Transcript Document
Using Bond Graph Modeling to Optimize a Mousetrap Car Lester Weitman 4/17/2006 7/21/2015 1 What’s a Mousetrap Car? www.docfizzix.com PROJECT GOALS: Design car for 9TH grade physics project, within given constraints: – Standard VICTOR® mousetrap – Only spring energy @ 1800 To maximize grade, car must travel: – 5 meters in under 2.75 sec; & – > 20 m total Physics dictates key parameters cannot improve both speed & distance. MODELING GOAL: Create BG of car and simulate speed & distance w/ varying: – Lever arm length – Wheel diameter – Axle diameter – String HOOP distance to spring axis – Wheel rotational inertia DIRECTIONALLY (not exactly) predictive! PREDICTED ISSUES RESOLVED: • Wheel friction: omit (verified no slip) • Wheel inertial mass: (used I=1/2Mω2 for measured evenly distributed mass) • Spring energy: (measure actual mousetrap x .1) DETERMINE PARAMETERS CALC: Moments of Inertia A solid cylinder of mass m= .03 kg 10 and radius R = cm will have a moment of inertia about its central axis: Icentral axis = 0.0001 kg m2 Note: lumped drive wheel mass & ignored smaller undriven wheels. Ref: http://hyperphysics.phy-astr.gsu.edu/HBASE/icyl.html#icyl CALC: Spring Constant (K) Used sample data from DocFizzix.com to be very conservative: – PE = ½ K Ø2 – DocFizzix’s PE = .8 J K= .08 Nm/rad – Actual Model PE = 7.2 J K= .72 Nm/rad – DocFizzix’s Initial Torque = .135 Nm (used) CALC: Spring Constant (K) Lab #2 All Wound Up Data was collected by measuring force per angle with a force probe and a torsion wheel. Record the radius of the torsion wheel, or the length of the measuring arm in (D12) and record the force as well as angle in (C22-C?) and (B22-B?) BLUE ITEMS - Data that is measured and entered from experimen. DO NOT CHANGE BLACK NUMBERS Spring to drive axle = 0.18 Measure the distance from the center of the mousetrap's spring to the drive axle Actual lever arm length = 0.18 Measure the length of the leverarm Testing Arm Length (m) = 0.05875 Total Starting Potential Energy (J) = NOTE: If you are using Doc Fizzix's Torsion Wheel DO NOT change this value 0.8107 Spring Constant - Based on slope of Torque vs. Angle (Nm/rad) = 0.0795 Spring Constant - Based on average of all Spring Constants (Nm/rad) = 0.0778 Ref: http://www.docfizzix.com/shop/tools- software/t100df.shtml CALC: Spring Constant (K) K = slope of Spring Torque vs. Angle CALC: String Velocity L u P Ø Law of Cosines: u2=L2+P2-2LPcosØ ů= ω x LPsinØ / sqrt(L2 + P2 - 2LPcosØ) STRING VELOCITY = ů UNANTICIPATED ISSUES: Could not retrieve actual optimal car. Used 3 alternatives. MUCH harder to model & de-bug in 20-SIM than anticipated! Had to verify kinematics were correct. Had to experiment to find reasonable value for axle friction. (Initially ignored, until simulation car ran forever.) MODELING CHALLENGES: 1 2 3 Lumped mass of car MODELING CHALLENGES: 1. Lever stops when Ø = 1800 2. Axle should roll freely when lever stops, so car continues on own momentum. 3. To simplify analysis, treated car stationary & floor moving w/ mass of car (inversing sign of TF at wheel). ISSUES RESOLVED: HALLELUJAH! WORKING Bond Graph Model: WORKING Bond Graph Model: Modeled spring as MSe, setting it to ZERO when lever stopped. WORKING Bond Graph Model: Modeled axle as being connected to half of axle driven by string with a MR (rotational damper), setting it to ZERO when lever stopped. CONSTANTS USED: SAMPLE RUN: Parameters SAMPLE RUN: PRE-SCREENING EXPERIMENT: RUN 1 2 3 4 5 6 7 L .1 .1 .2 .08 .08 .08 .08 P .1 .1 .2 .08 .08 .04 .04 Rw .05 .05 .05 .05 .08 .08 .08 Ra .002 .002 .002 .002 .002 .002 .004 DISTANCE 22 24 22.7 20 46 45 45 5m in x s 3.18 3.69 4.5 2.9 3.4 1.9 1.6 Jw .0000125 .0000125 .00000125 .00000125 .00000125 .00000125 .00056 SIMULATION vs. REALITY Directionally matched performance of actual cars. Matched intuition: – Longer LEVER string pulls longer – Shorter P (hoop) string starts moving faster, then slows up Placeholder for Final OPTIMIZING DOE: 2IV4-1 FRACTIONAL FACTORIAL DESIGN RUN 1 2 3 4 5 6 7 8 L .08 .2 .08 .2 .08 .2 .08 .2 P .04 .04 .2 .2 .04 .04 .2 .2 Rw .05 .05 .05 .05 .1 .1 .1 .1 Ra .002 .006 .006 .002 .006 .002 .002 .006 DISTANCE 18.7 18.4 18.6 22.8 72 73 74 55 5m in x s 1.8 1.6 1.7 4.5 1.6 1.8 2.4 3.3 MiniTab Resuts: Velocity Pareto Chart of the Effects (response is VEL, Alpha = .10) A: B: C: D: B AB A D AC C AD 0.0 0.5 1.0 L P Rw Ra MiniTab Resuts: Distance NEXT STEPS: Verify MiniTab analysis for optimized values. CONCLUSIONS: 20-SIM difficult to de-bug for novice Need to recall tricks to simulate real world: – Attach “MR” • ZERO to “connect” • 999999 to “disconnect” Model only needed to be DIRECTIONALLY accurate! Model worked adequately in region of interest for “optimization”. THE END