Volume 11, Issue 2

2016 ARC- - CSI CONF.

ollision C The International Compendium for Crash Research

Volume 11 Issue 2

Motorcycle Crash Testing

Evaluation of Slam Stick for Data Acquisition Rollover Scene Evidence Low-speed, Rear-end Crash Analysis collisionmagazine.com Cover.indd 1

1/23/2017 2:11:42 PM

OVER 150 NEW VEHICLES

HVE Custom Vehicle Database The HVE Custom Vehicle Database is a hierarchical database of vehicles selectable according to vehicle type, make, model, year and body style. Every vehicle has been physically inspected; the source of each parameter is documented. Vehicle geometry has been measured using a Faro Arm or laser scanner.

The HVE Custom Vehicle Database contains several hundred vehicles; more than 25 new vehicles are added each year. Each vehicle includes the following properties: • Dimensions • Inertias • Stiffness Coefficients • Tires • Suspensions • Brake System • Steering System

• Drivetrain • ABS and Electronic Stability Control (if equipped) • Inter-vehicle Connections (if applicable) • Light System • Geometry (vehicle mesh)

Each vehicle has been validated using numerous handling and collision simulations.

HVE program results using these vehicles easily pass Federal standards for admissibility of scientific evidence.

Engineering Dynamics Corporation • 8625 SW Cascade Ave. Suite 200 • Beaverton, OR 97008 USA • 888.768.6216 HVE, EDSMAC, EDVSM, EDSVS, EDSMAC4, SIMON, DyMESH, HVE-CSI, and HVE DamageStudio www.edccorp.com

VISUAL REALITY

Cover.indd 2

are trademarks of Engineering Dynamics Corporation. All rights reserved.

7103-12/15

1/23/2017 2:11:42 PM

Volume 11 Issue 2.indd 1

1/23/2017 2:01:05 PM

Volume 11 Issue 2.indd 2

1/23/2017 2:01:05 PM

Contents

Volume 11 Issue 2

inside 6

Letter From The Editor

38

features 22

The Accelerations Present During The Trip Phase Of A Soil-Tripped Rollover Crash – An Update

by: Nathan A. Rose, Neal Carter, and Gray Beauchamp 30

My Turn At The Wheel: Principal Direction Of Force Real Or Abstract?

by: Erik Carlsson 38

50

Impact Speed Determination In A Head-On Crash Using Bayesian Networks

by: 2106 ARC-CSI Crash Team Boot Camp: Gary Davis, Andre Doria, Michael Morris, and Randy Eldridge 50

Evaluation Of The MIDE Slam Stick X As A LowCost Accelerometer And Data Acquisition System For Vehicle Crash Testing

by: David M. Hallman, Robert D. Anderson, Wesley Grimes, Michael DiTallo, Kenneth Salisbury and Kevin Vosburgh 64

Low-Speed Rear-End Crash Analysis MARC1 Use In Test Data Analysis And Crash Reconstruction

86

by: Rudy Limpert, Ph.D. 72

Piston-Type Bumper Isolator Compression Versus Delta-V In Bumper-To-Barrier And Bumper-tTo-Bumper Impacts

by: Robert D. Anderson, Michelle R. Hoffman Russell L. Anderson and Michael Rosenfield 86

The Longevity Of Scene Evidence From A Rollover - A Case Study

by: Nathan Rose and Neal Carter 96

Motorycycle Crash Testing: Advanced Boot Camp Was Born

by: Michael DiTallo, Thomas Green, Wes Grimes, Ken Salisbury, Brent Munyon, Tom Lawson, James Whelan, Eric Moody, Kevin Vosburgh, and Eric Paul

96

case study 8 110

Evaluating Crash Data From Late Model GM Vehicles by: W. R. Rusty Haight Case Problem Solution www.collisionpublishing.com

Volume 11 Issue 2.indd 3

Collision Magazine - Volume 11 Issue 2 3

1/23/2017 2:01:09 PM

Industry Partners IAARS

is one of the oldest associations dedicated to expanding the knowledge and training of automobile crash Reconstructionists. IAARS was started in 1980 as a resource for members to use when the inevitable questions arose while trying to reconstruct a crash. We have members from several different countries as well as many different states across the U.S. The annual seminars not only provide instruction on specific topics by experts in the field, they provide face-to-face discussions on how other people approach reconstructing a crash. The old saying "two heads are better than one" is definitely true in accident reconstruction.

Society of Accident Reconstructionists, and electronic delivery of bi-monthly editions of Accident Reconstruction Journal. Members also receive discounts at our training seminars.

Our association has "old-timers" that have worked thousands of accidents. If you have a question, chances are they have encountered a similar situation in the past and can give some insight on how they worked it out. We also have members very familiar with the ins and outs of new technology (Crash Data Retrieval, Photogrammetry) and are willing to help with any questions or concerns you may have.

Membership in IAARS has other advantages, including a discount (from IAARS) on a subscription to Collision Magazine; access to electronic delivery of the quarterly publication of the

IAARS enjoys working with all of the associations and organizations that are involved in helping get answers to how and why collisions occur and how to make our roads and highways safer.

Hope to see you at one of the conferences.

We have annual conferences with training opportunities across the country. IAARS was the first association to do actual crash testing in 1982 with T-Bone and Head-on collisions and we continue to try and have a dynamic full scale vehicle crash or testing at each of our seminars. As a member you have full access to the setup and test with all of its data. As a member you have access to a vast archive of videos and data from many different and sometimes unique tests. Some of these tests include vehicle to vehicle, truck to vehicle, motorcycle to vehicle and vehicle We have three levels of membership. Our highest level, Felto pedestrian crashes, semi rollovers, tip-over testing and brake low, includes a peer review of a reconstruction project. This efficiency testing. SCARS/IAARS 2017 conference will be held differs from many other associations in that with this process for membership it can help the member as peer review is a spe- from July 23 to July 28. IAARS members will receive a $100 rebate upon submission of a copy of their attendance certificate. cific area of judicial consideration when admitting an expert opinion in court. This review, along with training and experi- Non-members who attend the conference will have an opportuence, go a long way if you happen to come across a stickler for nity to join IAARS for $50; if accepted, dues for 2017 and 2018 a judge. We offer opportunities for a member to publish work will be waived for those new members. Check www.scarsrecon. org to see if updated information about that conference has been or research in the reconstruction field and an opportunity to posted. present materials at our annual conferences.

The ARC Network

Sincerely, Butch Thompson President, IAARS

www.iaars.org

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

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!

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.

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)!

The ARC Network offers various membership levels to fit your needs and budget. Membership starts at just $89.00 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 User's Summit and a free copy of Greg Russell's Excel Crush

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

www.accidentreconstruction.com

For information on becoming a Collision Industry Partner send an email - sbaker@collisionpublishing.com 4

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 4

www.collisionpublishing.com

1/23/2017 2:01:09 PM

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

Collision Publishing Inc PO Box 892885, Temecula, CA 92589 Toll Free: 800-280-7940 E-mail: sbaker@collisionpublishing.com Web: www.collisionmagazine.com ISSN: 1934-8681

COLLISION STAFF Scott Baker Sean Haight Tonya Baker Carla McLeod

Owner, Managing Editor Senior Editor Advertising Manager Subscription Services

All rights reserved, © 2017 Collision Publishing Inc. 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. Dis-

senting 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 e-mail to the editor at editor@collisionpublishing.com or sending a "regular mail" letter to us at PO Box 892885, Temecula, CA 92589. By

sending correspondence to either our e-mail 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 e-mail 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 re-emerge, the existence of this publication and data disk in that universe cannot be guaranteed.

ADVERTISER INDEX

PAGE

4N6XPRT Systems 120-121 ARC-CSI Crash Conference 63 ARC Network 109 Bertan W. Morrow 36 Collision Magazine 62 Collision Safety Institute 119 Crash Data Group 95 EDR Experts 21 EDR Summit 85 Engineering Dynamics Inside front FARO 1, 37 Laser Technology 20 MapScenes 2 MapScenes 45-48 Northwestern CPS 7 Sensefly 49 Vericom Computers 29 Visual Statement 82-83 Visual Statement Back Cover

Included with this issue of Collision

is the 2016 ARC-CSI Crash Conference data; including all the photos, video, instrumented crash testing data, scaled diagrams, speaker PowerPoints, and all other supporting data in an easy to use duallayer data DVD. Over 8GB of data!

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

BACK ISSUES Next ARC-CSI Crash Conference If you would like purchase back issues of Collision Magazine, please visit www.collisionmagazine.com or call us toll free 800-280-7940. www.collisionpublishing.com

Volume 11 Issue 2.indd 5

September 18-21, 2017 www.arccsi.com

Collision Magazine - Volume 11 Issue 2 5

1/23/2017 2:01:09 PM

Letter from the editor No, we didn't "skip a year." Observant readers of Collision Magazine will recognize that the timing of the CDR User's Summit (now the EDR Summit) and ARC-CSI Crash Conference have changed. I talked a little about that change in the last issue and now it's set.

The CDR User's Summit has been moved from January to March 6-8, 2017 and the ARC-CSI Crash Conference has been moved from May to September 18-21, 2017. There are other changes, too.

Since EDR technology is growing at such a rapid pace, we are re-branding the “CDR Summit” to the “EDR Summit”. Thus, we are expanding the topics to include additional tools used in the collection of EDR data.

The EDR Summit will deliver the next step in advanced Event Data Recorder (EDRs) technology for vehicle crash analysis and investigation. Most importantly, this summit brings together industry experts from around the world to present on timely topics and case studies. The presentations will focus on EDR data found in light trucks, passenger cars, SUVs, motorcycles, heavy commercial vehicles, autonomous vehicle safety and vehicle infotainment systems. EDR has become the one of the most reliable sources of evidence for road crash cases and the automotive EDR market is projected to more than double in the next five years.

The second change was timing. Given the potential for bad weather impacting travel (as it has in the past) and feedback from attendees regarding the timing of government budget years and other considerations, we made the decision to move the Summit to March.

upcoming 2017 ARC-CSI Crash Conference will be held at the South Point Hotel with far more to offer attendees than the Palace Station had in the past. More on that in a coming issue of Collision.

For now, we want to be able to focus on the exciting new expansion of the EDR Summit in terms of topics and speakers. Most of the speakers and topics for the 2017 EDR Summit have already been posted on line (www. edrsummit.com) and the lineup with CDR Tool data related topics, big truck EDR related topics and presentations on non-CDR Tool systems including motorcycles illustrates the new, expanded and improved focus.

Naturally, this means that the timing of Collision - traditionally following the two conferences - is shifted correspondingly. So, no, we didn't skip a year, we just shifted the timing of the release of the magazine to finalize all of the new contracts, get around holiday mailing and correspond more appropriately with the new and improved timing of the EDR Summit and ARC-CSI Crash Conference. More importantly, if you follow us on Facebook, we needed some addtional time to add over 1000 additional files to the 2016 ARC-CSI Crash Conference Data DVD included in this issue!

Managing Editor

Moving the Summit to March meant that, if we would be trying to get the ARC-CSI Crash Conference ready for May, we'd have a short window between the two so, as a result, the ARC-CSI Crash Conference will be offered in September. The crashes will still be held at the Las Vegas Motor Speedway; however, we decided it was time for a hotel upgrade. The

6

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 6

www.collisionpublishing.com

1/23/2017 2:01:09 PM

Take a Deeper Dive ADVANCED RECONSTRUCTION COURSES Northwestern’s advanced reconstruction courses offer an in-depth view of common reconstruction challenges. Work with our experienced faculty and reconstructionists from around the world. Receive ACTAR credit.

To register go to nucps.northwestern.edu.

