Transfer Functions. Convenient representation of a linear, dynamic model. A transfer function (TF) relates one input and one output: ( ) system

Similar documents
Note 10. Modeling and Simulation of Dynamic Systems

Numerical Heat and Mass Transfer

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE

( ) ( = ) = ( ) ( ) ( )

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix

Implicit Integration Henyey Method

APPENDIX A Some Linear Algebra

NUMERICAL DIFFERENTIATION

Lecture 12: Discrete Laplacian

Appendix B. The Finite Difference Scheme

Difference Equations

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4)

ELASTIC WAVE PROPAGATION IN A CONTINUOUS MEDIUM

Digital Signal Processing

PHYS 705: Classical Mechanics. Calculus of Variations II

Salmon: Lectures on partial differential equations. Consider the general linear, second-order PDE in the form. ,x 2

Inner Product. Euclidean Space. Orthonormal Basis. Orthogonal

2.3 Nilpotent endomorphisms

PHYS 705: Classical Mechanics. Canonical Transformation II

Lecture 21: Numerical methods for pricing American type derivatives

8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS

2 Finite difference basics

Norms, Condition Numbers, Eigenvalues and Eigenvectors

EEE 241: Linear Systems

Chapter 12. Ordinary Differential Equation Boundary Value (BV) Problems

Introduction to Vapor/Liquid Equilibrium, part 2. Raoult s Law:

LECTURE 9 CANONICAL CORRELATION ANALYSIS

Supplementary Notes for Chapter 9 Mixture Thermodynamics

n α j x j = 0 j=1 has a nontrivial solution. Here A is the n k matrix whose jth column is the vector for all t j=0

Process Modeling. Improving or understanding chemical process operation is a major objective for developing a dynamic process model

Math1110 (Spring 2009) Prelim 3 - Solutions

Finite Element Modelling of truss/cable structures

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems

Global Sensitivity. Tuesday 20 th February, 2018

Modeling of Dynamic Systems

Trees and Order Conditions

Canonical transformations

Week 5: Neural Networks

Week3, Chapter 4. Position and Displacement. Motion in Two Dimensions. Instantaneous Velocity. Average Velocity

November 5, 2002 SE 180: Earthquake Engineering SE 180. Final Project

Physics 181. Particle Systems

Lecture 13 APPROXIMATION OF SECOMD ORDER DERIVATIVES

Structure and Drive Paul A. Jensen Copyright July 20, 2003

A new Approach for Solving Linear Ordinary Differential Equations

Solution Thermodynamics

Open Systems: Chemical Potential and Partial Molar Quantities Chemical Potential

Linear Regression Analysis: Terminology and Notation

Quantum Mechanics for Scientists and Engineers. David Miller

The Multiple Classical Linear Regression Model (CLRM): Specification and Assumptions. 1. Introduction

Energy Storage Elements: Capacitors and Inductors

U.C. Berkeley CS294: Beyond Worst-Case Analysis Luca Trevisan September 5, 2017

Modelli Clamfim Equazioni differenziali 7 ottobre 2013

MAE140 - Linear Circuits - Winter 16 Final, March 16, 2016

Numerical Solution of Ordinary Differential Equations

COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS

MEM 255 Introduction to Control Systems Review: Basics of Linear Algebra

Hidden Markov Models & The Multivariate Gaussian (10/26/04)

Module 3: Element Properties Lecture 1: Natural Coordinates

Module 1 : The equation of continuity. Lecture 1: Equation of Continuity

NON-CENTRAL 7-POINT FORMULA IN THE METHOD OF LINES FOR PARABOLIC AND BURGERS' EQUATIONS

Linear Approximation with Regularization and Moving Least Squares

College of Computer & Information Science Fall 2009 Northeastern University 20 October 2009

Inductance Calculation for Conductors of Arbitrary Shape

Fundamental loop-current method using virtual voltage sources technique for special cases

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography

Consistency & Convergence

Lecture 10 Support Vector Machines II

The Feynman path integral

Chapter 8. Potential Energy and Conservation of Energy

Modelli Clamfim Equazione del Calore Lezione ottobre 2014

Estimation of the composition of the liquid and vapor streams exiting a flash unit with a supercritical component

