Lecture 13 APPROXIMATION OF SECOMD ORDER DERIVATIVES

Similar documents
Lecture 12 APPROXIMATION OF FIRST ORDER DERIVATIVES

One-sided finite-difference approximations suitable for use with Richardson extrapolation

2.29 Numerical Fluid Mechanics

NUMERICAL DIFFERENTIATION

Chapter 3 Differentiation and Integration

2.29 Numerical Fluid Mechanics Fall 2011 Lecture 12

Numerical Heat and Mass Transfer

Lecture 2: Numerical Methods for Differentiations and Integrations

Solution of Linear System of Equations and Matrix Inversion Gauss Seidel Iteration Method

Polynomial Regression Models

Report on Image warping

Numerical Simulation of One-Dimensional Wave Equation by Non-Polynomial Quintic Spline

Application of B-Spline to Numerical Solution of a System of Singularly Perturbed Problems

Implicit Integration Henyey Method

The KMO Method for Solving Non-homogenous, m th Order Differential Equations

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

Solving Fractional Nonlinear Fredholm Integro-differential Equations via Hybrid of Rationalized Haar Functions

CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE

Advanced Circuits Topics - Part 1 by Dr. Colton (Fall 2017)

Integrals and Invariants of Euler-Lagrange Equations

PART 8. Partial Differential Equations PDEs

Solution for singularly perturbed problems via cubic spline in tension

( ) [ ( k) ( k) ( x) ( ) ( ) ( ) [ ] ξ [ ] [ ] [ ] ( )( ) i ( ) ( )( ) 2! ( ) = ( ) 3 Interpolation. Polynomial Approximation.

Difference Equations

Formal solvers of the RT equation

Feb 14: Spatial analysis of data fields

Convexity preserving interpolation by splines of arbitrary degree

Integrals and Invariants of

Lecture 12: Discrete Laplacian

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

Lecture 5.8 Flux Vector Splitting

Section 8.3 Polar Form of Complex Numbers

2 Finite difference basics

New Method for Solving Poisson Equation. on Irregular Domains

Maejo International Journal of Science and Technology

Canonical transformations

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

Numerical Solution of One-Dimensional Heat and Wave Equation by Non-Polynomial Quintic Spline

The exponential map of GL(N)

Module 3: Element Properties Lecture 1: Natural Coordinates

Restricted divisor sums

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

1 Matrix representations of canonical matrices

Applied Mathematics Letters

Formulas for the Determinant

COMPOSITE BEAM WITH WEAK SHEAR CONNECTION SUBJECTED TO THERMAL LOAD

A new Approach for Solving Linear Ordinary Differential Equations

The Finite Element Method

1 Introduction We consider a class of singularly perturbed two point singular boundary value problems of the form: k x with boundary conditions

1 GSW Iterative Techniques for y = Ax

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

Numerical Methods. ME Mechanical Lab I. Mechanical Engineering ME Lab I

Chapter 4: Root Finding

TR/95 February Splines G. H. BEHFOROOZ* & N. PAPAMICHAEL

Higher Order Wall Boundary Conditions for Incompressible Flow Simulations

Linear Feature Engineering 11

Irregular vibrations in multi-mass discrete-continuous systems torsionally deformed

DUE: WEDS FEB 21ST 2018

12. The Hamilton-Jacobi Equation Michael Fowler

A Local Variational Problem of Second Order for a Class of Optimal Control Problems with Nonsmooth Objective Function

Section 3.6 Complex Zeros

Numerical Solution of Ordinary Differential Equations

ISSN: ISO 9001:2008 Certified International Journal of Engineering and Innovative Technology (IJEIT) Volume 3, Issue 1, July 2013

Expected Value and Variance

= z 20 z n. (k 20) + 4 z k = 4

FTCS Solution to the Heat Equation

1-Dimensional Advection-Diffusion Finite Difference Model Due to a Flow under Propagating Solitary Wave

Finite Differences, Interpolation, and Numerical Differentiation

Lecture 3 Stat102, Spring 2007

PHYS 705: Classical Mechanics. Calculus of Variations II

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X

6.3.4 Modified Euler s method of integration

More metrics on cartesian products

THE STURM-LIOUVILLE EIGENVALUE PROBLEM - A NUMERICAL SOLUTION USING THE CONTROL VOLUME METHOD

THE CHINESE REMAINDER THEOREM. We should thank the Chinese for their wonderful remainder theorem. Glenn Stevens

Inner Product. Euclidean Space. Orthonormal Basis. Orthogonal

Finite Element Modelling of truss/cable structures

STATIC ANALYSIS OF TWO-LAYERED PIEZOELECTRIC BEAMS WITH IMPERFECT SHEAR CONNECTION

Chapter 13: Multiple Regression

The Fundamental Theorem of Algebra. Objective To use the Fundamental Theorem of Algebra to solve polynomial equations with complex solutions

MMA and GCMMA two methods for nonlinear optimization

One Dimensional Axial Deformations

Grid Generation around a Cylinder by Complex Potential Functions

Solution of the Navier-Stokes Equations

Lecture 2 Solution of Nonlinear Equations ( Root Finding Problems )

Combined Wronskian solutions to the 2D Toda molecule equation

Isogeometric Analysis with Geometrically Continuous Functions on Multi-Patch Geometries. Mario Kapl, Florian Buchegger, Michel Bercovier, Bert Jüttler

The optimal delay of the second test is therefore approximately 210 hours earlier than =2.