CLASSES OFFERED INCLUDE: • CDR Data Analyst April 3 – 7, 2017 • Advanced Crash Reconstruction Utilizing Human Research Factors (40 ACTAR CEUs) May 15 – May 19, 2017 • Heavy Vehicle Crash Reconstruction (39 ACTAR CEUs) May 15 – May 19, 2017 • Traffic Crash Reconstruction Refresher (23 ACTAR CEUs) May 31 – June 2, 2017 • Motorcycle Crash Reconstruction (39 ACTAR CEUs) August 21- August 25, 2017

nucps.northwestern.edu 800-323-4011

Volume 11 Issue 2.indd 7

1/23/2017 2:01:09 PM

dy

tu e S

Cas

Evaluating Crash Data from late model GM vehicles w. R. Rusty Haight Collision Safety Institute

T

he case problems for this issue of Collision are composed of two parts, based on data sets from crashes involving two recent model GM vehicles. The theme for the case problem set is: “this is a reading intensive technology.” The following pages include select data tables and graphs from the larger reports and, as the reader can see, some pages and data tables one might normally find have been left out to keep the space in Collision manageable but still provide the reader sufficient information from which to evaluate and work the case problem as presented. Undeniably, other analysis might be undertaken; however, for the limited scope of these case problems, there is sufficient information provided.

We should start the analysis with the idea that to be able to effectively use CDR Tool retrieved data, one needs to first answer the question: is the retrieved data in this report from the crash under investigation? In other words, can you associate the data you have, as it is presented, to the crash you’re working on?

The Verano Case Problem The first case problem involves a set of data from a 2013 Buick Verano. For this case problem, you may assume that you’ve seen the vehicle and it has some

88 Collision Collision Magazine Magazine - Volume - Volume 11 Issue 10 Issue 2 2

Volume 11 Issue 2.indd 8

side damage, specifically damage to the right side of the vehicle just to the rear of the B pillar. You have examined the vehicle and find no airbags deployed; however, the driver and passenger pretensioners have deployed and when you turn the key “on” during your vehicle examination and imaging, you see the SIR warning lamp is “ON.” One can see that the data set for the Verano reflects a single event recovered and it is labeled (on the first page of the report) a “Non-Deployment.” In the “System Status at Event” data table, it reads: “Event Record Type = Non-Deployment.” However, we also see that there is a B0052 “DTC present at Time of Event” and we find this passage associated with that DTC data element in the Data Limitations: “…- For Deployment Events, DTC B0052 (Deployment commanded) shall be recorded with the remainder of the data for this event even though it occurred after Event Enable. …”

Then, in the Event Data (Record 1) data table we find the data element “Pretensioner Deployment, Time to Fire, Driver (Driver Pretensioner Time From Time Zero to Deployment Loop #1 or Loop #2 Com-

www.collisionpublishing.com www.collisionpublishing.com

1/23/2017 2:01:11 PM

mand Criteria Met) (msec) = 5ms” (and the same for the passenger pretensioner).

So, all this would seem to beg the question: 1. Is this data set, the crash record retrieved using the CDR Tool, really representative of a “Deployment” or a “Non-Deployment” Event? (You might imagine yourself on the stand having to explain this so a simple “yes” or”no” wouldn’t be sufficient here.)

Answering this question leads us toward an evaluation of whether or not the data is from the crash you’re working on.

The Corvette Case Problem The previous case problem (for the Buick Verano) in this issue sets the stage for this more complex analysis of a crash involving a 2014 Corvette from which the second data set presented here was retrieved. For this case problem, in addition to the CDR Tool retrieved data, we are provided with a video of the crash from a nearby home security camera (select screen captures are provided in this issue of Collision) as well as a rough diagram of the motion of the Corvette into the crash (prepared for Collision magazine by Tom Potter).

From an examination of the vehicle and scene, we find that damage to the right front wheel/tire of the Corvette, damage to its right rear wheel and tire, some damage to the rear of the vehicle at the rear bumper and damage to the left side of the car, at about the A pillar from contact with a light pole (as sen in the video). We find that there were no airbags deployed although the driver and front passenger seat belt pretensioners have been deployed.

From an evaluation of the scene and vehicle information available (setting apart from the CDR Tool report for the moment) together with a review of the security video footage, we would conclude that the Corvette was driving through a left hand curve in the road and began to yaw counter clockwise from its lane of travel across the oncoming lanes of travel toward the far curb. It struck the curb on the far side of the opposite lanes of travel first with its right front wheel/tire. Then, continuing to rotate counterclockwise, it struck an embankment beyond the curb with the back end of the car. The yaw largely stopped or at least slowed and the Corvette moved more laterally, right side lifting more than left, with the left side

presented until it struck a light pole with the left side of the car at about the A pillar. It came to rest at the base of the pole.

The CDR Tool report shows three recorded events, two Non-Deployments and a Deployment. They are recorded in three records (Event Record 1, 2 and 3).

While there may be a number of methods one might use to evaluate this data in terms of the order of steps in that evaluation, these questions are posed for this case problem not so much to guide the analysis in a specific order but ultimate to guide the solution as presented later in this issue of Collision.

1. What is the chronological order of the events recorded? Meaning, the data is presented as “Event Record 1 (then 2 and 3) but is that the chronological order in which they occurred?

2. In the data table “System Status At Time of Retrieval” we see these data elements: “Deployment Event Counter,” “Multi-Event, Number of Events (Event Counter)” and “OnStar Notification Event Counter.” How do these data elements relate to the events under investigation? Meaning, it would appear that there have been 4 lifetime events observed by the airbag control module (Multi-Event, Number of Events (Event Counter) = 4) although there are only three recorded. You should be ready to have to explain which events are recorded and, since only three are recorded, what happened to the 4th event?

3. Taken in light of the previous case problem involving the Verano, the data in Event Record 3 for the Corvette is described as a “Deployment;” however, like the data in the Verano case, the only devices commanded for a deployment appears to be the driver and passenger belt pretensioner. One should expect to have to address that apparent conflict or at least “difference” in terms of the naming convention. Like the Verano data set: is this data set, the crash record retrieved using the CDR Tool, really representative of a "Deployment" or a "Non-Deployment Event?”

Editor's Note: The following pages are excerpts from the related CDR reports. The case study solution can be found at the end of this issue.

www.collisionpublishing.com

Volume 11 Issue 2.indd 9

Collision Magazine - Volume 11 Issue 2 9

1/23/2017 2:01:12 PM

10

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 10

www.collisionpublishing.com

1/23/2017 2:01:14 PM

2013 Buick Verano www.collisionpublishing.com

Volume 11 Issue 2.indd 11

Collision Magazine - Volume 11 Issue 2 11

1/23/2017 2:01:15 PM

12

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 12

www.collisionpublishing.com

1/23/2017 2:01:16 PM

2014 Chevrolet Corvette www.collisionpublishing.com

Volume 11 Issue 2.indd 13

Collision Magazine - Volume 11 Issue 2 13

1/23/2017 2:01:17 PM

14

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 14

www.collisionpublishing.com

1/23/2017 2:01:18 PM

2014 Chevrolet Corvette www.collisionpublishing.com

Volume 11 Issue 2.indd 15

Collision Magazine - Volume 11 Issue 2 15

1/23/2017 2:01:20 PM

16

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 16

www.collisionpublishing.com

1/23/2017 2:01:21 PM

2014 Chevrolet Corvette www.collisionpublishing.com

Volume 11 Issue 2.indd 17

Collision Magazine - Volume 11 Issue 2 17

1/23/2017 2:01:22 PM

2014 Chevrolet Corvette 18

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 18

www.collisionpublishing.com

1/23/2017 2:01:22 PM

www.collisionpublishing.com

Volume 11 Issue 2.indd 19

Collision Magazine - Volume 11 Issue 2 19

1/23/2017 2:01:23 PM

NEXT-GEN MAPPING FOR CRASH SCENES TruPulse® 200x + QuickMap 3D® on Android + Collects and reports real-time data as easily as a smartphone app + Measures with high grade accuracy and is ideal for larger scenes + Outputs field data into CAD format that can be quickly sent by email

For more options or to schedule a demo, call: 1.877.696.2584 or visit: www.lasertech.com/Collision

Volume 11 Issue 2.indd 20

1/23/2017 2:01:23 PM

,

n

Volume 11 Issue 2.indd 21

1/23/2017 2:01:23 PM

ed w e i v e R Peer

The Accelerations Present during the Trip Phase of a Soil-Tripped Rollover Crash – An Update

I

Nathan A. Rose, Neal Carter, and Gray Beauchamp Kineticorp, LLC

ntroduction In Volume 11, Issue 1 of Collision, we examined longitudinal accelerations present during the trip phase of a fullscale rollover crash test reported by Asay and Wooley in 2010 (Test #2 in SAE Paper 2010- 01-0521). The trip is the phase during which lateral forces applied to the tires or wheels on the leading side of the vehicle cause the trailing side tires to lose contact with the ground. Eventually, these forces cause the center of gravity to roll past the leading side tires and the vehicle begins to roll over. Historically, analysis of the trip phase has focused on the lateral forces that generate the rollover (see Cooperrider, 1990 and 1998, for instance). However, the vehicle in Asay’s test experienced both longitudinal and lateral deceleration during the trip phase. The brakes were not applied on the test vehicles during these tests until well after the rollover had commenced, so the longitudinal accelerations had to have been generated from the tire interaction with the soil.

22

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 22

www.collisionpublishing.com

1/23/2017 2:01:24 PM

In our first article on this topic, we reported multiple sets of speed calculations using the principle of conservation of energy. Some of these calculations neglected the longitudinal deceleration and others incorporated them. These calculations demonstrated that, at least for this test, it was important to include the longitudinal deceleration during the trip phase in order to calculate an accurate vehicle speed at the beginning of the loss of control. Of course, a conservative speed estimate could be obtained without including the longitudinal deceleration.

In this particular test, which utilized a four-wheel drive 1991 Isuzu Rodeo, the test vehicle was shifted into neutral and towed up to and released from a speed of 73.5 mph (118.3 kph). After release, the vehicle was steered, using an automated steering controller, with a sharp left steering input of approximately ¼ turn that caused the vehicle to travel to the left across the roadway and to yaw counterclockwise. One second later, the vehicle was

steered back to the right until the steering stop for the vehicle was reached. This steering input was maintained for the remainder of the test. The vehicle then continued off the left side of the road into the dirt, but reversed its yaw direction, developing a significant slip angle as it yawed in a clockwise manner. The vehicle deposited furrows in the dirt and then began rolling over. The vehicle rolled 7 times over 181 feet (55.3 meters).

During a course on rollover crash reconstruction in September 2016, Wes Grimes pointed out to us that the longitudinal forces present during the trip phase in this test were likely due to the fact that the front tires of the test vehicle were steered significantly to the right during the trip.1 Because of the steer input, a lateral force applied to these front tires would apply both a lateral and longitudinal force to the vehicle. Figure 1 contains nine frames from the video of this test that depict the trip phase. The rightward steer of the front wheels is evident in these frames of video.

www.collisionpublishing.com

Volume 11 Issue 2.indd 23

Collision Magazine - Volume 11 Issue 2 23

1/23/2017 2:01:25 PM

Figure 1: Video Frames of the Trip Phase To test the hypothesis that the longitudinal forces applied to the Isuzu during the trip phase were due to the significant rightward steering input, we used PC-Crash crash simulation software. This software allowed us to easily incorporate the effects of steering on the speed loss during the trip phase and also to easily incorporate the threedimensional characteristics of the terrain. The PC-Crash user’s manuals describe the specific models used by the software when simulating rollovers and those models will not be discussed here. A number of prior studies have validated or used PC-Crash for modeling all three phases of a rollover crash, including Berg [2003], Steffan [2004], Viano and Parenteau [2004], Schubert [2006], Ootani and Pal [2007], Rose [2009], Andrews [2009], Kiefer [2011], Heinrichs [2013], and Richardson [2013].

The physical evidence diagram included in Figure 2 was used as a background image for the simulation. A three dimensional terrain was used for the simulation, which was based on Asay’s survey data. Manufacturer specifications 24

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 24

provided the vehicle dimensions and the vehicle weight reported by Asay was used (4,266 pounds). Moments of inertia were estimated based on formulas in MacInnis [1997] and Allen [2003]. PC-Crash’s TM-Easy tire model was used with the default parameters. Representative shape parameters for the vehicle body of the Isuzu were measured from a computer model and entered for the simulation.

