SOLVING PROBLEMS OF GLOBAL IMPORTANCE www.ara.com Lowering Pavement Evaluation Costs Using Big Data Bill Buttlar UIUC Bill Vavrik ARA T.H.E. 2016 www.ara.com 2015 2014 Applied Research Associates, Inc. 1
SOLVING PROBLEMS OF GLOBAL IMPORTANCE The evolution of pavement data collection Manual Vehicle based data collection Film Digital Current 3D Systems PaveVision Smart Phone Smart Roads www.ara.com 2015 Applied Research Associates, Inc. 2
SOLVING PROBLEMS OF GLOBAL IMPORTANCE Pavement Evaluation Data Collection Continues to Evolve www.ara.com 2015 Applied Research Associates, Inc. 3
4
SOLVING PROBLEMS OF GLOBAL IMPORTANCE Pavement Evaluation Data Collection Continues to Evolve www.ara.com 2015 Applied Research Associates, Inc. 5
SOLVING PROBLEMS OF GLOBAL IMPORTANCE Pavement Evaluation Data Collection Continues to Evolve www.ara.com 2015 Applied Research Associates, Inc. 6
SOLVING PROBLEMS OF GLOBAL IMPORTANCE The Current State of the Art is 3D www.ara.com 2015 Applied Research Associates, Inc. 7
SOLVING PROBLEMS OF GLOBAL IMPORTANCE The analysis potential increases Range 3D Intensity www.ara.com 2015 Applied Research Associates, Inc. 8
SOLVING PROBLEMS OF GLOBAL IMPORTANCE The next generation is now here www.ara.com 2015 Applied Research Associates, Inc. 9
SOLVING PROBLEMS OF GLOBAL IMPORTANCE Advanced Computer Vision Is Changing Data Analysis www.ara.com 2015 Applied Research Associates, Inc. 10
SOLVING PROBLEMS OF GLOBAL IMPORTANCE PaveVision is low cost with the ability for automated data analysis www.ara.com 2015 Applied Research Associates, Inc. 11
SOLVING PROBLEMS OF GLOBAL IMPORTANCE And it will continue to evolve www.ara.com 2015 Applied Research Associates, Inc. 12
Road User s Opinion of Roughness http://www.wsj.com/articles/nations-crumbling-roads-put-a-dent-in-drivers-wallets-1438365456 13
Road and Highway Conditions Our Roadway Network o Approx. 4 million miles paved roadways o Approx. 94% are surfaced with asphalt 32% of major roads are in poor condition o Incur costs of $67 billion a year $170 billion needed annually to improve o $85 billion for existing pavements 14
Pavement Roughness Defined in engineering practice as surface unevenness which adversely affects ride comfort Expressed by a numerical scale called the International Roughness Index (IRI) Quarter-car Model Inertial Profiler 15
Current Roughness Measurement Systems Inertial Profiler First developed by Elson Spangler and William Kelley Modern inertial profilometers require four basic sub-systems: Accelerometers Height sensors Distance or a speed sensor Computer hardware and software International Cybernetics Corporation (ICC), Automatic Road Analyzer (ARAN), and K.J. Law profilometer 16
Limitations/Challenges of IRI Unable to provide distress information Multiple pavement sections can have the same IRI value Data collected at low speeds can generate false peaks in the profile Create false spikes in the IRI parameter Accelerometer sensitivity affects pavement roughness 17
Motivation VDOT reported a contractor is employed to gather roughness data at an annual cost of $1.8 million Data are collected once-every-five years for secondary roads For small transportation agencies such counties and cities with low operating budgets, pavement condition data collection frequency may be limited Thus, M&R decisions are often performed using outdated data Infrequent roughness measurements also preclude the identification of rapidly developing distress features on pavements, such as potholes There is a need for a pavement roughness data collection system which is: Economical and simple Easily accessible Crowd-source based, having the potential to save agencies millions of dollars Providing data for more intelligent route selection 18
Research Objectives Development of a smartphone application, Roughness Capture, to measure pavement roughness Initial validation of IRI predictions using Roughness Capture, comparing cell-phone based IRI values to those obtained using an industry-standard inertial profiler 19
Hypothesis: Vertical Acceleration of Vehicle Pavement surface irregularities causes the vehicle wheels to move up and down with respect to the road surface, causing the vehicle cab to accelerate (although cab movement is dampened by suspension) Roughness Capture has been used to collect vertical acceleration data in the vehicle cab - It is hypothesized that vehicle cab acceleration measured with smart phones can be combined with vehicle dynamics models to arrive at accurate measures of pavement IRI 20
