Introduction to Chemical Engineering Calculations

Lecture 2.

Regression and Nonlinear Axes

Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
2 Regression and Nonlinear Axes

What is regression analysis?

A technique for modeling and analyzing the relationship

between 2 or more variables. Usually, 1 variable is
designated as the independent variable.

Purpose: Establish a mathematical relation between the

variables.

C(t) = e t

 2  a f(x)sin nπxdx  sin nx sinh ny

u(x,y) =   0 
n=1 a sinh (nπb/a)  a  a a

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
2
2 Regression and Nonlinear Axes

Linear Regression

An equation of a line is used to establish the relationship

between 2 variables x and y.

y = mx + b

where

y,x = dependent and independent variables, respectively.

m, b = slope and intercept, respectively.

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
3
2 Regression and Nonlinear Axes

Fitting a Straight Line to a Set of Data Points

Consider a rotameter calibration data:

Rotameter Reading Flow Rate (mL/min)

(independent variable) (dependent variable)

10 20.0

30 52.1

50 84.6

70 118.3

90 151.0

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
4
2 Regression and Nonlinear Axes

Fitting a Straight Line to a Set of Data Points

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
5
2 Regression and Nonlinear Axes

Fitting a Straight Line to a Set of Data Points

Calculating the slope and intercept using the method of

least squares:

Δy Sxy - Sx Sy
slope : m= =
Δx Sxx -  Sx 2

SxxSy - SxySx
intercept : b= 2
Sxx -  Sx 

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
6
2 Regression and Nonlinear Axes

Fitting a Straight Line to a Set of Data Points

Calculating the S-values:

1 n
S x (average of x values) =  xi
n i 1
1 n
S y (average of y values) =  y i
n i 1
2 1 n 2
S xx (average of x values) =  xi
n i 1
1 n
S xy (average of xy values) =  xi y i
n i 1

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
7
2 Regression and Nonlinear Axes

Fitting a Straight Line to a Set of Data Points

For the previous data set:

1
Sx = 10  30  50  70  90  = 50
5

1
Sy =  20.0  52.1  84.6  118.3  151.0   85.2
5

1 2
S xx =
5

10  302  502  702  902 = 3300 
1 10  20    30  52.1   50  84.6  
S xy =    5572.8
5   70 118.3    90  151.0  

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
8
2 Regression and Nonlinear Axes

Fitting a Straight Line to a Set of Data Points

Calculating the slope and intercept:

5572.8 -  50  85.2 
m 2
= 1.641
3300 -  50 

b
 3300  85.2  -  5572.8  50 
 3.15
2
3300 -  50 

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
9
2 Regression and Nonlinear Axes

Fitting a Straight Line to a Set of Data Points

Therefore, the mathematical relation for flow rate in terms

of the rotameter reading is:

y = 1.641x + 3.15

At any given value of x, a corresponding value for y can

easily be calculated using this equation.

For x = 25, y = 1.641(25) + 3.15 = 44.175

for x = 250, y = 1.641(250) + 3.15 = 413.40

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
10
2 Regression and Nonlinear Axes

Two-Point Linear Interpolation

Given the following set of data:

x 1.0 2.0 3.0 4.0

y 0.3 0.7 1.2 1.8

Supposed we want to know two values of y when:

1. x = 2.5
(x is between the range of data points: interpolation)

2. x = 5.0
(x is outside the range of data points: extrapolation)

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
11
2 Regression and Nonlinear Axes

Two-Point Linear Interpolation

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
12
2 Regression and Nonlinear Axes

Two-Point Linear Interpolation

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
13
2 Regression and Nonlinear Axes

Two-Point Linear Interpolation

If points 1, 2, and A lie on the same line, then

Slope (1-A) = Slope (1-2)

If slope is defined as y/x, then,

y - y 1 y 2 - y1
=
x - x1 x 2 - x1
Rearranging the equation,

x - x1
y = y1 +  y 2 - y1 
x 2 - x1

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
14
2 Regression and Nonlinear Axes

Two-Point Linear Interpolation

Two-point interpolation can be used when extracting values

from a tabular data.

Consider the following tabular data:

x 1 2 3
y 1 4 8

Find the value of y when x = 1.3.

1.3 - 1
y=1+  4 - 1 = 1.9
2- 1

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
15
2 Regression and Nonlinear Axes

Fitting Nonlinear Data

For simple non-linear equations, linearization can be done

by manipulating or plotting the data so that it can be
expressed analogously as a linear equation.