SIO 224. m(r) =(ρ(r),k s (r),µ(r))

Bernoulli Numbers and Polynomials

MAE140 - Linear Circuits - Winter 16 Midterm, February 5

8.6 The Complex Number System

Modelli Clamfim Equazioni differenziali 22 settembre 2016

Expected Value and Variance

Errors for Linear Systems

Linearity. If kx is applied to the element, the output must be ky. kx ky. 2. additivity property. x 1 y 1, x 2 y 2

SINGLE OUTPUT DEPENDENT QUADRATIC OBSERVABILITY NORMAL FORM

APPENDIX 2 FITTING A STRAIGHT LINE TO OBSERVATIONS

MAE140 - Linear Circuits - Fall 10 Midterm, October 28

Operating conditions of a mine fan under conditions of variable resistance

Section 8.3 Polar Form of Complex Numbers

Handout # 6 (MEEN 617) Numerical Integration to Find Time Response of SDOF mechanical system. and write EOM (1) as two first-order Eqs.

Spin-rotation coupling of the angularly accelerated rigid body

Prof. Dr. I. Nasser Phys 630, T Aug-15 One_dimensional_Ising_Model

1 Derivation of Rate Equations from Single-Cell Conductance (Hodgkin-Huxley-like) Equations

6) Derivatives, gradients and Hessian matrices

Thermal-Fluids I. Chapter 18 Transient heat conduction. Dr. Primal Fernando Ph: (850)

COEFFICIENT DIAGRAM: A NOVEL TOOL IN POLYNOMIAL CONTROLLER DESIGN

DETERMINATION OF TEMPERATURE DISTRIBUTION FOR ANNULAR FINS WITH TEMPERATURE DEPENDENT THERMAL CONDUCTIVITY BY HPM

A Hybrid Variational Iteration Method for Blasius Equation

Adiabatic Sorption of Ammonia-Water System and Depicting in p-t-x Diagram

9 Derivation of Rate Equations from Single-Cell Conductance (Hodgkin-Huxley-like) Equations

Module 2. Random Processes. Version 2 ECE IIT, Kharagpur

Numerical Transient Heat Conduction Experiment

Chapter Newton s Method

MATH 5630: Discrete Time-Space Model Hung Phan, UMass Lowell March 1, 2018

Lecture 3. Ax x i a i. i i

Transcription:

Transfer Functons Convenent representaton of a lnear, dynamc model. A transfer functon (TF) relates one nput and one output: x t X s y t system Y s The followng termnology s used: x y nput output forcng functon response cause effect 1

Defnton of the transfer functon: Let G(s) denote the transfer functon between an nput, x, and an output, y. Then, by defnton where: G s Y s X s L y( t) Y s L x( t) X s 2

Development of Transfer Functons Example: Strred Tank Heatng System Fgure 2.3 Strred-tank heatng process wth constant holdup, V. 3

Recall the prevous dynamc model, assumng constant lqud holdup and flow rates: dt Vρ C = wc T T + Q dt (1) Suppose the process s ntally at steady state: ( 0 ) =, ( 0 ) =, ( 0) = ( 2) T T T T Q Q where T steady-state value of T, etc. For steady-state condtons: 0 = wc T T + Q (3) Subtract (3) from (1): dt V ρ C = wc T T T T Q Q dt + (4) 4

But, dt dt = ( T) because s a constant (5) d T dt T Thus we can substtute nto (4-2) to get, dt VρC = wc T T + Q dt ( ) (6) where we have ntroduced the followng devaton varables, also called perturbaton varables : T T T, T T T, Q Q Q (7) Take L of (6): ( = ) = Vρ C st s T t 0 wc T s T s Q s (8) 5

