Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

CHE 411 Lesson 7 Note

Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 38

⮚ NONLINEAR

LINEARIZATION OF MODELS
NONLINEAR ⮚ STATE-SPACE
MODELS
MODELS: THE REPRESENTATION

STATE-SPACE ⮚ LINEARIZATION
PROCESS
FORMULATION
◼Nonlinear Models: Example 15 – The
three Interacting Cascade Tanks
◼State-Space Models (SSM)
Representation
◼Example 16: SSM Example - Non-
CONTENTS Interacting Tanks
◼Linearization Process
✔Single-variable
✔One state one input
✔Multiple State multiple Input
◼ Consider 3 Interacting Cascade Tanks below
with each tank interacting with one another

NONLINEAR
MODELS
◼ The outflows F1, F2 and F3 from Tank 1, 2 and
3 are related to the height of the cascaded
tanks in the following ways:

NONLINEAR (104)
MODELS

(105)

(106)
◼ Writing mass balances for each of the tank, we
have
Tank 1

NONLINEAR
MODELS
(107)

Tank 2
(108)

Tank 3
NONLINEAR
MODELS

(109)
◼ Equations (107), (108) and (109) represent the
dynamic model equations for the 3-cascade tanks
◼ They are all nonlinear ordinary differential
equations (since each involves the dependent
variable raised to power other than 1(0.5 in this
case)
NONLINEAR ◼ Most of chemical process models are nonlinear,
MODELS but they are often linearized to perform some
analysis such as stability, robustness, etc. These
analysis involves the use of: Laplace transforms,
State-space models, Bode-plot, Nyquist plot,
Transfer-function analysis, etc., but can all be
implemented with linear systems.
◼ Hence, the need for SYSTEM LINEARIZATION
◼ Most practical system dynamic models are of the
general form:
(110)

◼ Where f(x, u) is, in general, a nonlinear function


STATE-SPACE vector.
MODEL ◼ A linear model is a subset of the more general
FORMULATION modeling equation (110).
◼ The form of linear model that we discuss in this
section is known as a state-space model.
◼ First we show how to write state-space models for
systems that are inherently linear.
◼ Then we show how to approximate nonlinear systems
with linear models.
◼ Model Example 16 illustrates the form of a
state-space model.

STATE-SPACE ◼ Consider 2 tanks in series where the flow out


of the first tank enters the second tank (see
MODEL
Figure below).
FORMULATION:
◼ Our objective is to develop a model to describe
how the height of liquid in tank 2 changes with
EXAMPLE 16 – time, given the input flowrate F0(t).
NONINTERACTING
◼ We assume that the flow out of each tank is a
TANKS
linear function of the height of liquid in the
tank.
STATE-SPACE
MODEL
FORMULATION:
◼ A material balance around the first tank yields
EXAMPLE 16 – (assume a constant density and ):
NONINTERACTING
TANKS (111)

◼ where A1 is the const x-sectional area for tank 1


◼ β1= flow coefficient
◼Writing a material balance around the second
tank ( since ) we have:
STATE-SPACE (112)
MODEL
FORMULATION: ◼where A2 is the const x-sectional area for tank2,
β2= flow coefficient
EXAMPLE 16 –
◼Substituting for F1 in (112), we obtain
NONINTERACTING
TANKS (113)

◼We can write both (111) and (113) in the


following matrix form:
(114)

STATE-SPACE
MODEL which has the general form:
FORMULATION: (115)
CONTINUED
where:
◼ The state and input vectors are (notice that the
input is a scalar):

◼ The additional equation that is normally


STATE-SPACE
associated with a state-space model is
MODEL
(116)
FORMULATION:
CONTINUED where y is a vector of output variables.
◼ Generally, output variables are variables that can
be measured (at least conceptually) or are of
particular interest in a simulation study.
◼ Here, we will consider the case where both tank
heights are outputs
◼ So let:

◼ The matrix-vector form is:


STATE-SPACE
MODEL
FORMULATION: where
CONTINUED

◼ If we also consider the input, F0, to be the third


output variable, we have the following
relationship
◼ Which is the form of (116) with
STATE-SPACE
MODEL
FORMULATION:
CONTINUED

◼ If we also consider the input, F0, to be the


third output variable, we have the following
relationship
◼ Model Example 16 illustrated a specific case
of a state-space model. In general, a state-
space model has the following form:

GENERAL FORM OF
STATE SPACE
MODELS

◼ Which has n state variables (x), m input


variables (u) and r output variables (y).
◼ The relationship is normally written in the
matrix form:

GENERAL FORM OF
STATE SPACE
MODELS

◼ Which has the general state-space form:

(117)
◼ Linear systems are easier to understand (than
nonlinear models) and are necessary for most
common control system design methods.
◼ A general single variable nonlinear model is:

LINEARIZATION
(118)
OF NONLINEAR
SYSTEMS – A ◼ The function of a single variable, f(x), can be
SINGLE VARIABLE approximated by a truncated Taylor series
approximation around the steady-state operating
SYSTEM
point (xs):
(119)

◼ Neglecting the quadratic and higher terms, we


obtain
(120)
◼Note that (from eqn. (118)):

LINEARIZATION (121)
OF NONLINEAR
◼by definition of a steady-state, so:
SYSTEMS – A
SINGLE VARIABLE (122)
SYSTEM
◼Where the notation is used to indicate the
partial derivative of f(x) with respect x, evaluated
at the steady state. Since the derivative of a
constant (xs) is zero, we can write:
(123)
◼which leads to:
(124)
LINEARIZATION
OF NONLINEAR
◼The reason for using the expression above is that
SYSTEMS – A we are often interested in deviations in a state
SINGLE VARIABLE from a steady-state operating point.
SYSTEM ◼Sometimes the ‘ symbol is used to represent
deviation variable, .
◼We can see that a deviation variable represents
the change or perturbation (deviation) from a
steady-state value, i.e.
(125)
◼This can be written in a state-space form:

