Volume 8, Issue 1

2013 CDR SUMMIT DVD INCLUDED

C

2013 Vol.8 Iss.1

ollision C The International Compendium for Crash Research

Volume 8 Issue 1

Collision: The International Compendium for Crash Research Volume 8, Issue 1 - Spring 2013

Cover_Outside.indd 1

SPRING 2013

Hyundai & Kia EDR what is it... or not?

Validation & Use of EDR Data from a Non-CDR Supported Vehicle in a Criminal Prosecution Case Accuracy of GPS Speed & Location Data Measured in Emergency Vehicles collisionmagazine.com 5/3/2013 9:40:34 AM

Crash site evidenCe

A 2010 Chevrolet Equinox SUV attempted to make a left turn in front of a 2003 Dodge Ram pickup. The pickup was in the middle northbound lane when it struck the left-rear side of the SUV, causing both vehicles to rotate CCW to their final rest positions.

The right-rear passenger in the SUV suffered significant facial injury, but there was no physical evidence on the back of the right front seat. How did the facial injury occur?

For more information on the capabilities of HVE please visit edccorp.com

HVE directly answers the following questions using Newton’s laws of motion:

Questions

Answers

Impact Speeds

SUV: 22 – 25 mph Pickup: 42 – 45 mph

Peak Impact Force

SUV: Fx= -37,850 lb, Fy= 66,350 lb, Fz= -1,560 lb Pickup: Fx= -71,400 lb, Fy= -26,700 lb, Fz= 1,880 lb

Peak Impact Acceleration

SUV: 18 – 23 g Pickup: 12 – 15 g

Peak Angular Velocity

SUV:

~ -340 deg/sec

PDOF

SUV:

~ -61 deg

Occupant Analysis (Right-Rear)

Seat Belt Status: Unbelted General Motion of Occupant: Left/Forward Facial Contact Detected: Left Front Seat Headrest

Required Technologies

SIMON 3-D Vehicle Dynamics Simulation Model DyMESH 3-D Collision Model GATB 3-D Occupant/Pedestrian Simulation Model

Other software may claim to answer these questions. Have you closely evaluated their claims and the basis for their results?

To learn more about HVE and all of the available tools for crash investigators, call 503.644.4500 or visit www.edccorp.com. Our professional sales staff will help you get started using HVE today!

Knowledge = Power 7096-0820

Engineering Dynamics Company, LLC • 550M Ritchie Hwy #128, Severna Park, MD 21146 USA • Phone 503.644.4500 • www.edccorp.com

A SOFTWARE SOLUTION FOR EVERY SITUATION:

Total station w/onboard data collection

Robotic total statio n w/Windows CE

Get EVR

□ Reliable □ Compatible See Exhibit A

Evidence Collection Forensic Mapping

Total station or GPS w/data collector

3D Laser Scanner

Evidence Collection Software?

tion Animacapabilities?

Do we

Yes

No Get Forensic CAD

MapScenes Evidence Recorder • Software for tablets and Windows CE devices • Connect with a total station or GNSS receiver to take measurements and collect evidence • Built for the professional reconstructionist

+ Animation Module

Cloud + Point Module

MapScenes Forensic CAD • A powerful and feature-rich program that provides law enforcement professionals and accident reconstruction specialists with confidence that the evidence they present is precise, compelling, and irrefutable • Produce scaled diagrams for use in cour t, creating a strong visual case for the prosecution or defense. Contains flexible options for point cloud/LiDAR data visualization and anim ation tools • Now with point cloud and animation addon modules • Built specifically for forensic www.MapScenes.com investigators and accident scene reconstructionists 1.800.668.3312

is registered in the U.S. Patent and Trademark Office. All other brand ® MapScenes names, product names or trademarks are property of their respective owners.

Exhibit A

want

Spring 2013 Volume 8 Issue 1 Cover photo courtesy of Shawn Gyorke

Contents

View the complete article index of all published issues of Collision Magazine at: collisionmagazine.com/articleindex.html

inside

10

04

Letter From the Editor

06

NAPARS: Letter From the President

07

Organization Partners Advertiser Index

05

features 10

Further Developments Regarding the Dynamic Modeling of Motor Vehicle Collision Response Using the SDOF Approach

by Jai Singh

34

Hyundai and Kia Crash Data - A Preliminary Overview by: Rusty Haight, Shawn Gyorke, and Sean Haight

42

The Effect of Tinted Headlights. A Look Into the Level of Light Diminishment Headlights Experience When a Tint is Applied to the Headlight

by Jeff Cardita

48

Laser Scanning for Crash Reconstruction

by Joel Salinas

60

Validation and Use of EDR Data from a Non-CDR Supported Vehicle in a Criminal Prosecution Case

by Wesley Vandiver, Isaac Ikram and Bryan Randles

78

Investigation of Traffic Crashes Involving the Inhalation of Difluoroethane

by Adam M. Hyde and Roger W. Barrette

88

Accuracy of GPS Speed and Location Data Measured in Emergency Vehicles

by David M. Little, Cst. Rob Joiner and Cpl. Stephen Hilliard

112

Reader Commentary on a Previously Published Article

www.collisionpublishing.com

42

60

88

Collision Magazine - Volume 8 Issue 1 3

Letter from the editor

A

s we put this issue of Collision to press, there's a lot to look back on and even more to look forward to in the issue to come later this year after the ARC-CSI Crash Conference in Las Vegas.

Looking back, the CDR User's Summit in Houston this past January was a success hosting an International group of attendees and featuring more participation made directly by auto manufacturers than in years past. In comparison, a small uninformed few failed to grasp the significance and benefit of hearing directly from Bosch staff, from the most involved OEs directly and from those involved in the development of relevant CDR standards but, in the end, that's ultimately more illustrative of their position in the community than it is a statement on the value of the annual CDR Summit.

While still early, other important changes are coming for the 2014 CDR User's Summit including an all-new venue. The new venue, North Houston Hilton, is a significant upgrade for the attendee experience, while continuing to offer a low group room rate and free shuttle service from Houston's Bush Intercontinental Airport (and nearby locations).

Moreover, contacts with OE reps, government representatives and others as well as an effort to take a look at crash data compared to crash test results from Hyundai and Kia vehicles will be the center piece of the next CDR User's Summit.

Hyundai and Kia vehicles. While this is becoming a faster-evolving area than ever before, and without taking too much from the preliminary look found later in this issue, it seems like there are several as yet unaddressed "considerations" left to resolve with the new tools for these vehicles than may have been anticipated.

Yet another "look ahead" is the "Yaris Project" which we plan to feature as an upcoming special supplement in Collision Magazine.

An expansion of a 1993 SAE paper, the "Yaris Project" will bring crash reconstruction technology from the 90's to today together in one extensive set of crash tests and other data anticipated to be as timely as it is relevant in many ways. From CDR-related data to finite element modeling to other types of simulations, the "Yaris project" will be, like the 1993 SAE paper has been for some 2 decades now, something that will be referenced for years to come.

This sample of what has passed and what will come is, I hope, an illustration of the continued value and relevance of Collision Magazine, the CDR User's Summit and the ARC-CSI Crash Conference. To those involved in the industry and who recognize that, while there are many opportunities for training and refResearch which essentially started at the 2013 erence material, Collision Publishing remains CDR Summit will be part of not only the your central "Crash Research Point of ConARC-CSI Crash Conference in Las Vegas but tact." also a follow up in the next issue of Collision.

In another "look ahead," in this issue, there's a first-of-it's kind preliminary review of crash Managing Editor data from from a non-Bosch CDR Tool for

4

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

BACK ISSUES

Collision Publishing LLC PO Box 3383 Kirkland, WA 98083-3383 Toll Free: 800-280-7940 Email: editor@collisionpublishing.com Web: www.collisionpublishing.com Web: www.collisionmagazine.com ISSN: 1934-8681

If you would like to purchase back issues of Collision Magazine, please visit www.collisionmagazine.com or call us toll free 800-280-7940

COLLISION STAFF Scott Baker Sean Haight Tonya Baker Carla McLeod

Owner, Managing Editor Senior Editor Advertising Manager Subscription Services

All rights reserved, © 2006-13 Collision Publishing LLC. The opinions and conclusions expressed in this publication and attached data disk and throughout the articles attributed to specific authors are the opinions and conclusions of the authors noted and not necessarily the editorial staff or anyone else for that matter. While some articles have been reviewed for content, the accuracy of reprinted models or equations cannot be fully guaranteed. It is the responsibility of the reader to apply critical thinking to an individual review of the content and make their own personal judgements as to its value to them, individually.

At the end of the day, facts belong to everybody, any other opinions to us. The distinction is yours to draw...otherwise, the opinions expressed herein are not necessarily those of any employer, not necessarily ours and probably not necessary. Dissenting opinions, discussion or conclusions which may

express an adverse position to those expressed herein by specifically cited authors can be addressed in writing by sending an email to the editor at editor@collisionpublishing.com or sending a “regular mail” letter to us at PO Box 3383, Kirkland, WA 98083-3383. By sending correspondence to either our email or snail mail

addresses listed in this publication you are agreeing that: (1) we are by definition, “the intended recipient” (2) all information in the email is ours to do with as we see fit and make such financial profit, political mileage, or good joke as it lends itself to and, (3) this overrides any disclaimer or statement of confidentiality that may be included on your original message. The entire physical universe, including this publication and attached data disk, may one day collapse back into an infinitesimally small space. Should another universe subsequently reemerge, the existence of this publication and data disk in that universe cannot be guaranteed.

ADVERTISER INDEX

PAGE

4N6XPRT Systems ARAS360 ARC Network CDR User's Summit Collision Safety Institute Crash Data Group (CDR)

Inside Back Cover 1 47

CrashTest-Service.com Engineering Dynamics Eos Systems: Photomodeler Laser Technology, Inc. MapScenes NAPARS Northwestern University VCRware Vericom Computers Visual Statement

77 41 76 8 Inside Front Cover 116 87 2, 55-58 6 9 77 59 32-33, Back Cover

Included with this issue of Collision

SUBMIT AN ARTICLE If you would like to submit an article to be published in Collision please email us your article for review or call us toll free 800-280-7940. Email: editor@collisionpublishing.com

is the 2013 CDR User's Summit data DVD. The CDR User's Summit DVD includes pdf's of the speaker presentations and any other supporting data compiled in a easy to review, menu-driven data DVD.

ADVERTISING RATES & INFO If you would like to advertise your products or services in Collision Magazine, rates, availability and other information are available online at: www.collisionmagazine.com www.collisionpublishing.com

Next CDR User's Summit January 27-29, 2014 www.cdrsummit.com Collision Magazine - Volume 8 Issue 1 5

Premier Industry Partner Letter from the President

• •

“No Event Data Recorders! Where in the Constitution do you find the authority to do this, you tyrannical busybodies...”

“I am concerned that these could be Orwellian surveillance devices. They would not just create databases of how people drive, but when and where...”

Notably, most of those statements are anonymous.

But there are also a small number of realistic and well reasoned comments from consumer safety groups, OEMs, and others that explain how improvements to the current Part 563 Rule are a better compromise. Likewise there are suggestions on lowering the 5 MPH minimum dV to better record pedestrian and low speed crashes, expanding coverage and data recorded, and better defining the data points. But sadly these constructive comments are few.

Late last year NHTSA issued an NPRM (Notice of Proposed Rulemaking), “Federal Motor Vehicle Safety Standards: Event Data Recorders” and opened the floor for a short period of public comment. In a nutshell, NHTSA proposes to expand regulation of light vehicle EDRs using an FMVSS standard, rather than the current regulations in Part 563. While I won’t wade into the raging debate here there were interesting, informative, and many downright humorous views presented in the 1,000+ public comments. The vast majority were of the following type: •

“The assault on privacy and liberty in this country has reached pre-revolutionary levels. You people are would-be fascists, and you can shove this monitoring system up your rectums...”

We should all make an effort to more aggressively monitor these emerging proposals and remain actively engaged in the debate. You can find a searchable index of NPRM’s posted at www.regulations. gov, and sign up to receive emails when comments are made. While comments on this specific NPRM are closed, more will surely follow. NAPARS intends to monitor the issues most affecting our membership and “keep the information flowing”.

NOTICE: Plans for the NAPARS Fall 2014 conference are currently being discussed. Any suggestions about the conference location, venue, topics, and crash testing are not only welcome, but greatly encouraged. Email us at President@NAPARS.org.

Chuck Veppert President, NAPARS

The National Association of Professional Accident Reconstruction Specialists is a non-profit organization whose members have joined together to share the challenge of dealing with the complex problems of accident reconstruction and to upgrade and ultimately professionalize the accident reconstruction field.

6

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

www.napars.org

Industry Partners IAARS is one of the oldest associations dedicated to

expanding the knowledge and training of accident reconstructionists. Formed in 1980, we have members from several different countries as well as many different states across the U.S. We have three levels of membership. Our highest level, Fellow, includes a peer review of a reconstruction project. This peer review feature can be very beneficial to a member seeking expert witness qualification in court. One of the criteria a judge can consider when deciding if a person qualifies as an expert is a peer review of his or her work. We offer opportunities for a member to publish work or research in the field of accident reconstruction. This, also, can be very important when qualifying as an expert witness in court proceedings.

Membership in IAARS has other advantages, including a dis-

ASPACI is a non-profit society whose members have

joined to share in the challenge of analysing the complicated issues of road collision investigation.The Association seeks to promote a professional approach to collision investigation and to encourage the highest standard of ethics, integrity and honesty among its members.

Membership to ASPACI is open to all persons engaged in collision investigation and its allied fields. The association members include Engineers, Police Officers, Independent Consultants, Scientists, Insurance Corporations, Manufacturers and Safety personnel. ASPACI is not a Trade Association or Political Lobby group, nor is it a Police organisation.

The ARC Network

was founded in 1998 and since has grown to be the largest and most respected organization for Accident Reconstruction and Traffic Investigation.

The ARC Network is more than just an organization it is a research and communications Internet portal for everything accident reconstruction. The ARC Network provides research tools, Expert directory, event calendar, industry news feeds, education directlory and a Member's Only area that gives access to various online databases and information.

The ARC Network offers various membership levels to fit your needs and budget. Membership starts at just $79.00 per year.

However, that is where the value of the ARC Network starts.

With your membership you not only get access to everything in the ARC Network, you also receive a, $50 discount to the annual ARC-CSI Crash Conference, a $50 discount to the CDR

count (from IAARS) on a subscription to Collision Magazine, member discounts at our training seminars and others listed on our web page, www.iaars.org.

We have annual training opportunities at different locations across the country. We try to have some sort of actual vehicle testing at each seminar, including full scale vehicle crashes. As a member, you would have access to a vast archive of results from many different and sometimes quite unique tests.

Sincerely, Dan Lofgren President, IAARS www.iaars.org

A recent training initiative advertised by ASPACI labeled the "Australian Crash Data Retrieval (CDR) Summit" may be confused with the actual CDR User's Summit hosted by Collision Publishing in Houston in January each year. The ASPACI offering is not a "CDR Summit" in the same context as that sponsored by Collision Publishing, it is an ad-hoc collection of courses, only some of which are associated with Crash Data Retrieval and is not endorsed by Bosch, the manufacturer of the Crash Data Retrieval System, Crash Data Group, or Collision Publishing.

User's Summit and a free copy of Greg Russell's Excel Crush Stiffness spreadsheet program. Extend your savings and sign up as a Police Agency member or Platinum member and receive a complimentary subscription to Collision Magazine!

In addition, as a member you have the choice of being listed in the most-visited Expert Directory on the Internet. Our Expert directory boasts a #1 ranking in Google searches (a feat worth much more than the cost of membership)!

Scott Baker President The ARC Network (800) 280-7940 www.accidentreconstruction.com arc@accidentreconstruction.com

For information on becoming a Collision Industry Partner send an email - editor@collisionpublishing.com www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 7

Joe O. Montgomery July 27, 1949 - March 30, 2013

T

he crash reconstruction community lost a truly larger than life presence with the passing of Joe Owen Montgomery, 63, on March 30, 2013.

Joe was born July 27, 1949 in Greenville, Texas.

He graduated from Mesquite High School and East Texas State University. Joe served as a Dallas Police Officer for 15 years, and then taught Collision Reconstruction for the Law Enforcement Security Training Division of TEEX. He was a founding member and served on the boards of the Texas Association of Accident Reconstruction Specialists (TAARS past President), and past Chairman of the Accreditation Commission on Traffic Accident Reconstruction (ACTAR).

He and two partners established Complete Accident Reconstruction Service, Inc. (CARS, Inc) in 1984. Mr. Montgom-

ery and one other partner are founding owners of United Transportation Training, Inc, which was formed in December of 2003.

Joe was a consummate gentlemen and one of the true leaders in the crash reconstruction community. Over the years, many came to know Joe to be a principled driving force behind not only TAARS but later ACTAR and, of course, at TEEX. those who were honored to work with him, or enjoyed his presence as an instructor learned much from him.

ACTAR's early development and success are largely a function of his even-tempered and calm, measured approach to problem-solving. Joe's leadership and pursuit of excellence was the reason the TEEX AI/AR programs became such a significant force in the community over the years and his influence has clearly touched AI/AR programs everywhere. Joe positively impacted more lives in more ways than can every really be counted. He brought out the best in the people around him and, both figuratively and literally, his reach and influence cannot be understated.

In so many ways, Joe's passing leaves a true void in the community. Joe's loss to his family, his wife Jackie, and to those who knew and admired him and to the community at large cannot be underestimated.

For more than 75 years, the Northwestern University Center for Public Safety has provided professional development to law enforcement professionals throughout the United States and abroad. NUCPS offers the world-renowned School of Police Staff and Command, as well as courses and programs in police management, police operations, police motorcycle operator and instructor training, crime scene investigation and forensics, crash investigation, transportation engineering and more.

nucps.northwestern.edu/collision nucps@northwestern.edu • 800-323-4011

Crash Investigation and Reconstruction

Upcoming Courses

The concept of scientific crash investigation has been a part of the Northwestern University Center for Public Safety (NUCPS) since its inception in 1936.

We literally wrote the book on crash investigation and reconstruction. In addition to writing the books, we continue to be world leaders in providing training in traffic crash investigation and reconstruction through our many courses of study.

We offer the most comprehensive program of its kind, with the course series beginning at basic crash investigation and culminating in advanced traffic crash reconstruction.

Unless otherwise noted, all classes will be held at NUCPS’s state-of-the-art facilities on the Northwestern University campus in Evanston, Illinois. For an up-to-date listing of crash investigation and reconstruction courses scheduled on campus and throughout the country, please visit nucps.northwestern.edu/collision.

ONLINE Crash Investigation 1 Online, Self-paced Start any time

Crash Investigation 2 Online, Self-paced Start any time

EVANSTON, ILLINOIS Motorcycle Crash Reconstruction September 9–13, 2013

Crash Investigation 1 September 9–20, 2013

Crash Investigation 2 September 23–October 4, 2013 * Register for CDRT1 and CDRT2 at the same time and save $25.

Register now for any on-campus and select online courses and save 20% on subsequent registrations from the same agency for the same course.

Vehicle Dynamics October 7–11, 2013

CDR Technician Level 1 October 10, 2013

Traffic Crash Reconstruction 1 October 14–25, 2013

Traffic Crash Reconstruction 2 October 28–November 1, 2013

CDR Data Analyst November 4–8, 2013

Traffic Crash Reconstruction 3 February 24–28, 2014

Math and Physics Workshop for Crash Reconstruction April 7–11, 2014

Heavy Vehicle Crash Reconstruction May 12–16, 2014

Pedestrian Vehicle Crash Reconstruction May 19–21, 2014

Understanding Heavy Vehicle EDRs September 8–12, 2014

CDR Technician Level 2 October 11, 2013

NOW AVAILABLE FROM NORTHWESTERN UNIVERSITY CENTER FOR PUBLIC SAFETY

TRAFFIC CRASH RECONSTRUCTION,

2ND EDITION

Purchase your copy today. Spend a total of $500 or more and receive a 5% discount. nucps.northwestern.edu/tcrbook 800-323-4011, extension 1-5670

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 9

FURTHER DEVELOPMENTS REGARDING THE DYNAMIC MODELING OF MOTOR VEHICLE COLLISION RESPONSE USING THE SDOF APPROACH Jai Singh, BS, MS, ACTAR Biomechanical Engineering Analysis & Research, Inc.

10

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

I

ntroduction: The utilization of the single degree of freedom approach to the simplified modeling of motor vehicles involved in collinear collisions and collisions that are effectively collinear (i.e. those with minimal rotational effects) has been ubiquitous in the accident reconstruction literature over the past five decades and represents a methodology that is generally well-accepted in the accident reconstruction community. The basic tenets of this modeling approach are bipartite. The first aspect consists of lumping the mass of each collision partner into a nodal mass located at the static center of mass of the collision partner. The second aspect consists of utilizing the appropriate simplified uniaxial structural model for relating the force applied in a collision and the corresponding structural deflection experienced by the impacted region of the collision partner, which can be directly related to the displacement experienced by the collision partner during the impact. We first consider the general case of the two body collinear collision as shown in Figure 1. www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 11

Figure 1: The two collision partner collinear collision.

For the purposes of simplicity, the collision partners and their corresponding parameters are denoted through the use of subscript notation in which the subscript one denotes the first collision partner and two denotes the second collision partner. The displacement experienced by each collision partner, measured with respect to an inertial reference frame, is denoted as u1(t) and u2(t). The single dot and double dot notations are used to denote the first time derivative of displacement (i.e. velocity) and the second time derivative of displacement (i.e. acceleration). The force developed during the collision is denoted as Fc(t).

The collision, itself, occurs over a finite time duration. The first moment in time at which segments of each collision partner attempt to occupy the same region of physical space is the start of the collision and is denoted as occurring at t = to. The terms closure and loading are used synonymously herein. The terminus of closure, defined by the achievement of common velocity between the collision partners, is contemporaneous with the start of the separation, or unloading, phase, and occurs at t = tc. The terminus of the separation phase, defined by the time at which the collision force returns to zero magnitude, is denoted by t = ts. The duration of the closure phase is thusly ∆tco = tc – to. The duration of the separation phase is ∆tsc = ts – tc. The collision duration is ∆tso = ts – to. In the absence of any significant net externally applied forces during the collision duration, the system under consideration during the collision is that of the coupled two collision partners.

The coupled equations of motion are determined through the direct application of Newton’s Second Law.

[1]

[2]

The most lucid and clear derivation, starting from these basic equations, of the relationship between the closing velocity and internal work absorbed , and between separation velocity and internal work recovered, have been previously presented by Carpenter and Welcher (2001). The relationship between closing velocity and internal work absorbed during closure for the system is: 12

Collision Magazine - Volume 8 Issue 1

[3]

The relationship between separation velocity and internal work recovered is:

[4]

In equations (3) and (4), the displacement differential du is that of the displacement, u(t) = u1(t) – u2(t), which is work conjugate with the collision force. The derivation of the term between the equalities makes no assumption regarding the shape of the force-deflection response and holds apt regardless of the shape of the force-deflection response. We introduce the time-varying structural deflection of each collision partner during the collision. Using the definitions of position, displacement, initial length and uniaxial compressed length, the relationship between deflection and displacement are:

[5]

[6]

In equations (5) and (6), the displacement uIF(t) is the time-varying displacement of the common collision interface that develops between the impacted regions of the collision partners. The natural constraint condition is that which is determined by summing the deflections.

[6]

Equations (3) and (4) can thereby be rewritten as:

www.collisionpublishing.com

[7]

[8]

It is of critical importance to note that the deflection of each collision partner is a function of time and that the collision force Fc(δ(t)) is also a function of time. The deflection of each collision partner at t = tc are the peak values of deflection for the uniaxial modeling approach under the conditions of monotonically increasing loading.

Therefore, at t = tc, δ1(tc) = δd1 and δ2(tc) = δd2. Another result of the condition of monotonically increasing force with increasing deflection is that the magnitude of the collision force has a peak value at the terminus of closure.

Therefore, at t = tc, Fc(tc) = Fcp. From the point of peak deflection and peak force, the unload curve progresses in a monotonically decreasing manner until the terminus of separation. The residual damage depths (i.e. crush) are the values of deflection present at the terminus of the separation phase. Therefore, at t = ts, δ1(ts) = δs1 and δ2(ts) = δs2. The residual damage depths are NOT time parametric parameters. A plot of force versus deflection will yield a curve that is parametric in time. This time parametric response maps to a single point on the non-time parametric plot of the barrier equivalent velocity (BEV) associated with the internal work absorbed as a function of residual damage depth or peak collision force as a function of residual damage depth. Suggestions that such a plot can simply be shifted along the abscissa or ordinate, to equate to the time-parametric dynamic response are generally nonsensical (i.e. a force-crush plot is not the same as a shifted

force-deflection plot). The shape of the relationship between a plot of peak force as a function of residual damage depth and the corresponding shapes of the load and unload curves only provide for similitude under a very strict set of conditions. There are no portions of the load curve or the unload curve that can have a tangent stiffness with slope of zero or ∞. Both are nonsensical, even as modeling simplifications. There is nothing within the actual physical system or simplified model that provides a mechanism by which the collision force stays constant for a magically changing deflection or by which the deflection stays constant for a magically changing magnitude of collision force. The latter of the two approaches, unfortunately, has been found in the literature since the 1960s in the context of modeling the first segment of the unload response (Emori, 1968). The corresponding modeled accelerationtime histories have a magic drop in the acceleration over a zero time duration. It is clear that the tangent stiffness at any point on the unload curve will be substantially greater than the tangent stiffness of any point on the load curve but the former cannot be infinite.

S

implified equations for aligned vehicle to FRMB testing Collinear collision testing involving an aligned, full-width engagement, impact between the front or rear of a test vehicle to a fixed, rigid, massive barrier (FRMB) greatly simplifies the work-energy relationships.

This configuration is shown in Figure 2 and the subscripts one and two are used to denote the test vehicle and FRMB, respectively.

Figure 2: Collision configuration for aligned impact between test vehicle and FRMB type testing. www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 13

First, because the barrier does not undergo any rigid body motion during the collision (i.e. it is fixed) and because the deflection of the barrier face is infinitesimally small in comparison to the deflection of the test vehicle, the deflection-displacement relationships become:

[10] [11]

[12]

[13]

The final simplification to the work-energy relationships comes from noting that m2 is substantially greater than m1. Using this fact (i.e. taking the limits of equations (12) and (13) as m2 » ∞), results in the final form of the workenergy relationships for this type of collision test.

To reiterate, thus far, no modeling assumptions have been made regarding the shape of the loading curve or the shape of the unload curve in regards to the derivation of equations (14) and (15).

G

eneral considerations based on NCAP data The New Car Assessment Program (NCAP) test conducted for a 2010 Honda Crosstour AWD EX-L (Test No. 6857) will serve as the baseline collision test for the remainder of this study. The test vehicle mass was 2032 kg and the BEV was 15.6 m/sec. All accelerometer derived data is generated from test vehicle affixed and test vehicle local coordinate system (LCS) oriented accelerometers such that the LCS x-direction (longitudinal) acceleration-time history is considered. Accelerometer based data was derived from the left rear floor (FLRR) and right rear floor (FLRR) test vehicle affixed accelerometers (denoted as x-direction rear crossmember affixed accelerometers in the contractor’s report).

The total barrier force data is generated by summing the values of the individual load cells at all times t = ta. It is critical that total barrier force data not contain substantive drop outs of failures from individual load cells. The case where such is present is easily determinable using the

Figure 3: Total barrier force time history for NHTSA Test No. 6857.

14

Collision Magazine - Volume 8 Issue 1

[15]

[9]

Furthermore, because the FRMB is stationary throughout the collision, the closing and separation velocities reduce to the velocities of the test vehicle at the start of closure and at the terminus of separation. The speed of the test vehicle at the start of closure is its BEV and is a parameter directly related to the internal work absorbed.

[14]

www.collisionpublishing.com

freely available NHTSA Load Cell Analysis (v. 1.6.4.2) program or the freely available NHTSA PlotBrowser (v.

10.11.10.27) program coupled with the test data in UDS1992 format. First, the force-time history can be readily checked for each load cell. Load cells exhibiting a flat line response are those that failed during the collision test.

Secondly, the EV5 instrumentation file can be checked.

Signals from failed load cells, as displayed in the Load Cell Analysis program, have their corresponding columns infilled in red. Thirdly, the total energy can be checked. The sum of the work and kinetic energy (i.e. the total energy of the system) should remain at an approximate constant value. Fourthly, the “momentum” velocity solution can be checked. This solution, should be approximately the same as that derived from the vehicle fixed accelerometers distal to the impact. A clear indication of load cell failure is when the momentum velocity solution at the terminus of separation has a positive value. NHTSA’s Load Cell Analysis program is used herein and automatically filters the accelerometer and barrier load cell data using a SAE CFC60 filter.

