PHY 306 Project Report Double Pendulum by Arnab Dhabal and Raziman T V (Y7081) (Y7355) Mentor: Dr. M.K. Verma 1 Contents 1 Introduction 3 2 Theory 3 3 Construction 5 4 Numerical Analysis 5 5 Future work 7 6 Acknowledgement 8 References 8 4.1 Non-Chaotic Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 4.2 Chaotic Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2 1 Introduction A double pendulum consists of one pendulum attached to another[1]. This is a simple mechanical system that shows chaos for some initial conditions. A double pendulum has two degrees of freedom and a four dimensional state space. A pair of double pendulums can serve as a good demonstrative apparatus to show chaos, as visibly identical initial conditions result in very dierent evolution of the system. In this project we build a double pendulum and demonstrate chaotic and non-chaotic evolution of the system based on dierent initial conditions. A numerical treatment of an ideal undamped double pendulum is also done to verify experimental results. 2 Theory Here is a schematic diagram of the double pendulum. Both limbs of the double pendulum are restricted to move in the vertical plane. The upper limb of the double pendulum has length L. Its centre of mass is at a distance l from the point of suspension and the moment of inertia about the centre of mass is I . The centre of mass of the lower limb is situated at a distance l from its point of suspension and its moment of inertia around the centre of mass is I . The masses of the limbs are m and m respectively. Here we do the analysis for an ideal - ie, undamped - double pendulum. This is a conservative system. Equations of motion are derived here using the Lagrangian formalism. Translational kinetic energies of the centres of mass of the two limbs are given by: 1 1 2 2 1 T1,trans = T2,trans = = = 2 1 2 2 2 m1 x˙1 + y˙1 1 2 ˙ 2 2 m1 l1 θ1 1 2 2 2 m1 x˙2 + y˙2 1 1 2 ˙ 2 2 ˙ 2 2 m2 L θ1 + 2 m2 l2 θ2 + m2 Ll2 cos(θ1 − θ2 )θ˙1 θ˙2 Rotational kinetic energies of the limbs around their respective centres of mass are given by T1,rot = T2,rot = 3 1 ˙2 2 I 1 θ1 1 ˙2 2 I 2 θ2 Hence the total kinetic energy of the system is 2 2 2 2 T = 12 m1 l12 θ˙1 + 12 m2 L2 θ˙1 + 12 m2 l22 θ˙2 + m2 Ll2 cos(θ1 − θ2 )θ˙1 θ˙2 + 12 I1 θ˙1 + 12 I2 θ˙2 2 The gravitational potential energies of the two limbs are V1 V2 −gm1 l1 cos(θ1 ) −gm2 L cos(θ1 ) − gm2 l2 cos(θ2 ) = = Hence the total potential energy of the system is V = −gm1 l1 cos(θ1 ) − gm2 L cos(θ1 ) − gm2 l2 cos(θ2 ) and the Lagrangian is L = where = c1 = c2 c3 c4 c5 = = = = T −V 2 2 c1 θ˙1 + c2 θ˙2 + c3 θ˙1 θ˙2 cos(θ1 − θ2 ) + c4 cos(θ1 ) + c5 cos(θ2 ) m1 l12 2 m2 l22 2 + I1 2 I2 2 + m2 L2 2 + m2 Ll2 g (m1 l1 + m2 L) gm2 l2 The evolution of the system is determined by the Euler-Lagrange equations d ∂L dt ∂ θ˙i − ∂L ∂θi =0 For the current system, this gives us two coupled second order ordinary dierential equations 2 c4 sin(θ1 ) + 2c1 θ¨1 + c3 θ¨2 cos(θ1 − θ2 ) + c3 θ˙2 sin(θ1 − θ2 ) 2 c5 sin(θ2 ) + 2c2 θ¨2 + c3 θ¨1 cos(θ1 − θ2 ) − c3 θ˙1 sin(θ1 − θ2 ) = 0 = 0 Decoupling the second order parts, 2 2 θ¨1 = 2c2 c4 sin(θ1 )+c23 θ˙1 sin(θ1 −θ2 ) cos(θ1 −θ2 )+2c2 c3 θ˙2 sin(θ1 −θ2 )−c3 c5 cos(θ1 −θ2 ) sin(θ2 ) c23 cos2 (θ1 −θ2 )−4c1 c2 θ¨2 = 2 2 2c1 c5 sin(θ2 )−c23 θ˙2 sin(θ1 −θ2 ) cos(θ1 −θ2 )−2c1 c3 θ˙1 sin(θ1 −θ2 )−c3 c4 cos(θ1 −θ2 ) sin(θ1 ) c23 cos2 (θ1 −θ2 )−4c1 c2 4 3 Construction The experimental apparatus was constructed in the Physics Workshop at IIT Kanpur. The apparatus consists of two identical double pendulums supported by the same structure. Two double pendulums are used to see how the evolution is dierent for nearly identical initial conditions. The upper limb of each double pendulum is made out of two thin rectangular aluminium rods. Two rods are used to maintain symmetry. The rods are 30 cm long, 2.5 cm wide and 5 mm thick. the two rods of the upper limb were separated by 8 cm. The lower limb of each double pendulum is an aluminium rod 25 cm long but identical otherwise to the rods making up the upper limbs. The upper limbs are suspended from a mild steel rod of diameter 6 mm. A similar mild steel rod is xed to the lower end of either upper limb from which the lower limbs hang. All free joints are connected using ball bearings to minimise damping. The steel rod from which the upper limbs are suspended is supported by two vertical aluminium rods of square cross section of side 2 cm, xed to the mild steel base. A heavy base was used to counter the large hammering torques that come about when the lower limbs ip over. 4 Numerical Analysis The two second order ODEs were solved using Runge-Kutta 4th order integration method by converting them to four rst order ODEs in θ , p , θ , p . 