Bernoulli Numbers and Polynomials

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

By Samuel Schechter. (1) H = i-î -1 with 1 ^ i,j g n. [a, b,j

CALCULUS CLASSROOM CAPSULES

Lecture 10 Support Vector Machines II

ORDINARY DIFFERENTIAL EQUATIONS EULER S METHOD

Physics 4B. A positive value is obtained, so the current is counterclockwise around the circuit.

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

Snce h( q^; q) = hq ~ and h( p^ ; p) = hp, one can wrte ~ h hq hp = hq ~hp ~ (7) the uncertanty relaton for an arbtrary state. The states that mnmze t

Review of Taylor Series. Read Section 1.2

Construction of Serendipity Shape Functions by Geometrical Probability

Cubic Trigonometric B-Spline Applied to Linear Two-Point Boundary Value Problems of Order Two

Transcription:

COMPUTATIONAL FLUID DYNAMICS: FDM: Appromaton of Second Order Dervatves Lecture APPROXIMATION OF SECOMD ORDER DERIVATIVES. APPROXIMATION OF SECOND ORDER DERIVATIVES Second order dervatves appear n dffusve terms of transport/conservaton equatons. To obtan a fnte dfference appromaton of the second dervatve at a pont, we can ether use the appromaton of the frst dervatves or etend any of the technques descrbed n the prevous lecture for the frst order dervatves. We dscuss applcaton of each of these approaches n the sequel.. USE OF APPROXIMATIONS OF FIRST ORDER DERIVATIVE To obtan the second order dervatve at a pont, one may use appromaton of frst order dervatves. For eample, an appromaton of the second order dervatve can be obtaned usng the forward dfference formula f f f (.) We can use a dfferent formula (say, BDS) for the appromaton of the frst order dervatves n the precedng equaton whch results n f f f f f f f f On a unform grd, Eq. (.) reduces to (.) f f f f (.) Appromaton (.) s frst order accurate. A better appromaton for the second order dervatve can be obtaned usng CDS at ponts halfway between nodes,.e. f f f (.) CDS s also used for appromaton of the frst order dervatves n Eq. (.),.e. f f f f f f and (.5) Dr K M Sngh, Indan Insttute of Technology Roorkee NPTEL.

COMPUTATIONAL FLUID DYNAMICS: FDM: Appromaton of Second Order Dervatves Substtuton of Eq. (.5) nto Eq. (.) yelds the followng formula for the second order dervatve f f f f (.6) On a unform grd, Eq. (.6) reduces to Eq. (.).. TAYLOR SERIES EXPANSION Usng Taylor seres epanson about, functon values at grd ponts and can be epressed as f f f f f H (.7)!! f f f f f H (.8)!! Multply Eq. (.8) by and add t to Eq. (.7) to elmnate the frst order dervatve. Rearrangement of the resultng equaton leads to the followng relaton for the second order dervatve: f f f f (.9) f H whch s dentcal to Eq. (.6) obtaned usng CDS. The leadng term n the truncaton error of precedng appromaton s formally of frst order on a non-unform grd, and of second order on unform grds. However, even for a non-unform grd, the truncaton error s reduced as n a second order scheme wth the grd refnement (Ferzger and Perc, 00). A general procedure on unform grds On unform grds, we can defne the dfference appromaton for the second order dervatve as (Chung, 00) f af bf cfdf ef... (.0) Coeffcents a, b, c, d, e,. can be determned from Taylor seres epanson for the functon values nvolved at the RHS around pont. Usng ths approach, the three pont central dfference formula (.) can be derved startng wth the relaton Dr K M Sngh, Indan Insttute of Technology Roorkee NPTEL.

COMPUTATIONAL FLUID DYNAMICS: FDM: Appromaton of Second Order Dervatves f af bf cf (.) Taylor seres epanson for f and f lead to the followng relaton: af bf cf ( abc) ( cb) f ( bc) f f (.) f f ( cb) ( b c)... 6 From the precedng equaton, t s clear that Eq. (.) wll represent an appromaton for the second order dervatve f and only f a bc0, b c 0and bc. Clearly, bcand a. Thus, we obtan the followng central dfference appromaton f f f f f (.) whch s second order accurate. Smlarly, we can derve the followng one-sded formula f f f f f (.) whch s only frst order accurate (Chung, 00). Further, usng ths approach, we can easly derve a hgher order central dfference appromaton usng functon values at fve ponts (for detals, see Eample. below) gven by f 0 f 6( f f ) ( f f) O (.5) Eample. Derve a fve pont central dfference formula for the second order dervatve on unform grd usng Taylor seres epanson and Eq. (.0). Soluton Fve pont central dfference formula for the second order dervatve can epressed as f af bf cf df ef Taylor seres epansons for f, f, f and f, we get a b c d e b c d e af bf cf df ef f f f f bcd e bc8d 8e 6 0 5 f f bc6d 6e bcd e 5 6 f bc6d 6e H 6 70 () () Dr K M Sngh, Indan Insttute of Technology Roorkee NPTEL.