Based on this physical evidence diagram, five vehicle positions were established for the final clockwise yaw during which the trip occurred. Figure 3 depicts these positions along with four vehicle positions for the prior counterclockwise yaw. During the test, speed and positional data had been collected for the Isuzu using a 100Hz VBOX III GPS sensor that was attached to the roof at the centerline. A non- contact radar speed sensor was also attached to the rear bumper and accelerometers and rotation rate sensors were attached near the center of gravity. To determine the actual speed at the beginning of the final clockwise yaw, the VBOX data was aligned with the positions

www.collisionpublishing.com

1/23/2017 2:01:25 PM

Figure 2: Evidence Diagram for Test #2 in SAE 2010-01-0521

Figure 3: Loss of Control and Trip Phase Positions Determined from the Scene Evidence in accordance with the alignment data reported by Asay.

This process resulted in a speed for the first position of the final yaw of 68.9 mph. Asay [2010] had reported a speed for the Rodeo at the end of the trip phase (Position 5) of 50.7 mph. However, Arndt [2011] has reevaluated Asay’s analysis and reported a revised speed at the end of the trip phase of 44.4 mph. Arndt’s corrected speed of 44.4 mph was used for the fifth position.

To generate the simulation in PC-Crash, the vehicle was placed at Position 1 and a speed of 68.9 mph was entered.

The yaw rotational rate and velocity direction at Position 1 were varied during the simulation process to optimize the match of the subsequent vehicle motion with the tire marks and furrows. During the simulation process, the results were judged based on how closely the calculated vehicle motion followed the tire marks and furrows of the final clockwise yaw that preceded the roll phase.

According to technical specifications for the Isuzu, the steering ratio was 18.7:1 and the minimum turning circle for the vehicle was 40.9 feet. For a rightward steering input, this implies a maximum input at the steering wheel of approximately 506 degrees. This would produce a rightward steer at the left front tire of approximately 27 degrees and a rightward steer at the right front tire of approximately 35 degrees. This steering input was entered into PC-Crash for the simulation. Brake factors of 1% were entered for

each tire of the Isuzu to represent rolling resistance. No additional braking was applied during the simulation. Friction zones were constructed for the trip phase in order to generate the known speed loss over the trip phase while also matching the tire marks and furrows and the known roll angle and roll angular velocity of the vehicle at the end of the trip. Figure 4 is a screen capture from the simulation that shows the friction zones that we utilized. This approach is similar to the one described by Grimes in his 2006 study, though that study utilized HVE rather than PC-Crash.

Figure 5 shows the vehicle motion in the optimized simulation. In our previous article, we reported based on the sensor data from the test that the roll angle of the Isuzu at the end of the trip phase was approximately 61 degrees, the roll angular velocity was approximately 400 degrees per second, and the average lateral and longitudinal accelerations over the final yaw were 1.09g and 0.47g. Both of these accelerations were oriented such that they would contribute to decelerating the vehicle. In our PC- Crash simulation, the speed of the vehicle at the end of the trip phase was 44.56 mph, close to the speed of 44.4 mph determined by Arndt. The roll angle at the end of trip was approximately 63.8 degrees and the roll velocity was 382 degrees per second, reasonably consistent with the values determined from the sensor data. In the simulation, the average lateral acceleration during the final clockwise yaw

www.collisionpublishing.com

Volume 11 Issue 2.indd 25

Collision Magazine - Volume 11 Issue 2 25

1/23/2017 2:01:25 PM

Figure 4: Friction Zones used in the PC-Crash Simulation

Figure 5: Vehicle Motion in PC-Crash Simulation rotation of the vehicle was 1.24g and the longitudinal acceleration was 0.36g, both of which acted to decelerate the vehicle.

After optimizing this simulation with the rightward steering input, we attempted to generate a simulation of the trip phase with the rightward steering input removed.

Without this steering input, we were not able to generate a simulation that generated a satisfactory and simultaneous

26

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 26

match with the tire marks, the furrows, the vehicle speeds, and the roll angle and roll angular velocity at the end of trip. One of our attempts is shown in Figure 6. In this simulation, the match with the tire marks and furrows was acceptable, though certainly not perfect. The speed loss during the trip, however, was significantly underestimated with the speed entering the roll phase being more than 11 mph too high. This indicates a significant drop in the overall deceleration level from the simulation that included the

www.collisionpublishing.com

1/23/2017 2:01:25 PM

Figure 6: Vehicle Motion in Simulation without Rightward Steering Input rightward steering input. The roll velocity at the end of the trip phase was also significantly underestimated at 230 degrees per second, rather than the actual 400 degrees per second.

The longitudinal accelerations in the simulation without steering were significantly less than in the previous simulation. The magnitude of the longitudinal accelerations in this simulation were consistent with the level that would be expected from rolling resistance and from traversing the slope of the roadway shoulder during the trip (less than 0.1g), but they were not anywhere near the level that actually existed during the test. Together, the simulation with steering and the one without steering confirmed that the longitudinal accelerations that we observed in the sensor data of the actual test were largely, though probably not completely, due to the significant rightward steering input.

D

iscussion and Conclusions While the match of the optimized simulation to the sensor data was not perfect, it was adequate to demonstrate that the longitudinal accelerations observable in the sensor data from this test were largely due to the significant rightward steering input. This is a helpful finding, because it indicates that there is not a need to develop any new models or formulas in order to incorporate these longitudinal forces into modeling of the trip phase. Accident reconstructionists already have a number of methods for incorporating the effects of steering on the deceleration rate of a yawing vehicle. Simulation is, of course, one of these [Rose, 2016]. Carter [2012]

discusses several other calculation techniques that would allow a reconstructionist to calculate the effects of steering on the deceleration rate of a yawing vehicle.

One difficulty, of course, would be knowing which direction, and the extent to which, a driver was steering during the trip phase. This is where simulation has a distinct advantage over the type of calculations one might carry out in a spreadsheet. In modeling the trip phase of the Rodeo test, the match with the physical evidence was significantly improved by including the rightward steering input. This is an indication that simulation can help detect the presence of a steering input during the trip phase. In cases where a vehicle begins yawing on asphalt before it enters the soil, striations in the tire marks on the asphalt may also be useful in determining which direction and the extent to which the driver was steering during the trip phase [Beauchamp, 2009 and 2016]. In some instances, data from an event data recorder might also give the reconstructionist knowledge of the steering inputs.

R

eferences 1. Allen, R., Klyde, D., Rosenthal, T., and Smith, D., "Estimation of Passenger Vehicle Inertial Properties and Their Effect on Stability and Handling," SAE Technical Paper 2003-010966, 2003, doi:10.4271/2003-01-0966.

2. Andrews, Stanley B., et al., “A Comparison of Computer Modeling to Actual Data and Video of a Staged Rollover Collision,” Enhanced Safety of Vehicles (ESV) Conference, 2009, Paper Number 09-0346.

www.collisionpublishing.com

Volume 11 Issue 2.indd 27

Collision Magazine - Volume 11 Issue 2 27

1/23/2017 2:01:26 PM

3. Arndt, Mark, “Drag Factors from Rollover Crash Testing for Crash Reconstructions,” Proceedings of the ASME 2011 International Mechanical Engineering Congress & Exposition IMECE2011 November 1117, 2011, Denver, Colorado, IMECE2011-65537.

4. Asay, A. and Woolley, R., “Rollover Testing of Sport Utility Vehicles (SUVs) on an Actual Highway,” SAE Technical Paper 2010-01-0521, 2010, doi:10.4271/2010-01-0521.

5. Beauchamp, Gray, Hessel, David, Rose, Nathan A., Fenton, Stephen J., “Determining Steering and Braking Levels from Yaw Mark Striations,” Paper Number 2009-01-0092, Society of Automotive Engineers, 2009, published in the SAE International Journal of Passenger Cars – Mechanical Systems 2(1):291-307.

6. Beauchamp, G., Thornton, D., Bortles, W., and Rose, Nathan A., “Tire Mark Striations: Sensitivity and Uncertainty Analysis,” SAE Int. J. Trans. Safety 4(1):121127, 2016, doi:10.4271/2016-01-1468.

7. Beauchamp, G., Pentecost, D., Koch, D., and Rose, Nathan A., “The Relationship Between Tire Mark Striations and Tire Forces,” SAE Int. J. Trans. Safety4(1):134-150, 2016, doi:10.4271/2016- 01-1479.

8. Berg, Alexander, et al, “Rollover Crashes – Real World Studies, Tests and Safety Systems,” Enhanced Safety of Vehicles (ESV) Conference, 2003.

9. Carter, Neal, Gray Beauchamp, Nathan A. Rose, “Comparison of Calculated Speeds for a Yawing and Braking Vehicle to Full-Scale Vehicle Tests,” Paper Number 2012-01-0620, Society of Automotive Engineers, 2012.

10. Cooperrider, N., Thomas, T., and Hammoud, S., “Testing and Analysis of Vehicle Rollover Behavior,” SAE Technical Paper 900366, 1990, doi:10.4271/900366.

11. Cooperrider, N., Hammoud, S., and Colwell, J., “Characteristics of Soil-Tripped Rollovers,” SAE Technical Paper 980022, 1998, doi:10.4271/980022.

12. Grimes, W., Balasa, J., Hunter, E., and Stevens, D., “Analyzing The Trip-Phase of Soft-Soil Rollovers,” SAE Technical Paper 2006-01-1558, 2006, doi:10.4271/2006-01-1558.

13. Heinrichs, B., Goulet, J., Fix, R., and King, M., “Measuring and Modeling Suspensions of Passenger Vehicles,” SAE Technical Paper 2013-01-0774, 2013, doi:10.4271/2013-01-0774.

14. Kiefer, Aaron, Bilek, David, Moser, Andreas, Webb, Andrew, “A Comparison Study between PC-Crash Simulation and Instrumented Handling Maneuvers,”

28

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 28

SAE Technical Paper Number 2011-01-1121, 2011, doi:10.4271/2011-01-1121.

15. MacInnis, Duane, Cliff, W., and Ising, K., "A Comparison of Moment of Inertia Estimation Techniques for Vehicle Dynamics Simulation," SAE Technical Paper 970951, 1997, doi:10.4271/970951.

16. Ootani, R. and Pal, C., “Effective Numerical Simulation Tool for Real-World Rollover Accidents by Combining PC-Crash and FEA,” SAE Technical Paper 2007-01-1773, 2007, doi:10.4271/2007-01-1773.

17. Ootani, R. and Pal, C., “Soil Trip Rollover Simulation and Occupant Kinematics in Real World Accident,” SAE Technical Paper 2007-01-3680, 2007, doi:10.4271/2007-01-3680.

18. Richardson, Shane, et al., “Simulation of Quad Bike (ATV) Rollovers Using PC-Crash to Evaluate Alternative Safety Systems,” Enhanced Safety of Vehicles (ESV) Conference, 2013, Paper Number 13-0286.

19. Rose, Nathan A., Beauchamp, Gray, “Analysis of a Dolly Rollover with PC-Crash,” SAE Technical Paper 2009-01-0822, 2009, doi:10.4271/2009-01-0822.

20. Rose, Nathan A., Carter, Neal, “The Accelerations Present during the Trip Phase of a Soil- Tripped Rollover Crash,” Collision: The International Compendium for Crash Research, Volume 11, Issue 1, 2016.

21. Rose, Nathan A., Neal Carter, Gray Beauchamp, “Post-Impact Dynamics for Vehicles with a High Yaw Velocity,” SAE Technical Paper Number 2016-011470, doi:10.4271/2016-10-1470.

22. Schubert, P., “Real-World Rollovers Reconstructed from Interviews and Measurements,” SAE Technical Paper 2006-01-0060, 2006, doi:10.4271/2006-010060.

23. Steffan, Hermann, “How to Use PC-Crash to Simulate Rollover Crashes,” SAE Technical Paper 2004-010341, 2004, doi:10.4271/2004-01-0341.

24. Viano, David C., Parenteau, Chantal S., “Rollover Crash Sensing and Safety Overview,” SAE Technical Paper Number 2004-01-0342, 2004, doi:10.4271/2004-01-0342.