1 θ˙1 = pθ˙ 1 θ˙2 = pθ˙ 2 = = 2 θ1 θ2 pθ1 2c2 c4 sin(θ1 )+c23 p2θ sin(θ1 −θ2 ) cos(θ1 −θ2 )+2c2 c3 p2θ sin(θ1 −θ2 )−c3 c5 cos(θ1 −θ2 ) sin(θ2 ) 1 2 c23 cos2 (θ1 −θ2 )−4c1 c2 pθ2 2c1 c5 sin(θ2 )−c23 p2θ sin(θ1 −θ2 ) cos(θ1 −θ2 )−2c1 c3 p2θ sin(θ1 −θ2 )−c3 c4 cos(θ1 −θ2 ) sin(θ1 ) 2 1 c23 cos2 (θ1 −θ2 )−4c1 c2 5 The time step used for integration was 0.001 seconds. As the expression in the denominators of the ODEs for p˙ never approaches zero and energy conservation bounds the values of p , the integration is not singular. Hence this time step is good enough for evolving the system based on RK4 routine. Approximate values of c s were found for the current apparatus and were used in the numerical simulation. Here we tabulate the normalised values (normalised wrt c = 1) θi θi i 3 c1 2.16 c2 0.27̇ c3 1 c4 172.5 c5 32.6̇ Time evolution of the system was noted for dierent initial conditions. To see if a given initial condition results in chaos, evolution of the system for nearly identical initial conditions (diering in one of the angles by 0.0001 radians) was also done and the results were compared. Some initial conditions were found to be non-chaotic while some others were found to be chaotic. As an example, we show here two sets of initial conditions with the same energy (E = 0), one of which results in non-chaotic evolution of the system while the other shows chaos. 4.1 Non-Chaotic Evolution A set of initial conditions that shows non-chaotic evolution is θ = θ = , p = p = 0. Total energy of the system in this case is zero. Evolution of θ and θ with time is shown for two closely separated (4θ = 0.0001) initial conditions. 1 1 2 π 2 θ1 10 12 θ2 2 2 2 1.5 1.5 1 1 0.5 0.5 Angle (rad) Angle (rad) 2 0 0 -0.5 -0.5 -1 -1 -1.5 -1.5 -2 -2 0 2 4 6 8 10 12 14 16 18 20 0 Time (s) 2 4 6 8 14 Time (s) 16 18 20 Evolution of θ Evolution of θ We see that the two evolution curves virtually lie on top of each other. This shows that the system is non-chaotic for this inital state 1 2 6 4.2 Chaotic Evolution Another initial state of the system that has the same value of energy is θ = , θ = , p = p = 0. Even though the initial energy is the same, the evolution of the system is qualitatively very dierent. Evolution of θ and θ with time is shown for two closely separated (4θ = 0.0001) initial conditions. 1 π 2 2 3π 2 θ1 35 40 θ2 2 2 2 100 1.5 0 1 -100 0.5 -200 Angle (rad) Angle (rad) 1 0 -300 -0.5 -400 -1 -500 -1.5 -600 -2 -700 0 5 10 15 20 Time (s) 25 30 35 40 0 5 10 15 20 Time (s) 25 30 Evolution of θ Evolution of θ We see that the evolutions for the initial states are very dierent, even qualitatively. This shows that the evolution is chaotic. 1 5 2 Future work The current apparatus provides scope for a lot more of work. The theoretical treatment could be expanded to include damping. This would give a more realistic model for the system. Once this is developed, at least for the non-chaotic initial conditions, comparison could be done between theoretical and experimental evolution of the system. Experimental evolution of the system could also be studied quantitatively. Two possible ways for analysis involve image processing and the use of motion sensors. Tips of the limbs could be marked with easily identiable - for example reective - material. A video could be made of the evolution with a camera having suciently high frame rate. Tracking software could be used to nd the values of the state variables from the videos. Alternatively, motion sensors such as accelerometers and gyroscopes could be attached to the limbs to record the positions as a function of time. Also, in the current work we have not investigated the sets of initial conditions that result in chaos. A more detailed analysis could be done to see what initial angular momenta and energies result in chaos. 7 6 Acknowledgement We would like to thank Dr. M.K. Verma for guiding us through the project. We also express our heartfelt gratitude for Mr. Omprakash and others at the Physics Workshop for their eort in constructing the experimental apparatus. References [1] Wikipedia. Double pendulum wikipedia, the free encyclopedia, 2010. [Online; accessed 8-November-2010]. 8
© Copyright 2026 Paperzz