LINEARIZATION (126)
OF NONLINEAR
SYSTEMS – A ◼where
SINGLE VARIABLE
◼We have shown how to linearize a single
SYSTEM variable equation. Next, we consider a system
with one state and one input.
◼Similarly, consider a function with one state
variable and one input variable
(127)
◼Using a Taylor series expansion for f(x, u):
LINEARIZATION
OF NONLINEAR
SYSTEMS – ONE
STATE AND ONE
INPUT SYSTEM

◼Truncating after the linear terms, we have


(128)

LINEARIZATION
OF NONLINEAR ◼Realizing that
SYSTEMS – ONE
STATE AND ONE
INPUT SYSTEM
◼Using deviation variable:
(129)

◼Which can be written as


LINEARIZATION
OF NONLINEAR (130)
SYSTEMS – ONE
STATE AND ONE ◼Where
INPUT SYSTEM

◼If there is a single output that is a function of


states and inputs, then:
(131)
◼Again, performing a Taylor series expansion for
g(x, u) and truncating quadratic and higher
terms:

LINEARIZATION (132)
OF NONLINEAR
SYSTEMS – ONE
◼Since g(xs, us) is simply the steady state value of
STATE AND ONE
the output (ys), we can write:
INPUT SYSTEM
(133)

◼or
(133)
◼Where

LINEARIZATION ◼Using deviation notation:


OF NONLINEAR (134)
SYSTEMS – ONE
STATE AND ONE ◼Example 17 illustrates the application of
INPUT SYSTEM LINEARIZATION to a one-input, one-state
nonlinear system
◼Consider a Nonlinear Tank Height Problem

(135)

LINEARIZATION ◼Where h is the state variable, F0 is the input


OF NONLINEAR variable, β and A are parameters.
SYSTEMS. ◼From (135), the right hand side is defined as the
function
EXAMPLE 17: AN
(136)
APPLICATION OF ◼Using Taylor’s Series expansion for (136):
LINEARIZATION I
◼Noting that:

LINEARIZATION
OF NONLINEAR
◼Using the deviation variable:
SYSTEMS.
(137)
EXAMPLE 17: AN
APPLICATION OF ◼where
LINEARIZATION I
(138a)

(138b)
◼Substituting we have:

(139)
LINEARIZATION
OF NONLINEAR
◼Using state-space variable notation:
SYSTEMS.
(140)
EXAMPLE 17: AN
APPLICATION OF ◼In state-space form, (140) is written as
LINEARIZATION I
(141)
◼Or simply as
(142)
◼Refer to the models of nonlinear interacting
tanks in Equations (107), (108) and (109)
(143)
LINEARIZATION (144)
OF NONLINEAR
SYSTEMS.

EXAMPLE 18: AN (145)


APPLICATION OF
LINEARIZATION II ◼We assume that only the heights of Tank 1 and 2
are measured.
◼Then note that there are 3 state variables, one
input variable and 3 output variables
◼The output, in deviation variable form is:

(146)
LINEARIZATION ◼Let
OF NONLINEAR (147)
SYSTEMS.
(148)
EXAMPLE 18: AN ◼Using Eqn. (143), the linearization gives
APPLICATION OF
LINEARIZATION II
(149)
◼Linearization of Eqn.(144) also gives:

LINEARIZATION
(150)
OF NONLINEAR
SYSTEMS.
◼And linearizing Eqn. (145), we have
EXAMPLE 18: AN
APPLICATION OF
LINEARIZATION II
(151)
◼Recall that f1, f2 and f3 are respectively:

(152)

LINEARIZATION
OF NONLINEAR (153)
SYSTEMS.
(154)
EXAMPLE 18: AN
◼Using Eqn. (152) in Eqn. (149), we obtain:
APPLICATION OF
LINEARIZATION II
◼ ,

LINEARIZATION ◼Using Eqn. (153) in Eqn. (150), we have:


OF NONLINEAR
SYSTEMS.

EXAMPLE 18: AN
APPLICATION OF
LINEARIZATION II
,
◼Employing Eqn. (154) in Eqn. (151), we have:

LINEARIZATION
OF NONLINEAR
SYSTEMS.

EXAMPLE 18: AN
APPLICATION OF
LINEARIZATION II
◼Substituting the obtained steady-state
coefficients into the respective eqns. (149), (150)
and (151), we obtain
(155)

LINEARIZATION
OF NONLINEAR
SYSTEMS.
(156)
EXAMPLE 18: AN
APPLICATION OF
LINEARIZATION II
(157)

◼ Eqns. (155) – (157) constitute state-space equations


for the nonlinear model
◼In matrix form, using the notations defined in
(146)-(148) we write:

LINEARIZATION (158)
OF NONLINEAR
SYSTEMS.

EXAMPLE 18: AN ◼
In a more closed-form, Eqn. (158) is written as
APPLICATION OF
LINEARIZATION II (159)

◼Where matrices A and B are as obtained through


the steady-state coefficient evaluation earlier.
◼ In this Section nonlinear models are derived as
addendum to hitherto developed models (which
were all linear equations).
◼ Most practical systems in Chemical Engineering
are nonlinear.
◼ Engineers usually prefer linear systems because
NOTES they can easily be analyzed using most of the
available methods.
◼ Therefore, linearization of nonlinear systems
becomes necessary and procedures were derived.
◼ State-space formulations were developed for
different categories of systems with examples

You might also like