Evaluate T ( t = ) 0. By defnton, T T T. Thus at tme, t = 0, T 0 = T 0 T (9) But snce our assumed ntal condton was that the process was ntally at steady state,.e., T( 0) = Tt follows from (9) that T 0 = 0. Note: The advantage of usng devaton varables s that the ntal condton term becomes zero. Ths smplfes the later analyss. Rearrange (8) to solve for T ( s) : K 1 T s Q s T s τs+ 1 τs+ 1 = + (10) 6

where two new symbols are defned: 1 V ρ K and τ 11 wc w Transfer Functon Between Q and T Suppose T s constant at the steady-state value. Then, T() t = T T ( t) = 0 T ( s) = 0. Then we can substtute nto (10) and rearrange to get the desred TF: T s K = Q s τ s+ 1 (12) 7

Transfer Functon Between and T T : Suppose that Q s constant at ts steady-state value: Qt = Q Q t = 0 Q s = 0 Thus, rearrangng Comments: ( s) T 1 = T s τ s+ 1 (13) 1. The TFs n (12) and (13) show the ndvdual effects of Q and on T. What about smultaneous changes n both Q and? T T 8

Answer: See (10). The same TFs are vald for smultaneous changes. Note that (10) shows that the effects of changes n both Q and T are addtve. Ths always occurs for lnear, dynamc models (lke TFs) because the Prncple of Superposton s vald. 2. The TF model enables us to determne the output response to any change n an nput. 3. Use devaton varables to elmnate ntal condtons for TF models. 9

Propertes of Transfer Functon Models 1. Steady-State Gan The steady-state of a TF can be used to calculate the steadystate change n an output due to a steady-state change n the nput. For example, suppose we know two steady states for an nput, u, and an output, y. Then we can calculate the steadystate gan, K, from: y y K = u u 2 1 2 1 (4-38) For a lnear system, K s a constant. But for a nonlnear u, y. system, K wll depend on the operatng condton 10

Calculaton of K from the TF Model: If a TF model has a steady-state gan, then: K = s 0 lm G s (14) Ths mportant result s a consequence of the Fnal Value Theorem Note: Some TF models do not have a steady-state gan (e.g., ntegratng process n Ch. 5) 11

2. Order of a TF Model Consder a general n-th order, lnear ODE: n n 1 m d y dy dy d u an + a n n 1 + a n 1 1 + a0y = bm + m dt dt dt dt m 1 d u du bm 1 + + b 1 1 + b0u (4-39) m dt dt Take L, assumng the ntal condtons are all zero. Rearrangng gves the TF: G s Y s = = U s m = 0 n = 0 bs as (4-40) 12

Defnton: The order of the TF s defned to be the order of the denomnator polynomal. Note: The order of the TF s equal to the order of the ODE. Physcal Realzablty: For any physcal system, n mn (4-38). Otherwse, the system response to a step nput wll be an mpulse. Ths can t happen. Example: ay 0 = b du 1 + bu 0 and step change n u (4-41) dt 13

3. Addtve Property Suppose that an output s nfluenced by two nputs and that the transfer functons are known: Y( s) 1 and 2 Y s = G s = G s U s U s 1 2 Then the response to changes n both U1 and U2can be wrtten as: = + Y s G s U s G s U s 1 1 2 2 The graphcal representaton (or block dagram) s: U 1 (s) U 2 (s) G 1 (s) G 2 (s) Y(s) 14

4. Multplcatve Property Suppose that, Y s U2 s = G2( s) and U2 s U3 s = G3( s) Then, Y s = G s U s and U s = G s U s Substtute, Or, Y s 2 2 2 3 3 = Y s G s G s U s 2 3 3 2 3 3 2 3 U3 s = G s G s U s G s G s Y s 15

Lnearzaton of Nonlnear Models So far, we have emphaszed lnear models whch can be transformed nto TF models. But most physcal processes and physcal models are nonlnear. - But over a small range of operatng condtons, the behavor may be approxmately lnear. - Conclude: Lnear approxmatons can be useful, especally for purpose of analyss. Approxmate lnear models can be obtaned analytcally by a method called lnearzaton. It s based on a Taylor Seres Expanson of a nonlnear functon about a specfed operatng pont. 16