Footnotes: 1. Grimes had previously reported techniques for using simulation to analyze the trip phase of soil tripped rollovers in his 2006 SAE paper titled “Analyzing the Trip-Phase of Soft-Soil Rollovers” (Paper Number 2006-01-1558).

www.collisionpublishing.com

1/23/2017 2:01:26 PM

www.collisionpublishing.com

Volume 11 Issue 2.indd 29

Collision Magazine - Volume 11 Issue 2 29

1/23/2017 2:01:26 PM

rn u T My

eel h W he T t a

Principal Direction of Force Real or Abstract?

Erik Carlsson

30

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 30

www.collisionpublishing.com

1/23/2017 2:01:26 PM

Excerpt from a transcript of a deposition: Attorney: What was the PDOF at the collision?

Deponent: It hadn’t been determined yet when the vehicles collided.

Attorney: What do you mean, hadn’t been determined yet?

Deponent: PDOF is a specific angle an accident reconstructionist assumes or determines when analyzing a traffic accident.

Attorney: Why would he assume an angle?

Deponent: To simplify the reconstruction of accidents where the colliding vehicles change direction during the impact, and the collision force therefore covers a range of angles instead of having a specific direction.

I

believe many accident reconstructionists agree with the explanation given by the deponent quoted above, though I doubt that many would be as straight forward as a defense expert witness was when asked at his deposition how he had established the PDOF he used in his computer simulation: You pick a number to start with, then change, and change, and change, until you get it right.

Naturally, by selecting a single direction of force rather than to use a range of directions, we greatly simplify the reconstruction of traffic accidents where the trajectories of the vehicles play an important role in the analysis. But if we presume that a single PDOF can truly enable an accident reconstructionist to determine an accident vehicle’s movements at a collision, can the same PDOF also be used to determine the movements of occupants inside the vehicle during the impact, if they are assumed to be unrestrained?

This article is about a fatal traffic accident case where a defense expert witness postulated that the PDOF he claimed applied to the collision force also applied to the inside of the accident vehicle, a heavy truck. He could therefore determine that the truck driver, who he claims was not wearing the seatbelt, ended up on the passenger side of the cab when the truck hit a concrete wall with the its right front corner. [The driver was burned to death when he jumped out of the truck after it was engulfed in flames following impact with a concrete road divider.] www.collisionpublishing.com

Volume 11 Issue 2.indd 31

Collision Magazine - Volume 11 Issue 2 31

1/23/2017 2:01:27 PM

T

he Accident: A truck pulling two loaded boxtype trailers was northbound in the center lane on New England Throughway, north of New York City. When a car driver changed from left lane to the center lane, the truck driver swerved to the right in an attempt to avoid a collision. The truck, having a short wheelbase, rapidly crossed the right lane and entered the breakdown lane. Before the driver managed to line up the truck with the road, the truck’s front end struck the road wall, a concrete barrier. [The highway passes over a railway at the accident site.] The speed of the traffic at the time was some 60 mph according to other northbound drivers, who had to stop and wait for the road to be cleared after the accident.

Following the impact with the concrete wall, the truck scraped along the wall for some distance, then crossed all three travel lanes and struck the left side concrete road divider. The truck next moved to the right, crossed the center lane, and came to rest in the right lane. The two trailers, of which the rear one had overturned, came to rest blocking two of the three travel lanes and the breakdown lane.

The passenger side fuel tank, on which the footsteps were mounted, was located below the passenger door and was torn apart at the impact with the concrete wall, causing diesel fuel to pour out. When the truck seconds later struck the road divider after having crossed the road, the driver side fuel tank was crushed. The fuel cascaded out of the tank and ignited, engulfing the truck in flames. The driver jumped out of the burning truck, but his cloth caught fire and he was burned to death while in clear view of vehicle occupants in the left lane behind the truck.

Photo 1 shows the burned-out truck and the two trailers at their points of rest. (The blue tarp seen close to the road divider covers the deceased driver.)

Photo 2: Still from a video recording by News Channel 7 in New York Photo 2 shows that the tire marks began to cross the road well behind the position of the fire truck. Seen in the photo are some irregular dark spots caused by diesel fuel that was flowing out from the damaged fuel tank on the passenger side.

Following are some photos derived from a video recording

by News Channel 7 in New York.

Photo 3: Still from a video recording by News Channel 7 in New York

Photo 1: Still from a video recording by News Channel 7 in New York 32

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 32

Photo 3 shows the marks on the concrete wall where it was struck by the truck (about a car length behind the white car stopped in the breakdown lane.) The scrape marks on the concrete wall extends all the way to the left edge of the photo.

www.collisionpublishing.com

1/23/2017 2:01:27 PM

40-degree crab angle at which the cab approached the barrier. This angle was sufficient for the dummy to clear the steering wheel at the test.

Photo 4: Scrape marks on the concrete wall Photo 4, copied from the Police Report, shows the scrape marks on the concrete wall. They extend well beyond the left edge in the photo. Laying on the ground is what appears to be the passenger side footsteps, torn away from the fuel tank brackets.

What made the defense expert conclude that the truck driver was not wearing his seat belt? The expert claimed that he could tell by looking at the seat belt hardware that it was not deformed, which he said it would have been, had the driver been wearing his seat belt.1 However, the energy-absorbing steering wheel was not deformed either, which it presumably would have been, had the driver been unrestrained. Not so, the defense expert claimed. The driver did not hit the steering wheel, but passed to its right, ending up on the passenger side of the cab.

A revealing test? To ‘prove’ that the defense expert’s opinion was correct, the defense team produced a video that showed the result of a full-scale impact test with a cab of the type used on the accident truck, occupied by an unrestrained dummy. The test was set up with the cab mounted in a fixed position on the test sled, with the cab facing 40 degrees to the left relative the direction in which the test sled was moving. This angle was the expert’s claimed PDOF, 40 degrees from the right when the accident truck hit the concrete wall. The test speed was 20 mph, which according to the defense expert was the truck’s Δ-V at the impact. [Speed and angle rounded to even numbers.] At the defendant’s cab test, the unrestrained test dummy continued straight forward when the test sled struck the barrier, passed to the right of the steering wheel and continued over the center console, ending up on the passenger side of the cab. This, of course, is also what could be expected since the unrestrained dummy was not seated straight behind the steering wheel, but off to its side due to the cab’s

A deliberately misleading test? I believe the testing staff and the defendant truck maker’s own engineers knew perfectly well that the cab test could not even closely be assumed to show what happened inside the truck cab at the accident. The accident driver was of course seated straight behind the steering wheel, not off to its side and traveling partially sideways, as the test dummy was at the cab test. The truck driver, if unrestrained, would initially have moved straight towards the steering wheel at the start of the impact. When the truck’s front end at the impact began to move left (relative its pre-impact direction) and slow down, the driver, if unrestrained, would have continued forward without change of direction and speed while the steering wheel in the meantime was moving to the left and slowing down. The net result would have been that the driver’s chest approached the steering wheel and moved in a curve to the right, relative the steering wheel, during the impact. It can therefore also be concluded that the driver, if unrestrained, could not have escaped hitting the steering wheel since the test dummy barely cleared the steering wheel at the cab test, where the dummy’s movement from the onset of the impact moved at a 40-degree angle to the right from the direction in which the test cab was facing at the test.

Even worse! There is another important fact about the vehicle dynamics in this accident that made the cab test even more unreal. The truck’s counterclockwise rotation at the impact with the concrete barrier could not have exceeded the angle at which the truck hit the barrier. This angle was about 10 degrees, as established by the investigating police officers from the tire marks. [Interestingly, this is also the approximate angle the defense expert showed in his animation of the accident!] To this fact comes yet another factor that was not accounted for in the defense expert’s analysis of the accident. The truck’s point of rotation when the truck rotated the 10 degrees at the impact was not its Cg (or the driver’s seat!), but the truck’s front axle! When the truck’s right front wheel hit the concrete wall, the movement in easterly direction of the truck’s front end on the northbound road ceased immediately. The truck’s rear end was then pushed against the concrete barrier by the two trailers, of which the front one at the moment of impact approached the barrier at a greater angle than that of the truck itself. Hence the truck’s front axle was the center point of the truck’s counterclockwise rotation, which ended when the truck’s outside rear tire hit the concrete wall.

The truck’s counterclockwise rotation with the front axle as the swiveling point means that the cab moved suddenly

www.collisionpublishing.com

Volume 11 Issue 2.indd 33

Collision Magazine - Volume 11 Issue 2 33

1/23/2017 2:01:27 PM

some distance to the right at the impact with the concrete wall. It follows that the driver, if unrestrained, moved some distance to the left relative the cab. This reduced the driver’s net movement to the right relative the cab during the impact, making the defendant’s cab/dummy test even more unreal as the test cab only moved in a straight line and stopped rapidly without any rotational or lateral acceleration of the test sled.

Computer simulation programs – tools or toys? While it is difficult to believe that the defense expert really thought he could tell that the driver of the accident truck was not wearing his seatbelt by merely looking at the seatbelt hardware on the burned-out truck, what came next in his findings is completely out of the blue. The expert ‘showed’ with his computer simulation that the truck and the front trailer jackknifed while crossing the road after the first impact. It was therefore the passenger side fuel tank that was crushed at the impact with the road divider, not the driver side, he claimed.

Picture 2: The jackknifed truck approaching the road divider

Following are six pictures copied from his computer animation:

Picture 3: Truck has turned around a full 180 degrees and is now parallel with, and sliding along, the road divider

Picture 1: The truck and the front trailer have started to jackknife Picture 1 intends to show that the truck and the front trailer are jackknifing after having struck the concrete wall seen behind the overturning rear trailer. The car presumably represents the one that caused the truck driver to swerve to the right.

Picture 3 intends to show that the truck has turned around a full 180 degrees and is now parallel with, and sliding along, the road divider. The shown angle between the truck and the trailer is completely unreal. Notice that the cab and hood widths, as depicted, are only a fraction of the widths shown in Picture 1.

Picture 4 intends to show that the truck has started to recover from the jackknifed position while still moving backwards and trapped between the road divider and the trailer, as shown in Picture 3. [The road divider did not end at the accident site, but continued along the highway.]

Picture 2 intends to show the jackknifed truck approaching the road divider in what appears to be a fully jackknifed position, with the left rear corner of the cab against the side if the trailer.

34

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 34

www.collisionpublishing.com

1/23/2017 2:01:27 PM

Picture 4: The truck has started to recover from the jackknifed position

Picture 5: The truck managed to move away from the road divider

Picture 6: The truck as having recovered fully from its jackknifed position, and is now able to move forward again.

A self-canceling jackknife? The defense expert’s claim that the truck and the front trailer jackknifed and that the truck then, after having hit the road divider with its passenger side, somehow could turn around while moving backwards, is completely unreal. But it is not only the truck's mysterious turn-around from its jackknifed position while moving backwards, as seen in Pictures 3 and 4, that is unreal. As seen in Picture 2, the truck's rear wheels would be the leading wheels when in the jackknifed position, and they would be pointing to the right, as seen facing north.

The truck and the front trailer, if jackknifed, would therefore have moved to the right, not left. Any suggestion by the expert that the truck’s rear wheels were locked during the jackknifing event and that the truck therefore did not move in the direction the rear wheels were pointing would have been contradicted not only by the tire marks seen in the air photos, but by his own claim in his report. He said in the report that he found at his examination of the wreck at a salvage yard that the brakes on the truck’s rear axle were out of adjustment, and that the brakes therefore where not working properly at the time of the accident. 2 Was there a purpose of the claim about the driver ending up on the passenger side of the cab? It may perhaps appear as the defense expert’s claimed conclusion about the driver not wearing the seatbelt was irrelevant. However, the defendant truck maker’s legal team probably thought their expert’s findings might be useful. What brought the case to court was the allegation by the legal team for the estate of the driver that the truck was unnecessarily dangerous due to the location of the two fuel tanks, one under each cab door, and with the fuel tanks carrying footsteps that protruded to the outside of the truck. The defendant’s legal team most likely planned to argue at trial that the driver himself caused the second and fatal impact by failing to wear the seatbelt, and that this is why he could not regain control of the truck and prevent the second impact that caused the fire that led to his demise.