Original equation: y = mx2 + b

Linearized Form: y = mw + b (with w = x2)

Original equation: sin y = m(x2 – 4) + b

Linearized Form: u = aw + b
(with w = x2 – 4 and u = sin y)

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
16
2 Regression and Nonlinear Axes

Fitting Nonlinear Data

A mass flow rate M(g/s) is measured as a function of

temperature T(0C).

T(0C) 10.0 20.0 40.0 80.0

M 14.76 20.14 27.73 38.47

There is reason to believe that m varies linearly with the

square root of T:

M = aT1/2 + b

Determine the values and units of a and b.

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
17
2 Regression and Nonlinear Axes

Fitting Nonlinear Data

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
18
2 Regression and Nonlinear Axes

Fitting Nonlinear Data

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
19
2 Regression and Nonlinear Axes

Fitting Nonlinear Data

Linearizing the data set:

T 10.0 20.0 40.0 80.0

T1/2 3.162 4.472 6.325 8.944

M 14.76 20.14 27.73 38.47

The values of a (slope) and b (intercept) can be determined

using the method of least squares as discussed previously.

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
20
2 Regression and Nonlinear Axes

Fitting Nonlinear Data

Calculate for SX, SY, SXX, and SXY:

X (T1/2) Y (M) X2 XY

3.162 14.76 10.00 46.68

4.472 20.14 20.00 90.07

6.325 27.73 40.00 175.38

8.944 38.47 80.00 344.09

SX = 5.726 SY = 25.275 SXX = 37.50 SXY = 164.05

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
21
2 Regression and Nonlinear Axes

Fitting a Straight Line to a Set of Data Points

Calculate the slope (a) and intercept (b):

(164.05) - (5.726)(25.275)
slope : a= 2
 4.100
(37.50) -  5.726 

(37.50)(25.275) - (164.05)(5.726)
intercept : b= 2
 1.798
(37.50) -  5.726 

Apply dimensional consistency to get the units of a and b:

a = 4.100 g/(s-0C1/2) and b = 1.798 g/s

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
22
2 Regression and Nonlinear Axes

Nonlinear Axes: The Logarithmic Coordinates

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
23
2 Regression and Nonlinear Axes

Nonlinear Axes: The Logarithmic Coordinates

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
24
2 Regression and Nonlinear Axes

Nonlinear Axes: The Logarithmic Coordinates

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
25
2 Regression and Nonlinear Axes

Nonlinear Axes: The Logarithmic Coordinates

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
26
2 Regression and Nonlinear Axes

Nonlinear Axes: The Logarithmic Coordinates

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
27
2 Regression and Nonlinear Axes

Nonlinear Axes: The Logarithmic Coordinates

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
28
2 Regression and Nonlinear Axes

Nonlinear Axes: The Logarithmic Coordinates

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
29
2 Regression and Nonlinear Axes

Nonlinear Axes: The Logarithmic Coordinates

1. If y versus x data appear linear on a semilog plot, then

ln(y) versus x would be linear on a rectangular plot, and
the data can therefore be correlated by an exponential
function:

y = aebx
2. If y versus x data appear linear on a log plot, then ln(y)
versus ln(x) would be linear on a rectangular plot, and
the data can therefore be correlated by a power
function:

y = axb

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
30
2 Regression and Nonlinear Axes

Nonlinear Axes: The Logarithmic Coordinates

A plot of F versus t yields a straight line that passes

through the points (t1 = 15, F1 = 0.298) and (t2 = 30, F2 = 0.0527)
on a log plot. Determine the mathematical equation that
relates F and t.

If linear on a logarithmic plot, then,

F = atb

Linearizing the equation,

ln (F) = ln (a) + b ln(t)

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
31
2 Regression and Nonlinear Axes

Nonlinear Axes: The Logarithmic Coordinates

Determine a and b:

Δ  ln(F) ln(F2 /F1 ) ln(0.0527/.298)

b= = = = -2.50
Δ  ln(t) ln(t 2 /t1 ) ln(30/15)

To determine a:

ln (a) = ln (F1) – b ln(t1) or ln (a) = ln (F2) – b ln(t2)

Using the first data point,

ln (a) = ln (0.298) – b ln(15) = 5.559

a = exp (5.559) = 260

SLIDE
Prof. Manolito E Bambase Jr. Department of Chemical Engineering. University of the Philippines Los Baños
32