It is important to note that even the uniaxial response of the test vehicle under the test conditions under consideration is most indicative of a multiple degree of freedom (MDOF) system. This is readily evidenced by even a cursory examination of the total barrier force time history, as shown in Figure 3.

This event, temporally, occurs within the closure period.

Any corresponding simplified SDOF modeling approach, utilizing monotonically increasing functions, however, shifts the occurrence of peak collision force development to the terminus of closure. In turn, the modeled peak collision force, at the terminus of closure, is greater than the actual force at the terminus of closure and the internal work absorbed is more heavily weighted towards the terminus of closure. Use of the actual value of collision force at peak deflection coupled with the monotonically increasing closure force-deflection modeling approach will result in a substantive underestimate of the internal work absorbed during closure. These facts are important to remember in the context of the modeling simplifications. It is also important to remember that the deflection, equal to the displacement of the test vehicle for this type of collision, is a function of time. This is clearly shown in Figure 4.

The “momentum” solution is derived by the simple inverse mass scaling of the total barrier force data. This acceleration is actually the dynamic center of mass acceleration of the test vehicle. If the vehicle itself is viewed as a system then the total barrier force represents the external force acting on the vehicle. The sum of individual mass weighted accelerations from the test vehicle divided by the total vehicle mass equals the acceleration of the dynamic center of mass of the test vehicle.

The peak collision force response is secondary to the stiff engine of the test vehicle impacting against anterior structures of the vehicles interposed between it and the FRMB.

[16]

Figure 4: Test vehicle displacement (deflection) for NHTSA Test No. 6857. www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 15

Integrating this equation once, with respect to time, provides the velocity solution and integrating again, with respect to time, provides the displacement and deflection solutions. The Load Cell Analysis program does not provide the “momentum” acceleration solution, but this can easily be obtained by importing the total barrier force data into the appropriate external software program and dividing each discrete total barrier force data point by the test vehicle mass. The general shape of the force-deflection response, determined by the parametric plot of total barrier force verses displacement (deflection) at each time t = ta, is shown in Figure 5. The force-displacement (deflection) plot not only provides force and displacement (deflection) data but also encodes time data as both the force and the displacement (deflection) are time parametric quantities.

In the ideal case, the total barrier force would be zero before and up to time t = to. At time t = tc, the displacement (deflection) of the test vehicle would be maximum and the velocity of the test vehicle would be zero. At time t = ts the total barrier force would be zero, the acceleration experienced by the test vehicle due to the collision force would be zero and the magnitude of the negatively signed velocity of the test vehicle would have a maximum value. Under ideal conditions one could use any singular parameter to determine the start of the collision, the time of the terminus of closure and the time of terminus of separation. This ideal situation, however, is rarely the case when working with actual test data. The total barrier force data generally tends to exhibit a certain degree of oscillatory behavior about a non-zero mean even under conditions of zero collision load and the time zero defined as per the test data

is rarely equal to the time at which the total barrier force exhibits the start of the first principal change in response, indicative of the start of the collision. The signal processing methods for correctly scaling and time shifting the response parameters are beyond the scope of the subject study. However, the displacement (deflection) and total barrier force, at the start of the collision are taken as being zero with the resultant (δo, Fc(δo)) at t = to being (0, 0). The time at which closure terminates can be found based on the velocity solution or from the displacement solution. Each vehicle fixed sensor considered along with the momentum solution provides for one solution for each method.

The solution derived from each sensor, for each method, should provide a substantially similar, but not necessarily exactly equal result for the time at which the terminus of closure occurs. Because the data being considered is discrete, the velocity based solution will generally not be one based on finding a data point at which the velocity is zero.

Instead, the solution is found as lying between the last data point for which velocity is positive and the first data point for which velocity is negative. Because of the high sampling rate (10 kHz) and corresponding small time step size (0.001 sec), linear interpolation can be used between these two points to determine the solution at which the velocity is zero. The displacement (deflection) time history can simply be searched to find the maximum displacement (deflection) and the corresponding time. For the terminus of separation, the barrier based and acceleration based solutions require a similar interpolation based approach.

The velocity based approach generally requires the simple finding of the time at which the velocity-time history has

Figure 5: Force-displacement (deflection) plot for NHTSA Test No. 6857 with displacement (deflection) derived from the double integration of the FLLR x-direction accelerometer signal.

16

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

a minimum value. This approach, however, does not work in cases for the “momentum” derived solution in which the total barrier force response under zero collision load approaches the magnitude of the frictional drag experienced by the test vehicle following separation from the barrier.

Overcoming these instrumentation related issues is, in the view of the subject author, a requirement for working with actual dynamic data and can provide a challenging but not insurmountable issue.

O T

bjectives The objective of the subject study consists of redeveloping and evaluating the closed-form analytical solutions for the simplified modeling of the uniaxial collision response for the SDOF model based on the reduction of NCAP test data. heory The simplified SDOF modeling approach based upon use of monotonically increasing analytic closed-form curves for the loading phase and monotonically decreasing analytic closed-form curves for the unloading phase requires the consideration of three factors each. For the loading phase, the factors are the peak deflection, δd1 = δ1(tc), the modeled force at peak deflection Fc(δ1(tc)) and the internal work absorbed, IWA. For the unloading phase, the factors are the residual deformation, δs1 = δ1(ts), the collision force at the terminus of separation, Fc(δ1(ts)) = 0 and the IWR. Also, a C0 continuity condition exists at the contemporaneous terminus of closure and start of separation such that the collision force modeled at the terminus of closure from the loading phase

model must equal the collision force modeled at the start of separation from the unloading phase model. Therefore, for the loading phase, there are initially three unknowns.

For the unloading phase, there are initially two unknowns.

The degree to which any model can exactly fit all unknowns depends on the number of modeling parameters available. Any a priori desire to force the force-deflection response to fit a particular presumed relationship between BEV and residual damage depth or peak force and residual damage (e.g. linear) is entirely eschewed herein. Such relationships must derive from the models for the loading and unloading phases of the force deflection response. It should be amply clear that the use of models that are not ubiquitously uniform in regards to linearity for the loading phase and unloading phase, separately, will not produce a ubiquitously uniform linear relationship between BEV and residual damage depth or peak collision force and residual damage depth. As noted previously, and as shown in Figure 6, the tangent stiffness of any segment of the force-deflection response, cannot have an infinite slope or zero slope. Even as a modeling simplification, this is entirely nonsensical, and has no basis within the physical system or within the modeling methodology.

We consider and develop two modeling approaches for full or segmented loading/unloading response modeling. The simplest approach is that of using the first two terms of a presumed polynomial expansion between collision force and deflection. In simpler terms, this is the linear model (i.e. use of the zero and first order deflection terms). With respect to a reference point (δo, Fco), the force-deflection relationship for any δ such that δ ≥ δo, is given by the fol-

Figure 6: Nonsensical approaches to modeling the tangent stiffness of any segment of the force-deflection response.

The plot on the left depicts a region of infinite stiffness and the plot on the right depicts a region of zero stiffness. www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 17

lowing equation in which k is the linear slope (i.e. stiffness) of the relationship:

For any two points on the force-deflection curve over which equation (17) is operative, the following relationship holds.

[26]

[17]

Equation (19) can be rewritten as shown by equation (27) the resultant substituted into equation (25) to provide a form of the internal work which is a function of the deflections at each point and the force at each point.

[18]

[27]

Equation (18) can be rewritten as: [19]

[28]

The second model under consideration is the power law model. The force-deflection relationship with respect to the reference point is:

Under the case in which the reference deflection and force are both zero (i.e. the origin of the force-deflection response), equation (28) simplifies further to:

[20]

The model parameters a and b are the constant and exponential terms, respectively. The relationship between collision force and deflection, as defined by equation (20), is log-linear.

[21] For any region where equation (20) is operative, the forcedeflection relationship at time t = ta and at time t = tb within this region, is defined as:

[22] [23]

[29]

For the power law model, the internal work is:

[30]

Each power term on the right of the final equality can be separated into the multiple of the difference between the deflection at that point and the reference deflection, raised to the b power and the difference between the deflection at that point and the reference deflection, raised to the first power.

Solving equation (22) for the reference force value and substituting the result into equation (23) yields the following singular relationship.

[24]

The internal work associated the traversal of the forcedeflection response from δa to δb at corresponding times ta and tb is obtained by integrating the force-deflection response over the differential in deflection with the limits set to δa and δb. For the linear model, the internal work is:

Equation (25) can be rewritten as: 18

Collision Magazine - Volume 8 Issue 1

[31]

Using equations (22) and (23):

[32]

[33]

[25]

www.collisionpublishing.com

In the case where the reference deflection and force are zero, equation (33) simplifies to the following form:

[34]

F

ull loading or unloading phase modeling using singular functions The simplest modeling approach is to model the entire loading phase or unloading phase response using a single function. This approach, as well as the subsequently discussed segmental modeling approach, requires the determination of to, tc and ts. Once these values are determined for each sensor and method, the values of δd1 and δs1 for each combination are known parameters.

Because the modeling approach for the loading phase is based on the use of monotonically increasing functions, the measured value of the total barrier force at t = tc is not used. Instead, the modeled force Fc(δd1(tc)) = Fcp is used. The internal work absorbed, the third parameter, is a matched parameter that is utilized in the derivation of Fcp and in its quantification along with the quantification of the model parameters. The separation phase model is based on the form of the model and matching the points (δd1, Fcp) and (δs1, 0). The internal work recovered, for the single linear unload model is a modeled parameter, as the shape of the modeled unload response is fixed between the two points defining the curve. For the power law model, the internal work recovered is an input parameter that is used for quantifying the model parameters.

The single linear model for the loading phase is determined by using equation (28) with the reference deflection and the deflection and collision force at t = ta = to and by setting t = tb = tc.

[35] The stiffness for the linear model for the loading phase, by equation (19), is:

straint equation to solve for the three unknowns consisting of the two loading model parameters and the modeled force at peak deflection. The first point on the load curve with non-zero collision force value and non-zero deflection, however, can serve as the point for t = ta. This can generally be taken to be at t = 0.001 sec (i.e. the first point after the start of the collision). Equations (22), (23) and (34) become:

[38]

[39] [40]

Solving equation (38) for the model parameter aload and substituting the result into equation (39):

[41]

Substitution of equation (41) into equation (40) leads to a nonlinear equation in bload that is explicit in terms of the internal work absorbed and that can readily be solved numerically. This approach differs from that taken by Singh and Perry (2008) in which equation (24) was used and the internal work absorbed was not included as an explicit constraint.

[42]

Once the solution for bload is determined using equation (42), the result is back-substituted into equation (41) to determine Fcp and then aload can be solved for by use of equation (38), (39) or (43).

[43]

[36]

The resulting single power law model for the loading phase force-deflection response can then be written as:

The resulting single linear model for the loading phase force-deflection response can then be written as:

[44]

[37]

For the single power law model for the full loading phase, the origin of the force-deflection curve serves as the reference point but cannot serve to provide a force-based con-

The general shape of the modeled response under each modeling approach is shown in Figure 7.

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 19

Figure 7: Full loading response modeling using a single linear function (left) and a single power law function (right) For the single linear model for the unload response, the unloading stiffness, the single model parameter, is fully defined by the boundary points.

[44]

The modeled internal work recovered is determined from equation (45).

[45]

[47] Equation (30) can also be rewritten as:

[48]

Solving equation (47) for aunload and substituting the result into equation (48) leads to a solution for bunload in terms of the known parameters.

The unload force-deflection response can be written as either equation (46) or equation (47):

[46]

[47]

For the single power law model for the unload phase, the replacement of t = ta with t = tc and the replacement of t = tb = ts coupled with t = to = tc allows equation (24) to be written in the following form:

20

Collision Magazine - Volume 8 Issue 1

[49]

The solution for aunload is then obtained by back-substitution of the solution from equation (49) into equation (47).

The operative form of the unload phase model based on the single power law function can be written as:

[50]

For the model based on a singular loading curve coupled with a single linear loading curve, the continuity condition

www.collisionpublishing.com

allows for the derivation of a closed form analytic relationship for δd1 in terms of δs1. Equations (33) and (46) can both be solved for Fcp and the results equated.

[51]

Substitution of equation (51) into equation (36) provides a relationship between the peak collision force and the residual damage depth without the explicit inclusion of peak deflection.

[55] Equation (55) cannot be written in terms of a closed form analytic solution for δd1 in terms of δs1 but can be readily solved numerically for known model parameters. This relationship, however, can be derived from the collision testing using the definition of the coefficient of restitution.

[56]

[52]

The relationship between the peak collision force and residual damage depth is:

Substitution into equation (35) allows for the determination of the relationship between BEV and residual damage depth.

[57]

[53]

Equation (53) clearly shows why the Campbell b1 linear curve fit parameter is not the same as the circular frequency for a SDOF mass-spring system. Equations (52) and (53) are only operative in the case in which the loading function is uniformly linear and the unloading function is uniformly linear. Changes in the slope of the load or unload curve, such as in the case with multilinear loading or unloading, will change the stiffness terms in equations (52) and (53) and thereby make the relationship between Fcp and δs1 or BEV1 and δs1 multilinear. While linearity may be present for a portion of the loading response and a portion of the unloading response, the assumption of uniform ubiquitous linearity, the only solution by which uniform ubiquitous linearity is present as per the Campbell and CRASH3 curve fit modeling assumptions, is problematic for a very important reason. This assumption fixes the coefficient of restitution as being constant for all cases in which δs1 > 0. From the definition of the coefficient of restitution and equations (35), (45) and (51):

The relationship between BEV1 is not readily reducible secondary to the use of the coefficient of restitution in the derivation between the peak deflection and the residual damage depth. Using this definition, however, and expanding the internal work recovered in terms of the model terms while expanding the internal work absorbed in terms of BEV1 leads to the following result.

S

[58]

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 21

[54]

This result is clearly unrealistic in light of actual test data.

In the case in which a singular power law function is used to model the loading phase and a singular power law function is used to model the unloading phase, the continuity condition yields:

egmented loading and unloading phase models Neither the full loading response nor the full unloading response need be modeled with singular curves. One may, instead, utilize a mixed modeling methodology in which the response transitions at determinate transition points. Mastandrea and Vangi (2005) treated the initial region of the loading response as being linear, correlating to the bumper stiffness, and utilized the collision force value from this transition point on the load curve to create a transition point on the unload curve and thereby bifurcate the unload curve into two differing regions of linear unloading. The Load Cell Analysis software calculates a value for initial stiffness for NCAP tests.

The approach taken is to perform a moving window linear regression analysis on CFC60 filtered and zero truncated force-deflection data such that the corresponding best fit line has a starting point within the first 200 mm of deflection, has a minimum length of 150 mm and has a coefficient of determination of greater than 0.95. Nusholtz et al. (2005) took a similar approach, in regards to fitting an

initial region of linear stiffness, by performing windowed linear fits over the first 250 mm of deflection and the first 400 mm of deflection. For the subject modeling approach, however, the requirement is that the modeled force-deflection response start at (0, 0) and the transition points lie on the actual force-deflection response. One method of determining an appropriate set of points for consideration is through the first derivative test of the force-deflection response. The inflection points of the original response have a first derivative with respect to deflection of zero. Once the inflection points are determined, they can be modeled as discussed above with the total internal work absorbed serving as the constraint during loading and the internal work recovered serving as either the modeled result or input for the segmental internal work recovered.

W

orked Example The worked example, as noted previously, is that for NHTSA Test No. 6857. Filtered displacement (deflection), velocity, acceleration and total barrier force time history data was exported into a series of text files from the NHTSA Plot Browser software and imported into a symbolic mathematics software package (Mathematica v. 8.0; Wolfram Research; Champaign, Illinois). The analysis for the parameters at the terminus of closure and the terminus of separation were conducted using each method described previously and for the “momentum” based solution in addition to the accelerometer solutions. For the velocity based solution to the terminus of closure, the bounding velocities for the momentum, FLLR and FLRR solutions were, respectively, (1.17 x 10-2 m/sec, -6.69 x 10-2 m/sec), (1.38 x 10-2 m/sec, -3.27 x 10-2 m/sec) and (3.60 x 10-2 m/sec, -2.19 x 10-2 m/ sec). The corresponding times that bounded t = tc were (79.3 msec, 79.4 msec), (83.2 msec, 83.3 msec) and (81.6 msec, 81.7 msec). The velocity based deflections at the terminus of closure, for each method, differed by 1 x 10-5

m between the successive indices. Correspondingly, only the first indexed value is provided. These solutions were 6.80 x 10-1 m, 7.77 x 10-1 m and 7.74 x 10-1 m. The actual total barrier force at the bounding values for closure, for each method, was (3.77 x 106 N), (3.42 x 106 N, 3.41 x 106 N) and (3.64 x 106 N, 3.63 x 106 N).

The actual peak deflections were either their corresponding values at the first or second index for the velocity based method for the momentum and FLLR solution but differed by 2 x 10-5 m for the FLRR solution. The solutions for the peak deflection based velocity at the terminus of closure were -6.69 x 10-3 m/sec, 1.38 x 10-2 m/sec and 3.28 x 10-2 m/sec. The peak deflection based solutions for the time at which closure terminated were 79.4 msec, 83.2 msec and 82.2 msec. The force values at the terminus of closure were 3.77 x 106 N, 3.43 x 106 N and 3.58 x 106 N. The barrier based time at which separation terminated was 146.5 msec with corresponding separation velocities of -2.50 m/sec, -1.56 m/sec and -1.85 m/sec and corresponding residual deformation values of 5.40 x 10-1 m, 6.67 x 10-1 m/sec and 6.44 x 10-1 m/sec. The velocity based solutions for the separation velocity were -2.50 m/sec, -2.14 m/sec and -2.44 m/sec. The corresponding times at which separation reached a terminus were 145.3 msec, 113.3 msec and 115.1 msec. The velocity based residual deformation values were 5.43 x 10-1 m, 7.29 x 10-1 m and 7.12 x 10-1 m. The acceleration based times for the terminus of separation were between 146.5 and 146.6 msec, between 114.3 and 114.4 msec and between 103.5 and 103.6 msec.

The corresponding separation velocities were -2.50 m/sec (differences on the order of 10-5 m/sec), -2.12 m/sec (differences on the order of 10-4 m/sec) and between -2.29 m/ sec and -2.30 m/sec. The residual deformation was 5.40 x 10-1 m (differences on the order of 3 x 10-4 m), 7.26 x 10-1

Figure 8: Loading phase force-deflection response for the velocity based (left) and displacement based (right) solutions.

22

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

Figure 9: Unloading responses with the first row being velocity based loading and the second row being displacement based loading. The first column depicts barrier based unloading. The second depicts velocity based unloading. The third column depicts acceleration based unloading. In each chart the solutions (left to right) are momentum, FLLR and FLRR (blue, brown and red respectively). m (differences on the order of 2 x 10-4 m) and between 7.40 x 10-1 m and 7.39 x 10-1 m. The measured centerline post-collision shortening of the test vehicle was 5.71 x 10-1 m. The reported time of separation was 114 msec with a corresponding separation velocity of -2.28 m/sec. Figure 8 shows the loading closure phase force-deflection response for the velocity and displacement based methods for the momentum, FLLR and FLRR solutions.

Three methods were used for modeling the unloading phase for each loading phase. The results are shown in Figure 9.

The unloading responses for the accelerometer derived velocity and acceleration based solutions revealed a not unexpected response of the terminus of the curve not ending at zero or steady state total barrier force. The internal work absorbed was 2.49 x 105 J in all cases. The internal work recovered for the barrier, velocity and acceleration methods, in units of 103 J, was (6.34, 2.44, 3.48), (6.34, 4.64, 6.05) and (6.34, 4.58, 5.34). The corresponding coefficients of restitution x (10-1) were (1.60, 0.995, 1.18), (1.60, 1.37, 1.56) and (1.60, 1.36, 1.47). The internal work absorbed and peak deflection were used as the target values, with the reference value being (0, 0), for the full loading response linear model. The calculated force at peak deflection from the linear load model was then used in conjunction with the method based residual damage depth and zero barrier force point for determining the unloading stiffness. The force at peak deflection and the internal work recovered

represent the modeled parameters for the linear loading followed by linear unloading response. The modeling results are presented in Table 1.

The singular linear loading phase models resulted in a modeled force at peak deflection, which was approximately twice the value of the actual force at peak deflection.

Using the actual force values at peak deflection would have resulted in an incorrect internal work absorbed prediction.

Correspondingly, the modeled internal work recovered was generally one order of magnitude higher than the actual internal work recovered. The linear unload stiffness was substantially greater than the loading stiffness, which was an expected result. The results for singular power law model for the loading phase are shown in Table 2 (using the force at peak deflection as an input) and Table 3 (treating the force at peak deflection as a modeled parameter).

The models generated, as shown in Table 2, had extremely rapid rise times (almost reaching peak value within the first 0.05 m) followed by a gradual increase to the terminus of closure. This response was highly unsuitable for the subject formulation and was dropped from the potential solution set.

The modeled peak force values were greater than those found for the linear model. The results for the singular power law model for the unloading phase are shown in Table 3.

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 23

Loading Method

Momentum

FLLR

FLRR

Momentum

FLLR

FLRR

Stiffness (x10 N/m) Stiffness (x10 N/m)

Stiffness (x10 N/m) Force (x10 N)

Force (x10 N) Force (x105 N)

Velocity

1.07

0.823

0.829

7.3

6.4

6.42

Displacement

1.07

0.823

0.829

7.3

6.4

6.42

Unloading Method

Momentum

FLLR

FLRR

Momentum

FLLR

FLRR

Stiffness (x 106 N/m)

Stiffness (x 106 N/m)

Stiffness (x 106 N/m) IWR (x 104 J)

IWR (x 104 J)

IWR (x 104 J)

6

6

6

5

5

VBBB

5.2

5.81

4.94

5.13

3.52

4.17

VBVB

5.32

13.3

10.3

5.02

1.54

2

VBAB

5.11

12.7

18.4

5.22

1.61

1.12

DBBB

5.2

5.82

4.94

5.13

3.52

4.17

DBVB

5.32

13.3

10.3

5.02

1.54

2

DBAB

5.11

12.7

18.4

5.22

1.61

1.11

Table 1: Stiffness values for full loading and unloading phase modeling using singular linear curves.

For the unloading response, under, Method, the first two letters refer to the loading model (VB = velocity based and DB = displacement based) while the last two letters refer to the unloading model (BB = barrier based, VB = velocity based and AB = acceleration based).

Method

Momentum

FLLR

FLRR

Momentum

FLLR

FLRR

a (x 105)

a (x 105)

a (x 105)

b (x 10-2)

b (x 10-2)

b (x 10-2)

Velocity

3.82

3.49

3.77

3.3

7.17

13.5

Displacement

3.82

3.49

3.68

3.22

7.18

11.1

Table 2: Model parameters for full loading phase modeling using singular power law functions in which the force at peak deflection was used in the parameter quantification.

Method

Momentum

FLLR

Momentum

FLLR

FLRR

a (x 10 )

a (x 10 )

a (x 10 )

Force

Force

Force

b

b

b

(x 10 N)

(x 10 N)

(x 105 N)

1.45

1.02

1.03

8.61

7.37

7.4

1.36

1.3

1.31 8.61

7.37

7.4

6

Velocity Displacement

FLRR 6

6

1.45

1.02

1.03

1.36

1.3

1.31

5

5

Table 3: Model parameters for full loading phase modeling using singular power law functions and the corresponding modeled force values at peak deflection.

Method

Momentum a

FLLR a

Momentum b

FLLR b

FLRR b

29

VBBB

2.20 x 10

3.22 x 10

VBVB

1.44 x 1021

3.13 x 10

18.1

32.0

26.7

4.42 x 1014

6.93 x 1013

17.7

6.67

6.61

3.12 x 10

21

1.17 x 10

15

VBAB

2.72 x 10

18.4

7.09

3.82

DBBB

2.20 x 10

21

3.22 x 10

36

29

3.14 x 10

18.0

32.0

26.7

DBVB

1.44 x 1021

4.42 x 1014

6.97 x 1013

17.7

6.67

6.62

DBAB

3.12 x 10

1.17 x 10

2.75 x 10

17.7

7.09

3.82

21

21

FLRR a 36

15

11

11

Table 4: Model parameters for full unloading phase modeling using singular power law functions.

Under, Method, the first two letters refer to the loading model (VB = velocity based and DB = displacement based) while the last two letters refer to the unloading model (BB = barrier based, VB = velocity based and AB = acceleration based).

24

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

The quantified model parameters based on the momentum solution were generally on the order of 1021 for the aunload coefficient and 101 for the bunload coefficient. The accelerometer based solutions for the model parameter, aunload, were on the order of 1029 to 1036 for the a coefficient for the barrier based separation determinations and on the order of 1011 to 1015 for the velocity and acceleration based separation determinations. The bunload coefficients were approximately twice the value of the momentum solutions for the accelerometer based solutions and less than half the value of the momentum based solutions for the velocity and acceleration based determinations.

One subset of the models generated (velocity based closure solution and barrier based separation solution) are shown in Figure 10 for the single linear-linear function modeling approach and in Figure 11 for the single power law-power law function modeling approach. The actual and modeled unload responses terminate at the same value for each plot.

The scale of the ordinate axis coupled with the relatively large change in deflection over a relatively small change in force (with respect to the scale of the ordinate axis) for the terminal region of the unload curve obscures this fact.

For the segmented modeling approach, the first derivative test of the loading and unloading phase, separately,

was used to find a potential set of transition points. The numerical differentiation of the force-deflection response with respect to deflection was carried out using the Haar Wavelet transform (Karimi and Robbersmyr, 2010). The first point from each pair generated that bounded each crossing of the deflection axis for the unscaled first derivative were considered. For the loading phase with an initial region of linear loading, a linear model was generated between (0, 0) and each inflection point. The best fit model was determined using a deflection cut-off of 0.05 m and by the minimum value of the sum of squared error (SSE) terms between the linear fit and the actual data. For all cases, the solution mapped to the second critical point.

The second inflection point, the values at the terminus of closure and the difference between the total internal work absorbed and the internal work absorbed for the initial linear region were used for the power law model for the second region of the loading response. The results are shown in Table 5.

The stiffness of the initial linear region, as expected, was less than that of the full loading curve model (67% for the momentum solution and 87% for the accelerometer based solutions). The power law model coefficients were greater than their corresponding values for the full loading phase models (2.9 and 4.1 times as great for the aload and bload

Figure 10: Actual force-deflection response and modeled force-deflection response using single linear functions for the load and unload phase. From left to right, the solutions are for “momentum,” FLLR and FLRR. Shown on each plot are the loading portion of the actual force-deflection response (orange), the unloading portion of the actual force-deflection response (green) and the modeled force-deflection response.

Figure 11: Actual force-deflection response and modeled force-deflection response using single power law functions for the load and unload phase. From left to right, the solutions are for “momentum,” FLLR and FLRR. Shown on each plot are the loading portion of the actual force-deflection response (orange), the unloading portion of the actual force-deflection response (green) and the modeled force-deflection response. www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 25

coefficients, respectively, for the momentum solution and 2.9 and 18 as great for the accelerometer based solutions).

The visual difference between the full loading phase power model and the segmented linear-power model, however, were marginal. Figure 12 depicts the loading phase model for the three methods as well as the actual response for the “momentum” based solution with terminus conditions derived on a velocity basis.

Numerical differentiation of the unload response revealed two inflection points for each of the six curves. The first inflection point was within the first ten samples following the terminus of closure and correspondingly was not used. The second inflection point, however, was used. The deflection and force values corresponding to the second inflection point are shown in Table 6.

The magnitude of the collision force at the inflection point, for the unloading response, for each loading and Method

unloading response method, was substantially lower than the corresponding inflection point on the load curve. The stiffness values and segmental IWR for the linear fit to the first and second regions of the unloading response are shown in Tables 7 and 8.

The results of the first segment of the unload curve modeled using a power law function (peak values based on a singular linear load function) are shown in Table 9.

Both the first segment linear model and first segment power law model exceeded the actual IWR as shown in Table 10.