Another puzzle is why the defense expert tried to make believe that the truck jackknifed while crossing the highway (and then straightened itself while moving backwards, trapped between the trailer and the concrete road divider!) His reason for that opinion is probably only that his computer said so, based on the unreal data he used in his computations.3 It should be noted that the blue tarp seen in photo 1, which covers the body, makes it obvious that the driver jumped out of the truck through the driver side door, as also testified to by the eyewitnesses.

The true reason the driver was unsuccessful in preventing the second impact. When the truck hit the concrete barrier, the protruding lug bolts on the right front wheel

www.collisionpublishing.com

Volume 11 Issue 2.indd 35

Collision Magazine - Volume 11 Issue 2 35

1/23/2017 2:01:27 PM

cut deeply into the concrete barrier. The resulting force resisting the truck’s forward movement acted rearwards on the lug bolts, and at such a lateral distance from the front axle’s right side king pin that the front wheel turned to the right with a force that sheared off the adjacent tie rod end. Thus the right front wheel became disengaged from the truck’s steering system at the impact with the concrete wall.

The force that caused the tie rod end to shear off was transmitted to the left side front wheel through the steering link tie rod. The force naturally increased rapidly at the impact until the tie rod end sheared off. The duration of the force therefore had to very short. Nevertheless, the high force in the tie rod forced also the left front wheel to turn right at the impact. How much it turned to the right before the tie rod end sheared off will probably never be determined.

However, as can be deduced from the scrape marks seen in Photos 3 and 4, there was a significant delay before the driver managed to get the truck and the front trailer away from the concrete barrier. The left front wheel may very well have been turned substantially to the right at the impact, delaying the driver’s ability to get the truck to move away from the concrete wall. With steering only on the left front wheel, the truck’s steering response was badly reduced after the first impact, which very likely explains why the driver could not prevent the second impact.

Volume 11 Issue 2.indd 36

Case Result. As could be expected, the case settled, and the defense expert never had to be confronted in court with the puzzling issue of how it would be possible for a jackknifed truck and trailer to straighten themselves while the truck was moving backwards and trapped between the trailer and the concrete road divider!

Anyone who has some questions about what I have described in this article is welcome to contact me. My e-mail address is erikcarlsson@live.com Footnotes: 1. The belt was found unbuckled after the accident. This, however, does not indicate that the driver was not wearing the belt when the accident happened as the belt, if used, had to be unbuckled before the driver could jump out of the burning truck. (Naturally, the belt webbing was consumed by the fire.)

2. Tow truck drivers must deactivate spring-operated parking brakes on trucks (as on the accident truck) that are to be towed away with the rear wheels on the road.

3. There was a substantial amount of completely imaginary input data in his computations.

1/23/2017 2:01:28 PM

www.collisionpublishing.com

Volume 11 Issue 2.indd 37

Collision Magazine - Volume 11 Issue 2 37

1/23/2017 2:01:28 PM

n o i t a n i m r e t e D g d n i e e s p u S h t s c a a r p C m n I O s d k a e r in a H ian Netwo s e y a B

ed w e i v e R Peer

38

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 38

www.collisionpublishing.com

1/23/2017 2:01:30 PM

A

Gary Davis, Andre Doria, Michael Morris, and Randy Eldridge

s in years past the 2016 ARC-CSI conference offered a Crash Test Boot Camp experience, where a select group participated in the conference crash testing. This article analyzes the head-on crash test between a 2011 Ford Crown Victoria and a 2016 Volkswagen that was assigned to the Boot Camp team (2016 ARC-CSI Crash Test #2). The target impact configuration was a collinear head-on impact with impact speeds for both vehicles at 48.5 kph (i.e. 97 kph closing speed). Both vehicles were remotely operated by Phantom Driver systems and instrumented with IST accelerometers, V-Box video/GPS units, and data collection systems to record crash pulses and impact speeds. Vehicle specifications and test instrumentation data results are summarized in Table 1. www.collisionpublishing.com

Volume 11 Issue 2.indd 39

Collision Magazine - Volume 11 Issue 2 39

1/23/2017 2:01:31 PM

Vehicle

Model Year Make/Model

Scale Mass [kg]

1 2

2011 2016

1896 1515

Ford Crown Victoria Volkswagen Passat

Impact Speed [kph] V-Box Video -45.9 55.5 57.0

Delta-V [kph] IST -43.8 -58.4

Table 1: Vehicle Specifications and Test Instrumentation Collision Observations

Figure 1: Passat video views immediately prior to impact. GPS-based speed and time are indicated in red text. The lower right small frame shows the Passat Phantom Driver system at same moment as the large frame.

Figure 3: Aerial image of vehicular impact orientation from drone video data. The Passat is blue, the Crown Victoria is white.

Figure 2: IST accelerometer data derived (a) longitudinal and (b) lateral delta-V for the Passat (blue curve) and Crown Victoria (black curve).

40

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 40

www.collisionpublishing.com

1/23/2017 2:01:32 PM

The objective for this test was to estimate the vehicle impact speeds using conventional crash reconstruction methods and event data recorder (EDR) data, and compare these estimates with the instrumentation observations.

However, the desired collinear collision configuration was not entirely achieved due to slight steering motions preimpact resulting in a partial overlap impact (see Figure 3).

Such offset alignment led to significant post-impact rotation (see Figure 2b) compromising analyses that assumed co-linearity of the centers of mass. Further challenges plagued the crash as the V-Box system on the Crown Vic suffered a data loss and no pre-crash or crash pulse data were obtained from the Passat EDR. The Passat results were expected as the team had confirmed with previous low-speed collision testing that the subject Passat CDR reports would not display pre-crash nor impulse data despite the EDR incrementing the event count after impacts. The Crown Vic’s EDR did provide information; the Airbag Control Module yielded a longitudinal crash pulse (-47.8 kph delta-V), but the PCM pre-crash data were lost. Owing to these multiple issues, the test objective was modified to address the following question: Given the limited EDR data, to what extent can these be leveraged with additional data, either from the scene or from the vehicles, to produce defensible estimates of impact speeds?

The first source of supplemental data was estimates of overall delta-V computed from the vehicles’ crush. Scale drawings of both vehicles were provided to the Boot Camp team and were used to measure the crush profiles (see Figure 4). Table 2 details the crush measurements for both vehicles and the estimated delta-Vs. Prasad’s (1990) reformulation of the CRASH3 algorithm was used to estimate the delta-Vs. The stiffness coefficients (d0 and d1) needed for Prasad’s reformulation were computed from the A and B coefficients provided by 4N6XPRT Systems using the standard conversions (e.g. Brach and Brach 2010, chapter 8). It is noteworthy that delta-Vs estimated from crush were lower than those reported by the instrumentation for both vehicles, consistent with findings reported by Hampton and Gabler (2010).

Crush Profiles [mm] C1 C2 C3 C4 C5 1 (Crown Vic) 732 701 518 305 152 2 (Passat) 488 518 457 305 122

Figure 4: Crown Victoria damage profile and crush measurement locations.

The second source of supplemental data was the collisionrelated vehicle directional changes. In the field the directions would be estimated from a scene evidence diagram, but scarcity of evidence in the test scene diagram made it difficult to identify the vehicles’ directions. To serve as a proxy for scene-based direction measurements, an overhead video (e.g. drone footage) of the collision was loaded into the program Videopoint, coordinates for pre and post-impact trajectories for the vehicles were determined using Videopoint’s point and click feature, and pre-and post-impact directions were then estimated using linear regression (see Figure 5). Taking the Passat’s pre-crash orientation as defining the positive x-direction equated to post-impact directions of 73o for the Crown Victoria and -126o for the Passat.

Planar impact mechanics (PIM) was used to predict the available crash data (Crown Vic delta-V, crush delta-V’s, and post-impact directions) from conjectured values of the impact speeds. Detailed presentations of PIM can be found elsewhere (Brach and Brach 2006) and only a brief description will be given here. The PIM model requires vehicle masses, inertial moments, the location of the crash impact center with respect to each vehicle’s center of mass, the orientation of the impact surface, and parameters related to normal and tangential restitution be provided. Given these quantities and pre-impact linear and angular veloci-

Vehicle

C6 L 0 1585 0 1524

Stiffness Coefficients d0 [N1/2] d1 [N1/2/m] 71.60 717.21 68.92 1048.66

Delta-V [kph] -40.1 -50.1

Table 2: Crush parameters and crush-based delta-V. Stiffness coefficients were provided by 4N6XPRT Systems (http://www.4n6xprt.com/).

www.collisionpublishing.com

Volume 11 Issue 2.indd 41

Collision Magazine - Volume 11 Issue 2 41

1/23/2017 2:01:32 PM

Figure 5: Pre and post-impact trajectories estimated using Videopoint. ties for the involved vehicles, the model uses conservation of momentum to predict post-impact velocities, energy loss, exit directions, and the planar delta-V components for each vehicle. Scale drawings of the vehicles’ crush profiles were overlaid to identify the approximate location of the impact center (d and φ) and the orientation of the crush surface (Γ). The locations of the vehicles’ centers of mass and their inertial properties were taken from Expert AutoStats (http://www.4n6xprt.com/). Table 3 summarizes the input data for the PIM model and compares the measured and predicted collision values. The predictions utilized impact speed for Crown Vic and Passat of 45.9 and 55.5 kph, respectively, which were based on the V-Box (Passat) and Video (Crown Vic) data (see Table 1). The crush-based delta-Vs were predicted using the regression equation from Hampton and Gabler’s (2010) Figure 1, with the delta-Vs predicted by PIM as inputs. The Table 3 results suggest the crash measurements (e.g. Crown Vic delta-V, crush-based delta-Vs, and post-impact directions) should be informative regarding the vehicles’ impact speeds.

42

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 42

One way to assess the contribution a given data item provides to a reconstruction is to quantify the reduction in uncertainty provided by that item. That is, beginning with a broad range of potential impact speeds, how effectively does the addition of a measurement (e.g. delta-V, exit angle, etc) collapse the conceived impact speed range toward the actual speeds. Bayesian networks (BN), a statistical inference model, provide a framework for organizing and applying expert knowledge in situations where uncertainty cannot be ignored. BNs have been used, for example, to reason medical diagnoses, DNA profiles, and highway infrastructure (Pourret et al. 2008). Over the past several years one of the Boot Camp participants has constructed BN versions of several commonly-used crash reconstruction techniques including, most recently, a PIM model for two-vehicle crashes (Davis and Mudgal 2016). Beginning with a prior probability distribution that describes uncertainty before taking data into account, the BN then uses Bayes theorem from probability theory to update this prior distribution, producing what Bayesians call a posterior dis-

www.collisionpublishing.com

1/23/2017 2:01:32 PM

Parameter Crown Vic (Vehicle 1) Mass [kg] 1896 2 Yaw Inertial Moment [kg-m-sec ] 411 d [m] 1.77 φ [degrees] 18 θ [degrees] 0 Restitution 0.1 Γ [degrees] 20 μ% 100 PIM Predicted (Observed/Crush Estimated) Quantities Longitudinal ΔV [kph] -47.3 (-47.8) Crush ΔV [kph] -41.4 (-40.1) Exit angle [degrees] 76.6 (73)

Passat (Vehicle 2) 1515 298 1.58 6 0

--51.9 (-50.1) -115.8 (-126)

Table 3: Summary of quantities used to initialize the PIM model. Impact speeds for the Crown Vic and Passat were 45.9 and 55.5 kph, respectively (See Table 1).

Observations/crush-based estimates are shown in parentheses. tribution. This posterior distribution summarizes the uncertainty remaining after accounting for the available data.

To implement the BN it is necessary to specify measurement errors. EDR delta-Vs were assumed to be affected by normally distributed measurement error with a standard deviation of 3.5 kph (Niehoff et al 2005) while the exit angle measurement error was taken as normally distributed with a standard deviation of 10o. The crush-based estimates of delta-V were assumed to follow the regression equation identified in Figure 1 of Hampton and Gabler (2010). Arguments supporting these choices can be found in Davis and Mudgal (2016). Prior to accounting for the measurements, the uncertainty regarding both vehicles’ impact speeds was modeled as a uniform probability distribution bounded by 5.5 and 109.7 kph (i.e.. 5 – 100 ft/sec). Posterior distributions for the impact speeds were then simulated using the Bayesian freeware WinBUGS (Lunn et al 2013; http://winbugs-development.mrc-bsu.cam. ac.uk/).

