweaver Burke plot
Table continues
9
Lineweaver-Burk Plots
Numerical results
From the least squares fit,
10
121p k
1
]par[Kk
100717.0
]par[
194.0
r
1
(3.A.3)
Comparison of equations (3.A.2) and (3.A.3) shows:
k1 = 1/.00717=139.4, K2=1/(0.194*k1)=0.037,
r2=0.900
How Well Does It Fit?
11
0 0.1 0.2 0.3 0.4 0.5 0.60
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
0.11
1/R
ate
1/Concentration0 10 20 30 40 50 60 70 80 90
0
50
100
150
Rat
e
Concentration
Data
Lineweaver-Burk
Figure 3.A.1 A Lineweaver-Burk plot of the data in Table 3.A.1
Figure 3.A.2 The Lineweaver-Burk fit of the data in Table 3.A.1
Why Systematic Error?
We got the systematic error because we fit to 1/rp. A plot of 1/rp gives greater weight to data taken at small concentrations, and that is usually where the data is the least accurate.
12
Eadie-Hofstee Plot
Avoid the difficulty at low concentrations by instead finding a way to linearize the data without calculating 1/rp.
Rearranging equation (3.A.1):rp(1+K2[par])=k1K2[par]
(3.A.4)
Further rearrangement yields:
13
p221p rKKk
]par[
r
(3.A.5)
Eadie-Hofstee Plot
14
0 50 100 1501
2
3
4
5
6
7
Rat
e/co
ncen
trat
ion
Rate0 10 20 30 40 50 60 70 80 90
0
50
100
150
Rat
e
Concentration
Data
Eadie-Hofstee
r2=0.34
Figure 3.A.3 An Eadie-Hofstee plot of the data in Table 3.A.1
Figure 3.A.4 The Eadie-Hofstee fit of the data in Table 3.A.1
r2 Does Not Indicate Goodness Of Fit
Eadie-Hofstee gives much lower r2 but better fit to data!
15
Non-linear Least Squares
Use the solver function of a spreadsheet to calculate the best values of the coefficients by minimizing the total error, where the total error is defined by:
16
Data
2
2
21p ]par[K1
]par[KkrabsErrorTotal
(3.A.7)
Summary Of Fits
17
0 10 20 30 40 50 60 70 80 900
50
100
150
Rat
e
Concentration
Data
Non-linear least squares
Non-linear least squares
0 10 20 30 40 50 60 70 80 900
50
100
150
Rat
eConcentration
Eadie-Hofstee
Lineweaver-Burk
Figure 3.A.5 A nonlinear least squares fit to the data in Table 3.A.1
Figure 3.A.6 A comparison of the three fits to the data
Comparison Of Fits
18
Method k1 K2 Total error r-squared
Lineweaver-Burk 139 0.037 9643 0.910 (linear plot)
Eadie-Hofstee 267 0.0156 6809 0.344 (linear plot)
non-linear least squares 204 0.0221 4919 0.905 (non-linear)
Table 3.A.5 A comparison of the various fits to the data in Table 3.A.1
Note: 1)Results change according to fitting method2)there is no correlation between r2 and goodness of fit.
Summary
Fit data using some version of least squares
Results change drastically according to How You fit data
Caution about using r2
19
Class Question
What did you learn new today?
20