Because the second data point for the second unload region contains a zero value of Force (i.e. Fc = for  = s) a second force based equation could not be developed. Furthermore, because the IWR for both the linear and power modeling for the first segment already exceeded the actual

Momentum

FLLR

FLRR

Momentum

FLLR

FLRR

Transition

Transition

Transition

Stiffness

Stiffness

Stiffness

(d, F)

(d, F)

(d, F)

(x 10 N/m)

(x 10 N/m)

(x 105 N/m)

(mm, 105 N)

(mm, 105 N)

(mm, 105 N)

Velocity

(140, 1.00)

(132, 0.889)

(132, 0.889)

7.16

6.72

6.72

Displacement

(140, 100)

(132, 0.889)

(132, 0.889)

7.16

6.72

6.72

a (x 106)

a (x 106)

a (x 106)

Force

Force

Force

b

b

b

(x 10 N)

(x 10 N)

(x 105 N)

1.41

1

1.01

8.37

7.19

7.22

1.36

1.3

1.31

1.41

1

1.01

8.37

7.19

7.22

1.36

1.3

1.31

Velocity Displacement

5

5

5

5

Table 5: Model parameters for the loading response based on an initial linear model followed by subsequent power law modeling to the terminus of closure.

Method VBBB

Momentum

FLLR

FLRR

Momentum

FLLR

FLRR

δ (x 10-1 m)

δ (x 10-1 m)

δ (x 10-1 m)

F (x 104 N)

F (x 104 N)

F (x 104 N)

6.47

7.51

7.43

6.94

7.15

7.27

VBVB

6.47

7.65

7.59

7.02

13.6

13.9

VBAB

6.47

7.64

7.64

6.94

13

17.3

DBBB

6.47

7.51

7.43

6.94

7.15

7.21

DBVB

6.47

7.66

7.59

7.02

13.7

13.7

DBAB

6.47

7.64

7.65

6.94

13

17.1

Table 6: Deflection and force values at the inflection point chosen for bifurcation of the unloading response. Under, Method, the first two letters refer to the loading model (VB = velocity based and DB = displacement based) while the last two letters refer to the unloading model (BB = barrier based, VB = velocity based and AB = acceleration based).

26

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

Figure 12: Full loading phase (left) and loading response up to the transition point for the linear-power model. (P = full loading curve power law model, L-P = mixed linear-power law model, Actual = actual data and Linear = full loading curve linear fit).

Method

Momentum

FLLR

FLRR

Momentum

FLLR

FLRR

Stiffness

Stiffness

Stiffness

IWR

IWR

IWR

(x 107 N/m)

(x 107 N/m)

(x 107 N/m)

(x 104 J)

(x 104 J)

(x 104 J)

VBBB

1.99

2.21

1.79

1.10

0.730

0.900

VBVB

2.01

4.37

3.37

1.08

0.291

0.376

VBAB

1.99

4.10

5.01

1.10

0.316

0.220

DBBB

1.99

2.21

1.79

1.10

0.730

0.909

DBVB

2.01

4.42

3.28

1.08

0.286

0.388

DBAB

1.99

4.10

4.99

1.09

0.316

0.321

Table 7: Stiffness and IWR for the linear model of the unload curve between the terminus of closure/start of separation and the inflection point.

Method

VBBB VBVB VBAB DBBB DBVB DBAB

Momentum Stiffness (x 103 N/m) 6.47 6.73 6.32 6.47 6.73 6.32

FLLR Stiffness (x 103 N/m) 8.48 37 34.4 8.48 37.1 34.5

FLRR Stiffness (x 103 N/m) 7.38 29.3 68 7.34 29.1 67.9

Momentum IWR (x 103 J) 3.72 3.68 3.81 3.71 3.69 3.81

FLLR IWR (x 103 J) 3.02 2.49 2.48 3.02 2.52 2.47

FLRR IWR (x 103 J) 3.57 3.29 2.2 3.54 3.21 2.16

Table 8: Stiffness and IWR for the linear model of the unload curve between the inflection pointand the terminus of separation.

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 27

Method

VBBB VBVB VBAB

Momentum

FLLR

FLRR

Momentum

FLLR

FLRR

a

a

a

IWR

IWR

IWR

b (x 101)

b (x 101)

b (x 101)

(x103 J)

(x103 J)

(x103 J)

5.37 x 1013

8.80 x 1012

4.02 x 1011

9.41

6.7

8.33

4.7

6.51

5.22

6.11 x 1013

1.44 x 1017

3.40 x 1014

9.33

3.76

4.92

4.74

10.4

78.6 9.41

3.98

3.35

9.41

6.7

8.37

9.33

3.75

5.04

9.41

3.98

3.42

5.37 x 10

13

4.7 DBBB DBVB DBAB

4.28 x 10

5.97 x 1017

9.87

108

16

8.80 x 10

3.76 x 1011

4.7

6.51

65.1

6.11 x 1013

1.87 x 1017

2.31 x 1014

4.74

10.5

7.71

5.37 x 10

5.36 x 10

13

13

4.7

12

4.28 x 10

3.79 x 1017

9.87

106

16

Table 9: Model parameters and calculated IWR for first unload segment power law modeling.

Barrier based

Velocity based

Acceleration based

IWR

IWR

IWR

(x10 J)

(x10 J)

(x103 J)

Momentum

6.34

2.46

3.48

FLLR

6.34

4.64

6.05

FLRR

6.34

4.58

5.34

3

3

Table 10: Actual internal work recovered based upon each method of determining the terminus of separation and each signal.

Figure 13: Actual unload response (red) for the “momentum” solution using the velocity based terminus of closure determination and the barrier based terminus of separation determination. The first segment linear model is shown in blue. The first segment power law model is shown in green. The second segment linear model is shown in brown. The second segment power law model for the first segment linear model is shown in purple and for the first segment power law model is shown in yellow.

28

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

IWR, the difference between the actual IWR and the first segment IWR could not be used as a constraint. Figure 13 shows one potential final result of the combined loading and unloading response (momentum, velocity based terminus of closure and barrier based terminus of separation). The loading model shown is that of the single linear model. The unloading response is the mixed power law followed by linear unloading.

The total internal work recovered based on this modeling approach and the actual internal work recovered are shown in Table 13.

While the use of the actual force at peak deflection leads to an underestimate for the modeled internal work absorbed, when compared to the actual internal work absorbed, for the modeling functions that are monotonically increasing, the unload response can be considered separate of the loading response. We consider the first segment unload response based upon the linear and power law approach using the first inflection point and the actual force and deflection at the terminus of closure. The results for the first segment linear model are shown in Table 11.

Because both the first segment linear model and the first segment power law model predicted values for the internal work recovered that was less than the actual internal work recovered (except for the accelerometer and barrier derived solutions), the difference between these values could serve as an input for the power law model for the second phase of the unload curve. The model coefficients for second segment power law model are shown in Table 15.

For the segmental bilinear unload model based on Tables 11 and 8, the segmental internal work recovered are shown in Table 12.

Method

Modeling the first segment between the actual peak deflection and collision force values and the first inflection point using a power law model resulted in the solutions shown in Table 14.

Figure 14 depicts the actual unload curve for the “momentum” solution using the velocity based terminus of closure determination and the barrier based terminus of separation determination along with the segmental models.

Momentum

FLLR

FLRR

Momentum

FLLR

FLRR

Stiffness

Stiffness

Stiffness

IWR

IWR

IWR

(x 10 N/m)

(x 10 N/m)

(x 10 N/m)

(x 10 J)

(x 10 J)

(x 103 J)

VBBB

9.26

10.5

9.18

5.11

3.46

4.6

VBVB

9.35

17.8

15

5.03

1.19

1.67

6

6

6

3

3

VBAB

9.26

17

20.3

5.11

1.31

0.892

DBBB

9.26

10.6

8.95

5.11

3.49

4.55

DBVB

9.35

18.1

14.6

5.03

1.17

1.7

DBAB

9.26

17.1

19.4

5.11

1.32

0.891

Table 11: Stiffness and IWR for the linear model of the unload curve between the actual force and deflection at the terminus of closure/start of separation and the inflection point.

Method

First segment IWR (x 103 J)

Second segment IWR (x 103 J)

Momentum

FLLR

FLRR

Momentum

FLLR

FLRR

VBBB

5.11

3.46

4.6

3.72

3.02

3.57

VBVB

5.03

1.19

1.67

3.68

2.49

3.29

VBAB

5.11

1.31

0.892

3.81

2.48

2.2

DBBB

5.11

3.49

4.55

3.71

3.02

3.54

DBVB

5.03

1.17

1.7

3.69

2.52

3.21

DBAB

5.11

1.32

0.891

3.81

2.47

2.16

Table 12: Modeled segmental internal work recovered for the linear-linear unload model based on Tables 10 and 8.

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 29

Method

Modeled Total IWR (x 103 J)

Actual IWR (x 103 J) (% difference)

VBBB VBVB

Momentum

FLLR

FLRR

Momentum

FLLR

FLRR

8.83

6.48

8.17

6.34

2.46

3.48

-39.30%

-163%

-135%

6.34

4.64

6.05

-37.40%

(-20.7%)

(-18.0%)

6.34

4.58

5.34

-40.70%

(-17.2%)

(-42.1%)

6.34

2.46

3.48

-39.10%

-165%

-133%

6.34

4.64

6.05

-37.50%

(-20.5%)

(-18.8%)

6.34

4.58

5.34

-40.70%

(-17.2%)

(-42.9%)

8.71

3.68

4.96

VBAB

8.92

3.79

3.09

DBBB

8.82

6.51

8.09

DBVB DBAB

8.72 8.92

3.69 3.79

4.91 3.05

Table 13: Total modeled segmental internal work recovered for the linear-linear unload model based on Table 12 and the actual internal work recovered based on Table 10. The percent difference is calculated as 100 x (modeled – actual) / actual.

Figure 14: Full force-deflection response showing actual and modeled components. The transition between the actual load and unload curve is shown by the solid circle. The transitions between the modeled load and first segment unload curve is shown by the long dashed circle while the transition between the first segment and second segment unload curve is shown by the short dashed circle.

30

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

Method VBBB VBVB

Momentum

FLLR

FLRR

Momentum

FLLR

FLRR

a / b (x 101)

a / b (x 101)

a / b (x 101)

IWR (x103 J)

IWR (x103 J)

IWR (x103 J)

1.71 x 10

4.19 x 10

9

6.94 x 10

6.08

4.45

5.74

3.38

4.64

3.86

1.81 x 1011

1.91 x 1012

1.09 x 1011

6.03

2.58

3.49

3.4

6.15

4.93 6.08

2.72

2.4

DBBB

6.08

4.47

5.72

DBVB

6.03

2.55

3.54

DBAB

6.08

2.73

2.43

10

1.92 x 10

2.16 x 1012

3.38

5.97

6.1

1.71 x 1011

4.38 x 1010

6.00 x 109

3.38

4.66

3.81

1.81 x 1011

2.31 x 1012

7.45 x 1010

3.4

6.23

4.79

1.71 x 1011

1.30 x 1012

1.28 x 1012

3.38

6

5.9

1.71 x 10

VBAB

11

11

12

Table 14: Model parameters and internal work recovered for the power law model for the first segment of the unload curve between peak deflection, actual collision force and the first inflection point.

Method

First segment linear

First segment power

Momentum

FLLR

FLRR

Momentum

FLLR

FLRR

a / b (x 101)

a / b (x 101)

a / b (x 101)

a / b (x 102)

a / b (x 102)

a / b (x 102)

8.16 x 10

VBBB

3.53 x 10

VBVB

1.58 x 1011

4.29 x 108

1.03 x 108

2.05 x 1032

8.22 x 1010

9.44 x 109

3.36

2.91

2.31

1.46

0.494

0.402

11

36

3.55

3.57 x 10

VBAB

1.7

11

3.55

4.57 x 10

6.05 x 10

2.94

2.87

8

1.04 x 10

8

37

1.7

1.93 x 10

2.89 x 1010

0.526

0.439

11

8.32 x 10

DBBB

3.53 x 10

DBVB

1.58 x 1011

4.42 x 108

9.65 x 107

2.08 x 1032

7.89 x 1010

9.83 x 109

3.36

2.92

2.28

1.46

0.493

0.403

DBAB

3.57 x 1011

4.69 x 108

5.72 x 108

1.06 x 1037

2.05 x 1011

2.92 x 1010

3.55

2.92

2.85

1.7

0.529

0.44

11

36

3.55

1.46

Table 15: Model parameters for the power law model for the second segment of the unload curve.

R

eferences 1. Carpenter, NJ and JB Welcher (2001) Stiffness and crush energy analysis for vehicle collision and its relationship to barrier equivalent velocity (BEV). Society of Automotive Engineers Technical Paper No. 2001-01-0500.

2. Emori, R (1968) Analytical approach to automobile collisions. Society of Automotive Engineers Technical Paper No. 680016.

3. Karimi, HR and KG Robbersmyr (2010) Waveletbased signal analysis of a vehicle crash test with a fixed safety barrier. In: Proceedings of the 4th European Computing Conference, 183-188.

4. Mastandrea, M and D Vangi (2005) Influence of braking force in low-speed vehicle collision. Proceedings of IMechE. Part D: Journal of Automotive Engineering 219, 151-164.

5. Nusholtz, GS, L Xu, Y Shi and L Di Domenico (2005) Vehicle mass, stiffness and their relationship.

In: Proceedings of the 19th Enhanced Safety of Vehicles Conference. Paper No. 05-0413-O. 6. Singh, J and J Perry (2008) Multilinear and nonlinear power law modeling of motor vehicle force-deflection response for uniaxial front impacts. Accident Reconstruction Journal 18(6), 30-38, 62.

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 31

CRASH WITH US...

Easy to Use • 2D • 3D • Accurate • Accident & Crime Scene Testimonial When I first became interested in Accident Reconstruction, I researched a lot of programs. I found that Visual Statement was one of the best, if not the best. I was employed as a State Trooper at the time, and the department eventually purchased the program for the acting Accident Reconstructionists. [...] After starting my own business, I tried a couple of different software companies that were okay, but did not have the capabilities or the ease of use as EdgeFX. I finally buckled down and purchased the program and I am extremely pleased. [...] My recommendation is to go with Visual Statement !!!

Dale Donham Owner, Donham Accident Reconstruction, AR

visualstatement.com/demo

SIMPLY THE BEST RECONSTRUCTION SOFTWARE

• Diagramming • Reconstruction • Animations • Point Cloud

CALL US NOW TO

BOOK A LIVE DEMONSTRATION

OR TO FIND OUT MORE

1.888.828.0383

Hyundai and Kia Crash Data A Preliminary Overview By: Rusty Haight, Shawn Gyorke, and Sean Haight

B

ackground The Bosch Crash Data Retrieval (CDR) Tool was first offered for sale, commercially, in 2000 largely as a function of a relationship between General Motors and, at the time, a scan tool company Vetronix. In the ensuing years, Ford (in 2003) then Chrysler (in 2008) and finally Toyota (in 2011) reached agreements to allow access to crash data from modules in their vehicles to be part of the CDR Tool functionality.

In 2012, the provisions of Title 49 of (US) Code of Federal Regulations, part 563 (49CFR563 or what has become known as “the rule” or “part 563”) took effect.

“The rule” specifies that for “...vehicles manufactured on

34

Collision Magazine - Volume 8 Issue 1

or after September 1, 2012, if they are equipped with a ... device or function in a vehicle that records the vehicle's dynamic time-series data during the time period just prior to a crash event ... intended for retrieval after the crash event... (the) manufacturer ... shall ensure by licensing agreement or other means that a tool(s) is commercially available that is capable of accessing and retrieving the data stored in the EDR...” 1 In the run-up to September 2012, the (US) National Highway Traffic Safety Administration (NHTSA) had published the provisions of “the rule” in the Federal Register and had responded to several petitions for reconsideration as well as comments published pursuant to notices of proposed rule making (NPRM). During

www.collisionpublishing.com

that process, some modifications and clarifications were ultimately made to provisions of “the rule;” however, one area in particular that was never really thoroughly addressed was the passage found in 563.12 specifying that manufacturers: “... of a motor vehicle equipped with an EDR shall ensure by licensing agreement or other means that a tool(s) is commercially available that is capable of accessing and retrieving the data stored in the EDR that are required by this part. The tool(s) shall be commercially available not later than 90 days after the first sale of the motor vehicle ...”

When “the Final Rule” was published, the only commercially available tool designed to access recorded crash data was the Bosch CDR Tool. On the one hand, many anticipated then that, given the limited market for this type of tool/system and experience (which included at least one other potential tool supplier expressing interest in developing an alternative tool and then deciding against it as a function of the product development cost versus the market potential), the Bosch CDR Tool would become the tool widely adopted by other OEs who would be looking to be “563" complaint. On the other hand, “the rule” was written specifically to leave it to the OEs to find a way to “be compliant” inasmuch as NHTSA, a government agency, couldn’t specify a particular vendor/supplier for the tool which would enable OE compliance.

At one point prior to the implementation of the “final rule,” Toyota had announced publicly (and even went so far as to post to their web site) that they were developing a tool for access to data recorded in their vehicles. In fact, Toyota was using what became known as their “Read Out Tool (ROT)” for some time. Ultimately, they worked with Bosch to include Toyota line vehicle access as part of the Bosch CDR Tool functionality. Similarly, other OEs who, at one time or another and in one way or another,

expressed objection to “the rule” or suggested the probability that they would develop separate or parallel tools have ended up working to make it such that data from their vehicles has become accessible using the Bosch CDR Tool.

Notable exceptions; however, have emerged.

As of this writing, Land Rover and Jaguar (which were purchased from Ford by Tata Motors around the time of the 2009 auto industry “troubles”) have a process in place which requires someone looking to get data from their relevant system(s) to arrange the retrieval with them (the OE). In a rather unique interpretation of “the rule,” Rover and Jag then required that the retrieved “data” be sent to (at last report) the UK for translation to what one might hope to anticipate is a “compliant” report form/format. In “563,” we find a passage (at 563.12) specifying that manufacturers: “... of a motor vehicle equipped with an EDR shall ensure by licensing agreement or other means that a tool(s) is commercially available that is capable of accessing and retrieving the data stored in the EDR that are required by this part...(emphasis, authors).” Notably, in the strictest reading of this part of “563,” the commercial availability of access to a tool that “accesses” and “retrieves” data but does not generate a report would satisfy this requirement. By extension, or perhaps in application, one might point to the remainder of “the rule” which addresses (for example, at 563.8(a) Table III) that the “... data elements ... must be reported in accordance with the range, accuracy and resolution specified in Table III ... (emphasis, authors).” While there is, admittedly, still no requirement in “the rule” specifically calling for the retrieval tool to generate a report, or “when” a report might be generated relative to the access and retrieval, clearly that was anticipated by NHTSA as “the rule” was drafted although it wasn’t actually laid out in “the rule’s” final format.

Editor’s note: While the Bosch Crash Data Retrieval Tool remains the primary commercially available system used by collision reconstructionists to access and retrieve crash data stored in passenger cars, light trucks and SUVs, other options have emerged which allow access to crash data recorded in modules in those few vehicles not currently covered by the Bosch CDR Tool. This first-of-its-kind review is a brief comparison of what those other systems offer - and don’t offer - compared to that afforded by the Bosch CDR Tool.

In addition to this article, some limited information about these systems has already been incorporated into the Collision Safety Institute’s CDR Data Analyst course curriculum including a comparison of the data which may be retrieved by way of the other options to data from instru-

mented crash tests. However, since that course is designed to train and support users of the Bosch CDR Tool, not these other tools, the information there is, of course, relatively limited.

For a detailed and comprehensive look at data from tools such as those offered for non-Bosch CDR Tool supported vehicles - including comparisons to a wider range of fully instrumented crash tests - those interested in these systems should consider attending the CDR User’s Summit in Houston, TX in January, 2014 where the use of these tools and detailed reliability comparisons from crash tests such as those mentioned herein as well as other related issues will be more fully presented.

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 35

Another exception, and the larger focus of this narrative, is the emergence of a tool - or tools depending on one’s perspective - capable of allowing access to and data retrieval from Hyundai and Kia vehicles. Announced about the time “the rule” took effect, these OEs have opted to use a tool(s) made by Global Information Technology (“GIT”) to be “563 compliant.”

Since 2001, “GIT” has been the “OE” supplier for diagnostic tools for both Hyundai and Kia. Editorially, one might recall that the origin of what is now the Bosch CDR Tool was based on collaboration between then Vetronix and General Motors where Vetronix was the supplier of the Tech I and Tech II scan tools as used by GM dealerships. When the “GIT”-based “EDR” application(s) for Hyundai and Kia were announced, the Hyundai kit became available through GIT America’s Aftermarket Support division and the Kia-specific system from Snap-On Business Solutions.

The primary difference between the two tools is really a result of the nature by which the distribution takes place. The tools are composed generally as one finds the Bosch CDR Tool meaning there is software, and hardware including what, in the Bosch tool would be known as the “interface module” or what is called, by GIT, the “Vehicle Communication Interface (VCI).” Unlike the Bosch CDR Tool “interface module,” which has been designed specifically for the CDR tool application, the VCI is a near-universal component for GIT scan tool or diagnostics connectivity

associated with Hyundai and Kia vehicles going back to, at least 2010.

Apart from that, the only remaining difference between the VCI sold for Hyundai applications and that sold for Kia applications is the color of the VCI enclosure. The “blue enclosure” VCI is sold for Hyundai applications and the “red enclosure” is sold for Kia applications (interestingly enough, Snap-On tools typically have red enclosures, a theme comparable to the use of “Bosch green”).

In testing conducted to date, the authors have used both enclosures interchangeably accessing and retrieving data from Kia and Hyundai vehicles. For example, Hyundai data access and retrieval has been completed using the Hyundai version of the “EDR (system)” software and the Kia-designated VCI, and vice versa. The basic system cables (i.e.: for DLC access) are also interchangeable. These systems are not; however, sold together or currently sold such that a “piecemeal” assembly of a single tool capable of working with both systems is possible. It should not; however, come as a surprise that these systems are really so similar. Prior to GIT becoming the official OE supplier for diagnostic tools to Hyundai and Kia, the Hyundai Motor Company purchased more than 50% of what was then South Korea’s second-largest auto manufacturer: Kia Motors. Over time; however, that relationship changed at least outwardly - such that, by 2012, Hyundai retained only slightly more than 30% of Kia ownership. Nonetheless, as a function of other agreements, Hyundai remains the default representative of the two otherwise seemingly separate entities.

T

he Hyundai Tool The Hyundai-specific tool is available from GIT America’s Aftermarket Support division located in Tustin, CA. Through March 2013, the tool was offered at what was apparently an “introductory” price of about $2850 (including shipping). In April, 2013, that price increased to $4000 including shipping and one more direct-to-module cable than was included in the earlier version/release of the system. The tool is sold to allow retrieval from model year 2013 Hyundai vehicles which would provide for Hyundai compliance with “the rule.” There is a software subscription required which, in March 2013, was $195 (per year) for each system (Hyundai and Kia).

T

he Kia Tool The Kia-specific version of the tool is available from Snap-on Business Solutions. Snap-on is based in Richfield, Ohio. Through March, 2013, the Kia version of the tool was available for $3767 plus shipping.

In April 2013, the price of the tool increased to $4300 and 2 additional direct-to-module cables were added to the

36

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

list of components one receives with the tool. The tool is sold to allow retrieval from model year 2013 Kia vehicles which, similarly, would provide Kia compliance with “the rule.”

O

ther Commonalities Both tools are sold as an “EDR” rather than a crash or event “data retrieval tool.” The name of the device, for example, sold for Kias, is simply “EDR (Event Data Recorder)” and indicated in the user’s manual. Throughout the user’s manual there are references to the tool where it is referred to simply as “the EDR.” For example, in the Introduction of the user’s manual for the Kia version of the system, it reads: “Thank you for purchasing the EDR (Event Data Recorder). Read the instructions thoroughly for proper operation of your EDR (Event Data Recorder)” then later in the manual it describes the “purpose of EDR (Event Data Recorder).”

Stated there, the purpose of (the) EDR is “...to confirm and analyze to collect the data when colliding with a vehicle.” (sic) The manual goes on to say that the “Aim of EDR is below. 1. A detailed investigation of collision status 2. An accurate analysis of safety device performance” (sic) followed by a reference to “the rule.”

As previously noted, the GIT-built components of the system allow - unofficially - for some level of interchangeabil-

ity and there is nothing on either tool’s VCI which would suggest that component is specified for one OE or the other (apart, perhaps, from the color of the enclosure). There are differences noted between the direct-to-module cables provided with the two systems. Both systems are designed to connect direct-to-module with an airbag control module as the source of recorded data but the outward appearance of these direct-to-module cables is different, at least at the ACM connector end. The supplier(s) of those modules are those who also supply ACMs to other OEs (for example Continental and Delphi). The current direct-to-module cables, do not appear to be generally interchangeable.

T

he Software The underlying software used with the “EDR” (whether for the Hyundai or Kia vehicles) is demonstrably the same. The apparent difference is what module(s) within that software are enabled as it is shipped with which tool version. After installation of the software, two shortcut icons (one for Hyundai and one for Kia) appear on the end user’s desktop. In the user’s manual, where that document discusses “default (tool) behavior” and vehicle selection, one can see (in the manual’s graphic) options for both “Hyundai Motor Company - US” and “Kia Motor Company - US” whereas that is not the case in the end user version.

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 37

Another part of the user manual, for example, has been clearly simply copied from the user’s manual for the GIT diagnostic tool user manual which incorporates the VCI (module) and could therefore be used for either Hyundai or Kia vehicles. More telling is, perhaps, where in one section, it reads “... The VCI module is designed for seamless operation that will not disturb normal vehicle operation. (It) can put anywhere in the passenger compartment where will not disturb normal driving. When using the VCI module while driving, keep the VCI module as far away from clutch, brake or acceleration pedals as possible.” (sic) Clearly, this part of the tool is not “crash data retrieval-specific” in its underlying design.

In order to open the “EDR software,” the end-user must have the VCI powered up and connected to the laptop or desktop on which the software has been installed. Once the software starts and the “landing screen” is loaded, the only real outward difference between that seen for a Hyundai and that observed for a Kia is the color of the VCI enclosure (red or blue) in the on-screen graphic. On a related positive note, the software function is fairly elementary and largely icon driven with few choices. Icons for vehicle search, screen navigation, opening a stored PDF, saving a “test” and updating the software are all easy to locate onscreen and relatively self-explanatory.

Keeping in mind that the tool is specifically designed to enable Hyundai and Kia compliance with “the rule,” end users involved in litigation should be forewarned that there is no requirement for selecting a vehicle by VIN, entering a VIN as part of a normal data retrieval routine, and that, in the final output report, no VIN is found which would directly associate that data with a specific vehicle. As such,

“

... end users involved in litigation should be forewarned that there is no requirement for selecting a vehicle by VIN. 38

Collision Magazine - Volume 8 Issue 1

we believe that the translation process is free of restrictions which might otherwise be associated with a VIN for a specific vehicle and, by extension, the calibration of a specific module. As noted later, this assumption was addressed in testing.

As we review “the rule,” we are reminded that there is no provision found in that text requiring a VIN or that the data retrieved is associated with a specified vehicle (by VIN or otherwise). Simply put, given that the vehicle’s VIN is not found in “the rule’s” required data elements. As such, Hyundai and Kia do not fall short of any provision of “the rule” by omitting a VIN from either the software access steps or the end report generated by the software from their “EDR (tool).”

In litigation; however, especially in light of decisions including Oklahoma v. Ingram 2 and LaBorde v. Shelter Insurance 3 the courts have taken a rather dim view of a printout report alone: the only output option possible with the Hyundai and Kia systems. For example, in “LaBorde,” the plaintiff’s expert gave evidence that “... the printout is sufficient and accurate ...” while the defendant’s expert testified that “... the printout is not enough ...” In that case, and later on appeal, the court recognized the data from “event data recorders” is not new or novel but noted (1) the airbag control module (in that case) was no longer available (it had been lost), (2) “... the officer who removed the event data recorder from the vehicle had no training on how to properly remove the recorder...” and (3) for a variety of reasons, the “recorder” (presumably here, the court meant the airbag control module) was not treated as evidence and a proper chain of custody not maintained.

The “LaBorde” appellate court held that “... As to the printout itself, I find that it raises questions of reliability. The vehicle identification number on the printout was entered manually by the operator of the software program during the download process because the recorder itself does not contain the VIN number of the vehicle into which it is placed. ... In addition, I find that the printout cannot be verified since location of the event data recorder containing the original data is not known and the computer records that would show that the data was properly downloaded, interpreted and generated are no longer available ... In conclusion, I find that the acquisition and handling of the event data recorder evidence in this case necessarily renders the printout at issue unreliable and inadmissible at trial.”

As it relates now to the Hyundai and Kia tool(s) and the potential application in litigation, the connection between the data, airbag control module (“event data recorder”) and end report is even less “solid” than that found in the

www.collisionpublishing.com

“LaBorde” case (as it would be offered up against this decision). In “LaBorde,” the Bosch CDR Tool report included the VIN entered by the operator and the report was generated as a function of a translation of a secure file (the .CDRx file). While there are undeniably other functional/factual issues in that case which might have been more thoroughly (and correctly) addressed by the involved “expert(s),” a theme which could not be addressed any differently had this been a set of data retrieved from a Kia or Hyundai vehicle using one of these systems, is that the Kia or Hyundai end report is demonstrably even less secure and, in the context of the appellate court’s decision here, arguably less reliable than that found in the “LaBorde” case.