Table 4 summarizes the results of different data scenarios on uncertainty. The “No Data” scenario was essentially a simulation of the prior uncertainties regarding the impact speeds as uniform between 5.5 kph and 109.7 kph. The uncertainty reduction provided by a data configuration was evaluated by comparing its posterior standard deviations to the baseline “No Data” scenario results. As expected, the greatest reduction in uncertainty occurred when all three data items (exit angles, EDR delta-Vs, and crush del-

ta-Vs) were incorporated; the posterior standard deviations were reduced to 3.2 and 3.8 kph for the Crown Vic and the Passat, respectively. Almost as good was the uncertainty reduction provided by the single EDR delta-V from the Crown Vic’s ACM plus the exit angles. The EDR delta-V from the Crown Vic alone provided only marginal reduction in uncertainty, as did combining the delta-V from the EDR with those from the crush analysis. Understandably, using the EDR reported delta-V and crush-based deltaV conjointly effectively provided no new information to train the BN model due to those data being relatively similar. Note that the 95% confidence bounds tend to catch the impact speeds observed from test instrumentation, although in some cases the bounds are so wide as to be uninformative. Figure 6 illustrates the prior distribution for both vehicles and posterior distributions generated when all available data were considered. Compared to the flat prior probability distribution, the posterior distributions for both impact speeds are highly, but not perfectly, informative.

C

onclusion In this example delta-V data alone were not sufficient to reliably estimate the impact speeds. This is not surprising since in collinear head-on collisions delta-Vs can give estimates of closing speed but not, without supplemental information, individual impact speeds (Haight 2011). The delta-V from the EDR was more informative than those estimated via crush, largely because current evidence indicates that EDR delta-Vs have lower

www.collisionpublishing.com

Volume 11 Issue 2.indd 43

Collision Magazine - Volume 11 Issue 2 43

1/23/2017 2:01:32 PM

Data Scenario

Vehicle

EDR delta-V, crush delta-Vs, Crown Vic exit angles Passat EDR delta-V, exit angles Crown Vic Passat Crush delta-Vs, exit angles Crown Vic Passat EDR delta-V only Crown Vic Passat

BN-PIM Model Posterior Summary Mean St. Dev. 2.5%-ile 97.5%-ile 46.7 3.2 40.6 53.2 56.1 3.8 48.8 63.7 47.3 3.6 40.4 54.2 56.7 4.2 48.5 65.0 45.7 6.7 32.9 59.1 54.8 8.0 39.4 70.9 51.8 27.3 7.8 99.4 52.2 27.3 7.8 99.4

Observed 45.9 57.0 (55.5)1 45.9 57.0 (55.5)1 45.9 57.0 (55.5)1 45.9 57.0 (55.5)1

EDR delta-V, crush delta-Vs

Crown Vic 51.8 27.0 7.5 97.8 45.9 Passat 51.1 27.0 7.7 98.1 57.0 (55.5)1 No Data Crown Vic 57.6 30.1 8.0 107.2 45.9 Passat 57.8 29.9 8.0 107.2 57.0 (55.5)1 1 Parenthesized values in the “Observed” column are from the Passat V-Box system; Nonparenthesized values are from the vehicles’ video system (See Table 1).

Table 4: Summary of posterior distributions for impact speeds and observed impact speeds. All speeds are in kilometers per hour. measurement error variances than crush-based delta-Vs. As noted earlier, when estimating the impact speeds the greatest reduction in uncertainty resulted when all available data were incorporated, highlighting the value of methods that support data fusion.

Figure 6. Prior and posterior probability distributions for impact speeds.

R

eferences 1. Brach, R., and Brach, M. Analysis of collisions: Point mass mechanics and planar impact mechanics, Collision, volume 2, Issue 1, 2006, pp. 38-50.

44

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 44

2. Brach, R., and Brach, M. Vehicle Accident Analysis and Reconstruction Methods, 2nd edition, SAE International, 2010.

3. Davis, G., and Mudgal, A. Bayesian uncertainty quantification in planar impact crashes via Markov Chain Monte Carlo simulation, SAE Technical Paper 201601-1481, SAE International, 2016.

4. Haight, W., Using CDR system data in crash reconstruction, or what does the term “complete reconstruction” really mean?, Collision, volume 6, Issue 1, 2011, 82-94.

5. Hampton, C., and Gabler, C. Evaluation of the accuracy of NASS/CDS delta-V estimates from the enhanced WINSMASH algorithm, Annals of Advances in Automotive Medicine, 54, 2010, 241-252.

6. Lunn, D., Jackson, C., Best, N., Thomas, A., and Spiegelhalter, D. The BUGS Book: A Practical Introduction to Bayesian Analysis, CRC Press, 2013.

7. Niehoff, P., Gabler, H., Brophy, Chidester, C., Hinch, J., and Ragland, C. Evaluation of event data recorders in full system crash tests, Enhanced Safety of Vehicles paper 05-0271, 2005.

8. Pourret, O., Naim, P., and Marcot, B. Bayesian Networks: A Practical Guide to Applications, Wiley, 2008.

9. Prasad, A. CRASH3 damage algorithm reformulation for front and rear collisions, SAE Technical Paper 900098, SAE International, 1990.

www.collisionpublishing.com

1/23/2017 2:01:32 PM

www.collisionpublishing.com

Volume 11 Issue 2.indd 45

Collision Magazine - Volume 11 Issue 2 45

1/23/2017 2:01:33 PM

46

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 46

www.collisionpublishing.com

1/23/2017 2:01:34 PM

www.collisionpublishing.com

Volume 11 Issue 2.indd 47

Collision Magazine - Volume 11 Issue 2 47

1/23/2017 2:01:34 PM

48

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 48

www.collisionpublishing.com

1/23/2017 2:01:36 PM

Swiss Made

Reconstruct your crash scene in just one flight

Capture 3D scene data quickly & safely Produce accurate, visually appealing crash simulations Quick to transport & deploy Compatible with maps, CAD drawings & more Backed by global support

Schedule a demo today: senseFly.com/WhereToBuy

Join us on: www.collisionpublishing.com

Volume 11 Issue 2.indd 49

Collision Magazine - Volume 11 Issue 2 49

1/23/2017 2:01:36 PM

ed w e i v e R Peer

Evaluation of the MIDE Slam Stick X as a Low-Cost Accelerometer and Data Acquisition System for Vehicle Crash Testing.

David M. Hallman, Robert D. Anderson, Wesley Grimes, Michael DiTallo, Kenneth Salisbury and Kevin Vosburgh

50

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 50

www.collisionpublishing.com

1/23/2017 2:01:37 PM

bstract The purpose of this study was to evaluate the viability of the Slam Stick X (Figure 1), manufactured by MIDÉ as a self-contained data collection system and accelerometer for crash testing. Two Slam Sticks were purchased and installed in crash tested vehicles alongside the “typical” higher cost equipment used by the 2015 Southwestern Association of Technical Accident Investigators (SATAI) and 2016 ARC-CSI crash teams. The collected acceleration data sets from both systems were processed as usual and the filtered acceleration data and Delta V data from the various instrumentation was compared to the Slam Stick data.

A

The initial testing and data comparisons from the 2015 SATAI conference revealed significant differences between the collected and processed data from the Slam Stick and the reference equipment. The Slam Stick acceleration values were significantly lower than data collected utilizing the reference equipment. The Slam Stick Delta V graphs were also different than the normal or expected Delta V graphs. This difference was found to be an artifact of the type of accelerometers utilized in the Slam Stick.

The stock Slam Stick contains a piezo-electric accelerometer while the typical systems for vehicle crash testing use piezo-resistive or MEMS (Micromachined ElectroMechanical System) accelerometers. The important difference between the two is that piezo-electric accel-

erometers experience low frequency roll-off and piezoresistive accelerometers do not. The stock Slam Sticks were upgraded with a capacitive MEMS accelerometer

by MIDÉ and retested alongside IST 3C and IST 3D data collection equipment at the 2016 ARC-CSI Crash Conference. While there are still some challenges with the Slam Stick’s use, good agreement was found between the two data sets. ntroduction Crash testing and collecting acceleration data from crash tests is necessary to have a good understanding of the forces generated during vehicle impacts. This in turn provides crash reconstructionists and engineers with empirical correlations between accelerations, Delta V, PDOF and occupant injuries. High frequency data acquisition equipment that is reliable and robust is required to collect accurate acceleration data. The cost of the typical data acquisition equipment can be a significant barrier to entry for those who would benefit from being able to conduct crash testing and data collection for training and research purposes. Traditional data acquisition systems can cost $1,500 to $2,500 per channel with an initial expense of $7,500 to $15,000 depending on the system. This significant expense for equipment that will likely see limited use exceeds the resources of many individuals, companies and public agencies. The expense of the typical system led to the search for a viable, lower cost acceleration data measurement and collection option.

I

Figure 1: Slam Stick X with aluminum and plastic enclosures

www.collisionpublishing.com

Volume 11 Issue 2.indd 51

Collision Magazine - Volume 11 Issue 2 51

1/23/2017 2:01:38 PM

The search for a low cost triaxial accelerometer and standalone data collection combination for crash testing resulted in the purchase two stock Slam Stick X Shock & Vibration Data Loggers from MIDÉ via their website.

The stock Slam Stick has 16-bit resolution for sample rates up to 8.5 kHz and 14-bit resolution up to the maximum sample rate of 20 kHz. The stock Slam Stick X combines a 3-axis piezo-electric accelerometer and data acquisition system in one self-contained unit. It is also equipped with temperature and pressure transducers. It can be equipped to measure accelerations up to 2000 g. The battery can sustain up to 15 hours of run time and up to 8 GB of memory can store 4 billion data samples. The cost per channel of acceleration in 2015 was about $550 for a total initial expense of less than $1,700 per unit. Based upon published specifications it initially appeared to be a viable option for low cost data collection for crash testing when compared to the traditional data acquisition equipment.

The Slam Sticks were tested side by side with reference equipment in multiple staged crashes at the 2015 SATAI Fall Conference in Glendale Arizona. Data was collected utilizing the Slam Stick and the traditional equipment used by Biomechanical Research and Testing (BRT) out of California. The collected data was processed and a comparison between the two data sets was made. The acceleration data from the Slam Stick was found to be significantly different from the reference collected data. This led to a conversation with representatives of MIDÉ and the crash team’s realization that the stock Slam Stick X utilizes a piezo-electric accelerometer rather than the piezo-resistive or MEMS accelerometers used in the traditional automotive crash data collection.

After a technical discussion with MIDÉ, the previously purchased Slam Stick X’s were returned to MIDÉ and modified by adding a triaxial capacitive MEMS accelerometer. These non-production or upgraded Slam Sticks were tested side by side with Instrumented Sensor Technology (IST) 3C / 3D units in multiple staged motorcycle crashes during the 2016 ARC-CSI Crash Conference crashes in Las Vegas, Nevada. IST 3D and Slam Stick data were collected during all of the motorcycle crash tests conducted including motorcycle into barrier testing and motorcycle into car testing. The involved cars were also instrumented utilizing upgraded Slam Sticks and an IST 3C in those crashes where a car was involved. The collected acceleration data from both systems was again processed and compared. In the analysis of the ARC-CSI collected acceleration data, excellent agreement was found between the IST and Slam Stick acceleration and Delta V data sets. ethods and Materials Two stock ±100G Slam Stick Xs with the aluminum enclosure option were purchased from MIDÉ through their website for a total cost of $3,310.00 in 2015 and tested at the 2015 SATAI fall crash conference. The Slam Sticks in this example crash (Figure 2) from that testing were mounted to the trunk lid of the Honda with double sided foam tape. In this test, two Slam Sticks were mounted at right angles so that the X-axis of one Slam Stick and the Y-axis of the other Slam Stick were aligned with the vehicle’s longitudinal axis. The BRT accelerometers were mounted at or near the vehicle center of gravity.

