DIFFERENTIAL EQUATIONS OF MOTION OF A CONSTRAINED PARTICLE Kalju Kenk kkenk@staff.ttu.ee Department of Mechatronics Tallinn University of Technology Introduction To solve the problems of a constrained particle dynamics the mixed systems of differential and algebraic equations (differential-algebraic equations – DAE) have been usually used. But many mathematical difficulties arise by direct solving of DAEs. If the index of DAE is greater than one, the index reduction operations and the changes of the coordinates are recommended [1 ]. Often the singularities arise. In this paper the differential equations of motion in Cartesian coordinates are derived to solve the problems of dynamics of a particle having the holonomic and ideal constraints. Any singularity does not arise. To become the solutions the Matlab ODE solvers can be direct used. The motion of the particles on the frictionless cylindrical surface and on the frictionless parabolic curve are examined for example. Differential equations of motion Let the constraints of the particle are holonomic and workless. It is wellknown, that the dynamics of this particle can be examined by use of the equations (1) mx1 F1a F1c mx2 F2a F2c (2) mx3 F3a F3c 1 x1 , x 2 , x 3 , t 0 2 x1 , x 2 , x 3 , t 0 (3) (4) (5) where m - mass of the particle xi - Cartesian coordinates i 1,2,3 in an inertial frame xi - second time derivate of xi Fia - component on the i - axis of the resultant of the active forces Fic - component on the i - axis of the resultant of the constraint forces 0 - scalar constraints conditions 1,2 The equations from (1) to (5) constitute a DAE. The condition 0 represent a surface and consider the workless of the constraints 1 (6) xi where - respective scalar Lagrange factor. The equations from (1) to (3) can yet be rewritten in form (7) mxi Fia xi and the repetition of the index means the summation from 1 to 2. The equations (7) consist the unknown factors 1 and 2 . Does not exist any differential equation to calculate ones. Thus is suitable to eliminate them. The second time derivates of the equations from (4) to (5) give respectively 1 2 1 2 1 2 1 xi x i x j 2 x i 0 (8) x i x i x j x i t t 2 Fci 2 2 2 2 2 2 2 xi x i x j 2 x i 0 x i x i x j x i t t 2 where the repetition of the indices i and j means the summation from 1 to 3. From equations (7) Fia xi m m x i The substitution of (10) into (8) and (9) gives 1 1 1 2 1 2 f1 m xi xi m xi xi 1 1 2 2 2 2 f2 m xi xi m xi xi where Fia 1 21 21 21 f1 x x 2 xi 2 m xi xi x j i j xi t t F a 2 2 2 2 2 2 2 f 2 i x i x j 2 x i m x i x i x j x i t t 2 From (11) and (12) 1 D1 m D 2 D2 m D where 2 2 1 2 D1 f 1 f2 xi xi xi xi (9) (10) (11) (12) (13) (14) (15) (16) (17) 2 D2 f 2 1 1 1 2 f1 xi xi xi xi (18) 2 1 1 2 2 1 2 (19) D x i x i x j x j x i x i Thus the differential equations of motion (10) can be rewritten in form F a D xi i (20) m D x i If the DOF of the particle is equal to one, i.e. both 1 and 2 do not be identical to zero, only one of equations (20) is independent. Regardless of this, these equations can be numerically solved by use for example the ODE solvers of Matlab. If the DOF of the particle is equal to two, i.e. for example 2 is identical to zero, two of equations (20) are independent. In this case the equations (20) can be written as Fa f1 1 xi i (21) 1 1 x i m x j x j and again direct solved by ODE solvers. Example 1. The particle of mass m is constrained to move on the smooth cylindrical surface given by equation 1 x12 x 22 r 2 0 where r is the radius of the surface and the active force is the gravitation given by equations F1 mg, F2 0, F3 0 The initial values are given as x1 0, x 2 r, x 3 0 x1 0, x 2 0, x 3 v 0 Solution It is easy to become 1 2 1 2 1 0, t x i t t 2 1 2 x1 , x1 1 2 x2 , x 2 1 0 x 3 F a 1 2 1 2 1 2 1 f 1 i x i x j 2 x i 2 gx1 x12 x 22 2 m x i x x x t t i j i 1 1 4 x12 x 22 4r 2 x l x l and the differential equations of the motion (21) in form 3 x1 g gx12 x1 x12 x 22 r2 gx1 x 2 x 2 x12 x 22 x2 r 2 x3 0 In following the new coordinates 1,2,3,...,6 1 x1 , 2 x 2 , 3 x 3 , 4 x1 , 5 x 2 , 6 x 3 and the correspond system of first order equations 1 4 2 5 3 6 4 g 5 g12 1 42 52 r 2 g1 2 2 42 52 r2 6 0 are used. To numerical solve of these equations with initial values at t 0 and date values 10 0, 20 r, 30 0, 40 0, 50 0, 60 v0 g 9.81 ms-2, r 2 m, v0 1 ms-1 the Matlab solver ODE45 was used. The results are presented on the following figure 1. Figure 1 4 Example 2. The particle of mass m is constrained to move along the smooth parabolic curve given by equations 1 2 x 2 x12 0 2 x3 0 and the active force is the gravitation given by equations F1 mg, F2 0, F3 0 The initial values are given as x1 0, x 2 0, x 3 0 x1 0, x 2 0, x 3 0 Solution 1 2 1 2 1 0, t x i t t 2 1 1 1 2 x1 , 2, 0 x1 x 2 x 3 2 0, x1 2 0, x 2 2 1 x 3 D1 2 gx1 x12 , D2 0, D 4 1 x12 Consider above results the equations (20) can be written in form gx12 x1 x12 x1 g 1 x12 x2 gx1 x12 1 x12 x3 0 It is obviously that x 3 is identical to zero. In the following the new coordinates 1,2,3,4 1 x1 , 2 x 2 , 3 x1 , 4 x 2 and the correspond system of first order equations 1 3 2 4 g12 112 3 g 1 12 g1 12 4 1 12 with initial values at t 0 5 10 20 30 40 0 and the Matlab solver ODE45 are used. The results are presented on the following figure 2 . Figure 2 Conclusions The differential equations of motion of constrained particle are presented in Carthesian coordinates. Any change of the coordinates and usually recommended [2] determine of dependent and independent variables do not be necessary. The using of the proposed method to solve the problems of dynamics of body and body systems is a matter of following .papers. References 1. K.E. Brenan,, S.L. Campbell and L.R. Petzold. Numerical Solution of Initial-Value Problems in Differential-Algebraic Equations.SIAM, Philadelphia,1996 2. E. J. Haug. Intermediate Dynamics. Prentice Hall, New Jersey, 1992 6