COMPUTATIONAL FLUID DYNAMICS: FDM: Appromaton of Second Order Dervatves Equatons () and (v) ndcate that the coeffcents n Eq. () must satsfy the followng condtons: abcde 0 () bcd e 0 (v) bcd e (v) bc8d 8e 0 (v) bc6d 6e (v) Solvng Eqs. (v)-(v), we get a5 /, bc /, d e / (v) The truncaton error (TE) s gven by 6 6 f f bc6d6e 6 6 70 90 Therefore, the desred fve-pont central dfference formula s f 0 f 6( f f ) ( f f), TE~ O () (). POLYNOMIAL FITTING For any functon, an nterpolatng polynomal of degree n can be ft usng functon values at ( n ) data ponts and appromaton to all dervatves up to order n can be obtaned by dfferentaton. Use of quadratc nterpolaton loads to three pont CDS formula (.) or (.) for the second order dervatve. In general, truncaton error of the appromaton to second order dervatve obtaned by fttng a polynomal of degree n s of order ( n ). One order s ganed (.e., truncaton error s of order n) f grd spacng s unform and an even-order polynomals s used. For eample, polynomal of degree on unform grd leads to the fourth order accurate formula gven by Eq. (.5). Eample. Derve Eq. (.5) by fttng a polynomal of degree on unform grd. Also obtan the correspondng appromatons for the frst, thrd and fourth order dervatves. Soluton To obtan a central dfference appromaton, we can ft the followng th degree polynomal through fve ponts,,, and : 0 f a a a a a () Formal dfferentaton of the precedng equaton leads to the followng relatons: f f f f a, a, 6 a, a () To obtan the values of coeffcents a, we ft the nterpolaton curve () to the functon values at ponts,,, and, whch results n the followng set of lnear equatons: Dr K M Sngh, Indan Insttute of Technology Roorkee NPTEL.

COMPUTATIONAL FLUID DYNAMICS: FDM: Appromaton of Second Order Dervatves f a0 () f f a a a a (v) f f a a a a (v) f f a a a a (v) f f a a a a (v) Soluton of smultaneous Eqs. (v)-(v) yelds the followng values for coeffcents a : f 8 f 8 f f f 6 f 0 f 6 f f a, a f f f f f f 6f f f a, a (v) Thus, appromatons for the dervatves obtaned from two-sded th fttng on a unform grd are f a f 8f 8f f f f 6 f 0 f 6 f f a f f f f f 6a f f f 6f f f a degree polynomal () () () () Equaton () s the same as Eq. (.5) derved earler usng Taylor seres epanson..5 APPROXIMATION OF SECOND ORDER DERIVATIVE IN GENERIC TRANSPORT EQUATION The dffuson term n the generc conservaton equaton nvolves a second order dervatve of the form ( / )/. If Γ s constant, ths term becomes ( / ) and fnte dfference appromatons n the precedng secton can be used. Otherwse, we have to employ sutable fnte dfference appromatons for the frst order dervatves for nner and outer dervatves. The most popular approach s to employ the central dfference appromaton for both the nner and outer dervatves. Let us use the values of the nner frst order dervatve at ponts md-way between the nodes and central dfference formula to obtan the appromaton for the outer dervatve gven by (.6) Dr K M Sngh, Indan Insttute of Technology Roorkee NPTEL.5

COMPUTATIONAL FLUID DYNAMICS: FDM: Appromaton of Second Order Dervatves Central dfference appromaton of the frst order dervatves on RHS of the precedng equaton are and Combnng Eq. (.6) and Eq. (.7), we get On a unform grd, the precedng equaton smplfes to ( ) (.7) (.8) (.9) Note that f s a functon of, values at a pont mdway between the nodes can be evaluated usng smple average of functon values at the neghbourng nodes,.e., ( ) (.0) Other appromatons can be easly obtaned usng dfferent fnte dfference appromatons for the nner and outer dervatves. REFERENCES Chung, T. J. (00). Computatonal Flud Dynamcs. nd Ed., Cambrdge Unversty Press, Cambrdge, UK. Ferzger, J. H. and Perć, M. (00). Computatonal Methods for Flud Dynamcs. Sprnger. Dr K M Sngh, Indan Insttute of Technology Roorkee NPTEL.6