One may suggest that photographing, videotaping, writing a VIN number on an airbag control module, or even preserving an airbag control module (assuming it is and not subsequently “lost” as in “LaBorde”) may be sufficient to address these issues (and those other related issues raised in the “LaBorde” case); however, one only need think back across recent history to see how photos, video and even the presence of the “bloody glove” itself and live testimony associated with its retrieval from a crime scene in trial failed to otherwise impress a jury. In short then, it should be clear that while the Kia and Hyundai “EDR” tool, at least on its face, appears to meet requirements of “the rule,” its successful application and use in criminal or civil litigation remains far from certain.

Where the GIT system (whether for Kia or Hyundai) accesses data in an airbag control module, the end-user can print the output in a report format which essentially follows the format of Tables I and II as found in “the rule.”

In the simplest of applications and interpretations then, the Kia or Hyundai report format is relatively straightforward as it may display data called out in Tables I and II of “the rule.” Detail and clarity such as may be found in the “Data Limitations” section of a Bosch CDR Tool generated report are missing in that found for a Hyundai or a Kia although there is a section of a Kia or Hyundai report labeled “EDR Data Limitations” which is obviously loosely patterned after the basics of that seen in a Bosch CDR Tool generated report. Going back to case law such as “Ingram,” the lack of a secure “raw” data file (a .CDRx file) raises issues of subsequent translation (re-translation), in particular where that may relate to reliability or accuracy of earlier translations.

Where it appears that both the retrieval and translation process are free of restrictions which might otherwise be associated with a specific VIN for a specific vehicle and, by extension, the calibration of a specific module, that also suggests that the equivalent of “VIN spoofing” is a “viable” possibility. While history has demonstrated there are those who would suggest “VIN spoofing” a reasonable “research tool” (suggested by some with a “wink and a nod” knowing the propensity for its application in litigation overtly or otherwise), the authors have explored this possibility as a function of several successful data retrieval efforts including data from nearly a half-dozen instrumented crash tests associated with Hyundai and Kia vehicles. As might be anticipated with the system that does not require a VIN specific to a vehicle to function “properly,” retrieval of data from vehicles other than those specifically supported by this tool (i.e.: those in model year 2013) has been “successful.” That “success” is, at this writing, limited to the notion that “data has been retrieved” not that it has been successfully translated accurately.

Such as the crux of the “Ingram” case and an area, as it relates to litigation, not addressed or least not apparently contemplated as it relates to Kia and Hyundai reports.

The Kia and Hyundai software generate, after download, a printed report from a portable document file (PDF). That PDF could be opened again a later time and that same report could be reprinted. To be clear, that process does not address (1) later re-translation by subsequent versions of the software whether Kia or Hyundai or (2) the fact that the PDF document is not a secure document and can be edited. Particularly in light of the observation that (1) no VIN is required to initiate data retrieval (2), no VIN is required to translate retrieved data, and there is (3) no VIN stored or at least recovered using the Hyundai are Kia tool(s) it’s obvious that a number of legal issues may be raised (in the context of “LaBorde” and other similar cases).

While data has been successfully retrieved from 2011 and 2012 Hyundai and Kia vehicles at the time this document was prepared, for some of those vehicles where the data is a product of an instrumented crash test, as well as where the data was retrieved from a vehicle not involved in a crash test but rather an “real world” crash, it is clear that at least some of the crash pulse data, for example, is either (1) being translated incorrectly or (2) not supported in the format anticipated by the Hyundai are Kia software for model year 2013 and, therefore, the output is of questionable accuracy and/or reliability. In that regard, it will be useful later to identify characteristics of “VIN spoofing” (or its equivalent as it’s related to Hyundai and Kia vehicles) as a function of either testing the reliability of the systems or in an effort to identify where an individual may have used “spoofing” to access data not intended for retrieval of translation by the Hyundai Kia tools. Lacking a clear

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 39

VIN-to-module-to-data/report association, this may be an important attribute or benefit of such ongoing research by the authors not to mention its potential application in litigation.

D

ata Reliability Keeping in mind that this narrative is a preliminary review of the data retrieval process and data recovered from Hyundai and Kia vehicles and that a more detailed analysis will be included in a later issue of Collision magazine and at the CDR User’s Summit in January 2014 after more testing, a “first blush” review of data recovered from Hyundai and Kia vehicles compared to that obtained in instrumented crash tests suggests that the crash pulse delta-V and, perhaps, the maximum deltaV reported by the “EDR” software may be as much as 20% higher than that found in control instrumentation used in the crash tests. The reader is reminded that in “the rule,” at “Table III,” an accuracy of plus or minus 10% is required for “563 compliance” at 563.8(a) Table III.

Other arguably unsettling (preliminary) observations in Hyundai and Kia vehicle data sets have been made including significant differences between the maximum delta-V as reported by the Hyundai and Kia software versus what is observed by way of the instrumentation in crash tests beyond that which would normally be interpreted as the complete crash pulse and for the period designed to meet, for example, maximum values found in 563.8(a) Table III. While the authors would hope that skilled and competent users of recorded crash data would, by now, display better judgment than to use or be proponents of blatantly “spoofed” data (particularly in litigation), in these early stages of independent evaluation of these tool(s), the propensity to want to “rush to publication” without a thorough examination of the output side of the system(s) is historically unwise and exposes a lack of full comprehension of the effect of that action in favor of trying to be simply “first on the block” with information, no matter how incomplete or inept. For that reason, this preliminary review unashamedly omits detail of the analysis of data which has only been publically accessible for the last few months but which, as noted, includes at least a preliminary evaluation of several instrumented crash tests against the GIT tool retrieved data. As previously indicated, a full evaluation and detailed analysis will be presented later and at the CDR User’s Summit in January 2014.

Thus far, “pre-crash” data appears to be reasonably accurate compared to observations of “pre-crash” parameters such as “speed, vehicle indicated.” Where “Table III” calls for accuracy of “speed, vehicle indicated” to be within 1km/h out to 200km/h, it would appear that more moderate crash

40

Collision Magazine - Volume 8 Issue 1

test speeds, that requirement is being met. However, where “pre–crash” data may have to be compared to delta-V data in the analysis of a crash, there may be a disconnect given some of the trends thus far observed as it relates to the level of (in)accuracy found in the crash pulse data reported by the Hyundai and Kia software.

C

onclusions It would be easy to say that the Hyundai and Kia systems are generally sufficient to meet the requirements of “the rule” and that that’s “good enough.”

As noted, the preliminary review of retrieved data versus control instrumentation from crash tests (particularly as relates to delta-V) suggests GIT for Hyundai and Kia may need to take a second, longer look at data translation and/ or the data being recorded or reported as it relates to accuracy. This, of course, again raises the issue of the output report’s admissibility where there may later be a revision of data translation which would require another retrieval effort with that later software version, among other related concerns.

The security and traceability of retrieved data, report/file security (or lack thereof ) is very likely to be an issue as it relates to admissibility for the majority of end-users of the system(s). A lack of the secure file which can be reevaluated without the need for a subsequent retrieval of data from a module is likely to become an issue in litigation given rulings in previous cases (i.e.: “Ingram” and “LaBorde”).

Is not the authors’ intention to simply “condemn” the Hyundai or Kia tool(s) altogether rather, at this point, highlight - if only generally - some of the observations made regarding accuracy and potential issues at trial as well as some of the relative shortcomings of the GIT tool compared to the longer standing Bosch CDR Tool in that regard. In short, “reinventing the wheel” will clearly not be without a few “bumps in the road,” some of which are already visible.

As previously noted, this is an on-going effort and additional detailed analysis of retrieved data from Hyundai and Kia vehicles compared to crash test data will be made at the CDR User’s Summit in January 2014 available after a more complete analysis.

Compiled from a contextual reading of 49CFR563 from, largely, 563.3, 563.5(b), and 563.12 2 Oklahoma v. Ingram, (2009) Cherokee County District Court, CF-2006-403 3 LaBorde, et al, v. Shelter Mutual Insurance, et al, (2011) Third Circuit Court of Appeal, CW11-00956 1

www.collisionpublishing.com

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 41

The Effect of Tinted Headlights A look into the level of light diminishment headlights experience when a tint is applied to the headlights

Jeff Cardita

42

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

A

bstract Automobile owners have been customizing their vehicles since they started rolling off the assembly line. With advances in technology, vehicle customization options advance as well. A trend of late is the darkening of all lamp lenses to create a sleek all black vehicle exterior. One of the lens covers to be altered is the headlamp. However, there is a dark side to this headlamp lens tinting, namely a significant reduction of light emitted from the tinted headlamp. Many states prohibit the tinting of the headlamp lens by statute, but it would seem to be a common sense violation in all states.

This paper is designed to help understand the severity of tinting headlights and give a greater insight to the detrimental effects that applying a coat of tint will do to the measurable light output in dark conditions.

I

ntroduction When a vehicle is cleared from the production line at any major manufacture it has to pass a strict set of Federal Standards before being able to be operated on a roadway. With respect to vehicle lighting standards, “all lighting on motor vehicles is governed by Federal Motor Vehicle Safety Standard (FMVSS) No. 108. This is a federal

regulation promulgated by the National Highway Traffic Safety Administration under its general authority to regulate motor vehicle equipment granted by the federal Motor Vehicle Safety Act. The federal lighting standards apply to the original equipment on motor vehicles and to replacements for this original equipment.” 1 While Federal Standards are applied to the manufacture of new motor vehicles, once the vehicle is shipped, sold and in the hands of the new owner those standards are not applicable for enforcement at the federal level. Many states have adopted and signed as legislation, laws that restrict changing the vehicles equipment from that of when it was originally manufactured. An example of such legislation is found in the Arizona Revised Statues under the Transportation section. The multiple beam road lighting equipment arrangement law is found in Chapter 3, Article 16 under the equipment category. ARS 28-941.4 states A person shall not operate a motor vehicle with a colored transparent or translucent substance or material installed, affixed or applied on or in front of the head lamps, the auxiliary driving lamp or the auxiliary passing lamp or combinations of the head lamps, driving lamp or passing lamp that would obstruct, reduce or interfere with the visibility or effectiveness of the head lamps or that would change the color of light emitted. This paragraph does not prohibit the placement of clear transparent material mounted in front of the head lamps, fog lamps or auxiliary driving lamps. 2 In a recent fatal collision involving a vehicle v. pedestrian the need arose to determine the effect of having a tint on the lens of the headlight assembly. This collision took place at night and the vehicle involved had tint on both headlight lens assemblies. One of the many questions dealing with collision causation was, did the tinting on the headlight lenses prove to be a significant causation factor to warrant criminal prosecution. In order to fully understand the extent that the tint diminished the light output of the headlamps an experimental comparison study was conducted.

E

xperimental Procedures - Vehicles The vehicle used in the comparison analysis was the original vehicle involved in the fatal collision, hereafter referred to as “the subject vehicle.” It was compared side by side with an exemplar vehicle with matching headlight assemblies, hereafter referred to as “the exemplar vehicle.”

The subject vehicle was a 2003 Dodge Pickup Ram 1500 single cab, black in color. The Passenger side headlamp assembly was damaged in the collision with the pedestrian so it was disengaged during the tests. Battery power was www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 43

Figure 1: Subject vehicle (2003 Dodge Pickup Ram 1500) provided by a portable jump box which registered 12.78 volts during the tests. The following measurements were noted on the subject vehicle. Measurements are from the roadway.

Bottom of front bumper height Bottom of driver's side headlight assembly height Bulb of driver's side headlight height Top of driver's side headlight assembly height

14 inches 29 inches 33 inches 36 inches

Table 1: Subject vehicle measurements The exemplar vehicle was a 2002 Dodge pickup truck Ram 1500 Quad cab, silver in color. This vehicle did not have any damage to the headlight assemblies and was powered by the same portable jump box which registered 12.70 volts. The passenger side headlight assembly was disengaged during the tests. The same points were measure on this vehicle as well.

44

Collision Magazine - Volume 8 Issue 1

Bottom of front bumper height Bottom of driver's side headlight assembly height Bulb of driver's side headlight height Top of driver's side headlight assembly height

16 inches 31 inches 36 inches 39 inches

Table 2: Exemplar vehicle measurements

T

esting Environment The rural area where this test was performed was on a two lane roadway which had no street lights.

There was no outside light influence other than the moon which had minimal influence. This was also the same location of the fatal collision. The roadway was blocked off and no other vehicle traffic was permitted on the roadway other than the two test vehicles. Both vehicles were lined up next to each other facing the same direction.

One vehicle’s headlamp was on at a time.

T

esting Equipment The instrument used to measure the light output was an Extech LT300 serial number 08053441.

There were two separate lamp bulbs used in the tests. The first lamp bulb was the subject vehicle light that was on at the time of the collision. The second bulb was

www.collisionpublishing.com

a new bulb bought for the test which matched the subject lamp bulb. A certified ASE mechanic was on site and completed the installation of the headlight assemblies and the lamp bulbs. The light meter was fastened to a pole and the readings were taken at the middle of the actual headlight in each case.

E

xperiment Results Before each test an ambient light check was done and all the lights in control area were turned off.

The ambient light read 0.00 to 0.01 foot candles.

Distance 1 foot 10 feet 50 feet 100 feet Distance of "drop off" where the headlight is detected as opposed to the ambient light

Table 5: Experimental measurements for Test 2 (Low Beam)

Test One: This test was conducted using the subject vehicle and the original bulb. Only the driver’s side was illuminated and it had the tint on the headlight assembly. The results are as follows: Distance 1 foot 10 feet 50 feet 100 feet Distance of "drop off" where the headlight is detected as opposed to the ambient light

Light Reading 51.7 foot candles 0.95 foot candles 0.05 foot candles 0.01 foot candles 88 feet showed .02 foot candles

Table 3: Experimental measurements for Test 1 (Low Beam) Distance 1 foot 10 feet 50 feet 100 feet Distance of "drop off" where the headlight is detected as opposed to the ambient light

Light Reading 96.1 foot candles 1.16 foot candles 0.05 foot candles 0.02 foot candles 100 feet showed .02 foot candles

Distance 1 foot 10 feet 50 feet 100 feet Distance of "drop off" where the headlight is detected as opposed to the ambient light

Light Reading 176.8 foot candles 5.51 foot candles 0.26 foot candles 0.07 foot candles 214 feet showed .02 foot candles

Table 6: Experimental measurements for Test 2 (High Beam) Test Three:

Light Reading 94.6 foot candles 4.10 foot candles 0.20 foot candles 0.05 foot candles 179 feet showed .02 foot candles

Table 4: Experimental measurements for Test 1 (High Beam)

This test was conducted using the exemplar vehicle using the subject vehicle’s original bulb. Only the driver’s side was illuminated and it did not have tint on the headlamp assembly. The results are as follows: Distance 1 foot 10 feet 50 feet 100 feet Distance of "drop off" where the headlight is detected as opposed to the ambient light

Test Two: This test was conducted with the subject vehicle using the new lamp bulb. Only the driver’s side was illuminated and it had the tint on the headlight assembly. The results are as follows: www.collisionpublishing.com

Light Reading 431 foot candles 17.7 foot candles 1.01 foot candles 0.25 foot candles 300 feet showed .03 foot candles *

Table 7: Experimental measurements for Test 3 (Low Beam)

Collision Magazine - Volume 8 Issue 1 45

Distance 1 foot 10 feet 50 feet 100 feet Distance of "drop off" where the headlight is detected as opposed to the ambient light

Light Reading 608 foot candles 88.3 foot candles 4.06 foot candles 1.01 foot candles 300 feet showed .11 foot candles *

Table 8: Experimental measurements for Test 3 (High Beam) Test Four: This test was conducted using the exemplar vehicle using the new bulb. Only the driver’s side was illuminated and it did not have tint on the headlamp assembly. The results are as follows: Distance 1 foot 10 feet 50 feet 100 feet Distance of "drop off" where the headlight is detected as opposed to the ambient light

Light Reading 400 foot candles 16.2 foot candles 0.88 foot candles 0.22 foot candles 300 feet showed .03 foot candles *

C

onclusion: This study clearly indicates that by adding tint to the headlight lens the amount of light output diminishes significantly. Based on the amount of available light which is reduced by headlamp lens tinting it is reasonable to conclude that the drivers of these vehicles will have difficulty in understanding the environment in front of his vehicle during night time conditions.

R

eferences: 1. "Arizona Revised Statutes ." Arizona State Legislature . AZ. Retrieved October 10, 2012 from the World Wide Web: http://www.azleg.gov/FormatDocument.asp?inDoc=/ars/28/00941. htm&Title=28&DocType=ARS. (accessed October 10, 2012).

2. Fazzalaro, J. (2006, Febu 24). Limitations on headlight brightness . Retrieved from http://www.cga.ct.gov/2006/ rpt/2006-R-0198.htm

Table 9: Experimental measurements for Test 4 (Low Beam) Distance 1 foot 10 feet 50 feet 100 feet Distance of "drop off" where the headlight is detected as opposed to the ambient light

E

xperiment Analysis The results are alarming when calculating the amount of diminished light between the tinted lens and the no tint lens. Using the first distance on test one and the first distance on test three there is an 89% diminishment in light output in the tinted lens. The minimum reduction found about 75% light diminishment in the tinted lens. The difference in the first tests with the exemplar vehicle is a difference of only 7% on low beam at the first reading.

Light Reading 660 foot candles 77.0 foot candles 4.08 foot candles 0.99 foot candles 300 feet showed .11 foot candles *

S

pecial Thanks: Casa Grande Police Department, Pinal County Attorney’s Office, Tempe Police Department, Heritage Motors, Henry Brown Dodge Dealership, Ken Harmon IPTM, Officer C. R. Schmidlin, Officer B. Walsh, Sgt. A. Grijalva, Commander S. Sjerven Ofc. Springstroh, Harry Elkins, Certified Mechanic

Table 10: Experimental measurements for Test 4 (High Beam) * 300 feet was the end of the measuring tape that was used on the ground 46

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 47

Laser Scanning for Crash Reconstruction Joel Salinas, ACTAR

48

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

O

ne of the tools receiving quite a bit of attention these days in the crash reconstruction arena is the 3D laser scanner. The use of laser scanners is not new to this field of forensic work but it is one of the most rapidly growing markets. Although originally developed for survey use, much like the total station, it did not take long for those in the forensic field to recognize the advantages of this technology. Laser scanning lends itself well to documenting crash scenes and the environment, as well as both damaged and exemplar vehicles. The fact that the technology has been readily accepted by the courts makes it more appealing as a tool.

You have probably seen scanners in use at the crash testing sites at the ARC-CSI conference over the past couple of years. Articles on the use of this technology are also readily available in various publications.

For those unfamiliar with the technology a brief description of laser scanning follows. A more detailed description can be found by researching the manufacturer’s websites.

A 3D laser scanner is a tool for capturing the 3D measurement of data points. The scan data produces a point cloud where each point measured has an X,Y,Z coordinate. Depending on the scene or object being scanned, these point clouds consist of several million data points.

Currently there are two types of laser scanners being used for forensic applications. The time-of-flight scanner emits a pulse of laser light that bounces off of the target and is reflected back to the scanner. Measurements are based on the time it takes for the pulse to travel to and from the target. The phase-based scanner emits a continuous wave that is reflected off of the target and returns to the scanner. Measurements are based on the changing phases of the laser light.

Time-of-flight scanners have a longer range, up to 300 meters (~1000 feet), and can gather up to 50,000 points per second. Phase-based scanners have a shorter range, up to 187 meters (~600 feet), and can gather up to 1 million points per second. Typically, time-of-flight scanners with their longer range are used for larger outdoor scenes.

The scanner will measure 360° horizontally and approximately 270° vertically, producing a near full-dome scan. The scanner cannot measure the area directly below its position due to the base of the instrument blocking the laser’s path.

The first step in the typical scanning workflow is to allow the instrument to measure the scene. The next step involves the scanner making an additional pass to take a panoramic image using an internal camera. This image is then used to apply color to each point captured. The scan data can then be viewed with color applied from the camera image, in grey scale, or in a rainbow true color mode.

The panoramic image can be viewed separately from the scan. Measurements can be made within the point cloud or from the image itself. It is rare that a single scan will capture all the data needed. The scanner, like a camera, is a line-of-sight tool. In order to capture the scan data, the scanner needs to “see” the object. Normally, multiple scans from various positions are required to capture the entire scene. The scan data is processed and registered together using the scanner’s proprietary software. The reconstructed scan data is then cropped and cleaned before it is exported for use in a CAD or modeling program.

Figure 3 shows a scan of the interior of the bus. Using the MapScenes software tools, the roof of the bus is removed to view the inside passenger compartment.

For most major scenes, laser scanning is considered to be the quickest as well as the most accurate and reliable way to document both the roadway and the surrounding environment. Laser scanning is also efficient in that a typical scan takes 6-7 minutes followed by the time to capture the image. When considering the hazards faced while mapping a scene using traditional methods, laser scanning reduces one’s exposure to traffic by reducing the necessity of standing in the roadway. Hence the reasons that highway departments are adopting this technology for survey work.

Figure 4 shows the reconstructed scan of a four-way intersection. The level of detail captured in this scan allows a view from each side of the intersection as well as the position of all the traffic signals. The scene was scanned during normal traffic flow. Traffic noise was removed from the scan. All scans were taken from the sidewalks.

Don’t throw out the total station yet. The scanner cannot wholly replace the total station. There are instances where you will need both tools to do a complete documentation of the scene. For example, the scanner may not pick up faint tire marks on the roadway surface if there is not enough contrast between the two. The tire marks will show up in the image captured by the scanner but not necessarily in the scan. In these instances use the scanner in conjunction with the total station and shoot some control points that are easily identified within the scan. When you need to combine the scan data with the total station data, the two sets can be aligned using the control points.

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 49

Practice with the scanner will allow the operator to develop a complete understanding of its benefits as well as its limitations. In other words, what it can and cannot capture in the field.

Law enforcement has a distinct advantage when scanning crash scenes. They are able to close the entire roadway or to control traffic while measuring the scene. Working in the private sector, you are faced with having to deal with traffic passing through the scans. This typically produces “noise” within the scan which can be removed during the post-processing phase. What cannot be removed are the vehicles captured in the photo images taken by the scanner.

Figure 1: Leica C10 scanner is being used to scan an exemplar bus

The Leica C10 scanner has the option of using an external camera to take the images. When using the internal camera, the C10 takes 260 images for a 360°-degree scan.

When using an external camera with the recommended 8mm fisheye lens, only six photos are needed to complete the panoramic image. The operator can wait for a break in

Figure 2: A bus scan viewed in MapScenes PointCloud software 50

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

Figure 3: A scan of the interior of the bus

Figure 4: The reconstructed scan of a four-way intersection www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 51

“

Figure 5: A TIN generated from scan data of the roadway and sidewalks

One of the benefits of scanning a scene is that all of the available data is preserved for you as well as others involved in the case 52

Collision Magazine - Volume 8 Issue 1

traffic and capture these six photos in a matter of seconds, allowing for a much cleaner image and a better looking point cloud.

One of the benefits of scanning a scene is that all of the available data is preserved for you as well as others involved in the case. Unlike with the total station, you do not have to be selective on what you measure or how many measurements you take. Let the scanner run and take your measurements from the point cloud. Scan data captures the environment as it existed at the time the scan was made. You can revisit the scene as often as needed on your computer. The high level of detail captured by the scan data can be used to show the line of sight for drivers or pedestrians, as well as witness viewpoints. It can also reveal any visual obstructions created by the environment. The large numbers of points captured by the scanner allow an accurate TIN, a network representing the roadway surface or other areas, to be easily created. The TIN, or triangulated irregular network, can be created in either Leica’s Cyclone software or in MapScenes PointCloud.

The laser scanner excels when used for documenting damaged and exemplar vehicles. Instead of obtaining a single set of measurements for the damage profile, multiple measurements can be made as needed in the office. Interior scans of vehicles can aid in determining how occupants were injured and in determining driver and passenger positions.

Figure 7 shows the Profile Slice tool. The vehicle point cloud can be sliced horizontally. The slicing tool can be positioned higher or lower www.collisionpublishing.com

Figure 6: A scan of a damaged vehicle

Figure 7: Profile Slice tool www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 53

depending on where the measurements need to be made. Figure 8 shows a topdown view of the horizontal slice applied.

A drawing can be made showing the damage profile at this position. Figure 9 shows the damage profile transferred onto the CAD screen.

If you have a high-profile case that will require a professional animation, you are likely to find that forensic animators prefer to work with scan data, especially when they need to create 3D models of both exemplar and damaged vehicles as well as the environment.

Figure 8: A top-down view of the horizontal slice

An important consideration when working with scan data is what type of deliverables you are trying to create. Deliverables can range from an initial 2D drawing used to complete the preliminary crash analysis, to 3D models of the vehicles or scene, to a 3D animation. You need CAD software that can handle the entire point cloud and that incorporates features that make working with the point cloud easier.

Another consideration is how to present the scan data in the courtroom. Nothing captures the jurors’ attention like a visual presentation. By using simple tools such as a fly-through animation, you can give the jury a visual tour of the scene from many perspectives.

Figure 9: The damage profile transferred onto the CAD screen

About the author: Joel Salinas, ACTAR Joel Salinas is a recently retired Police Lieutenant. He has been documenting crash and crime scenes for the past 29 years. He’s had extensive training in forensic mapping, laser scanning and collision reconstruction. He runs a business providing forensic mapping, laser scanning and collision reconstruction services, as well as training in each of these fields. He uses a Leica C10 scanner. He is a cer-

54

Collision Magazine - Volume 8 Issue 1

When it comes to using 3D laser scanners for crash reconstruction, we are just scratching the surface. The more this tool is used, and as users collaborate to share experiences and ideas, the better this technology will become.

tified MapScenes instructor and a Leica dealer through Collision Forensic Solutions. He can be reached at JSAFM@sbcglobal.net For more information about Leica scanners visit: www.leica-geosystems.com/hds/ For more information about MapScenes software visit: www.mapscenes.com

www.collisionpublishing.com

ACCIDENT RECONSTRUCTION AND FORENSIC MAPPING SOLUTIONS DESKTOP MAPPING & ANIMATION SOFTWARE | EVIDENCE COLLECTION SOFTWARE & HARDWARE | TRAINING

WHEN YOU’RE LOOKING FOR THE BEST ACCIDENT RECONSTRUCTON SOFTWARE, YOU NEED TO LOOK AT

NOW WITH ADD-ON MODULES MapScenes Forensic CAD with your choice of point cloud and animation options to ensure you get precisely what you need.

FORENSIC CAD Law enforcement and accident investigators use MapScenes Forensic CAD. MapScenes Forensic CAD is the most precise desktop forensic mapping software available in the world. This powerful and feature-rich application provides law enforcement professionals and accident reconstruction specialists, like you, with confidence that the evidence they present is precise, compelling, and irrefutable.

Turn the page for more information about our powerful animation and point cloud modules.

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 55

ANIMATION FOR THE REAL WORLD

ANIMATION MODULE Full-featured forensic animation software for creating precise and compelling animations.

The animation module gives you incredible flexibility to accurately animate your scenes - look at your data in ways you never thought possible.

Our rendering capabilities give you the tools you need to analyze, edit, and build engaging animations.

Timeline and advanced object movement, animated cameras, real surfacing, vaulting and other physics tools - you will be amazed at the possibilites!

POINT CLOUD MODULE An incredible point cloud visualization tool to assist with mapping and animation.

Imagine being able to walk through a point cloud, in real-time, in order to complete your line work or conceptualize your dramatic re-creation.

With the point cloud module you can draw lines right in the point cloud viewer, which will show up in your CAD drawing. You can also include the point cloud itself in your final animation for added realism at NO extra effort. Mapping through to animating, your efficiency will soar!

LET THE SCENE SPEAK FOR ITSELF

is registered in the U.S. Patent and Trademark Office. All other brand ® MapScenes names, product names or trademarks are property of their respective owners.

www.mapscenes.com

1.800.668.3312

THE REST OF THE

SYSTEM

EVIDENCE COLLECTION SOFTWARE:

EVIDENCE COLLECTION HARDWARE:

CERTIFIED TRAINING:

Reduce errors, save time, and eliminate frustration.

Incredible hardware options and support.

When accuracy and precision are non-negotiable – there is MapScenes.

Evidence Recorder is the world’s best selling evidence collection software built specifically for the professional reconstructionist.