M

Figure 2: 2015 SATAI Crash Test 52

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 52

www.collisionpublishing.com

1/23/2017 2:01:39 PM

Figure 3: DTS Slice Micro and Nano, and custom Slice Rack

Figure 4: TDAS Pro Signal Input Module (SIM), and TDAS Pro 4 and 8 SIM Racks Acceleration data was collected with the Slam Sticks at 10 kHz and processed and compared to the data from Diversified Technical Systems (DTS) Slice in a custom rack, as shown in Figure 3, and DTS Tiny Data Acquisition System (TDAS) Pro in a 4 module rack, as shown in Figure 4.

These DTS systems are currently used by Biomechanical Research and Testing (BRT) and Biomechanics Analysis (BA) for crash data collection. Both of these DTS systems use external piezo-resistive accelerometers arranged in a triaxial configuration on the sides of a metal cube, or inside the walls of a rectangular enclosure mounted to a base plate.

eration data was collected from motorcycle-to-car (Figure 5), and motorcycle-to-barrier tests. Data in the cars was collected using a triaxial IST 3C (Figure 6) and one or two upgraded Slam Sticks with custom ± 200 G capacitive MEMS accelerometers which sampled at 3.2 kHz. Data was collected from the motorcycles using a triaxial IST 3D, one or two upgraded Slam Sticks and GPS speed data was collected with a VBOX sport. The IST 3C / 3D used in this study is a stand-alone 10 bit piezo-resistive triaxial accelerometer which samples at 3.2 kHz per channel available in a ±2G to ±500G range for at an approximate cost of $8,500.00. The IST 3C / 3D units used in this testing were ±50G.

In motorcycle crash tests conducted before, during and after the 2016 ARC-CSI Crash Conference crash day, accelwww.collisionpublishing.com

Volume 11 Issue 2.indd 53

Collision Magazine - Volume 11 Issue 2 53

1/23/2017 2:01:41 PM

Figure 5: ARC-CSI crash test, Slam Sticks / ISTs on Cycle and in car.

Figure 6: IST EDR-3

The Slam Sticks in both the cars and on the motorcycles were attached with double-sided foam tape directly to the top of the IST (Figure 7). Given their low mass of only 2.3 oz for the aluminum enclosure, the tape was adequate to affix the Slam Sticks for all crashes. The two Slam Sticks, were mounted adjacent to each other so that they were roughly aligned the X-axis of the motorcycle or car.

complete instrument package was bolted to the motorcycle frame aft of the fuel tank (Figure 7). This method of mounting was usually sufficient; however, on some of the tests the self-tapping mounting bolts pulled out and the instruments came loose during the collision.

esults The accelerations and the processed Delta V data from the Slam Stick collected at the 2015 SATAI conference were found to be significantly different from the data collected by the described reference equipment.

R

Figure 7: Instrumentation mounted to the motorcycle Ideally, the accelerometers should be mounted so that their x-axis is level and coincides with the vehicle’s longitudinal axis. However, due to the geometry of the motorcycles, which generally lack flat horizontal surfaces to mount instrumentation on, the motorcycle instruments were not always earth level. The instrumentation’s angle relative to horizontal was documented on each motorcycle for later data analysis. The motorcycle Slam Sticks and IST’s were mounted on the same plane so that one-to-one comparisons could be made regardless of the orientation of the combined instrumentation package. Typically, the 54

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 54

The IST’s were set to trigger at 3 G, to record pre-crash data, and to record enough post-crash data to potentially capture the acceleration of the motorcycle tipping and hitting the ground post impact. Data collection for the Slam Sticks was activated prior to the crash test and data was collected continuously until the Slam Sticks were shut off after the crash. The collected data from the Slam Sticks was then filtered to 90 Hz to match the built-in mechanical filter of the IST units. The filtered Slam Stick acceleration data was then compared to the IST data. The crash pulse acceleration data from both systems was integrated so that the Delta V could also be compared.

As seen in the graph in Figure 8 from the Honda / Dodge Truck crash, the Slam Stick acceleration data seems to correlate well at first, but the magnitude of the acceleration reported by the Slam Stick seems to drop off, and then over-shoot the BRT accelerometer in the positive direction as the vehicle acceleration drops off. This causes the corresponding Delta V to return back towards zero. At the greatest reported Delta V value, when compared to the data collected by BRT, the Slam Stick under-reported the actual Delta V by approximately 7 mph.

www.collisionpublishing.com

1/23/2017 2:01:43 PM

Figure 8: Honda Prelude Data from SATAI 2015 The filtered accelerations and the processed Delta V data from the crash data collected at the 2016 ARC-CSI crash conference, using the upgraded Slam Sticks, were found to be in good agreement with the data collected from the IST 3C / 3D systems. As shown in Figure 9, the filtered X-axis accelerations from the 2010 Harley Fat-Bob overlay very well and the Delta V data match relatively closely from both systems. The processed Slam Stick data shows a slightly increased Delta V over the IST data. Shifting the acceleration data so that the pre-impact acceleration data reads zero may further improve the agreement.

Data from the Slam Sticks and IST 3C mounted in the cars was also processed and compared (Figure 10). The motorcycle in this crash was travelling at 30 to 40 miles per hour when it struck the stationary instrumented vehicle.

Figure 10 shows the Y-axis acceleration data from a car which was struck on the passenger side by a motorcycle.

The maximum acceleration for the vehicle in this crash is relatively low with indicated maximums of approximately

5 G. As seen in the figure, there is relatively good agreement in the acceleration pulse and the resulting Delta V values between the two systems. This indicates that the Slam Stick could also potentially be a viable option for data collection in low speed crash tests. More testing is necessary to further evaluate this device for low speed testing.

The maximum Slam Stick recorded acceleration in the motorcycle to barrier crashes conducted was 95 G with a 38 mile per hour Delta V. One of the issues discovered after the completion of the motorcycle barrier testing was clipping of the acceleration data collected by the IST 3D mounted on the motorcycles. The IST contained a ±50 G accelerometer which clipped the data at approximately 57 G as indicated in Figure 11. By comparison, the Slam Stick contained a 200 G accelerometer and therefore that crash pulse is not clipped. In several of the motorcycle barrier impacts, accelerations in the X direction were found

www.collisionpublishing.com

Volume 11 Issue 2.indd 55

Collision Magazine - Volume 11 Issue 2 55

1/23/2017 2:01:43 PM

Figure 9: Motorcycle data from 2016 ARC-CSI crash

Figure 10: Car Y-axis data collected from motorcycle impact with car 56

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 56

www.collisionpublishing.com

1/23/2017 2:01:44 PM

Figure 11: IST Data Clipping at 57 G (50 G IST Accelerometer)

to be in excess of the capability of the IST’s ±50 G accelerometers.

more than adequate to report peak and average accelerations, as well as Delta V.

When acceleration data is clipped, a portion of the crash pulse in excess of the clipping is missed. This results in a subsequently lower Delta V values when the acceleration data is processed, as also demonstrated in Figure 11.

The original Slam Sticks came in two versions, a Slam Stick X and C. The C version is equipped with a triaxial capacitive MEMS accelerometer, however, the range is limited to ±16 g, which is too low for most automotive crash applications. However, the Slam Stick X can be equipped with a triaxial piezo-electric accelerometer with ±25g, and ±500g or even higher ranges upon request. Certainly, ranges at or above ±100g are adequate for most automotive crash applications.

iscussion SAE J211specifies that for vehicle motion, data should be filtered with a Class 60 filter (100 Hz cutoff). However, if the data is going to be integrated, such as to obtain ∆V from acceleration, the data should be filtered with a Class 180 filter (300 Hz cutoff). It is also specified that data must be collected at a frequency that is 10 times the Channel Frequency Class. As such, SAE J211 requires that a data acquisition system for crash tests be capable of collecting acceleration data at a frequency of at least 1800 Hz. Clearly, with its capability to record up to 20 kHz, the Slam Stick’s data acquisition frequency is far

D

The acceleration reported by a data acquisition system depends on both the range and the resolution in terms bits.

For example, a system 16-bit resolution can report 216 or 65,535 different levels of acceleration. So for a ± 200 g accelerometer, a 16-bit system’s resolution is 400 g/65,535 = 0.006 g, a 14-bit system’s resolution is 400g/16,384 = 0.02 g, and a 12-bit system’s resolution is 400 g/4096 = 0.1 g

www.collisionpublishing.com

Volume 11 Issue 2.indd 57

Collision Magazine - Volume 11 Issue 2 57

1/23/2017 2:01:44 PM

(DTS). Even though the Slam Sticks have a 16-bit resolution up to 8500 Hz, the output of the capacitive MEMS accelerometer in the ARC-CSI testing appeared to have been collected with a 12-bit system because the accelerometer used for the modification was a digital accelerometer with a 0.1 g resolution.

The Slam Stick was originally designed by MIDÉ as a relatively low cost device to monitor a variety of commodities during transportation to various locations around the world. Occasionally these items are damaged in transit but it can be difficult to determine exactly where and when the damage occurred and therefore who is responsible for the damage. The Slam Stick can be set to record events in excess of a specified shock load and also the time and date that the shock load occurred. The party responsible for damaging the cargo could be more readily determined when the date and time of the potentially damaging shock load is known. Shippers monitoring their cargo are typically only concerned with the peak accelerations and not the entire duration of the acceleration event. That makes a piezo-electric accelerometer a good, low-cost choice for monitoring cargo.

A piezo-electric accelerometer, like those utilized in the original configuration of the Slam Sticks, is also known as an AC-response accelerometer. They are commonly used for shock and other high frequency recording applications because of their ability to record very high magnitude and/ or fast changing accelerations. They have good sensitivity, and very low noise levels. However, in order to produce a

signal, they require rapid changes in acceleration, because they experience low-frequency roll-off, as shown in Figure 12. In this respect, piezo-electric accelerometers electrically resemble a high-pass RC filter, in that AC current is allowed to pass through, but DC and/or low frequency current is blocked. Because they do not measure static or vibrations below a few Hz they are not suitable for applications where obtaining velocity, such as ∆V, and displacement is important (Walter, Hanly).

As seen in Figure 12 from MIDÉ’s data sheet, for the piezoelectric accelerometers that the Slam Sticks were originally equipped with, the signal becomes attenuated below approximately 10 Hz. The effect of the inability to report low frequency acceleration became evident in the SATAI results which produced an unacceptable level of underreporting of both acceleration and Delta V values, as shown in Figure 8. The SATAI crash test results exemplify the fact that applications that require integrating the acceleration for velocity, such as automotive crash testing, require DCresponse accelerometers, which are capable of measuring static and low frequency accelerations.

There are two main types of DC-response accelerometers: piezo-resistive and capacitive MEMS. The capacitive MEMS are good for measuring low frequency vibration, motion and constant acceleration, and they are low cost.

However, they have a limited band width and typically have smaller ranges, 200 g or less, for example. The piezoresistive accelerometers are more expensive, have low sensitivity, and temperature dependent output, they have a

Figure 12: Slam Stick Frequency Response from MIDE website 58

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 58

www.collisionpublishing.com

1/23/2017 2:01:44 PM

wide band width and high ranges that make them the best type of accelerometer for impact testing (Hanly). Perhaps it is not surprising that the reference equipment utilizes piezo-resistive accelerometers.

Due to the utilization of piezo-electric accelerometers, it initially appeared that the Slam Stick was not going to be a useful data collection tool for crash testing. However, the testing and data comparisons performed at SATAI coupled with close communication with the manufacturer resulted in a modified or upgraded version of this tool that is more appropriate for crash testing. For the current nonproduction version of the Slam Stick, the data collection frequency was limited to 3200 Hz, which is well above the 1800 Hz SAE J211 requirement.

There are still some hardware and software issues to overcome, but the acceleration data recorded with the upgraded capacitive MEMS accelerometers now appears comparable to the data recorded by more expensive equipment. While this analysis shows promise, there are still some limitations and challenges with the modified Slam Sticks. They can be set to trigger and collect data for certain g-levels; however, no pre-event data is collected when a g-level trigger is used. That means that any crash data before the trigger is not collected and the start of the event below the trigger threshold is not recorded. As such, in their current configuration, the g-level triggering is not appropriate for automotive crash testing.

