Detection of Faults on Off-Road Haul Truck Tires. M.G. Lipsett D.S. Nobes

Transcription

1 University of Alberta Mechanical Engineering Department SMART Meeting 14 October 2011 Detection of Faults on Off-Road M.G. Lipsett D.S. Nobes R. Vaghar Anzabi A. Kotchon K. Obaia, A. Munro (Syncrude) Topics: of haul truck tires o Reliability Model of haul truck tires o Simulation of Haul Truck Tire Dynamics o Finite Element Modeling of a rolling tire 2 of 24 1

2 : Tires are prone to premature failure Safety and production risks A range of failure modes Reliance on periodic physical inspection rather than condition monitoring Some fault modes can develop between inspection intervals Early detection often allows for repairing rather than scrapping tire 3 of 24 : To determine the specifications and requirements of the fault detection system of off-road trucks tires To study the feasibility of a non-contact method for extracting features from the surfaces of tires (preferably not on the vehicle itself) To test candidate monitoring techniques at a laboratory scale Test a fault detection system under field conditions 4 of 24 2

3 Structure of a radial-ply tire [Samuel, 1981] Failure area Tread area Shoulder area Sidewall area Bead area Liner area Failure modes Cuts Delamination Separations Cuts Separations Cuts Separations Radial splits Separations Flange erosion Cracking Splits Lifting liner Wrinkled liner 5 of 24 a) Tread cut b) Tread delamination c) Tread separation d) Sidewall cut Photographs of typical tire failures 6 of 24 3

4 Fault Modes of Interest Haul truck tire lifetime data Syncrude Canada Ltd. database of tire failures, Haul truck tires failure modes histogram 7 of 24 Current Tire Periodic visual inspections by technicians Wireless pressure/temperature sensors can be mounted inside the tire or on the rim to give some indication of tire inflation and heat generation X-ray inspection can be used for initial factory QA, and possibly in tire shop, with safeguards Tire Inspection in Syncrude repair shop (Winter 2011) 8 of 24 4

5 Candidate C.M. for Current Tire Periodic visual inspections by technicians Pressure or temperature sensors mounted inside the tire X-ray inspection (initial factory QA, possibly shop) Option1: Optical for Tire (Stationary sensor) Use cameras or other instruments to measure features on the tire surface as the vehicle is driving or stopped at specific locations crusher dump, fuel & lube module, etc. Option2: Add Sensors onto Vehicle to Improve Observability Estimate change in tire stiffness as function of tire rotation angle & relate stiffness change to fault(s) requires model-based analysis and knowledge of environment stressor function (random vertical displacement of road) Find local hot spots or tire geometry changes using chassis-mounted or tire-mounter sensors 9 of 24 Option1: Image-based techniques (visual, thermal) Non-contact method Low-risk environment for workers No requirement to remove equipment from service for inspection Determination of 3D Deformations of Tire Surfaces [O. Erne, 2007] 10 of 24 5

6 Optical Monitor deformation in regions of the tire Digital Image Correlation produces strain map Extract features for classification Faults potentially detectable and identifiable using simple visible features (wide cracks, bulges, change in tire ride height from pressure/stiffness change) Monitor temperature changes in tire (local hot spots) Thermal imaging (thermography) Can potentially detect faults that cause surface temperature change (hot spots), or show temperature gradients 11 of 24 Optical Challenges Strain map of 27 cm diameter pneumatic tire Deformation through 180 degree rotation Ensuring observability (occluded surfaces, surface fouling by mud/snow) Sensor performance in field conditions with moving tires, ruggedness Intuitive fault classification for decision-making (derate vehicle, remove from service for inspection/replacement) 12 of 24 6

7 Option2: Other Approaches Estimate change in tire stiffness as function of tire rotation angle Add sensors onto vehicle to improve observability (although this would require significant engineering and sensor reliability could be an issue) tire deformation monitoring using on-board sensors strut pressure monitoring to assess tire irregularities on good ground surfaces supplier systems for pressure and temperature 13 of 24 Reliability Model Reliability Modeling Algorithm 1. Collecting Data of Component Lifetimes 2. Estimating Probability Function 3. Selecting Candidate Distribution (Normal, Weibull, Exponential, ) 4. Estimating Parameters of Distribution (maximum likelihood method) 5. Conducting Goodness-of-fit Test (chi-squared test) 6. Choosing Reliability Function Collecting Data Estimating Probability Function Picking Distribution Estimating Parameters Goodness of fit test Accept Reject Final Distribution Function Finding reliability function algorithm 14 of 24 7

8 Reliability Model Model for All Chosen distribution for the data set 3-parameter Weibull distribution: Estimating Parameters Maximum Likelihood method Goodness-of-fit test Acceptance Fitted distribution to all failure data (3-parameter Weibull) 15 of 24 Reliability Model Seasonal Effect Tires fail more often in spring and summer than in winter Road conditions and summer heat may be causal factors 16 of 24 8

9 Reliability Model Models for Most Common Tread cut & separation (19% of all failures) Sidewall cut (18% of all failures) Time of year for early mortality failures will also be checked Impact tread (15% of all failures) 17 of 24 Dynamical Model Simulation of Haul Truck Tire Dynamics Vehicle dynamics can help in understanding effects of underinflation and road conditions The model allows change in tire stiffness as function of tire rotation angle to be estimated This estimate can then be linked to possible faults Strut pressure can be used, otherwise additional sensors must be added Quarter-car model physical system Change of tire stiffness vs. rotation angle 18 of 24 9

10 Dynamical Model a) Without Tire failure b) With Tire failure Effect of tire fault on dynamic response of the system Simulated data showing the effect of an artificial fault on motion of the axle 19 of 24 Finite Element Modeling of a rolling tire FEM Model Finite-element modeling (FEM) can give a good representation of how a tire behaves (deformations, etc) when damaged Non-linearities make tire modeling challenging: Geometrical (large deformations, ) Material (Complex material, damage, ) Force boundary conditions (pressure, loading, ) Displacement boundary conditions (rough road, seasonal effect, ) ABAQUS FEM software features allow for nonlinear analysis and crack modeling: Modeling hyperelastic materials Large deflections Modeling Contact condition (tread pattern, vehicle weight, road friction effect) Footprint analysis J-integral calculation for crack modeling User subroutines 20 of 24 10

11 Preliminary modeling based on reported literature FEM Model 2D Model 3D Model 21 of 24 Tire Model Development Verification of models Include terramechanics model and boundary conditions Constitutive relationships of tires Relate models to physics of failure Employ reduced-order models (for fault detection and identification) Expand usfeulness of reliability models using fault data from other sites (LTUG) ideally including relevant operating information as well as maintenance history Relate reliability models to causal factors such as road quality, tkph, time of year, etc., to understand root causes 22 of 24 11

12 Fault Detection System Development Test coupons from failed tires Create artificial faults on tires at lab scale Extract fault features (bulges, surface flaws) Test a set of candidate fault detection methods to combine features for good diagnostics Some on-board or dispatch information will likely improve fault diagnosis (strut pressures, tire pressures & temperature) Test robustness in expected field conditions (lighting, environmental conditions, particularly dust and mud) Work with end-users on reporting requirements (start with sending images, then move to features, likely faults, and severity estimates) Develop a field version Conduct field trials for fault detection and identification ( ) 23 of 24 If you are interested in this project (or anything else our group is doing), please contact me at mlipsett@ualberta.ca Thank you 24 of 24 12