Using a Windows Mobile or tablet device, Evidence Recorder allows you to experience the power, precision and productivity gains associated with being able to visually verify your ongoing field evidence collection – on-the-spot, right in the palm of your hand. Then, for maximum results, it can seamlessly integrate with the MapScenes desktop software.

Evidence Recorder is supported with more total stations, data collectors, and GNSS receivers than ever.

Training is essential in all facets of law enforcement. MapScenes acknowledges its responsibility to the industry by ensuring that all trainers and educational courses are kept up to date through a base educational standard, preserving quality and integrity.

Combine with your existing hardware, or choose from one of our hardware bundles. Evidence Recorder is included on all hardware configurations and offers the absolute best in value.

We have set an industry benchmark by developing a comprehensive framework that provides investigators with the knowledge and skills required to use the MapScenes system easily, efficiently, and expertly.

Choose from numerous options, including: Nautiz X7, Algiz 7, Archer field PC, or Allegro MX.

THE

SYSTEM:

Evidence Collection Software Evidence Collection Hardware Desktop Mapping Software

is registered in the U.S. Patent and Trademark Office. All other brand ® MapScenes names, product names or trademarks are property of their respective owners.

www.mapscenes.com

Training

1.800.668.3312

Validation and Use of EDR Data from a

Non-CDR Supported Vehicle in a Criminal Prosecution Case Wesley Vandiver Orange County District Attorney’s Office

Isaac Ikram and Bryan Randles Biomechanical Research and Testing, LLC

C

ase Summary In the early morning hours of December 18, 2010, a father was driving his teenaged son to a high school debate competition in the family’s green 2000 Buick Le Sabre. They were traveling northbound on Knott Avenue in the City of Buena Park, California.

Driving conditions were less than optimal due to darkness and falling rain. At approximately 5:34 AM, they entered the intersection of Knott Avenue and Crescent Avenue. As they reached the center of the intersection, their Buick was struck on the passenger side in a violent collision that resulted in their deaths.

The 18-year-old driver of a blue 2009 Mitsubishi Lancer GTS had been warned about the dangers of driving while intoxicated. As he was driving westbound on Crescent Avenue approaching an eventual collision with the Buick, his blood alcohol concentration was 0.13% and he had been smoking marijuana. He had recently purchased the 2009 Mitsubishi Lancer, and later admitted to investigators that he verified the governed speed of the vehicle by reaching 120 miles per hour on the freeway several days prior to this incident.

and door. Following the near perfectly-perpendicular collision, the left side of the Mitsubishi and the right side of the Buick slammed together before both vehicles traveled in a West-northwest direction toward their eventual points of rest. The Buick, missing its right-front wheel and hub, left a trail of scrapes as it traveled over a curb and sidewalk, knocked down a metal sign post, crossed an adjacent cul-de-sac and penetrated a cinderblock wall before coming to rest in the front yard of a residence. The Mitsubishi left no post-collision evidence that could be seen by the on-scene investigators. Through later analysis of photographs and video evidence, its post-impact motion was determined.

The surviving driver of the Mitsubishi was the only known living eyewitness to the collision.

Despite significant investigative effort, no other eyewitnesses were located who could verify the phasing of the traffic signals at this intersection or the pre-collision speeds of the involved vehicles.

A security camera at an adjacent gas station captured the collision. The frame rate and lack of clarity in the video prevented any meaningful analysis regarding pre-collision speeds. HowThere was no evidence of pre-collision braking by ever, the video did capture the impact between either driver. The front of the Mitsubishi crashed the vehicles and provided important information into the Buick in the area of its right-front fender regarding the post-impact motion and trajectory 60

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

of the Mitsubishi. It also showed that the northbound traffic signal for the Buick had been green for over one minute when the Buick entered the intersection and that the westbound signal for the Mitsubishi had been red for that same period of time.

The minimal amount of physical evidence that could be identified and documented by on-scene investigators was later overlaid onto a collision site survey. The additional evidence available in scene photos and still captures from the security video were used to determine and verify the post-collision motion for the vehicles. Without any further analysis, one thing was plainly obvious - the Mitsubishi had been traveling at an extremely high rate of speed when it impacted the Buick. The extent of the vehicle deformation was a significant clue. However, most telling was the degree to which the Mitsubishi deviated the trajectory of the Buick despite the Buick’s nearly 800-pound weight advantage.

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 61

62

Figure 1: Collision Scene Photograph

Figure 2: Collision Scene Photograph

Figure 3: Collision Scene Photograph

Figure 4: Buick at Point of Rest

Figure 5: Mitsubishi at Point of Rest

Figure 6: Buick Damage

Figure 7: Buick Damage

Figure 8: Buick Damage

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

Figure 9: Buick Damage

Figure 10: Mitsubishi Damage

Figure 11: Mitsubishi Damage

Figure 12: Mitsubishi Damage

Figures 13-16: Still Captures from Security Video www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 63

Figure 17: Collision Scene Map

Figure 18: Impact Sequence The impact speed of the Mitsubishi was obviously greater than that of the Buick. There was sufficient evidence documented and pieced together from the previously-mentioned sources to complete a traditional speed analysis.

Using the available data suggested an at-impact speed for the Buick of 35 to 40 miles per hour and for the Mitsubishi of 80 to 85 miles per hour. The speed limits for Knott Avenue (Buick) and Crescent Avenue (Mitsubishi) at this location were 40 miles per hour and 45 miles per hour, respectively.

64

Collision Magazine - Volume 8 Issue 1

E

vent Data Recorders The 2000 Buick Le Sabre is supported by the Bosch Crash Data Retrieval (CDR) System. In April 2011, a few months after the collision, the Buick’s ACM was downloaded. The recorded data in the Deployment Event verified the manner in which the Buick was being driven at the time of the collision - lawfully and cautiously. The recorded data suggested the Buick approached the eventual collision at a steady speed of approximately 37 miles per hour, slightly below the posted speed limit.

www.collisionpublishing.com

Figure 19: Buick CDR Data The 2009 Mitsubishi Lancer is not supported by the Bosch CDR System. In July 2006, the National Highway Traffic Safety Administration (NHTSA) published data related to the prevalence of EDR’s in motor vehicles sold in 2004.

A table from that publication details that Mitsubishi sold 161,523 vehicles in the United States in 2004, representing a 1.5% market share, with 100% of those vehicles having some level of EDR capability.

At the time of this investigation, the EDRs in Mitsubishi vehicles were not accessible utilizing the Bosch CDR System or any other tool(s) not owned and operated by the manufacturer. The exception to this was the 20072010 Mitsubishi Raider, which was covered by the Bosch CDR System as a result of sharing a common platform with the Dodge Dakota produced by Chrysler. With this exception, Mitsubishi fell within a category of several vehicle manufacturers that have produced automobiles with EDR capability, but may record crash data that can only be retrieved utilizing internal proprietary hardware and software.

Research was done into the recording capabilities for the 2009 Mitsubishi Lancer. With the assistance of personnel from Mitsubishi Motors North America, it was learned that this vehicle was equipped with an airbag control module (ACM), internally referred to as an airbag electronic control unit (ECU), that had event data recorder functionality. The crash data recording for this vehicle was summarized by Mitsubishi personnel as follows: •

When the front airbags deploy, the following will be recorded as major data

•

Longitudinal Deceleration (-100 ms to +149 ms)

•

Vehicle Speed (-2.3 sec. to 0 sec.)

•

Gas Pedal/Throttle Status (-2.3 sec. to 0 sec.)

•

Brake Lamp Switch Status (-2.3 sec. to 0 sec.)

•

(Note: Time 0 = Time when the ECU sends the firing signal)

•

Seat belt status is not recorded

•

If the crash does not involve a frontal collision involving frontal airbag deployment, the above data will not be recorded

Unlike traditional data parameters and formatting that collision investigators have become familiar with from the Bosch CDR System, the proprietary Mitsubishi system recorded 2.3 seconds of pre-crash data in 100 millisecond increments. In May 2011, five months after the collision, Mitsubishi personnel facilitated the retrieval of crash data from the Mitsubishi’s ACM. The six-page Mitsubishi EDR report contained data relative to the Lancer’s precrash speed, engine RPM, braking, accelerator and throttle positions, as well as longitudinal acceleration. The “Speed Vehicle Indicated” data showed the Mitsubishi being accelerated from 127 kilometers per hour (79 miles per hour) to 137 kilometers per hour (85 miles per hour) in the 2.3 seconds prior to the airbag deployment command. The driver had steadily applied the accelerator pedal to a level exceeding 60-percent (except for the final zero millisecond reading).

Of concern was the data recorded for the longitudinal acceleration. The report indicated zero Gs from -100 milliseconds to -82 milliseconds then 123.53 Gs from -81 milliseconds to the end of the recording period. With the exception of the longitudinal acceleration data, all other data elements appeared consistent with one another and ultimately consistent with the known evidence and traditional analysis. The longitudinal acceleration data, although reasonable up to the point of -82 milliseconds, was subject to some anomaly for the remainder of the recording period. After reviewing the data and the recording process for this module, it was concluded that the anomaly was consistent with a power loss during the data writing phase of the process. In other words, the capacitors within the ACM ran out of power during the writing of the longitudinal acceleration data. The data obtained from the Mitsubishi EDR were very valuable evidence in a criminal case that included vehicular homicide charges against the driver

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 65

Figure 20: NHTSA - Estimate of the number of EDRs in MY2004 light vehicles with a GVWR of 3855 kg or less of the Mitsubishi. However, given the circumstances of this case, it was decided that testing would be performed to verify the accuracy of the Mitsubishi EDR data and attempt to verify the cause of the anomaly observed in the longitudinal acceleration data.

E

DR Validation Testing The formulated testing plan included obtaining an exemplar 2009 Mitsubishi Lancer and generating events within the ACM under circumstances substantially similar to those of this collision - specifically, under heavy acceleration at speeds in the range of 80 to 90 miles per hour. The challenges in this plan included the fact that the Lancer only recorded data for frontal deploy66

Collision Magazine - Volume 8 Issue 1

ment events and it was reasoned that each module could therefore only be used once. It was decided that the safest and most efficient manner in which to conduct the testing was with the test vehicle being run on a dynamometer while several new ACMs were subjected to “deployment events.”

The test vehicle matched the subject 2009 Mitsubishi Lancer involved in the case in every realistic way - year, make, model, engine, transmission, wheels, tires, etc.

They were produced in the same factory approximately six weeks apart from one another. The vehicles were manufactured with a 2.4L I4 DOHC 16-valve MIVEC (Mitsubishi Innovative Valve-Timing Electronic Control) System

www.collisionpublishing.com

Figure 21: Mitsubishi EDR - Speed Vehicle Indicated Data

Figure 22: Mitsubishi EDR - Engine RPM Data www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 67

Figure 23: Mitsubishi EDR - Accelerator Pedal Data

Figure 24: Mitsubishi EDR - Brake Data

Figure 25: Mitsubishi EDR - Longitudinal Acceleration Data 68

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

engine. The transmissions were each a 6-speed Sportronic ® CVT with INVEC-III, which operates as an automatic transmission with continuously variable ratios but can be shifted by the operator utilizing steering-wheel-mounted paddle shifters. During the testing the vehicle was operated in automatic mode.

The test vehicle runs took place upon and were monitored by an AVL 4-Wheel Dynamometer with overhead display and analog output. The speed measurement system included a digital optical encoder with a resolution of 10,000 PPR (pulses per revolution). The speed measurement system accuracy is +/-0.05 miles per hour with a maximum speed of 125 miles per hour. A Vericom VC4000DAQ was used for data acquisition. During the testing, data elements from the CAN bus stream were collected via a CAN cable connected to the vehicle’s OBDII port. Dynamometer speed was captured via a direct connection to the Vericom. Prior to the testing, the voltage/speed linearity was checked yielding an r-squared value of one. Twelve Mitsubishi ACMs were purchased and used during a total of 17 test runs.

A wiring harness was constructed to extend the “reach” of the test vehicle’s original ACM wiring harness to a workstation where ACMs could be connected to the vehicle and deployment events could be initiated. All deployable restraints in the vehicle were manually disconnected.

Resistors were placed into the wiring harness to replicate the resistance of the disconnected restraints. A controller was spliced into the ACM power supply allowing for the cutting of power to the module at a predetermined time following an impact. The impacts to the ACMs were monitored by means of a tape switch and accelerometer attached to the ACMs. To generate deployment events, the ACMs were impacted with a mallet with damping applied via a foam insert between the mallet and ACM. A two-camera Video VBOX system was used to capture the vehicle’s speedometer and the dynamometer’s overhead display. Additionally, a strobe, which was wired to a tape switch attached to the ACM, was placed within the camera’s field of view.

The initial tests included acceleration in steps up to approximately 90 miles per hour to monitor the agreement and

Figure 26: Test Vehicle

Figure 27: New, Unused Mitsubishi ACMs

Figure 28: Testing Equipment and Set-up

Figure 29: Testing Equipment and Set-up

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 69

Figure 30: Testing Equipment and Set-up

Figure 31: Testing Equipment and Set-up

Figure 32: Speed - OBDII vs. Dynamometer

Figure 33: RPM - EDR vs. OBDII

Figure 34: RPM - EDR vs. OBDII

Figure 35: Speed - OBDII, EDR and Speedometer vs.

Dynamometer

Figure 36: Speed - OBDII, EDR and Speedometer vs. Dynamometer 70

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

synchronization of the speed data reported via the OBDII and the dynamometer. Wheel speed was monitored with the dynamometer and compared to the data broadcast over the CAN bus. Periods of acceleration, steady-state speed and braking were recorded. In the one complete data set from this phase of the testing, output from the OBDII matched closely with the wheel speed measured by the dynamometer during all phases. The correlation coefficient (R), root mean square error (RMSE) and average error were determined. When determining the average error, the time interval between 10 and 135 seconds was evaluated. This interval started and ended with a wheel speed of approximately 10 mph. This interval was chosen to prevent artificial inflation of the reported % error due to a reduction in the denominator size. For the data shown in the figure, R = 1 and RMSE = 1.39 mph. The average error was determined to be 3.9%. However, exclusion of the deceleration phase reduced the error to 1.4%.

In later tests, the vehicle was accelerated up to approximately 80 miles per hour, held at that speed momentarily and then rapidly accelerated. It was during this rapid acceleration that the deployment event was initiated and data recorded by the ACMs. Output from the dynamometer and OBDII port was sampled at 100 Hz by the Vericom.

Since the data were collected by the same system, their signals were inherently synched with one another. The strobe, which was activated by a tape switch attached to the side of the EDR module, was within the field of view of the camera monitoring the speedometer. This provided a means to synchronize the speedometer readings with the EDR recording. In order to align the EDR data with the OBDII and dynamometer output, RPM recordings were used. Since hard acceleration was occurring during the artificial deployment, visual alignment of the RPM data

Figure 37: Speedometer vs. Dynamometer @ -2300 milliseconds

could be achieved, and the time offset determined. After synchronizing the datasets, dynamometer and OBDII data were clipped and decimated to match the duration and sampling rate of the EDR recording. This allowed direct comparison of speed recordings at 24 discrete time intervals. Speedometer readings were obtained through video analysis of the individual tests. This involved working backwards in 100 millisecond increments from the flash for a 2.3 second duration.

Of six complete datasets, there were five with induced deployment events between 80 and 90 miles per hour, and one with a deployment event between 60 and 70 miles per hour. Comparison between the EDR and dynamometer data yielded the following average results: R = 0.88, RMSE = 3.53 mph, error = 4.34%. Comparison between the OBDII and dynamometer yielded the following average results: R = 0.62, RMSE = 1.17 mph, error = 1.43%.

Comparison between the speedometer and dynamometer yielded the following average results: R = 0.80, RMSE = 3.21, error = 3.9%.

In comparing the various speed values - the speed recorded by the EDR, the speed reported by the dynamometer, the speed broadcast via the vehicle’s OBDII port and the speed displayed on the vehicle’s speedometer - the dynamometer, based upon its high level of accuracy and precision, was deemed to be the standard to which all other sources of data should be compared. The expected outcome was that the OBDII broadcast would closely match the speedometer reading and EDR recording. This was based on the assumption that they were utilizing a common source of speed data (i.e., wheel speed sensors, vehicle speed sensor, etc.). However, what seemed apparent during the testing and was later verified through data analysis is that in this vehicle, the four sources of speed readings were derived

Figure 38: Speedometer vs. Dynamometer @ "Time zero"

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 71

Figure 39: Default Data - Longitudinal Acceleration

Figure 40: Default Data - Speed

72

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

Figure 41: Default Data - Engine RPM from one of two sources. The OBDII broadcast, which received its speed signal from the transmission output speed sensor, closely matched the speed reported by the dynamometer. The average error for the OBDII stream was less than 1.5%, which in our tests used for analysis amounted to an error of less than 1.9 km/h (1.2 mph). The speed displayed by the speedometer closely matched the speed reported by the vehicle’s EDR. These two received their speed signals from the front wheel speed sensors (averaged). The disagreement between these two values was less than 0.8%, which in our tests amounted to less than 1.3 km/h (0.8 mph). However, when we compared the two groups to one another, there was a clear difference. On average, the speedometer and EDR reported speeds that were 3.94% and 4.34% higher than the dynamometer, respectively. This pattern was consistent throughout the testing.

The EDR report for the Mitsubishi Lancer reports the pre-crash speed under the heading “Speed Vehicle Indicated.” Taken literally, this reported speed in the vehicle’s EDR shows strong correlation to what is being indicated on the vehicle’s speedometer. These values tended to be within 1% of each other. In terms of true wheel speed, they appear to consistently over-report the actual speed by approximately 4%.

The final goal of the testing was to address the progressive process of data being recorded to determine the reliability of any data that is written if that process is interrupted by a power loss. At the start of the testing, connection to the OEM module in the test vehicle revealed no file present.

Connection to a new, unused module revealed, as expected, no file present. The new, unused module was installed in the test vehicle which was then run on the dynamometer to a speed of approximately 128 kilometers per hour (80 miles per hour). Attempting to image that module was again unsuccessful, yielding no recorded file. The dynamometer test was repeated with an impact to the module in an attempt to simulate a collision event. Attempting to image that module was again unsuccessful yielding no recorded file. It was only when the module was impacted with a sufficient impulse to generate a deployment event that the EDR awakened and recorded crash data.

In subsequent tests, the power supply was cut to the module(s) at various intervals following the impact(s) in an attempt to close in on interrupting the writing process while in progress, with the goal of generating a partially written file. Distinct phases of the writing/recording process were revealed. If the power supply was cut too quickly, no data was written and as was the case with an insuf-

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 73

Figure 42: Default Data - Accelerator Pedal

Figure 43: Data Writing Process and Power Loss Description from Actual Crash

74

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

“ ”

ficient impact. There was no recorded file to download. If a longer delay was allowed before cutting power, a file was generated indicating a deployment event; however the recorded data for pre-crash elements (speed, RPM, throttle, pedal, etc.) and the crash pulse (longitudinal acceleration) were all displayed as maximum values (default data). For example, longitudinal acceleration was reported as 250 Gs and vehicle speed was reported as 255 kilometers per hour.

As expected, the task of actually cutting power during the brief window of time necessary to generate a partially written file was a low-probability endeavor. However, two tests were apparently very close to achieving that goal. In those two tests, the power was cut at a time that allowed for the deployment command to be sent, but precluded the writing of any crash data - all reported crash data was default maximum values. The maximum data values reflect the fact that the gates for the recording addresses on the Electrically Erasable Programmable Read-Only Memory (EEPROM) were open, but no data was written. A blank EEPROM has all the gates fully open, giving each cell a value of 1. If each bit has a value of 1, each byte will reflect a maximum value.

When test data was compared to analysis of the data from the Mitsubishi in this case, a second step in data writing related to the longitudinal acceleration was revealed.

The maximum value reported during testing was 250 g’s.

However, if actual longitudinal acceleration data was written (any amount), the maximum reportable value was 123.53 g’s. This was due to the fact that the default maximum value lacks a zero-g bias level. The accelerometer can report positive or negative acceleration on the longitudinal axis. Since the accelerometer operates by converting a voltage into an acceleration reading, the span of available voltage is divided between positive and negative acceleration when actual longitudinal acceleration data is being written. When the power was cut and a zero-g bias level was not received by the EEPROM, the default 250-g span of available accelerometer readings remained.

C

onclusions In the end, the testing quantified the accuracy that could be expected in the pre-crash speed reported by the EDR in a 2009 Mitsubishi Lancer, specifically under the circumstances of this collision - that is, at high speed and under hard acceleration. The EDRreported speed very closely matched the speedometer-reported speed and slightly overestimated the actual vehicle speed. In this case, the original calculated speed estimates were therefore corroborated by the EDR data from each vehicle. Additionally, the testing cast light on the data writing process and the difference between written data and default data. The consequences of the power loss dur-

The testing quantified the accuracy that could be expected in the pre-crash speed reported by the EDR in a 2009 Mitsubishi Lancer, specifically under the circumstances of this collision

ing the actual collision in this case were better understood and the process by which a crash file is generated could now be explained, particularly one that included both written crash data and default data.

In April 2013, the EDR data from both the Buick and the Mitsubishi were ruled to be admissible in the Superior Court of California, County of Orange. These were presented as evidence in the criminal trial for the driver of the Mitsubishi. Additionally, the validation testing and results were ruled as admissible and were offered as supporting evidence in the people’s case. On April 18, 2013, after deliberating for approximately 50 minutes, the jury returned a verdict of guilty on two counts of California Penal Code Section 187 - Murder.

A

cknowledgements The authors wish to thank the management and technical staff at Mitsubishi Motors North America, Inc. for their generous contributions of time, resources and technical knowledge. We would also like to thank Orange County (California) District Attorney Tony Rackauckas for providing the resources and support for this project.

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 75

®

B

FAST. EASY. VALIDATED. ACCURATE.

Brach Engineering • www.brachengineering.com • support@brachengineering.com • 574.273.8805

Investigation of Traffic Crashes Involving the Inhalation of Difluoroethane Adam M. Hyde and Roger W. Barrette, MSE Major Crash Assistance Team of Lake County, Illinois

C2H4F2 78

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

For decades people have been inhaling household products to get their next “high.” Some of the most common are spray paint, super glue, gasoline and common household solvents. The practice is commonly referred to as “huffing,” which is a slang term defined as inhaling a substance to become intoxicated, or volatile substance abuse (VSA).

It can easily be done by inhaling fumes off a rag covered in solvent or by inhaling fumes from spray paint contained in a paper bag, as shown in Figure 11. Many states have a law to address those driving under the influence of these intoxicating compounds (DUI) or other non-traditional DUI cases. When there is a death or great bodily injury, the charges are enhanced to felonies such as Aggravated DUI or Reckless Homicide. The “intoxicating compound” then becomes an element of the offense for which a prosecutor has to prove. This is most easily done by toxicological testing of the person’s blood, urine, or other bodily fluids or tissue. However, with the intoxicating compound, the toxicologist performing the test must know or have a good idea of what volatile substance they are looking for.

I

nhalation Abuse Inhalants are volatile substances that produce chemical vapors that can be inhaled to induce a psychoactive, or mind altering effect2. The term “inhalants” is used to describe a variety of substances whose main common characteristic is that they are rarely, if ever, taken by any route other than inhalation. This definition encompasses a broad range of chemicals that have different pharmacological effects. Precise categorization of inhalants is difficult. One classification system lists four general categories of inhalants; volatile solvents, aerosols, gases, and nitrites, which are based upon the forms in which they are often found.

Substances commonly used by inhalant abusers fall into these categories:

•

Volatile solvents are liquids that vaporize at room temperature. They are found in a number of products used for common household and industrial purposes. These include paint thinners and removers, correction fluid, glues, and felt-tip markers.

•

Aerosols are sprays that contain propellants and solvents. They include spray paints, hair sprays, deodorant sprays, and fabric protector sprays.

•

Gases include medical anesthetics as well as gases used in household or commercial products including ether, chloroform, and halothane. The most common abused gas is nitrous oxide, or “laughing gas.” This gas can be found in whipped cream dispensers.

•

Nitrites often are considered a special class of inhalants. Unlike most other inhalants, which act directly

on the central nervous system (CNS), nitrites act primarily to dilate blood vessels and relax the muscles.

These are primarily sexual enhancers and are commonly known as “poppers” or “snappers.” Amyl nitrite is used in certain diagnostic procedures and has been prescribed in the past to treat patients with heart pain.

Nitrites are now prohibited by the Consumer Safety Commission but can still be found today, sold in small bottles.

Inhalants can be breathed in through the nose or the mouth in a variety of ways including sniffing, spaying aerosols directly into the mouth or nose, bagging, huffing, or inhaling from balloons. Inhaled chemicals are absorbed rapidly into the bloodstream through the lungs and are quickly distributed to the brain and other organs. Within seconds of inhalation, the user experiences intoxication along with other effects similar to those produced by alcohol. Alcohol-like effects may include slurred speech, the inability to coordinate movement, euphoria, and dizziness.

Because the intoxication lasts only a few minutes, abusers frequently seek to prolong the high by inhaling repeatedly over the course of several hours. With successive inhalations, abusers can suffer loss of consciousness and possibly death. At the least, they will feel less inhibited and less in control. Although the chemical substances found in inhalants may produce various pharmacological effects, most inhalants produce a rapid high that resembles alcohol intoxication, with initial excitation followed by drowsiness, disinhibition, lightheadedness, and agitation. If sufficient amounts are inhaled, it may produce anesthesia and lead to unconsciousness or death from asphyxiation (high concentrations of the fumes which displace available oxygen in the blood).

Figure 1: Example of inhaling fumes off a rag covered in solvent or by inhaling fumes from spray paint contained in a paper bag

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 79

D

ifluoroethane In the last few years, an increase in abuse has been seen with inhaling aerosol computer spray cleaner such as Dust-Off® or Endust®, shown in Figure 2. These products are commonly used to dust a computer keyboard or electronics, and are easily accessible and low cost. Inhaling dusting spray is commonly referred to as “dusting,” and the effects of it are nothing short of tragic. Most of the duster products on the market contain 1,1-Difluoroethane (DFE) or Freon 152a. DFE is an organofluorine compound with the chemical formula C2H4F2 (carbon-hydrogen-fluorine), a fluorinated hydrocarbon, also known as Freon. It is a flammable, odorless and colorless propellant used in many aerosol products on the market including hairsprays, cleaners, spray paint, etc.

In essence, users obtain their high by inhaling the dusting spray directly into their mouth or nose. The inhalation of DFE produces a quick short high similar to that of alcohol intoxication, but is more unpredictable than alcohol.

Where alcohol has a dose related effect (the consumption

Figure 2: Aerosol computer spray cleaner such as Dust-Off® or Endust®

is related to the level of intoxication), you can’t gauge a dose with huffing DFE. Figure 33 shows a driver inhaling a keyboard cleaning aerosol through the mouth. They immediately feel the effects, as quick asphyxiation takes place because the body is being deprived of oxygen. However, the effects of dusting only last two to three minutes, but are known to cause unconsciousness, disorientation, dizziness, seizure-type symptoms and cardiac arrest. When dusting while driving a motor vehicle, the effects are often deadly for a few minutes of euphoria. A search of www. youtube.com show many examples of people inhaling keyboard cleaner aerosol and demonstrate the effects.

A 2011 national study by the National Institute on Drug Abuse revealed over 13% of children in the eighth grade used inhalants4, with other national studies showing 17% of drug users reported inhalants were their “gateway drug,” or the first drug they used5. Dusting is often combined with other drugs, such as cannabis or other synthetic narcotics, which increase the effects. Symptoms of inhalant abuse include these signs of acute toxicity: •

CNS signs – euphoria, dizziness, excitability, hallucinations, stupor, lethargy, agitation, ataxia, headache, loss of consciousness

•

Symptoms typically resolve in 30 minutes in most cases

•

Respiratory depression, asphyxiation

•

Cardiac arrhythmias

•

Sudden Sniffing Death (SSD) – the user can die the 1st, 10th, or 100th time he or she uses an inhalant

Falcon Safety Products, Inc. the maker of “Dust Off®” shown in Figure 4, adds a bittering agent to their duster to discourage inhalant abuse. This may produce a very bitter taste in the mouth or if exposed to evaporating gas. Others have anti-inhalant formulas to help prevent inhalation of the product, shown in Figure 5.

Q

Figure 3: A driver inhaling a keyboard cleaning aerosol through the mouth 80

Collision Magazine - Volume 8 Issue 1

uantification of Difluoroethane in Biological Fluids Two toxicological case reports were located which quantified difluoroethane in blood and urine, and other biological fluids and tissues. The method of detection and quantification is beyond the scope of this report; however, Gas Chromatography/Mass Spectrometry (GC/MS) seem to be the method of choice6. The case reports give an insight into the quantification of DFE from the time of inhalation and can serve as a reference base line.

www.collisionpublishing.com

knowledge of the extent of use before the time of the crash, or how recent the inhalation occurred.

