Lunar Rover Suspension

1
Lunar Rover Suspension

• Final Project Presentation
• ME 6105 Modeling Simulation in Design
• Georgia Institute of Technology

April 26, 2007
2
Team Members

• Stephanie Thompson
• 2nd year MSME student
• Advisor Dr. Farrokh Mistree
• Research Material Design

• Nathan Young
• 1st year MSME student
• Advisor Dr. Mervyn Fathianathan
• Research Adaptive Systems

• Robert Thiets
• 1st year MSME student
• Advisor Dr. Bert Bras
• Research Alternative Fuels

3
Presentation Outline

• Introduction
• Project Overview
• Objectives Hierarchies
• Influence Diagram
• Dymola Model
• Key Assumptions
• Modeling Strategy /Description
• Animation
• Problems and Solutions
• Simulation Time Reduction
• Preference Elicitation
• AIAA Modeling Conference
• Questions

4
Project Overview

• NASAs Vision for Space Exploration includes a
  goal to return to the moon by the year 2020 as a
  launching pad to manned exploration of Mars
• One important task for both lunar and martian
  exploration is the design and development of
  unpressurized and pressurized rovers for use in
  both environments
• In this project, we will concentrate on the
  design of the suspension for a manned lunar rover
  with extensibility to the martian landscape
• Specifically, we seek to provide a recommendation
  for target values of effective spring rates and
  dampening coefficients

5
Fundamental Objectives Hierarchy
6
Influence Diagram
7
Dymola Model Key Assumptions

• Rover suspension and terrain components
• The rover never breaks contact with the terrain
• The rover has a perfectly rigid and mass-less
  wheels/tires
• The rover travels at a constant velocity
  regardless of the terrain
• The terrain only influences the vertical position
  of the contact points between the models.
• Zero friction between the terrain and the
  suspension

8
Dymola Model Strategy

• Two main subsystems
• Rover Suspension
• Test Frame
• Time Varying Terrain Signal
• Amplitude (m) vs. Time (s)
• Terrain Signal Split
• Left to Right Time Delay
• Front to back Time Delay
• Sensor
• Needed to measure settling time
• Fixed World

9
Dymola Model - Strategy

• Rover Chassis
• MacPherson based Suspension geometry
• Point mass which simulates both payload and
  variation in gravitation
• Torsion Springs / Dampers
• Connection Nodes

10
Dymola Model - Strategy

• Test Frame
• Four Linear Actuators
• Four connection nodes
• Four signal input nodes
• Fixed displacement
• Cut Frames

11
Dymola Model Terrain Input
Amplitude (m)
Time (s)
12
Dymola Model - Animation
13
Simulation Time Reduction

• Kriging Interpolation
• Used as a surrogate model to estimate the utility
  of a given combination of spring and damper
  values
• Interpolation is based of a random data set
  generated by the model
• Requires substantially less simulation time vs.
  running the model
• Error of estimation is known. If error is above a
  set threshold, the model can be run to generate
  additional data points, reducing the error of the
  interpolation
• Kriging Model Created in Model Center by Tom
  Groshans

14
Simulation Time Reduction
15
Preference Elicitation
Preference Equation

• A Acceleration
• T Travel
• ST Settling Time

16
Preference Elicitation
1.
U(T0.5, A, ST0) U(T0.5, A0.5, ST0.5)
U(0.101 m, A , 4.5 s) U(0.3 m, 1.03 m/s2,
0.96 s)
A 0.8 m/s2 U(A) 0.768
2.
U(T0.5, A, ST1) U(T0.5, A0.5, ST0.5)
U(0.101 m, A , 0 s) U(0.101 m, 1.03 m/s2,
0.96 s)
A 1.5 m/s2 U(A) 0.132
17
Preference Elicitation
3.
U(T, A0.5, ST0) U(T0.5, A0.5, ST0.5)
U(T, 1.03m/s2 , 4.5 s) U(0.101 m, 1.03 m/s2,
0.96 s)
T 0.085 m U(T) 0.653
4.
U(T, A0.5, ST1) U(T0.5, A0.5, ST0.5)
U(T,1.03 m/s2 , 0 s) U(0.1011 m, 1.03 m/s2,
0.96 s)
T 0.165 m U(T) 0.098
18
Preference Elicitation
5.
(T0, A, ST0.5) (T0.5, A0.5, ST0.5)
(0.3 m, A, 0.96 s) (0.101 m, 1.03 m/s2,
0.96 s)
A 0.75 m/s2 U(A) 0.821
6.
(T1, A, ST0.5) (T0.5, A0.5, ST0.5)
(0 m, A, 0.96 s) (0.101 m, 1.03 m/s2, 0.96
s)
A 1.1 m/s2 U(A) 0.419
19
Preference Elicitation

• Solution does not yield a result which is
  consistent with our expected preferences
• Solution is more or less invariant with regard
  to specific elicitation values
• New elicitation questions are used to solve
  problem

20
Preference Elicitation
1.
U(T0, A0.2, ST) U(T0, A0.5, ST0.5)
U(0.3 m, 1.4m/s2 , ST) U(0.3 m, 1.03 m/s2,
1.15 s)
ST 0.45s U(ST) 0.932
2.
U(T0.5, A0.8, ST) U(T0.5, A0.5, ST0.5)
U(0.1 m, 0.8 m/s2 , ST) U(0.1 m, 1.03 m/s2,
1.15 s)
ST 2.0 m/s2 U(ST) 0.117
21
Preference Elicitation
3.
U(T, A0, ST0.2) U(T0.5, A0, ST0.5)
U(T, 2.4 m/s2 , 1.7 s) U(0.101 m, 2.4 m/s2,
1.15 s)
T 0.08 m U(T) 0.692
4.
U(T, A0.5, ST0.8) U(T0.5, A0.5, ST0.5)
U(T,1.03 m/s2 , 0.7 s) U(0.1011 m, 1.03
m/s2, 1.15 s)
T 0.12 m U(T) 0.334
22
Preference Elicitation
5.
U(T0.2, A, ST0.5) U(T0.5, A0.5, ST0.5)
U(0.14 m, A, 1.15 s) U(0.101 m, 1.03 m/s2,
1.15 s)
A 0.85 m/s2 U(A) 0.710
6.
U(T0.8, A, ST0.5) U(T0.5, A0.5, ST0.5)
U(0.07 m, A, 1.15 s) U(0.101 m, 1.03 m/s2,
1.15 s)
A 1.1 m/s2 U(A) 0.419
23
Preference Elicitation
24
AIAA Modeling Conference

• AIAA Modeling and Simulation Technologies (MST)
  Conference and Exhibit
• 20 - 23 Aug 2007, Hilton Head, South Carolina
• Abstract for a paper based on this project has
  been accepted for this conference
• Addresses the design and development of flight
  simulation hardware, software, systems,
  innovative approaches, applications, and relative
  advances that keep modeling and simulation tools
  a viable, effective, and efficient engineering
  tool.

25
Questions ?
???
26
Means Objective Hierarchy
27
Original Influence Diagram