Double Integration vs. Inverse State-Space Model Double Integration Scheme Acceleration Data ʃʃ Vehicle Cab (Dashboard) Displacement Profile, y IRI Roughness Capture Quarter-Car Model (ASTM) Inverse State-Space Modeling Scheme Pavement Profile, y IRI Roughness Capture State Space Modeling of Suspension for Vehicle Actually Used, Solved Inversely Quarter-Car Model (ASTM) 21
State-Space Model Output Input Zs is known, y is unknown Therefore, it is inverse problem 22
Data Collection and Site Location Three test sites : County Highway 32, 9, and 23 Test sites were selected with wide variety of distresses County Highway 32 is with very low or no distresses, and County Highway 23 is very rough pavement 23
IRI Estimation using ProVAL and MATLAB Script 250 County Highway 23 32 9 MATLAB Script Esitimated IRI IRI (inch/mile) 200 150 100 50 Unity Line 0 0 50 100 150 200 250 ProVAL Estimated IRI, (inch/mile) 24
Inverse State Space: County Highway 32 300 Smartphone Measured IRI (inch/mile) 220 170 119 94 60 Honda CRV 60 94 119 170 220 300 Profiler Measured IRI (inch/mile) 25
Inverse State Space: County Highway 9 300 Smartphone Measured IRI (inch/mile) 220 170 119 94 60 Honda CRV 60 94 119 170 220 300 Profiler Measured IRI (inch/mile) 26
Inverse State Space: County Highway 23 Smartphone Measured IRI (inch/mile) 300 220 170 119 94 60 Honda CRV 20 inch/mile offset line from unity line 10 inch/mile offset line from unity line 60 94 119 170 220 300 Profiler Measured IRI (inch/mile) 27
Validation: Profiler vs. App Measured IRI 250 200 Average Inertial Profiler IRI over Total Length (in/mi) Average Smartphone Measured IRI over Total Length (in/mi) 150 IRI (in/mi) 100 50 0 IL-10E County Rd N 1200E County Rd 2500N County Rd 1000N County Road 9 County Rd 200E County Rd 800E County Rd 900E US-150 US-45 SB County Rd 2700N/300E US-136 IL-47 28
Effect of Different Vehicles on IRI Mazda 3 Honda CR-V Dodge Avenger Chevrolet Impala M1 (kg) 343 420 494 500 M2 (kg) 40 40 40 45 C1 (N*s/m) 1,500 1,400 1,550 1,500 K1 (N/m) 13,500 11,000 12,000 10,000 K2 (N/m) 200,000 198,000 200,000 200,000 29
IRI vs. Vehicle at County Highway 9 220 Profiler IRI Mazda 3 Honda CRV Chevrolet Impala Dodge Avenger IRI (inch/mile) 170 119 94 60 Very Poor Poor Fair Good Very Good 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 Distance (mile) 30
Average IRI with Vehicles at County Highway 9 Profiler IRI Average Smartphone IRI 220 IRI (inch/mile) 170 119 94 60 Very Poor Poor Fair Good Very Good 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 Distance (mile) 31
Moving Ahead For Progress (MAP-21) Performance Measure Data (1) % pavements on the Interstate Systems in Good condition 1. IRI (2) % pavements on the Interstate Systems in Poor condition 2. % Cracking 3. Rutting (3) % pavements on the NHS in Good condition 4. Faulting (4) % pavements on the NHS in Poor condition Surface Type All Pavements IRI (inch/mile) Rating < 95 Good Area 95-170 Fair Areas with a population <1 million 95-220 Fair Urbanized areas with population 1 million > 170 Poor Areas with a population <1 million > 220 Poor Urbanized areas with population 1 million 32
MAP-21 IRI Data: County Highway 9 Poor Profiler IRI Smartphone IRI IRI (inch/mile) 170 94 Fair Good 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 Distance (mile) 33
Surface Irregularity Detection IRI itself does not offer any idea of presence of surface irregularities Two different pavement profiles can generate same IRI values though distress types and locations are different Detection of surface irregularities location will give an idea of the severity of distresses 34
Localized Distress Detection 70 Number of Localized Distresses 60 50 40 30 20 10 Distresses Detected by Pavement Engineer Distresses detected by Smartphone App 0 County Rd 2500N County Rd 900E US-136 US-136 IL-47 US-150 County Rd 2700N/300E County Rd N 1200E IL-10E County Rd 200E County Rd 1000N County Rd 800E 35
Bump/Pothole Locations 36
Integration of IRI Data into Roadway Network Map Visualization of network condition is an outmost interest to transportation agencies Incorporation of pavement roughness values in the roadway Existing roadway network data Provide a link between PMS and GIS 37
IRI on Roadway Map using ArcGIS Very Good Good Fair Poor Very Poor 38
SOLVING PROBLEMS OF GLOBAL IMPORTANCE www.ara.com 2015 Applied Research Associates, Inc. 39
Acknowledgements NexTrans: USDOT Region V Regional University Transportation Center Applied Research Associates, Inc. 40