Figure 4: Falcon Safety Products, Inc. the maker of “Dust Off®” adds a bittering agent to their duster to discourage inhalant abuse.

The second case report involved a 24-year-old driver who was pronounced dead following a single vehicle accident8.

A surviving front seat passenger witnessed the decedent inhaling Dust-Off® keyboard cleaner just prior to losing control of the vehicle. The sudden death associated with traumatic injuries of traffic crashes may provide a snapshot of the pharmacokinetic characteristics of acute inhalant use. The sudden cessation of normal physiologic activity following a traumatic death will “freeze” conditions within the body to provide a snapshot in time of drug levels at that moment9. This case presents a uniquely witnessed observation of the apparent impairing effects and consequences of the acute inhalation of DFE and the operation of a motor vehicle.

The results of DFE quantification in the driver’s biological samples are listed in Table 18.

Postmortem 1,1 - Difluoroethane Levels Tissue Concentration

Figure 5: Example of anti-inhalant formulas to help prevent inhalation of the product The first case report involved an 18-year-old driver and a 17-year-old passenger of an automobile that were killed when their vehicle crossed the median of a 4-lane highway and collided with a minivan7. A can of airbrush propellant was found in the vehicle of the deceased driver. During the initial toxicological analyses, a blood alcohol concentration of 0.13 g/dL was found in the blood of the driver. When performing ethanol analysis by headspace gas chromatography, an unidentified peak was observed in the blood of both deceased. This peak was identified as difluoroethane, the propellant in the aerosol can found in the vehicle. The concentration of difluoroethane in the blood of the driver was 78 mg/L and 35 mg/L for the passenger. This 1997 paper was the first to quantify difluoroethane in a biological sample. The interpretation of this data is difficult without

Aortic blood

85.6 mg/L

Chest cavity blood

79.9 mg/L

Pulmonary arterial blood

40.3 mg/L

Femoral blood

29.8 mg/L

Vitreous fluid

21.2 mg/L

Brain tissue

11.7 mg/kg

Liver tissue

27.9 mg/kg

Urine fluid

71.0 mg/kg

Gastric contents

51.8 mg/total

Table 1: Driver’s DFE Levels.

As shown in Table 1, the levels of DFE in tissue sites displayed a significant variability. The blood obtained from the aorta, which contains recently oxygenated or ventilated blood, contained the highest DFE level of 85.6 mg/L. This is consistent with a very recent inhalation of the DFE compound into the lungs, where initially the inhalant is inspired by means of pulmonary ventilation10. It dilutes with the functional residual air and is distributed to the alveolar membrane (the thin barrier through which gases

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 81

are exchanged between the alveolar air and the blood in the pulmonary capillaries) or the blood-air barrier. Diffusion takes place here until equilibrium occurs in the pulmonary blood (blood which is carried to and from the lungs).

Alveolar air is the respiratory gases in the alveolus of the lung and can be analyzed for its content of oxygen, carbon dioxide, or other gases by collecting the last portion of air expelled by maximum exhaustion11. Then the gas is distributed through the peripheral arteries to the individual tissues. The venous blood from the tissues returns to the lungs carrying some fraction of its original gas concentration. In this manner, the alveolar, arterial, tissue and venous tension of the inhalant gradually rise toward eventual equilibrium10.

The lower blood levels were calculated in those sites further from the lungs and heart. The femoral blood (femoral artery in the groin) contained 29.8 mg/L, whereas blood in the pulmonary artery (carries deoxygenated blood from the heart to the lungs), just prior to being exhaled, contained 40.3 mg/L, or that fraction of its original gas concentration. This demonstrates the absorption of DFE through the body tissues while traveling from the site of introduction (lungs) to the more peripheral arterial blood sites with transportation of mixed blood back to the heart through the venous system and finally back to the lungs.

This variability in DFE levels throughout the circulatory system indicates a dynamic process of DFE distribution related directly to the acute inhalation8.

year where dusting was suspected as a contributing cause of the crash. Your authors are aware of at least three other local incidents involving deaths of young adults who were dusting to get high. Simple internet research reveals countless other incidents and tragic losses of life.

In 2011, a crash occurred in a northern Illinois suburb of Chicago. The driver of the bullet car, a Jeep Grand Cherokee, was traveling down a large four lane highway.

The Jeep approached a stop light controlled intersection and slowed for a red light. A small car was stopped at the stop bar in front of it, and the Jeep bumped it out of the way. The Jeep then drove diagonally across the intersection, against a red light, continuing as if there was no one driving the car. It struck multiple street signs in the grassy median and stopped approximately 200 feet away.

At that point some Good Samaritans ran to the Jeep to offer aid. The witnesses later reported to police that the driver looked “contorted” with his hands in the air, and he appeared intoxicated. The witnesses reported the driver suddenly regained consciousness and Jeep sped off down the highway at speeds in excess of 20 miles per hour over the posted speed limit.

A

pplication to Traffic Crash Investigations Dusting has crossed over into the world of traffic crash reconstruction, and awareness of the phenomenon is limited with many investigators. It should be noted though, that most deaths associated with volatile substance abuse (VSA) are attributed to the direct toxic effects delivered by the inhalant on sensitive tissues, with deaths caused by trauma related inhalant abuse (traffic crashes) representing only a small fraction9.

As part of the law enforcement community ourselves, drug training has focused on the common drugs of choice and new breakthrough drugs in non-traditional DUI arrests.

We have yet to attend any courses with regards to dusting, and have found the same sentiment with law enforcement from many other jurisdictions and states. In addition, medical examiner offices typically do not have details about the dusting phenomenon, and prosecutors are only exposed to the problem when a case lands on their desk.

Investigators from the Major Crash Assistance Team of Lake County (MCAT), a multi-jurisdictional crash response team comprised of over 30 jurisdictions north of Chicago, have investigated two fatal crashes within the last

82

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

Figures 6 and 7: Scene photographs

Meanwhile, the victim was having her Volkswagen GTI washed and dried at a full service car wash before departing to a weekend family event. A car wash attendant towel dried her car, signaled its completion, and the victim walked to her driver’s door.

By this time the Jeep had traveled about one mile down the roadway. 9-1-1 was called by witnesses, but police were not able to catch up to the Jeep. The Jeep approached the car wash, drove off the roadway into the right side ditch, and it struck a warning sign. The Jeep continued bulleting directly at the Volkswagen. The victim, who was reaching for her door handle, had no chance to get into her car. The Jeep struck and pinned her between the cars. Scene photo-

graphs are shown in Figure 6 and 712. Tragically, the victim died instantly at the scene. Video surveillance from the red light enforcement cameras, as well as local gas stations and the car wash confirmed all driving activities. Figures 8 and 912 are video screen captures of the Jeep at the red light controlled intersection and as it left the roadway and crashed into the Volkswagen.

A crash investigation was begun, and located on the front seat of the Jeep was a can of compressed dusting spray shown in Figure 10. During a subsequent interview with the driver, he admitted to smoking cannabis, but would not answer direct questions in regard to huffing. His blood and urine later confirmed cannabis in his system,

1

2

3

4

5

6

Figure 8: Video screen captures of the Jeep at the red light controlled intersection and as it left the roadway and crashed into the Volkswagen

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 83

1

2

3

4

5

6

Figure 9: Video screen captures of the Jeep at the red light controlled intersection and as it left the roadway and crashed into the Volkswagen but no testing for dusting compounds or DFE were completed during the initial investigation. The driver was later charged with Reckless Homicide and Aggravated Driving Under the Influence of Drugs. During litigation, the driver’s defense was that he had a seizure while driving and the effects witnessed were from the seizure, not dusting. In late 2012, the driver plead guilty and is serving a six year prison sentence for Driving Under the Influence of Drugs.

84

Collision Magazine - Volume 8 Issue 1

R

esearch and Protocol Development After the subsequent crash investigation, research was conducted specific to dusting. Many studies about inhalants were reviewed, including detailed information from the National Institute on Drug Abuse. Your authors learned about symptoms from dusting, and how amazingly quick and frightening the effects are. Also, that the most common compound contained

www.collisionpublishing.com

within the aerosol is a fluorine compound called Difluoroethane.

A year after investigating the first incident, your authors learned of testing procedures to detect Difluoroethane and other inhalants in blood. It was also learned that only certain laboratories across the United States have the proper testing equipment. Being that the substance is so volatile; it enters the system quickly and exits rapidly. From a crash investigation standpoint, this creates a magnitude of issues for criminal prosecution. After learning this new informa-

Figures 10 and 11: Can of compressed dusting spray shown on the seat of the Jeep

tion, your authors resubmitted the driver’s blood from the aforementioned investigation for additional toxicology testing, but Difluoroethane was not detected.

Working with the Illinois State Police Crime Laboratory, the Northeastern Illinois Regional Crime Laboratory, and the Lake County (Illinois) State’s Attorney’s Office, a first responder protocol was developed. The protocol can be used during crash investigations or during any other situation where volatile inhalant abuse is suspected. The key to the entire protocol is critical timing. Blood samples must be taken quickly after the crash and they must be refrigerated and submitted for specified testing within 14-days.

Specimen collection should be in a gray top tube (Sodium Fluoride/Potassium Oxalate) with a minimum amount of 3 mL for examination for identification and quantification. The specimen container must remain tightly sealed.

Stability at room temperature is 7 days, refrigerated is 14 days, and frozen is 14 days6.

The air canisters are often found in the vehicles at the time of the crash investigation. This is because the effects are immediate and the driver does not have time to discard the “evidence.” Unlike over consumption of alcohol, a person can leave the cans or drinks behind in a bar and have no physical evidence of consumption in the vehicle. This may be the investigators first clue of DFE inhalation. Therefore, a search of the vehicle at the crash scene should be done to identify possible DFE inhalation or any other signs of VSA. The blood/urine taken during the DUI investigation is then analyzed for DFE. No special process or additional samples are required prior to or up to this point. If there was an air canister or other volatile substance container located in the vehicle at the time of the crash investigation, this should also be given to the laboratory for analysis and comparison of the ingredients to that located in the biological specimen, Figure 11. Had responding law enforcement known this during the prior investigation, identification of the intoxicating compound may have changed the criminal prosecution and eventual case outcome. Figure 12 is an example of a toxicological report showing the presence only (qualitative analysis) of DFE in the blood specimen and DFE in the canister. In this case, the speci-

Figure 12: Toxicological report showing the presence only (qualitative analysis) of DFE in the blood specimen and DFE in the canister. www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 85

men would need to be sent to another lab that could do quantification, such as shown in Table 1.

Huffing or VSA Investigative Protocol: 1. Investigate the incident and obtain samples of blood and urine using DUI Test Kits, which contain the gray top tubes, after following appropriate procedures, within 2-hours after the crash 2. Refrigerate the samples as soon as possible 3. Contact crime laboratory toxicology department and discuss the immediacy of the request with the scientist 4. Transfer the DUI Kit samples and inhalant canister to the laboratory via hand delivery, as the canister cannot be mailed and the samples need to remain cool 5. Once arriving at the laboratory, inform the intake official of the special considerations and need for refrigeration 6. Follow up with the laboratory to ensure your request is being handled within the two week time frame. The sample must be tested within two weeks

C

ase Example-Post Protocol In another recent fatal crash investigation, a driver was traveling down a local roadway and lost consciousness. The vehicle crossed all traffic lanes, left the roadway on the opposite side of the road, and it struck a family walking down the sidewalk. Tragically, a young child was killed by the vehicle.

During this investigation, MCAT investigators were armed with the above protocol and followed it properly. Toxicology testing revealed the driver’s blood contained Difluoroethane. The testing supported evidence recovered, video surveillance of the crash, and statements by witnesses.

Criminal prosecution is still pending; therefore, no other information is available at this time.

C

onclusion In conclusion, crash investigators need to be prepared for inhalant abuse crashes. Arrange a method for quickly obtaining blood and urine samples from drivers involved in incidents where dusting is suspected. Coordinate with your evidence custodian to transfer the samples as soon as possible for testing. Contacting and arrange a testing procedure with your laboratories before an incident occurs. Last, disseminate the protocol to first line law enforcement for use in crash investigations and other matters. With dusting investigations, time is truly of the essence.

86

Collision Magazine - Volume 8 Issue 1

R

eferences

1. Photograph courtesy of Allen County Sheriff’s Department, Fort Wayne, Indiana

2. Inhalant Abuse. National Institute on Drug Abuse, NIDA Research Report Series, July 2012.

3. Swifferkillsdogs (Author Screen Name). (2011, December 14). Jimmy Huffing Dust Remover [Video file].

Retrieved from http://www.youtube.com/ watch?v=FjlazUNE2-8 4. U.S. Department of Health and Human Services. National Institute on Drug Abuse/University of Michigan Monitoring the Future Study. (July 2012). Inhalant Abuse (NIH Publication #12-3818) [Brochure]. Washington, DC: U.S. Government Printing Office. Retrieved from database.

5. U.S. Department of Health and Human Services. Substance Abuse and Mental Health Services Administration.

(2008). Trends in Adolescent Inhalant Abuse: 2002 to 2007 (DHHS Publication No. SMA 08-4343, NSDUH Series H-34) [Brochure]. Washington, DC: U.S. Government Printing Office. Retrieved from database.

6. NMS Labs Online Test Catalog. http://www.nmslabs.com/SearchResults. aspx?code=1611B 7. Broussard LA, Brustowicz T, Pittman T, Atkins KD, and Presley L. Two Traffic Related Fatalities Related to the Use of Difluoroethane. J. Forensic Sci. 42(6): 1186-1187 (1997).

8. Hahn T, Avella J, and Lehrer M. A Motor Vehicle Accident Fatality Involving the Inhalation of 1,1-Difluoroethane. Journal of Analytical Toxicology Volume 30: 638642, (2006).

9. Flanagan RJ, Street PJ, and Ramsey, JD. Volatile Substance Abuse: Practical Guidelines for Analytical Investigation of Suspected Cases and Interpretation of Results.

Technical Series No. 5. United Nations International Drug Control Program, Vienna, Austria, 1997.

10. Wegner M and Goff J. A Review – Uptake and Distribution of Inhalant Agents. AANAJ 47(1): 55-62 (1997).

11. Mosby’s Medical Dictionary, 8th Edition, Elsevier, 2009 12. Figures 6-10 courtesy of Lake Zurich Police Department, Lake Zurich, Illinois

www.collisionpublishing.com

I AM

CRASH INVESTIGATION It’s not like it looks on TV. Crash scene investigations are never cut and dry. Time is always of the essence.

No two scenes are alike, so I need my mapping equipment to be easy to run and versatile enough to handle any situation I’m up against.

We hear you.

Look into our TruSpeed® Sx laser, a more versatile and less expensive alternative to a total station.

See us at the ARC-CSI Crash Conference in Las Vegas and check out our presentation at the Motor Speedway.

Register for our May 21 Crash Scene Mapping Webcast.

lasertech.com/collision 877.696.2584

Accuracy of GPS Speed and Location Data Measured in Emergency Vehicles David M. Little, P.Eng. Baker Materials Engineering Ltd.

Cst. Rob Joiner Abbotsford Police Department

Cpl. Stephen Hilliard RCMP ICARS

88

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

A

GPS

bstract Modern emergency vehicles, such as police vehicles and fire trucks, are often equipped with a Global Positioning System (GPS) receiver that can determine the vehicle’s speed and location based on data the unit receives from satellites. This information is typically converted to one or more National Marine Electronics Association (NMEA 0183) format sentences or a proprietary Trimble (TAIP) sentence and then stored in a user-specified format in a laptop within the vehicle. This laptop - often referred to as a docking station or a mobile work station (MWS) - then forwards this data to a central dispatch by way of an on-board modem.

In the event that one of these vehicles is involved in a collision, the GPS data can be retrieved either from the onboard computer or from the central server to assist with a collision reconstruction.

Typically, the speed of the vehicle just prior to impact is of particular importance. The GPS data may also be relied upon to determine the path travelled by the vehicle and its direction prior to impact.

In order to properly apply the GPS data to an investigation, the reconstructionist must know the limitations of the information. In particular, one needs to know the magnitude of the potential error in the speed and location values. The subject

testing set out to quantify these potential errors for some GPS receivers commonly used in emergency vehicles. The testing also aimed to establish if there was any data in the NMEA sentences captured by the GPS receivers, but not always available in the on-board laptop, that would allow for a more precise determination of speed.

Two GPS receivers that are not used on emergency vehicles but that can be found during collision investigations (an after market vehicle navigation system and a personal heart rate monitor for jogging or cycling) were also included in this study.

I

ntroduction GPS technology allows a receiver (such as a handheld GPS unit or a laptop in an emergency vehicle) to receive data from a network of satellites in orbit around earth. This data can be interpreted by the receiver to establish its location (latitude, longitude and elevation) and speed (horizontal and sometimes vertical) relative to earth. This information is typically provided by the GPS receiver at 1 Hz and includes a date and time stamp in Coordinated Universal Time (UTC). The GPS receiver provides this data in a standard NMEA or TAIP sentence - string of numbers and letters that have a specific order and meaning. Two standard GPS NMEA sentences that provide information valuable to a reconstruction and a TAIP sentence are shown below:

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 89

$GPRMC,192358.00,A,4901.437763,N,12221.311778,W,000.000,091.95,200212,, ,A*78 $GPGGA,192359.000,4901.43776,N,12221.31177,W,1,10,00.8,+00051.0,M,,M,,*4F RLN52482000+490380269-1222777231+000108270000+000000000702000400050010 00120025002900000000000012;ID=0000;*41 The NMEA data fields are comma separated while the TAIP data fields are not. What the values represent and their units are as follows: NMEA $GPRMC: 192358.00 A 4901.437763 N 12221.311778 W 000.000 091.95 200212 A*78

GPS Time in hour, minute, second Data validity flag, A = valid current Latitude in degrees, minutes, decimal minutes N = North, S = South Longitude in degrees, minutes, decimal minutes W = West, E = East Horizontal velocity in knots Heading angle, in degrees from true north Date, in day, month, year Checksum

NMEA $GPGGA: 192359.000 4901.43776 N 12221.31177 W 1 10 00.8 +00051.0 M (empty set) M *4F

90

GPS Time in hour, minute, second Latitude in degrees, minutes, decimal minutes N = North, S = South Longitude in degrees, minutes, decimal minutes W = West, E = East Fix quality: 0 = invalid, 1 = GPS, 2 = DGPS, etc.

Number of satellites being tracked Horizontal dilution of precision (HDOP) Altitude above/below sea level units, Metres Height of Geoid units, Metres Checksum

Collision Magazine - Volume 8 Issue 1

TAIP RLN: GPS Time in seconds since midnight sssss.sss +490380269 Latitude in degrees +/-dd. ddddddd -1222777231 Longitude in degrees +/-ddd. ddddddd +00010827 Altitude in Feet +/-ffffff.ff 0000 Horizontal Speed in mph sss.s +0000 Vertical Speed in mph +/-sss.s 0000 Heading in Degrees hhh.h 07 Number of satellites being tracked 02000400050010 SV ID for each Satellite 00120025002900 IODE for each Satellite - Not Supported 0000000000 Not Used 0 Fix Mode 0 = 2D, 1 = 3D, 9 = no fix available 2 Age of Data 2=<10sec, 1=>10sec, 0=not available 52482000

As with many other jurisdictions across North America, all police forces in British Columbia, Canada, rely on Versaterm’s Mobile Data Terminal (MDT) software to transfer data to, from and among vehicles. The MDT program stores the GPS data in the on-board laptop in two different file formats. One of these contains the raw NMEA or TAIP sentences and the other contains converted data.

The raw NMEA/TAIP sentences are stored in the MDTGPS.LOG file. This information is stored for typically one day (until midnight local time) and is deleted the next time the MDT program is started on a subsequent day. Caution must be exercised if an investigator attempts to copy this file by turning on a laptop that has the MDT program set to auto start. There are several ways to avoid losing this data, including removing the hard drive and installing it in another computer where it is known that the MDT program will not auto start. Alternatively, the hard drive can be imaged or the laptop can be started in Windows Safe Mode to prevent the MDT program from auto starting.

www.collisionpublishing.com

Following is an example of the data stored temporarily on the laptop in the MDTGPS.LOG file. This example shows a combination of raw NMEA sentences as well as converted data: 02-20 111244.046 NMEAMessage.ParseData Data:$GPRMC,170739.000,A,4901.9012,N,12220.7954,W,12.32,221.54,200212,,,E*48 02-20 111244.046 GPSMessage.LogValues TimeOfData: 170739.000 02-20 111244.046 GPSMessage.LogValues Lat: 49.0316867 02-20 111244.046 GPSMessage.LogValues Long: -122.3465900 02-20 111244.046 GPSMessage.LogValues MSLAlt: 102.7 02-20 111244.046 GPSMessage.LogValues HSpeed: 36.7 02-20 111244.046 GPSMessage.LogValues VSpeed: 02-20 111244.046 GPSMessage.LogValues Heading: 221.54 02-20 111244.046 GPSMessage.LogValues Source: 02-20 111244.046 GPSMessage.LogValues AgeOfData: 02-20 111244.046 GPSMessage.LogValues SVIdIODE: 02-20 111244.046 GPSMessage.LogValues X: 547762 02-20 111244.046 GPSMessage.LogValues Y: 5431184 02-20 111244.046 GPSMessage.LogValues ID: 02-20 111244.671 NMEAMessage.ParseData Data:$GPGGA,170740.000,4901.8986,N,12220.7988,W,6,00,50.0,102.7,M,16.6,M,,0000*5F 02-20 111244.780 NMEAMessage.ParseData Data:$GPGSA,A,2,,,,,,,,,,,,,50.0,50.0,50.0*06 02-20 111244.874 NMEAMessage.ParseData Data:$GPGSV,1,1,02,06,47,293,17,16,23,265,09*7E The raw NMEA data in the previous example are prefixed with ‘$GP’ while the converted data can be identified by “GPSMessage.LogValues”. The units for the data are not shown in this file but are as follows1: •

Speed values in knots are converted to units of km/h.

•

Latitude and Longitude values (in degrees) are converted to universal transverse mercator (UTM) values known as Easting and Northing, or X and Y values, in metres.

Unlike the raw NMEA sentences, which are logged every second, the converted values are not logged as frequently.

The recording rates vary among computers, with recording rates as slow as once every ten seconds observed so far.2 The converted data are then saved in a different file (file name yyyymmdd.gps) and stored in another folder in the on-board laptop for typically 30 days. However, some of the data provided in the NMEA sentences, such as the number of satellites that the receiver was communicating with and the horizontal dilution of precision (HDOP, which is a measure of the geographical location of the satellites and helps define the accuracy of the location data), are not recorded in this file. Following is a typical character string for data stored in the local laptop for longer than one shift:

AB1T06|142011|120220142011|80446.000 |547876|5431263|+000183.73|21.7|0.0| 336.7| Vertical lines (|) separate the data fields. What the values represent and their units are detailed below: Call sign of member logged onto laptop 142011 MDT time when entry was written, in hhmmss 120220142011 MDT Date and Time, in yymmddhhmmss 80446.000 GPS satellite time, seconds from midnight UTC 547876 X position (or Easting) in metres, UTM 5431263 Y position (or Northing) in metres, UTM +000183.73 Altitude in feet above mean sea level 21.7 Horizontal speed, in km/h 0.0 Vertical speed, in km/h (not always measured) 336.7 Heading angle in degrees from true north AB1T06

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 91

The amount of information that is sent from the on-board laptop via the modem to a central server is further reduced, mainly due to the cost of sending data wirelessly. This frequency is controlled by the operation of the vehicle at the time. For example, a stationary police vehicle may send data to the central server only once every 60 seconds while a vehicle travelling will send that data approximately once every 15 seconds. A vehicle responding to a call may send information to the server approximately once every five seconds. For more information on the wireless network system employed in the province of British Columbia to connect various forces, the reader is directed to the article

by Brewer.3

the receiver than it actually is. Conversely, if the receiver is travelling towards the satellite, then the perceived signal frequency measured at the receiver increases. This change in frequency, the Doppler shift, from several satellites can be used to calculate the speed of the vehicle. This method is preferable because it provides an instantaneous speed and is independent of any errors in location measurements.

The second method for measuring speed is based on the time for the receiver to travel the distance from point A to point B. This provides an average speed between these two locations, which is less precise than the instantaneous speed measured by the previous method. As well, this speed calculation is subject to any errors in determining locations A and B. Testing by Chalko 4 and Fuch 5 revealed that Doppler speed measurements are approximately one order of magnitude more accurate than speed derived from position and time data. Consequently, it is important to be aware of which method the GPS receiver utilizes to calculate speed in order to assess the accuracy of the data.

“ ” One of the main parameters the reconstructionist seeks to extract from the GPS data is speed. The speed of the GPS receiver (and hence the vehicle) can be determined by one of two different methods.

One of the main parameters the reconstructionist seeks to extract from the GPS data is speed. The speed of the GPS receiver (and hence the vehicle) can be determined by one of two different methods. The receiver can measure its speed based on the Doppler shift in the signal it is receiving from the satellites, or it can calculate an average speed between two locations (often called “waypoints”, or “trackpoints”).

The first method employs the same principle as that used by a radar gun for measuring speed. If the GPS receiver (the vehicle) is travelling away from a satellite, then the frequency of the satellite’s signal is perceived to be lower at

92

Collision Magazine - Volume 8 Issue 1

The subject testing aimed to establish not only the accuracy of the speed and location data captured by the GPS receivers, but also to determine if there is a decrease in the accuracy as the information is converted from the NMEA sentences to that which is stored in either the local laptop or at the central server. As well, the study aimed to establish if there was any information in the NMEA sentences that is currently not recorded in the laptop for longer than one shift, but which may help to increase the accuracy or understanding of the data.

L

iterature Review Performance data for GPS receivers can be found in the manufacturer’s specifications. However, this information is sometimes more challenging to find than it would first appear, for a few reasons. First, the GPS receivers are often integrated into other components, such as laptops or modems, and the GPS receivers are not always identified. In such cases, the laptop or modem manufacturer may or may not reveal what GPS receiver they employed.

Second, the performance specifications of the GPS receivers are often detailed in different ways by different manufacturers, making a comparison among receivers more challenging. Sometimes, an accuracy value is provided with no defined level of probability. Where the probability level is provided, it is often described using one of the following terms: •

CEP. This acronym refers to the Circular Error Probability, which is a value that is accurate with 50% probability.

www.collisionpublishing.com

•

RMS. This refers to the Root Mean Square, which is a value that is accurate with 65% probability. This is equivalent to one Standard Deviation.

•

2DRMS. This is twice the Root Mean Square, which is a value that is accurate with 95% probability.

The third challenge with obtaining GPS performance data from the manufacturer is that the accuracy of the manufacturer’s claims is unknown.

Several published articles have addressed the accuracy of GPS Doppler speed determined by various receivers.

Witte and Wilson 6 tested a REB 2100 GPS engine consisting of a SiRF star II chipset made by RoyalTek. The authors concluded that the receiver measured speed within 0.2 m/s of the true speed 45% of the time and within 0.4 m/s 64% of the time. This accuracy did not fluctuate significantly even when the receiver was communicating with a low number of satellites or when the location of the satellites was not optimal (the HDOP was high).

Chalko4 conducted testing on a Locosys GT-11 GPS receiver. The author concluded that the receiver was capable of generating average Doppler speed errors of 0.05 knot with a standard deviation of 0.05 to 0.1 knot. When multiple receivers were used and the speed values were averaged, the accuracy could be increased to 0.01 knot.

Fuchs5 also conducted accuracy testing on a Locosys GT11 handheld GPS receiver. He concluded that the average Doppler speed error was 0.08 knot with a standard deviation of 0.03 knot.

Butler et al.7 tested a Track Systems Traqmate data acquisition unit. This device was designed primarily for race car data acquisition. It uses a Wide Area Augmentation System (WAAS) differential GPS signal with a stated positional accuracy of 3 metres over a wide variety of locations and times of year. The authors concluded that the GPS system was capable of resolving distances of 0.75 ft (0.23 m) uncertainty with 95% confidence, provided these locations were measured over short time periods. Doppler speed values were compared with those from a Vericom VC2000PC and the values from each system were found to be within 1 km/h of the other.

Positional accuracy testing on a Leadtek 9546 GPS receiver was conducted by the University of California, Berkeley, in April 2004.8 After almost 18 hours of stationary testing on two receivers, the standard deviation ranged from 0.00006 to 0.002 degree (7 to 175 metres)9, when five satellites were tracked by the receiver.