Lnearzaton (contnued) Consder a nonlnear, dynamc model relatng two process varables, u and y: dy dt = f ( y u), (4-60) Perform a Taylor Seres Expanson about u = u and y = y and truncate after the frst order terms, f f f ( u, y) = f ( u, y) + u + y (4-61) u y s s where u = u u, y = y y, and subscpt s denotes the steady state, ( u, y). Note that the partal dervatve terms are actually constants because they have been evaluated at the nomnal operatng pont, s. 17

Lnearzaton (contnued) Substtute (4-61) nto (4-60) gves: dy f f = f ( u, y) + u + y (A) dt u y s Because u, y s a steady state, t follows from (4-60) that s f ( u, y) = 0 (B) Also, because y y- y, t follows that, dy dy = (C) dt dt Substtute (B) and (C) nto (A) gves the lnearzed model: dy f f = u + y dt u y s s (4-62) 18

q Example: Lqud Storage System dh Mass balance: A q q (1) dt = Valve relaton: q = Cv h (2) h A = area, C v = constant Combne (1) and (2) and rearrange: dh 1 Cv = q h = f( h, q) (3) dt A A Eq. (3) s n the form of (4-60) wth y = h and u = q. Thus, we can utlze (4-62) to lnearze around h= h and q = q, q dh f f = h + dt h s q s q (4) 19

where: f = h s C v 2A h (5) f q s 1 = A Substtute nto (4) gves the lnearzed dh 1 Cv = q h dt A 2A h model: (6) (7) Ths model can be expressed n terms of a valve resstance, R dh 1 1 2 h = q h wth R (8) dt A AR C v 20

State-Space Models Dynamc models derved from physcal prncples typcally consst of one or more ordnary dfferental equatons (ODEs). In ths secton, we consder a general class of ODE models referred to as state-space models. Consder standard form for a lnear state-space model, x = Ax + Bu + Ed (4-90) y=cx (4-91) 21

where: x = u = the state vector the control vector of manpulated varables (also called control varables) d = y = the dsturbance vector the output vector of measured varables. (We use boldface symbols to denote vector and matrces, and plan text to represent scalars.) The elements of x are referred to as state varables. The elements of y are typcally a subset of x, namely, the state varables that are measured. In general, x, u, d, and y are functons of tme. The tme dervatve of x s denoted by ( = x t) Matrces A, B, C, and E are constant matrces. x d /d. 22

Example: CSTR Model Consder the prevous CSTR model. Assume that T c can vary wth tme whle c A, T, q and w are constant. Nonlnear Model: Lnearzed Model: where: 23

Example 4.9 Show that the lnearzed CSTR model of Example 4.8 can be wrtten n the state-space form of Eqs. 4-90 and 4-91. Derve state-space models for two cases: (a) Both c A and T are measured. (b) Only T s measured. Soluton The lnearzed CSTR model n Eqs. 4-84 and 4-85 can be wrtten n vector-matrx form: dc A a11 a12 c A 0 dt = + Tc dt a21 a22 T b2 dt (4-92) 24

Let x1 c A and x2 T, and denote ther tme dervatves by x 1 and x 2. Suppose that the coolant temperature T c can be manpulated. For ths stuaton, there s a scalar control varable, u T c, and no modeled dsturbance. Substtutng these defntons nto (4-92) gves, x 1 a11 a12 x1 0 u x = 2 a21 a + 22 x 2 b 2 A B (4-93) whch s n the form of Eq. 4-90 wth x = col [x 1, x 2 ]. (The symbol col denotes a column vector.) 25

a) If both T and c A are measured, then y = x, and C = I n Eq. 4-91, where I denotes the 2x2 dentty matrx. A and B are defned n (4-93). b) When only T s measured, output vector y s a scalar, y = T and C s a row vector, C = [0,1]. Note that the state-space model for Example 4.9 has d = 0 because dsturbance varables were not ncluded n (4-92). By contrast, suppose that the feed composton and feed temperature are consdered to be dsturbance varables n the orgnal nonlnear CSTR model n Eqs. 2-60 and 2-64. Then the lnearzed model would nclude two addtonal devaton varables, c A and T. 26