Chalko10 conducted testing on the speed accuracy in a Locosys GT31 GPS receiver equipped with a SiRF3 chipset.

This receiver had the unique ability to measure and record a speed dilution of precision (SDOP) value that can be used to assess the accuracy of any given speed value. Stationary testing was conducted for twenty two hours and the recorded speed errors were compared with the SDOP theoretical predictions. The average speed error was not reported but appeared from their Figure 1 to be around 0.04 m/s while the maximum did not exceed 0.14 m/s.

The maximum SDOP value did not exceed 0.30 m/s and in all cases, the SDOP values exceeded the actual speed error. On average, the SDOP values were over four times higher than the actual errors, indicating that this SDOP value provided a very conservative estimate of the speed error.

Chalko was specifically interested in a method by which to accurately record the maximum speeds obtained during speed sailing. The author noted that, since the Doppler speed measurement is three dimensional, any error in the horizontal speed is related to an error in the vertical speed. Since Speed Sailing occurs on an essentially flat surface, any non-zero speed in the vertical direction represents a horizontal speed error. This suggests that vertical speeds can be used to provide an indication of any errors in the horizontal speed, even when the SDOP values are not available.

Reust 11,12, reported on the accuracy of several Garmin vehicle navigation GPS devices (models Nuvi 255W, Nuvi 765, Nuvi 1200, Nuvi 1300 and StreetPilot 2720). The devices stored speed calculated based on the elapsed time between logged data points, not based on an analysis of the Doppler shift. The sampling frequency varied, depending on the motion of the receivers. Higher sampling appeared to be associated with changing speed or direction. If the receivers were stationary, speeds weren’t logged for at least 25 seconds.

While the author did not specifically address the speed error, he graphically compared the Garmin receivers’ speed with that measured by a more sensitive GPS device (the Racelogic VBOX III). The graphs showed that the Garmin receiver speeds were generally in agreement with the VBOX, although there were times when the receivers were in error by as much as 60 km/h. These large errors appeared to be associated with times when the vehicle was changing direction (highway on-ramp). As for the positional accuracy, Reust noted that the measured locations fluctuated by about 2 to 3 metres over 4 hours.

Bortolin et al 13 conducted testing on four vehicle navigation GPS receivers and compared them with an engineer-

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 93

ing grade GPS receiver (Racelogic VBOX III). The four GPS receivers tested were a Garmin Nuvi 265, a Garmin Aera 510, a Magellan eXplorist 210 and a Tom Tom One. Three of the receivers were capable of logging files while the fourth receiver (Tom Tom One) could not. Data from the Tom Tom were recorded with a video camera. The three receivers that logged files recorded the leg speed, or average speed, between two locations. They did not record the Doppler speed. The speed error ranges for these three varied considerably between units with the Garmin Aera 510 providing the best results. Similar to the testing by Reust, the authors noted that the speed errors increased when the vehicle was changing speed or direction.

While the published testing revealed that those GPS receivers calculating Doppler speeds produced very small speed errors (for collision reconstruction purposes), none of the receivers tested were the same models as those used by local police forces. Consequently, further testing was conducted on GPS receivers commonly found in police vehicles.

T

est Procedure Test equipment: A review of some of the GPS receivers in use by the Abbotsford Police, Delta Police and the Royal Canadian Mounted Police (RCMP) was first conducted. It was determined that some of the GPS receivers were embedded in the portable laptops while oth-

Modem or Laptop Make

Model

Make

er receivers were located in the modems installed elsewhere in the vehicles. One obvious difference between these two locations is that those receivers located in the external modems had a separate antenna located on the top of the vehicle, while the laptops with internal GPS receivers did not have external antennas, unless the laptops were docked within a vehicle.

Table 1 lists common GPS receivers currently in use by local law enforcement detachments, plus some that had recently been retired and others that are scheduled to be in wide use in the near future. The list has been split into those with internal and external modems. The list is not intended to be an exhaustive list; it is only intended to provide a general listing of those receivers most likely to be encountered during a collision investigation. Based on the equipment identified above, modems and laptops were obtained that had the same or similar internal GPS receivers. Table 2 outlines the equipment that was obtained for testing. Equipment #8, 9 and 10 were selected since it was anticipated that they would provide an accurate reference against which the other equipment could be compared.

Equipment #11 was chosen so the receiver positions could be compared with a known reference point on the ground.

Internal GPS Receiver Stated Accuracy Model Speed Distance

Laptops with internal modems and GPS Receivers Panasonic Toughbook CF29 LeadTek LR9548 Panasonic Toughbook CF30 LeadTek LR9548 or LR9101 Panasonic Toughbook CF31 LeadTek LR9101 External modems with built-in GPS Receivers Sierra Wireless MP555 Trimble Lassen SQ Sierra Wireless MP595 Trimble Copernicus Sierra Wireless MP890 Trimble Copernicus Sierra Wireless GX400 Qualcomm MSM6290 Trimble Trimfleet APU Trimble Copernicus II Cypress CTM-15x Trimble Copernicus Solutions Cypress CTM-200 Fastrax IT500 Solutions

0.1 m/s 0.1 m/s

10m 2D RMS,<5m 50% 10m 2D RMS,<5m 50%

0.1 m/s

10m 2D RMS,<5m 50%

0.06 m/s 0.06 m/s 0.06 m/s 0.06 m/s 0.06 m/s

<6m 50%, <9m 90% 3m 50%, 8 m 90% <3m 50%, <8m 90% 3m 68% <2.5m 50%, <5m 90% 3m 50%, 8 m 90%

0.1 m/s

3m RMS (65%)

Table 1: Legacy, current and future GPS receivers identified in use by local emergency fleet vehicles.

94

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

T

esting Data Logging: In general, the information provided by each GPS receiver was unique to that model. Because of this, the programs used to log the data also varied among receivers. Following is a description of the data from each receiver and how it was logged.

Equipment #1 was an RCMP Panasonic Toughbook CF31, which had a LeadTek LR9101 internal GPS receiver.

This receiver provided the following NMEA sentences at 1 Hz: $GPGGA, $GPRMC, $GPGSV and $GPGSA. The raw NMEA sentences were logged with the VisualGPS (Freeware) program for this testing.

This raw information was converted by the Versaterm system and was stored in the Toughbook laptop in a .gps file once every 2 or 3 seconds. The file could be opened, viewed and/or edited as a text file. This file was also retrieved from the main server, where it was stored as a text file. Data were stored on the server roughly once every 60 seconds while the vehicle was stationary and roughly once every 18 seconds while the vehicle was moving. This laptop was not docked in the vehicle. Consequently, it did

not have the benefit of using the vehicle’s roof mounted antenna.

The type and frequency of the GPS NMEA sentences outputted by Equipment #2 and #3 (Cypress Solutions CTM15x and CTM-200, respectively) are configured by the manufacturer to meet the requirements of the client. Typically, the output rate is once every five or six seconds. For our testing purposes, the output rate was increased by the manufacturer to 1 Hz and the data were provided in the $GPGGA and $GPRMC formats. This allowed for time, location, horizontal speed, number of satellites tracked and HDOP to be monitored. The data was logged using Tera Term software. Equipment #3 also provided a proprietary sentence ($PACCEL), which contained acceleration data.

The acceleration information was not analyzed as part of the subject testing.

The Trimble Copernicus II receiver in the Abbotsford Police Trimfleet APU modem (Equipment #4) provided the following NMEA sentences at 1 Hz: $GPGGA, $GPRMC, $GPVTG, $GPGLL, $GPGSV, $GPGSA and. $GPZDA. These sentences were logged in a text file with the Microsoft Windows HyperTerminal program.

Internal GPS Receiver

#

Make

Model

Make

1

Panasonic

Toughbook CF31

LeadTek LR9101

1 Hz

2

CTM-15x

Trimble

Copernicus

1 Hz

CTM-200

Fastrax

IT500

1 Hz

0.1 m/s

4

Cypress Solutions Cypress Solutions Trimble

Stated Accuracy Speed Distance 0.1 m/s 10m 2D RMS,<5m (50%) 0.06 m/s 3m 50%, 8 m 90%

Trimfleet APU

Trimble

Copernicus II

1 Hz

5

Panasonic

Toughbook CF30

LeadTek LR9548

1 Hz

0.06 m/s <2.5m (50%), <5m 90% 0.1 m/s 10m 2D RMS,<5m (50%)

6 7 8

Garmin Garmin Vericom

Foreunner 305 Nuvi 3790 VC4000PC

Garmin Garmin Garmin

18x

1 Hz Variable 5 Hz

9

Vericom

VC4000PC

Garmin

18x

5 Hz

ATR 0625

1 Hz

3

10 Wintec 11 Leica

G-Rays I WBT300 AtmeluBlox RX1250XC

Model

Frequency

0.1 knot RMS 0.1 knot RMS

3m RMS (65%)

15m 95% 15m 95% 2.5m CEP (50%)

Table 2: Equipment chosen for testing. Accuracy reflects that determined based on available manufacturer data.

Information was not captured during testing on the Wintec GPS (Equipment #10). www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 95

Data files were also extracted from the computer’s logged files (both the temporary - MDTGPS.LOG - and the converted - yyyymmdd.gps - files). The .LOG file was recorded at 1 Hz and included the same NMEA sentences as those logged by the HyperTerminal program. The .gps files were recorded once every five seconds while the vehicle was stationary and once every two seconds when the vehicle was moving. However, neither of these files could be generated at the same time as the raw GPS files (NMEA sentences) were being saved with the HyperTerminal program.

The Leadtek LR9548 receiver in the Abbotsford Police Panasonic Toughbook CF-30 (Equipment #5) provided the following NMEA sentences at 1 Hz: $ GPGGA, $GPRMC, $GPGSV and $GPGSA. These files were logged with the Microsoft HyperTerminal program. Data files were also retrieved from this laptop as MDTGPS. LOG (including the same NMEA sentences as logged by the HyperTerminal program) and yyyymmdd.gps files, where they were stored at 1 Hz. Again, however, these files could not be logged at the same time as the raw NMEA sentences were saved. This laptop was not docked with the vehicle. Consequently, it was relying on its internal antenna for GPS data reception.

Data stored by the Garmin Forerunner 305 (Equipment #6 – Software Version 2.9) were logged internally and retrieved with the Garmin Training Centre program, Version 3.6.5. This receiver did not record any NMEA sentences or any Doppler speed. It recorded the time, position (latitude, longitude and elevation) and the leg speed. The number of satellites in communication with the receiver and their signal strengths could be viewed on the screen

Figure 1: The white Chevrolet Suburban (foreground) was equipped with ten GPS receivers and stationed adjacent to a geodetic marker on the road surface (located under the tripod).

96

Collision Magazine - Volume 8 Issue 1

during operation, but this information was not logged.

The receiver could be set to record data at one of two different rates. One rate varied depending on changes in direction or speed (Smart Recording). The other rate, (the one used for this testing), was fixed at 1Hz. The data files logged in the Garmin Nuvi 3790 (Equipment #7) were extracted with the Garmin MapSource software, Version 6.16.3. Like the Garmin Forerunner (Equipment #6), there were no NMEA sentences or Doppler speeds recorded, only time, location and leg speed data were recorded. There was no user control for the rate of data logging. This receiver recorded at a variable rate.

The Vericom VC4000PCs with Garmin 18x GPS receivers (Equipment #8 and #9) stored GPS data internally at 5 Hz. The type of data stored depended on the original set-up of the device and on the software installed in the equipment. Equipment #8 had firmware version 1.61 installed and was capable of storing location (latitude, longitude and altitude), plus leg speed and distance. Equipment #9 had firmware version 1.45 and could only record leg speed and distance. The NMEA sentences were not logged, regardless of the software type, although some of this information (including HDOP and number of satellites) was shown on the screen during testing. The data were extracted with the Profile 5 software.

The data for the Wintec G-Rays I WBT300 Atmel-uBlox (Equipment #10) were logged with the VisualGPS program The receiver provided the following NMEA sentences at 1 Hz: $ GPGGA, $GPRMC, $GPGSV and $GPGSA. However, after testing was completed, it was determined that only the final few minutes of the data had

Figure 2: Equipment #1 was positioned on the hood of the vehicle for Test #1.

www.collisionpublishing.com

Figure 3: Remote antennas and two of the receivers were installed on the roof of the vehicle.

Figure 4: Data logging was conducted with several computers. Equipment #5 (Toughbook CF30 laptop) was located inside the vehicle adjacent to a metal cabinet (arrow).

Figure 5: The Vericom VC4000PC units and the Garmin Nuvi were mounted on the windshield and the Cypress Solutions modems were taped to the dash.

The antennas for all but the Garmin receiver were mounted on the roof.

Figure 6: Equipment #4 had been permanently installed inside the vehicle between the front seats. This modem/receiver was equipped with an external antenna.

been logged. It is suspected that the data were inadvertently overwritten after testing was completed.

because this receiver only recorded data roughly once every 2½ minutes when it was stationary.

Test 1 - Stationary Test: The first test was conducted with all of the GPS receivers stationary. In this situation, each receiver should report a speed of zero. Therefore, any nonzero speed values represent the magnitude of the speed error. It is important to note that even though the receivers are stationary with respect to earth, the receivers are all travelling very fast relative to the orbiting satellites.

Hence, the speed errors at zero speed (relative to earth) should be the same as when the receivers are travelling 100 km/h. This method was not effective for establishing the speed error in one of the Garmin receivers (Equipment #7)

Ten GPS receivers were installed in or on an unmarked police Chevrolet Suburban. For those receivers with external antennas (Equipment #2, 3, 4, 8, 9), the antennas were mounted on the top of the vehicle. As well, the Garmin Forerunnner (Equipment #6) and the Wintec receiver (Equipment #10) were mounted on the vehicle’s roof rack.

The Garmin Nuvi 3790 (Equipment #7) was mounted on the windshield, as it would be during normal operation.

One of the laptops with the internal receivers was left to stand alone on the hood of the vehicle (Equipment #1) and the second laptop (Equipment #5) was positioned inside

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 97

the vehicle, adjacent to a steel storage box. The mounting locations of the equipment are shown in Figures 1 to 6.

Each of the devices was turned on and the various data acquisition systems for the devices were set to log data. Once the final piece of equipment was logging data, the receivers were left to collect data for a minimum of one hour. For most of the receivers, data were collected for close to two hours.

During data logging, a Leica GPS survey instrument was used to locate each of the GPS receivers. The base station was placed vertically over a geodetic survey monument located on the road and measurements were taken with the rover placed directly on each receiver or its antenna.

Test 1 measured not only the fluctuation (error) in speed but also the accuracy of the location (latitude and longitude) values.

Test 2 - Moving Vehicle: Following Test 1, the instrumented vehicle was driven at varying speeds with all GPS receivers still being logged. The purpose of this test was to establish if the speed errors measured by Test 1 varied with changes in speed, changes in direction or due to interference from adjacent geographical features. Equipment #1 was placed inside the vehicle for this testing.

D

ata Synchronization In order for GPS systems to provide a reasonable level of accuracy, they have to measure time to a very small fraction of a second. Even an error of one thousandth of a second can translate into a distance error of over 300 km.14 For this reason, satellites are equipped

Equip #1 Equip #2 Equip #3 Equip #4 Equip #5 Equip #6 Equip #7 Equip #8 Equip #9

Panasonic Cypress Solutions Cypress Solutions Trimble Panasonic Garmin Garmin Vericom Vericom

Toughbook CF31 CTM-15x CTM-200 Trimfleet APU Toughbook CF30 Forerunner 305 Nuvi 3790 VC4000PC VC4000PC

with highly accurate atomic clocks. Thanks to some complex math, this same timing accuracy is available at the GPS receiver, even though it is not equipped with an atomic clock.

It was anticipated that this highly accurate time could be used to synchronise the data collected by the GPS receivers during test 2. However, even though the time was measured very accurately, most of the GPS receivers only recorded time to the units (1 second), while two of the receivers (the Vericom receivers) did not record the GPS time stamp at all, relying only on a user-entered time. Because of this, the receivers with satellite time stamps were synchronised with each other based on the recorded times (within 1 second) and the other receivers were adjusted to provide a “best fit” based on the speed values. As well, the Garmin Nuvi provided only an average speed over several seconds, so it was also adjusted to provide a best fit. Fortunately, Test 1 did not require an exact synchronization in order to measure the speed error of the different receivers.

R

esults Speed error: Test #1 - Stationary Vehicle The average value of the recorded speed was calculated for each device during Test 1. The speed values were taken directly from the logged NMEA data. Since the vehicle was stationary, the values represented the magnitude of the speed errors. The maximum errors and the standard deviation from the average errors were also calculated. Table 3 illustrates the magnitude of these errors, along with the resolution of each device.

Speed error Ave RMS km/h km/h 0.1783 0.103 0.0024 0.048 0.0004 0.006 0.0020 0.013 0.4791 0.396 0.1009 0.157 0.0067 0.009 0.0054 0.009

Max km/h 1.39 1.48 0.20 0.22 5.17 2.00 0.04 0.22

Resolution km/h 0.018 0.18 0.18 0.036 0.018 0.01 0.01

Table 3: Comparison of average, maximum and standard deviation of speed errors, plus the resolution for each device tested.

98

Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

As Table 3 illustrates, the average speed error ranged among receivers from less than 0.001 to 0.5 km/h. This would be considered very accurate for most collision reconstructions. Figure 7 shows the average and RMS error for each receiver in graphical form, for easier comparison.

As Figure 7 demonstrates, three receivers had considerably higher speed errors than the other receivers. Two of these receivers (Equipment #1 and #5) were not connected to external antennas while the third receiver (Equipment #6) was not recording Doppler speed, only an average speed calculated based on the time to travel between two locations. All of the other receivers (those with low speed errors averaging less than 0.01 km/h) were connected to external antennas.

Equipment #1 and #5 were both Panasonic Toughbooks with internal antennas. Equipment #1 was positioned on the hood of the vehicle and therefore was in a better position to “view” the satellites. Conversely, Equipment #5 was installed in the centre of the vehicle against a metal

storage container, which would have limited the receiver’s ability to receive the satellite signals. The maximum error associated with this device was 5 km/h, out of 6092 samples.

Table 4 compares the measured accuracy relative to the manufacturers’ stated accuracy. In all but one case, the measured errors (both average and RMS) were below the manufacturer’s claimed accuracy. The one case where this was not true was for the laptop that was not connected to an external antenna and was located inside the vehicle adjacent to a metal box. This receiver was poorly positioned to receive the satellite signals.

Test #2 - Moving Vehicle: During Test #2, the instrumented vehicle was driven from test site #1 onto a nearby freeway and then returned to test site #1, following the same route. The route was driven twice. Each lap consumed approximately 18 minutes and the maximum speed approached 120 km/h.

Figure 7: Average speed errors for the devices capable of recording data while stationary.

Note that Equipment #1 and #5 were not docked for this testing, so they did not have the benefit of the external antennas that they would otherwise use. www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 99

Speed Accuracy # 1 2 3 4 5 6 8 9

Make

Model

Stated

Panasonic Cypress Solutions Cypress Solutions Trimble Panasonic Garmin Vericom Vericom

Toughbook CF31 CTM-15x

0.1 m/s 0.06 m/s

0.36 0.22

Measured km/h km/h Average RMS 0.18 0.21 0.00 0.05

CTM-200

0.1 m/s

0.36

0.00

0.01

Trimfleet APU Toughbook CF30 Forerunner 305 VC4000PC VC4000PC

0.06 m/s 0.1 m/s

0.22 0.36

0.1 knot RMS 0.1 knot RMS

0.19 0.19

0.00 0.48 0.10 0.01 0.01

0.01 0.62 0.19 0.01 0.01

km/h

Table 4: Comparison of manufacturers’ stated speed accuracy with measured accuracy.

Figure 8 illustrates the speed values from each device during the first lap. A closer look at these speed values is shown in Figure 9. As indicated previously, the time stored in each GPS receiver typically did not record to a higher accuracy than one second so exact synchronization of the

various receivers was not possible. This prevented a precise comparison of the variance among speed values. However, the graph does reveal that the variations in speeds among receivers were similar to the low error values determined from Test #1. Certainly the recorded speeds were almost

Figure 8: Speed values recorded by each GPS receiver during Test #2.

100 Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

“”

always well within 1 km/h of each other for all but one receiver. The Garmin Nuvi had a considerably higher speed error than the other receivers.

For several of the GPS receivers, the speed values briefly spiked or dropped twice, as shown in Figure 10. These were transient errors that each of the receivers experienced at the same time. The location of the test vehicle when these fluctuations occurred was investigated and determined to have been immediately adjacent to a freeway overpass. Thus, the GPS units were most likely receiving a bounced or otherwise altered signal from the overhead roadway, resulting in errors of up to about 5 km/h. The fluctuations were observed twice because the vehicle returned along the same path that it departed, thereby passing under the overpass twice.

This speed fluctuation reveals the importance of reviewing the surrounding geography in order to establish if any of the speed values used in a reconstruction could have been affected by a bounced signal. It also indicates the need for further testing in a well developed urban area (with numerous tall buildings, such as in a downtown core).

Certainly the recorded speeds were almost always well within 1 km/h of each other for all but one receiver.

S

peed Conversion Error The MDTGPS.LOG files were extracted for Equipment #1, #4 and #5. These files record both the raw NMEA sentences as well as the converted speed values but are only saved for one day. The raw speeds from the NMEA sentences were compared with the speeds that had been converted and stored in the local police laptops.

Figure 9: Close-up of the speed values in the range of 100 to 120 km/h. www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 101

For Equipment #1 and #5, the speeds recorded in the laptop were all higher than the raw NMEA speeds by a factor of 1.609, as shown in Figure 11. Notably, this value is equivalent to the conversion factor from mph to km/h.

For Equipment #1, the speeds were also compared with that stored in the central police server. The speed values stored in the central server were also found to be high by a factor of 1.609. Thus, the speed values were being improperly converted from the data in the NMEA sentences

Figure 10: Several of the GPS receivers experienced brief fluctuations in speed (circled) at the same time and location, which was attributed to signal bounce or interference from an overpass.

Figure 11: Logged file showing speed conversion error.

102 Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

to that stored in the local laptop. There was no error in the time stamps recorded by the various files, as shown in Figure 12.

For Equipment #4, the converted data (including time, speed and location) were extracted from the local laptop even though they did not overlap with the test period. A time when the vehicle was travelling in a relatively straight

direction at a constant speed was identified and the average speed was calculated based on the distance travelled (per the UTM coordinates) for the recorded time. This speed was then compared with the average of the Doppler speed values recorded in the local laptop. This revealed that the Doppler speed values were being recorded correctly (without any conversion error) in the local laptop – see Figure 13.

Figure 12: Comparison of NMEA, MWS and Server data for the same GPS receiver.

Figure 13: The average speed between two points was calculated and compared with the recorded average Doppler speed for Equipment #4. www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 103

These three GPS devices (Equipment #1, #4 and #5) all stored the data in Panasonic Toughbook laptops. Equipment #4 and #5 belonged to the Abbotsford Police Department while Equipment #1 belonged to the RCMP. This revealed that the error was independent of which police force owned the laptops.15 Further investigation revealed that this error was caused by a conversion error identified in the MDT.ini file.

U

nrecorded Information As indicated previously, not all of the information in the NMEA sentences was being recorded in the local laptop (MWS) or on the central server. Furthermore, some of the NMEA information was being converted prior to being stored. The question arose as to whether any of this unrecorded information would be useful to better assess the accuracy of the speed values.

Two values that were observed by the GPS receivers in the police vehicles but that were not recorded in the local laptops were the number of satellites that each receiver had been communicating with and the horizontal dilution of precision (HDOP). It was anticipated that a low number of satellites and a higher HDOP would correspond with a higher speed error. To confirm this, these three values

were graphed against each other for Equipment #5, since this receiver had the highest speed error and was therefore expected to show any trends more readily.

Figures 14 through 16 show the relationship between the speed error, number of satellites and the HDOP. As the graphs show, a high speed error typically corresponded with a lower number of satellites but not a higher HDOP. Conversely, a low speed error could be found with either a high or low number of satellites and with either a high or a low HDOP. Hence, knowledge of these two unreported parameters (number of satellites and HDOP) would not significantly increase an investigator’s ability to confirm a low speed error.

The SDOP parameter identified by Chalko in the Locosys GT-31 receiver appears to provide much promise for establishing any potential speed error. However, this parameter was not recorded by any of the receivers we tested.

In light of the opinion expressed by Chalko regarding the potential to assess the horizontal speed error based on fluctuations in the vertical speed, we attempted to compare these two values. However, it was noted that none of the GPS receivers tested provided a vertical speed, even though

Figure 14: Speed error, number of satellites and HDOP were plotted against time.

104 Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

Figure 15: Speed error versus horizontal dilution of precision.

Figure 16: Speed error versus number of satellites www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 105

the files created in the local laptop and on the central server had a designated column for vertical speed. In some cases, the vertical speed was reported in both the on-board laptop and the central server as a blank field. Other times it was recorded as “0.0” and this value did not fluctuate.

A review of the standard NMEA sentences revealed that none of them provided a vertical speed. However, vertical speeds were found in files retrieved earlier from a Sierra Wireless MP595 modem equipped with a Trimble Copernicus GPS receiver. This modem had been removed from an RCMP vehicle that had been involved in a collision previously. The Trimble receivers were capable of providing a proprietary message (TAIP) that reported a vertical speed. This modem was originally slated to be tested but was not, for reasons not addressed here. Although two of the other receivers we tested used Trimble receivers, neither one had been configured to provide a TAIP message or a vertical speed.

As indicated previously, the raw NMEA sentences were observed by the GPS receivers every second. However, the GPS information that was recorded in the laptop for longer than one day was recorded at a rate of only once every several seconds (the time between recordings varied).

Considering a vehicle can travel a considerable distance in several seconds, it would be far better (for a collision reconstruction) to obtain the data recorded in the laptop at 1 Hz. Therefore, every effort should be made to secure the MDTGPS.LOG file following a collision before it is erased.

Equip #1 Equip #2 Equip #3 Equip #4 Equip #5 Equip #6 Equip #7 Equip #8 Equip #9

Panasonic Cypress Solutions Cypress Solutions Trimble Panasonic Garmin Garmin Vericom Vericom

Toughbook CF31 CTM-15x CTM-200 Trimfleet APU Toughbook CF30 Forerunner 305 Nuvi 3790 VC4000PC VC4000PC

L

ocation Error Test 1 - Stationary Vehicle: The geodetic location coordinates (latitude and longitude values) for each GPS receiver were converted to UTM values using the spreadsheet provided by the University Of Wisconsin – Green Bay 9. The difference between each UTM value and the surveyed location (which had been adjusted to match the known location of the geodetic marker on the road) was calculated and represented the location error for each reading. The Northing and Easting values were combined (as the root of the sum of the squares) and these values were averaged. The RMS and maximum errors were also computed. As shown in Table 5 and Figures 17 and 18, the average location errors ranged from approximately 2 to 5 metres while the maximum errors ranged from approximately 3 to 22 metres. These values were larger than the resolutions of each receiver.16 Once again, the larger errors were associated with the receivers that did not have external antennas.

The location files for several of the receivers (equipment #1 through #6) were converted using the GPS Visualizer program (http://www.gpsvisualizer.com/) and exported into Google Earth. Not all of the receiver files could be converted due to format incompatibilities. Figure 19 illustrates how the logged locations of each receiver drifted with time.

The measured location errors were compared with the manufacturers’ stated accuracies (see Table 6). Not all of the GPS receivers recorded locations or provided a level of accuracy. However, where the locations and accuracies were provided, the receivers measured locations with erLocation Error Ave RMS Max (metres) (metres) (metres) 1.65 2.49 15.65 1.62 1.94 3.65 1.86 2.08 3.40 2.32 2.47 4.20 5.14 6.17 21.91 1.94 2.22 5.89 2.53 2.57 3.17 -

Resolution Latitude Longitude (metres) (metres) 0.18 0.12 0.55 1.2 0.75 0.49 0.02 0.01 0.18 0.12 1.85 1.22 0.02 0.01 -

Table 5: The average, RMS and maximum errors are shown for each GPS receiver tested.

Average errors ranged from approximately 2 to 5 metres.

106 Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

Figure 17: The average and RMS location errors are shown for each GPS receiver.

Figure 18: The maximum location errors are shown for each GPS receiver. www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 107

Figure 19: With the GPS receivers stationary for at least one hour, the coordinates drifted with time, with Equipment #5 (purple) drifting the most. The locations are overlaid on a Google Earth image of the road where the instrumented vehicle had been parked.

Location Accuracy #

Make

Stated Error %

Model

Measured Error %

<2.5m <3m <5m <8m <10m <15m < 2.5m < 3m 50%

1 Panasonic

Toughbook CF31

2 Cypress Solutions

CTM-15x

50%

3 Cypress Solutions

CTM-200

65%

4 Trimble

Trimfleet APU

5 Panasonic

Toughbook CF30

6 Garmin

Forerunner 305

8 Vericom

VC4000PC

50%

95%

< 5m

< 8m

93.6%

90%

81.3%

< 10m < 15m 98.6%

100%

78.9% 90% 50%

59.7% 95%

100% 57.7%

91.2%

86.0% 99.3%

100%

95%

Table 6: Comparison of manufacturers’ stated location accuracy with measured accuracy.

108 Collision Magazine - Volume 8 Issue 1

www.collisionpublishing.com

100%

“”

rors smaller than the stated errors in all cases except one.

Again, this was the receiver that was not connected to an external antenna and was installed inside the vehicle adjacent to a metal cabinet.

Test 2 - Moving Vehicle: The paths for GPS receivers #1 through #7 were plotted with GPS Visualizer (http://www. gpsvisualizer.com/) and exported to Google Earth. Figure 20 shows the plotted path of the police vehicle as it was driven around a highway off-ramp. This route was driven twice, approximately 1 hr apart. Hence, two paths are shown for each receiver, except #4 and #5, which were not logging the raw data during one of these trips.

As the Figure shows, all but one of the receivers plotted a path close to each other and to the road. The other receiver (Garmin Nuvi) plotted a path that was not on the road and sporadically ‘jumped’ around. While the paths plotted by the majority of the receivers followed the road quite closely, the spread in the plotted paths between these receivers was sometimes wider than the full road width.

Based on the stationary and moving tests conducted, it was concluded that none of the receivers was sufficiently accurate to allow a determination of the lane in which the vehicle had been travelling.

As the Figure shows, all but one of the receivers plotted a path close to each other and to the road.

The overpass driven during our testing and shown in Figure 20 had been rebuilt in the spring of 2011 (less than one year prior to the subject testing in February 2012).

An aerial photograph of the previous overpass configuration (taken in 2009) was obtained from Google Earth and the test data were overlaid on this photograph (see Figure 21). The data from the Garmin Nuvi matched much more closely with the old highway and overpass, although the path was seen to ‘jump’ from one road to the next.

Figure 20: Overlap of receivers on the same vehicle being driven around an off-ramp twice. Most of the receivers followed a similar path, except for the Garmin Nuvi (blue), which jumped around. www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 109

Figure 21: The same GPS data in Figure 20 is shown overlaid on an earlier aerial photograph taken in 2009, prior to the overpass being rebuilt and prior to our testing.

This evidence suggests that the software in the Garmin Nuvi is designed to adjust the receiver’s location to match the nearest road, likely based on a road map stored within the device (which, in this case, was an outdated map).

While this would be beneficial for the intended use of a navigation unit, it may provide the reconstructionist with a location that appears more accurate than it is (provided the road has not been changed). These jumps to adjacent roads would also provide one explanation why the speed error in the Garmin Nuvi was quite large at times.

S

ummary and Conclusions 1. GPS receivers can record speed by one of two separate methods: Doppler shift (instantaneous speed measurement) and time from point A to B. The police GPS receivers and the Vericom devices tested all measured and recorded Doppler speeds. The Garmin vehicle navigation system and personal heart rate monitor recorded an average speed based on travel time between two points.

2. Doppler speed measurements were found to be very accurate for collision reconstruction, with errors typically well below 1 km/h. The average and maximum

110 Collision Magazine - Volume 8 Issue 1

speed errors for one receiver tested were 0.0004 km/h and 0.2 km/h, respectively. The average error was below the resolution for that receiver.

3. Physical obstructions, such as concrete highway overpasses, may cause transient spikes in the Doppler speed values.

4. The GPS receivers tested with external antennas provided more accurate speed and position measurements than the GPS receivers relying solely on internal antennas.

5. Some police laptops were equipped with software that was incorrectly programmed to convert the Doppler speed for storage in the local laptop (MWS). The speed values were recorded higher than the actual speeds by a factor of 1.609. Note that this is the conversion factor from mph to km/h.

6. 6. Location measurements were accurate to a few metres but were found to drift with time. None of the receivers tested was sufficiently accurate to allow a determination of the lane in which the vehicle was travelling.

www.collisionpublishing.com

7. All of the police GPS receivers tested relied on Versaterm’s Mobile Data Terminal (MDT) software to process data. This software, which is employed by numerous other emergency agencies across North America, is capable of storing two different GPS file types in the laptop. These files are deleted after a pre-programmed number of days. Investigators need to be aware of the time sensitive nature of these files.

8. The raw GPS information is very useful for identifying if any conversion errors occurred prior to the data being stored elsewhere in the laptop. As well, the raw NMEA sentences are recorded more frequently (at 1 Hz) than the rate at which the converted information is stored in the laptop. Hence, the investigator should always endeavour to obtain the raw MDTGPS.LOG file.

9. When the raw NMEA sentences are not available, a check can be performed of the speed conversion factor by comparing the average speed calculated based on the time to travel between two points with the average recorded Doppler speed during that interval. This will allow the investigator to establish if the recorded speeds were properly converted.

10. The measured speed and location accuracies were found to be better than the accuracies stated by the manufacturers in all but one extreme case, (where the GPS receiver was installed inside the vehicle adjacent to a steel cabinet and the receiver was not connected to an external antenna).

11. The Garmin Nuvi vehicle navigation system tested appears to adjust its location to fit an internal map. This can result in larger speed errors if the plotted path incorrectly jumps to an adjacent road.

R

eferences and Notes 1. PRIME Technical Services, “PRIME OffLine Search Results Legend”, Rev 1.2, 15 February 2011 2. Recordings at one, two, five or ten second intervals have been seen in different laptops. The setting for the frequency can be seen in the MDT.ini file.

3. Brewer, B., Sgt., “PRIME-BC A Canadian RMS Case Study”, Law and Order Magazine, April 2008 4. Chalko, T. J., “High accuracy speed measurement using GPS (Global Positioning System)” 2007, http:// nujournal.net/HighAccuracySpeed.pdf, accessed 17 March 2012.

5. Fuchs, M., “Handheld-GPS based Speed-Measurements”, 07. May 2007, available at http://www.gps-results.com/GPS_Speed.pdf, accessed 18 March 2012.

6. Witte, T.H., Wilson, A. M., “Accuracy of non-differential GPS for the determination of speed over Ground”, Journal of Biomechanics 37 (2004) 1891–1898 7. Butler, R.J., Breen, K.C., Fischer, W. R., Bedsworth, K. D., Haupt, N. R., “Using GPS Based Data Acquisition in Forensic Accident Reconstruction”, Collision, Volume 2, No 2, 2007.

8. http://firebug.sourceforge.net/gps_tests.htm 9. The spreadsheet provided by the University Of Wisconsin – Green Bay (at http://www.uwgb.edu/dutchs/ UsefulData/UTMConversions1.xls) was used to convert Latitude and Longitude values in degrees to UTM coordinates in metres.

10. Chalko, T., “Estimating Accuracy of GPS Doppler Speed Measurement using Speed Dilution of Precision (SDOP) Parameter”, 2009, http://nujournal. net/SDOP.pdf, last accessed 8 April 2012.

11. Reust, T., “Vehicle Navigation GPS Units Could Be the Overlooked EDR”, Collision, Volume 5, No 2, 2010.

12. Reust, T., “GPS Navigation Units: Recorded Data For Use in Accident Reconstruction”, Collision, Volume 6, No 1, 2011.

13. Bortolin, R., Hrycay, J., and Golden, J. “GPS Device Comparison for Accident Reconstruction”, SAE 2012-01-0997 14. Trimble GPS Tutorial, available at www.trimble.com/ gps/howgps.shtml, accessed 13 March 2012 15. This conversion error was also observed in a Panasonic Toughbook CF-30 laptop operated by the Delta Police Department and within Panasonic Toughbook laptops operated by the Vancouver Police Department.

16. The longitude resolution values will vary depending on the distance from the equator. These measurements were all taken near 49 degrees north latitude.

A

cknowledgments The authors wish to thank August Rinz (RCMP IT) for his technical assistance accessing and understanding the files stored in police laptops. We would also like to thank Matthew Murnaghan, P.Eng. (Cypress Solutions Inc.) for the use of two GPS modems and his patient advice regarding data logging.

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 111

y r a t men

com

mentum and angular momentum of the vehicles must be taken into account.

COLM and Newton’s Coefficient of Restitution, e, in Planar Collisions of Two Vehicles Raymond M. Brach, PhD, PE University of Notre Dame R. Matthew Brach, PhD, PE Brach Engineering, LLC The following presents comments in response to the article: Wilson, L. A., PE and S. H. Haight, “Dynamics of Vehicle-to-Vehicle Side Impact Crash Tests”, Collision, Fall 2012.

Despite ample science to the contrary [1 - 14], there appears to be continual serious misunderstanding and misuse of “COLM”, Conservation Of Linear Momentum, for analyzing and reconstructing vehicle crashes. Known limitations of this method are ignored. The Wilson and Haight article is one of the latest examples. This response is motivated by this and similar articles that have appeared in various publications over the last several decades and is intended to clarify the concept of COLM, its relationship to Newton’s Laws and its application (or misapplication) to impact mechanics and accident reconstruction.

Application of Newton’s second and third laws to shortduration contact of two bodies moving over a plane produces algebraic equations relating their linear and angular velocity changes to the contact impulse. This is referred to as planar impact mechanics. Appropriate manipulation of the planar impact equations can lead to the concept of Conservation of Linear Momentum in two, mutually perpendicular directions. Manipulation of the planar impact equations can also be used to similarly demonstrate conservation of angular momentum COAM. Thus, COLM is not a principle of mechanics but rather a special case, or result, of Newton’s laws.

•

•

The impulse and momentum equations that consider linear momentum alone are applicable only to the analysis of central collisions — that is, collisions between two point masses and to rigid body contact where the resultant impulse acts directly through the centers of gravity of the colliding objects. An example of the latter is a direct front-to-rear, zero-offset vehicle collision in a traffic lane.

To comprehensively and accurately apply Newton’s second law to the large majority of vehicle collisions, particularly offset/oblique collisions, both linear mo-

112 Collision Magazine - Volume 8 Issue 1

In other words, angular momentum must be included in any effective and accurate analysis or reconstruction of vehicle crashes — full planar impact mechanics must be used, not just COLM. In impact mechanics, a rigid body is an object that has mass distributed away from its mass center producing a rotational inertia, quantified as the moment of inertia. In a collision of two vehicles, the intervehicular crush surface is typically idealized as a flat vertical surface as indicated by the angle Γ in Figures 1 and 2. According to Newton’s third law, the forces and the corresponding impulses acting over this contact surface on each vehicle are equal in magnitude and opposite in direction. The resultant contact impulse typically is broken into two components, one normal (perpendicular) to the surface, Pn, and one tangential to the surface, Pt. The reason for choosing the specific components Pn and Pt is because the corresponding force components (that generate these impulse components) cause different physical effects, namely compression/rebound and shear/friction, respectively.

In formulating planar impact mechanics, two corresponding impact coefficients are defined to characterize the contact interactions of the vehicles in the normal and tangential directions. The coefficient of restitution, e, is used to characterize the amount of rebound of the vehicles normal to the contact surface. Another, independent, tangential impulse coefficient, μ = Pt / Pn [9], is used to characterize the effect of the tangential contact forces and impulse over the contact surface. Some formulations of planar impact mechanics [2,3] such as in the PC-Crash software use a tangential sliding velocity coefficient, s, in place of the impulse coefficient, μ.

With this foundation and clarification of planar impact mechanics and the two impact coefficients, several specific observations related to these topics and the article by Wilson and Haight can be made.

A. In a planar collision of two rigid bodies, Newton’s coefficient of restitution, e, is defined as the negative ratio of 1) the final, normal component of the separation velocity at the contact surface and 2) the initial, normal component of the approach velocity (commonly referred to as the closing velocity) at the contact surface. That is:

where vin, i = 1,2 is the initial normal velocity component of body i at the contact surface, and Vin, i = 1,2 is the final

www.collisionpublishing.com

normal velocity component of body i at the contact surface. Vin and vin are not the velocities or the velocity components of the vehicles’ mass centers. Not only does the restitution coefficient, e, characterize the rebound or separation velocity, but it can be shown to control the energy loss of the collision [1 - 14]. The tangential coefficient, μ, is also a parameter that controls energy loss [9,14]. The concept of “rebound factor” introduced in the Wilson and Haight article is given no rational justification, does not represent the restitution associated with a collision and is not shown in any way to be related to the collision’s energy loss.

B. The impulses and changes in velocity components at the contact surface of colliding rigid bodies (bodies of finite size with rotational inertia — not point masses) depend on (are functions of ) the mass center velocities, the angular velocities, the moments of inertia, the orientations of the bodies/vehicles at the time of impact and the orientation of the crush surface. The Wilson and Haight article does not take into account any crush surface geometry, vehicle orientations or vehicle dimensions.

C. The use of COLM (alone) is incapable of predicting the angular velocities of colliding bodies. Because the angular velocities enter into the calculation of both impact coefficients, COLM cannot be used to calculate the coefficient of restitution from experimental data of a non-central collision. Any coefficient of restitution computed using COLM from non-central collision data is meaningless and is unrelated to the true coefficient of restitution of the collision. In particular, negative values of the coefficient of restitution, such as presented in the Wilson and Haight paper, are unrealistic and physically impossible in vehicle collisions [9]. Moreover, an accurate calculation of the total kinetic energy loss in vehicle collisions must include the kinetic energy associated with the angular velocities. Since COLM neglects rotational effects, energy loss calculations for vehicle collisions (and any relation of this energy loss to crush energy) will be incorrect.

and that μ reaches its critical value, μ0 [9,14] (the final relative tangential velocity of the vehicles at the contact surface is zero). Because the common velocity occurs at the contact surface, results from COLM calculations are not capable of determining the conditions of the final velocity components of a collision.

Conclusions: A main conclusion drawn from the proper application of Newton’s laws is that the use of COLM in reconstructing the large majority of vehicle crashes, or for analyzing offset vehicle crashes (as done by Wilson and Haight), is completely inappropriate. The use of planar impact mechanics has been repeatedly validated for application to vehicle collisions; COLM has not. Development of planar impact mechanics has been available in published literature for decades [1 - 14] and has been implemented in commercial accident reconstruction software for many years. Practitioners in the field of crash reconstruction, particularly those in leadership and pedagogical roles and those in publishing, need to recognize the inadequacy of the inappropriate, but popular, use of “COLM”. They have a responsibility to stay up to date with technology and to lead the portion of the community who still use this method to the use of the appropriate (and readilyavailable) scientific and engineering tools.

Figure 1. COLM, point mass collision free body diagrams.

An Appendix on the attached data disk provides examples that illustrate why negative values of the coefficient of restitution are physically impossible in vehicle collisions.

D. As noted above, Newton’s third law applies to the equal and opposite resultant force and resultant impulse acting over the contact surface of the colliding vehicles. It bears no direct relation to final contact surface velocities (common velocities) as claimed by Wilson and Haight. Experimentally measured values of e and μ for vehicle-to-vehicle collisions [9,14] characterize the contact surface velocities.

The common velocity condition is that e = 0 (both vehicles have identical final normal velocity at the contact surface)

Figure 2. Planar Impact Mechanics free body diagrams.

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 113

References: 1. Goldsmith, W., Impact, Edward Arnold Publishers, Ltd., London, 1960.

2. Kudlich H., “Beitrag zur Mechanik des Kraftfahreug-Verkehrsunfalls” (Contribution to the Mechanics of Vehicle Traffic Accidents), Dissertation TU-Wien, 1966.

3. Slibar A., “Die mechanischen Grundsätze des Stoßvorganges freier und geführter Körper und ihre Anwendung auf den Stoßvorgang van Fahrzeugen” (The Mechanical Principles of the Impact Process of Free and Guided Bodies with Application to the Collision Process of Vehicles), Archiv für Unfallforschung, 2. Jg., H. 1, 1966, 31ff.

4. Brach, R. M., "Impact Analysis of Two-Vehicle Collisions", Paper 830468, SAE International Congress and Exposition, Detroit, MI, 1983.

5. Brach, R. M., "Analysis of Planar Vehicle Collisions using Equations of Impulse and Momentum", J Accd Anal & Prev, Vol 15, No 2, April 1983.

6. Brach, R. M., "Friction, Restitution and Energy Loss in Planar Collisions," Transactions ASME, J Appl Mech, Vol 51, No 1, March 1984, p 164-170.

7. Woolley, R., "The “IMPAC” Program for Collision Analysis," SAE Technical Paper 870046, 1987, doi:10.4271/870046.

8. Brach, R. M., "Restitution in Point Collisions", Chapter in: Computational Aspects of Impact and Penetration, edited by R. F. Kulak & L. Schwer, Elmepress International, Lausanne Switzerland, 1991.

9. Brach, R. M., Mechanical Impact Dynamics, John Wiley & Sons, New York, 1991, (also, Revised Edition, Brach Engineering, 2005, Granger, IN).

10. Ishikawa, H., “Impact Model for Accident Reconstruction — Normal and Tangential Restitution Coefficients,” Paper 930654, SAE International, Warrendale, PA, 1993.

11. Weaver, J. and R. M. Brach, "Limitations of Momentum Methods for Accident Reconstruction", Chapter in Forensic Accident Investigation, Michie Publishers, 1995 12. Stronge, W. J., Impact Mechanics, Cambridge University Press, 2000.

13. Brach, Raymond M. and R. Matthew Brach, “Analysis of Collisions, Conservation of Linear Momentum: Can we do Better?”, Collision Magazine, Vol 2, No 1, June , 2007.

14. Brach, Raymond M. and R. Matthew Brach, Vehicle Accident Analysis and Reconstruction Methods, 2nd Edition, SAE International, Warrendale, PA, 2011.

114 Collision Magazine - Volume 8 Issue 1

R

esponse to Comments by Dr. Raymond Brach and Dr. Matthew Brach Lawrence A. Wilson, P.E. and Sean H. Haight

This issue of Collision features commentary by Dr. Raymond M. Brach and Dr. R. Matthew Brach titled: “COLM and Newton’s Coefficient of Restitution, e, in Planar Collisions of Two Vehicles” (Brach comments).

The Brach comments address a previous work we presented in Collision Volume 7 Issue 2 titled: “Dynamics of Vehicle-to-Vehicle Side Impact Crash Tests.” The publishers of Collision have allowed us the opportunity to review the Brach comments and offer this response.

In short, we believe that a combination of misunderstanding and lack of clarity may be the cause of the disagreement. Regardless of the cause, we welcome the continued work of Dr. Brach and Dr. Brach to highlight this critical issue and we believe that in the end our understanding of the issue and intents are the same as theirs.

In their comments, Dr. Brach and Dr. Brach primarily take issue with two items in the original work, the first being the application of conservation of linear momentum (COLM) and the second being the concept of “rebound factor.”

Regarding the Misapplication of COLM Dr. Brach and Dr. Brach spend a considerable amount of time in their comments expressing concern over their perceived misapplication of COLM by the original authors. What the Brach comments appear to have overlooked is the clear language within the original work which specifically highlights the implausibility of negative coefficients of restitution: “Test 1 and Test 8 have negative values for coefficient of restitution … and therefore, COLM will not accurately predict the post-impact velocities of the vehicles.”

The original work was intended to compare the results of a series of vehicle-to-vehicle and MDB-to-vehicle crash tests. In the analysis of these tests, we identified two situations in which use of COLM would clearly be incorrect and decided to include a discussion of coefficient of restitution as an opportunity to highlight this fact. Specifically, the negative coefficients of restitution were presented and highlighted in red to draw attention to the very issue which seems to be one of the focal points of the Brach comments. In fact, the other seven tests reviewed in the original work did reasonably

www.collisionpublishing.com

approximate central collisions for which COLM provides a reasonable reconstruction.

We are in agreement with the Brach comments in that there continues to be a misapplication of COLM in the reconstruction community. While we welcome the extensive effort in their present comments to again clarify the limited situations to which COLM can apply, we believe that a misreading of our original work and/or a lack of clarity on our part has led to this misunderstanding. At the time of publication, we believed that the above quotation was enough for readers to understand that COLM would not be appropriate for some of the crash tests we analyzed in this paper.

Regarding the Rebound Factor Similar to the situation with the discussion of COLM, we believe that a combination of misunderstanding and lack of clarity may be contributing to the resulting objection in the Brach comments.

As previously stated, the intent of the original work was to compare the results of vehicle-to-vehicle and MDB-tovehicle crash tests. The discussion of “rebound factor” was presented to highlight the differences between the individual tests and the fact that the “coefficient of restitution” for a given collision is a result of the properties of each vehicle involved. As the paper clearly explains, the “rebound factor” is a means of identifying the contribution of each vehicle’s properties to the coefficient restitution. The paper in no way attempts to present the rebound factor as “the restitution associated with a collision” or “to be related to the collision’s energy loss” as suggested by the Brach comments.

This discussion highlights that in a given collision, each vehicle contributes to the resulting coefficient of restitution for the collision. For example, assume in one collision, the front of the bullet vehicle strikes the relatively stiff portion of the target vehicle (like the side of a vehicle near the firewall or front axle) and the resulting coefficient of restitution of this collision is calculated. If the same two vehicles were in another collision but the first vehicle (again the front of the vehicle) struck a relatively soft portion of the other vehicle (between the rear axle and the firewall), the calculated coefficient of restitution of this second example collision would most likely be lower than the coefficient of restitution of the first collision. When comparing the coefficients of restitution between the two collisions, the bullet vehicle will have substantially similar elastic properties and therefore, its contribution the coefficient of restitution will be similar. However, due to the difference in physical properties (stiffness and elasticity) of the impact points on

Com

men

the target vehicle, the struck vehicle could have appreciably different contributions to the coefficient of restitution in a given collision.

tar

y

The original discussion was not intended to re-write the laws of physics but suggest that the reader might do well to consider the specifics of a given collision. More specifically, the differences in physical properties (stiffness and elasticity) in side impacts between the axles and firewall when compared to the doors should be considered when conducting accident reconstructions.

Summary In conclusion, we would like to thank Dr. Brach and Dr. Brach for their continued contribution to the field of accident reconstruction and for their work expanding on the points introduced in our original work. We believe that a combination of misinterpretation and possibly a lack of clarity have resulted in the present situation but, in the end, it provides another opportunity to highlight an issue of serious concern in the accident reconstruction community which is the continued misapplication of COLM. We welcome any further discussion with Dr. Brach, Dr. Brach and the larger reconstruction community. We hope that their present comments and this response provide clarity to the original work and continued insight.

Editor's Note: The editors of Collision welcome commentary on previously published articles. Readers are encouraged to submit comments to the Collision publishing staff directly. Regarding this paper, the editors find no errors or misprints in the original article. There is no retraction of the original work presented by the authors. When the response from Wilson and Haight was presented, the Drs. Brach provided this final surrebuttal comment: "There is little or no benefit to present results from the inappropriate use of a method (COLM) to show that the method produces incorrect results."

www.collisionpublishing.com

Collision Magazine - Volume 8 Issue 1 115

Measure and Diagram Scenes and Vehicles from Photographs

PhotoModeler

®

· Quickly clear and measure scenes with your camera (ground-based or optionally aerial photography) · Measure crush using photos from your camera, or unknown cameras · Extract key evidence (distance, position, angles) from 3rd party photos · Model vehicles in 3D using a camera · Photo-based measurement for law enforcement, engineers, and reconstructionists since 1993 · Self-train with detailed video tutorials, attend public training held several times each year, or book a private course · Floating concurrent network licenses available for multiple users PhotoModeler and PhotoModeler Scanner share these common features: · Fully integrated support for digital and film cameras · Use 3rd party photos and unknown cameras with InverseCamera · On-photo dimension lines for basic diagrams · Export to other CAD, diagramming, or visualization packages (14 formats) · Sophisticated integration of control points from total stations or other sources · Map scenes over large distances using Project Merge · Extract photographic texture for realistic 3D models · Support for single photograph projects - great for forensics · Ortho-photo generation for that “birds-eye” view of your scene · Use the Surface Draw function to add 3D points, lines and curves traced on a single photograph

www.photomodeler.com

PhotoModeler: $1145

· Everything you need for scene diagramming, crush measurement, and more · Includes automation via Coded Target printing and recognition

PhotoModeler Scanner: $2595 · Adds textured-surface densepoint-cloud scanning

* Prices in USD and include free shipping by FedEx

®

Shoe print scan PhotoModeler Scanner

1-604-732-6658 toll free: 1-866-532-6658 email: sales@photomodeler.com

Expert AutoStats®

4N6XPRT BioMeknx™

Expert AutoStats® is a computer program containing over 42,000 vehicles cars, vans, utility vehicles, and light pickups. 1945-2012 model years are represented.

This program can be used alone or with our Expert Qwic Calcs® and Expert VIN DeCoder® programs. Also accessible from within REC-TEC.

(Lap top & D esktop W indow s 3 2 & 64 -bit O S C om p atible)

Enter:

M anufacturer and Year

Select:

M odel from pop-up window

Output:

Wheelbase Overall Length Height Wt. Distrib

Curb Weight Width Drive Wheels Front/Rear Track

M id-60's to present output also includes (when available): Front Overhang Rear Overhang Bumper Heights Hood Height Turning circle Wheel radius Bumper-to-Hood Bumper-to-Window

Expert AutoStats® is currently owned by over 700 private individuals and 300 Law Enforcement agencies in the U.S.A., Canada, and South America. Additional information, brochure, and order from can be downloaded from:

4N6XPRT BioMeknx is designed as a collection of the Biomechanical data of importance to the Accident Investigator into one easily accessible reference location. It is designed as a program to be useful to ALL accident investigators, not just the Biomechanical expert.

The program includes: 1. Over 65 printable illustrations and charts 2. 10 Separate categories or sections of information related to forensic biomechanics 3. Information on walking speeds of people from 17 months to > 65 years 4. Calculation modules for: - segment length from total body height, - segment weights and Centers of M ass (CoM ) based on total body weight, - Predicted M aximum Walking Speed based on total height 5. Information was obtained by review of over 65 scientific texts, several of which are now "out of print"

www.4n6xprt.com/4n6as.htm

www.4n6xprt.com/4n6bm.htm

PROGRAM ORDER FORM (Please Print) Contact Name: ________________________________________________ Company: ____________________________________________________ Street: ______________________________________________________ City:____________________ State:____ Zip:__ __ Country: __________ Phone: (_____) _______________ Fax: (_____) ___________________ E-M ail: ______________________________________________________ Expert AutoStats® 4N6XPRT BioM eknx™ Expert Qwic Calcs® 4N6XPRT Ped & Bike Calcs® 4N6XPRT StifCalcs® Expert VIN DeCoder® (pkg) Expert TireStuf® Order Processing

- $ 625.00 ea.

- $ 495.00 ea.

- $ 275.00 ea.

- $ 375.00 ea.

- $ 650.00 ea.

- $ 550.00 ea.

- $ 85.00 ea.

= ________ = ________ = ________ = ________ = ________ = ________ = ________

- Check = $5.00 / Credit Card = $10.00 / G ovt. P.O . = $15.00 = _______ Notarized Affidavit Requirement $25.00 per required Notarized Signature = _______

Q Deliver via electronic download (Email address Required) Q Ship on USB - $35.00 per disk per program California Sales Tax

- 8.50%

TOTAL (U.S. Dollars)

= 0.00 = ______ = ________

= ________

Enclosed is: Visa ___ MasterCard ___ Am. Ex. ___ Exp ____/____ - Check ___ - Gvt PO ___

Card Number: Name on Card: Signature:

_______________________________________________ _______________________________________________ ________________________________________________

Please M ake Checks*/Purchase Orders Payable to:

4N6XPRT Systems® 8387 University Ave - La M esa, CA 91942-9342 - U.S.A. FAX: (619) 464-2206 Phone: (619) 464-3478 Web: http://www.4n6xprt.com C hecks * M U ST * be draw n on a U.S. bank. O rders are shipp ed within 10 working days of receipt.

Prices subject to cha ng e without notice.

M ultip le orders and Package discounts availab le. Please call for details.

2014 ARC--CSI DATA DVD

C

2014 Vol.9 Iss.2

ollision C The International Compendium for Crash Research

Volume 9 Issue 2

FALL 2014

Collision: The International Compendium for Crash Research

Using Inequality Contraints in the Probability Analysis of

Motor Vehicle Accidents

Volume 9, Issue 2 - FALL 2014

Ignition Switch Data in CDR Reports

“Low Velocity” Testing collisionmagazine.com