Without g-level triggering with pre-trigger data, the Slam Sticks must be turned on prior to the crash and collect data continuously until shut off after the crash. During data processing, the analyst must pick out the correct data for processing. This is not difficult when the events are large acceleration events such as high speed barrier or vehicle impacts. It becomes more challenging when these events are small acceleration events such as low-speed impacts or off axis acceleration analysis. In the case of small acceleration events, it is important to keep track of the time interval from the Slam Stick activation until the impact or event of interest occurs in order to identify the correct subset of the acceleration data that is related to the event of interest.

Currently the Slam Sticks are not equipped with a way to trigger data collection externally. This means there is not a good way to synchronize collected data between the Slam Stick and other equipment mounted in a vehicle during a crash test.

onclusions As shown in this analysis, in order to collect appropriate acceleration data using the Slam Stick X, modified units with capacitive MEMS accelerometers are required. These accelerometers output a signal which will accurately reflect the actual data through the entire crash pulse. The acceleration data and Delta V values appear promising with good comparison to the IST data recorded in this study.

C

In order to collect appropriate acceleration data using the Slam Stick X, modified units with capacitive MEMS accelerometers are required The Slam Stick software works well for its intended use, but presents some difficulties for use in crash test data analysis.

The acceleration data in this study was exported from the Slam Stick software to a comma delimited form and processed utilizing other software. According to MIDÉ, when data is exported from the Slam Stick software, the values contained in the time column are truncated or rounded.

As a result, if the time interval between each data point is examined, it appears that the data was collected at unequal intervals. This is not the case as according to MIDÉ the data is collected at equal intervals. As filtering the data requires that the time interval to be constant, either the stated sample frequency was used to create a time-step for the data or a tool was used to read the data and create a uniform time interval set through data interpolation. The uniform data set was then used for all processing.

www.collisionpublishing.com

Volume 11 Issue 2.indd 59

Collision Magazine - Volume 11 Issue 2 59

1/23/2017 2:01:44 PM

It was also discovered that the acceleration data collection does not necessarily occur at the programmed frequency.

The Slam Sticks were all set to record data at 3200 Hz but in the data file header it was discovered that the actual data collection frequency was often close to 3200 Hz, and varied within a range of 3150 to 3250 Hz. A MIDÉ representative explained that once the sample rate is selected temperature and other variations affect the actual data collection rate. Not accounting for the actual sample rate affects the timing of features in the data as well the apparent collision duration and Delta V. Ideally, the actual sample rate should be accounted for which adds another step to the data processing.

A default feature of the factory Slam Stick software is mean removal for the displayed and exported acceleration data.

This involves automatically shifting the acceleration data by subtracting the average reading from each channel’s data set to account for temperature drift. Because this feature has the effect of zeroing the pre-impact acceleration, it is desirable for the original Slam Stick application. During automotive crash testing, however, data is collected over a relatively short time interval and this software feature could have the effect of reducing the magnitude of the reported acceleration and the subsequent Delta V. In reviewing the equipment mounting procedures utilized on the motorcycles tested, two potential changes for future testing would be to more robustly secure the instrumentation to the motorcycles and to find a way to level the instrumentation on the motorcycles. Self-taping bolt/ screws were used to secure the instrumentation package to the motorcycles and sometimes the screws pulled through the thin frame material. Through-bolts, with washers and bolts would secure the instruments more reliably. In some cases, the instrumentation was difficult to place on the motorcycles due to the overall size of the fabricated IST / Slam Stick / VBOX Sport (used for GPS speed) enclosure. This created situations where the instrumentation was attached at angles relative to the ground, which led to the horizontal and lateral accelerations having gravitational contributions. In all of these tests the location of the instrumentation as well as the angle were measured and documented.

During the ARC-CSI testing, there was significant pressure to stay on the time-schedule, which contributed to the need for easier and quicker setups being utilized instead of ‘cleaner’ data being captured.

In reviewing the equipment utilized for testing, higher capacity accelerometers are necessary to ensure that all of the crash data is collected. During the motorcycle to barrier tests, accelerations nearing 100 g were collected from several of the tests. This suggests that accelerometers with

60

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 60

capability in excess of 100 g should be utilized for motorcycle barrier testing. As the maximum Delta V of this testing was approximately 38 mph, even higher capacity accelerometers would be necessary for higher speed testing beyond what was performed here.

Data acquisition systems come in a variety of sizes, shapes, features and costs. Any data acquisition system used for automotive crash testing requires careful evaluation of the pros and cons prior to full implementation. This is necessary because, with the possible exception of low speed impact testing, there typically is not an opportunity to repeat a crash test without obtaining new vehicles. Because of this, new data acquisition systems are typically evaluated side-by-side with proven equipment whose robustness has already been demonstrated. Only after the new system has been thoroughly tested and proven will it be relied on for crash data acquisition.

Besides providing an example of the evaluation process for the Slam Stick, this exercise also has demonstrated the limitations of hardware previously thought to be appropriate. As an example, during this project it was discovered that one of the reference IST’s was equipped with an ±50g accelerometer from which the data was clipped in some crashes, and it was mechanically filtered at 90 Hz, which is a lower channel frequency class specified by SAE J211 for reporting acceleration and well below that for integrating to obtain Delta V. This example underscores the concept that even though some IST’s and Slam Sticks have an appropriate configuration for automotive crash testing, care must be taken to ensure that each one has the features required for a specific crash test.

Additional testing was performed by the ARC-CSI motorcycle crash team utilizing the Slam Stick’s with other traditional instrumentation for coefficient of friction testing. Future paper(s) are planned to address these tests and their results. uture Product MIDÉ has shared with the authors that the next evolutionary step for the Slam Stick, the Slam Stick S, is due to be available in late 2016.

F

The Slam Stick S will be equipped with analog piezo-resistive accelerometers with ±100 g and ±500 g ranges. As such, the accelerometers themselves will be appropriate for automotive crash testing. Upon request, MIDÉ can equip the Slam Stick S with analogue piezo-resistive accelerometers with other ranges.

The resolution will be 16-bit up to 8500 Hz, and 14-bit up to 20,000 Hz, which is much more than the 1800 Hz

www.collisionpublishing.com

1/23/2017 2:01:44 PM

required for our application. As the accelerometers will be analog, there will not be an apparent reduction in system resolution due the digital output of the accelerometer, as was observed in the modified Slam Sticks during the ARCCSI crash tests.

The g-level trigger feature will remain unchanged from that of the Slam Stick X and C models. Without pre-trigger recording, this feature will be inappropriate for most automotive crash testing applications.

It is anticipated that the sample rate variation issue for the Slam Stick S will be greatly reduced, so that the actual frequency will vary from the set frequency by only a couple Hz at most. This level of variation would be nearly imperceptible in acceleration overlays and Delta V since for a 3200 Hz sample rate there would only be a variation of less than half a sample for a crash pulse up to 200 millisecond duration.

Clearly, with these updated features, the Slam Stick S would be an affordable tool for automotive crash testing.

In communications with MIDÉ, it has been indicated that the development of their product line is expected to continue so that future embodiments of the Slam Stick may also have the ability to collect data from external sensors, receive time-zero signals and incorporated GPS, triaxial rotational rate and magnetometer for absolute orientation information. eferences 1. Beckage M & Pruitt S, “Principles of Dynamic Data Collection,” Diversified Technical Systems 2006.

R

2. Hanly S, “Accelerometers: Taking the Guesswork out of Accelerometer Selection,” MIDÉ, 2016.

Clearly, with these updated features, the Slam Stick S would be an affordable tool for automotive crash testing Thanks to BRT and Biomechanical Analysis for the reference comparison data they provided.

Thanks to Mr. Steve Hanly and MIDÉ Technology for their technical assistance with the Slam Stick.

A special thanks to all the members of the 2016 ARC-CSI motorcycle crash team (alphabetically): Kelley Adamson, Robert Anderson, Billy Cox, Team Lead Michael DiTallo, Jim Dunlap, Thomas Green, Wesley Grimes, Rusty Haight, Sean Haight, David Hallman, Thomas Lawson, Michael Merolli, Eric Moody, Jim Morgan, Brent Munyon, Eric Paul, William Redfairn, Tim Reust, Michael Rosenfield, Kenneth Salisbury, Kevin Vosburgh and James Whelan.

3. Surface Vehicle Recommended Practice, Instrumentation for Impact Test – Part 1 – Electronic Instrumentation, SAE J211 Rev March 2014, (http://standards. sae.org/j211/1_201403/).

4. Walter P.L., Ph.D., “Selecting Accelerometers for and Assessing Data from Mechanical Shock Measurements,” Technical Note (TN-24), PCB Piezotronics, Inc. cknowledgements The authors wish to thank SATAI, The ARC Network and Collision Safety Institute for making this testing possible during the 2015 SATAI Crash Conference and the 2016 ARC-CSI Crash Conference.

A

www.collisionpublishing.com

Volume 11 Issue 2.indd 61

Collision Magazine - Volume 11 Issue 2 61

1/23/2017 2:01:44 PM

Volume 11 Issue 2.indd 62

1/23/2017 2:01:45 PM

Volume 11 Issue 2.indd 63

1/23/2017 2:01:46 PM

ed w e i ev R l ria o t i Ed

Low-Speed Rear-End Crash Analysis MARC1 Use in Test Data Analysis and Crash Reconstruction Rudy Limpert, Ph.D.

I

ntroduction As indicated in Short Paper PCB1-2015, I am publishing another paper of low speed rear-end crash test analysis. I hope a detailed MARC1 analysis of specific crash tests may be of help to those who are new to MARC1 and those experts who may need a refresher. Crush depth measurements are not easily obtained from photographs.

In many tests crush depth values are not published, and worse, often not even measured. I suspect that most reconstruction experts do not analyze low-speed crashes, and hence, experts are not familiar what input data are required to calculate impact speed. Conducting MARC1 crash tests- on-paper will in many cases shed helpful light on the facts involved. Photographs and measurements of crush depth are always important for an accurate reconstruction in low-speed cases.

Low-Speed Rear-End Crash Test: Nissan Micra K11 v.

VW New Beetle

The crash damage of the Nissan and VW are shown in Figures 2 and 3 below. The crush damage of the Nissan seems to be more extensive than that of the VW. No crush measurements were published.

Figure 2: Nissan Frontal Damage

A Nissan Micra K11 rear-ended a VW New Beetle in an inline low speed test. The impact configuration is shown in Figure 1. The impact speed was 13.91 mph. The Nissan was braking at 0.59g deceleration during impact. The VW was stationary with its transmission in neutral.

Figure 3: VW Rear-end Damage.

Figure 1: Nissan-VW Rear-end Pre-Crash Impact Configuration 64

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 64

www.collisionpublishing.com

1/23/2017 2:01:46 PM

Published Crash Data

Crash Test-on-Pape Using MARC1 – W2

The AGU Crash Datenbank (www.agu.ch) data were published on the internet as shown below. When I first saw the data table, my curiosity was peaked. The data were exactly what a MARC1-2015 – W4 low-speed rear-end collision analysis requires as input data such as crush energy, energy equivalent speed and stiffness values (lb/ft). Crush depth data were not given. How accurate the data ranges were could only be answered through MARC1-W4 analysis,

We first run a simple crash-test-on-paper with MARC1W2., which does not include any braking during impact.

MARC1-W2 analyzes an inline collision with the use of conservation of energy across the impact, conservation of momentum across the impact (no external forces are considered such as tire-to-ground braking forces), and the coefficient of restitution.

The German-language data are summarized in Table 1 with imperial units.

The results show an acceptable data comparison when braking is excluded (W2) as shown below in Table 2.

The acceleration and velocity time histories are shown below for the reader to gain additional understanding of the low-speed inline crash test dynamics.

www.collisionpublishing.com

Volume 11 Issue 2.indd 65

Collision Magazine - Volume 11 Issue 2 65

1/23/2017 2:01:46 PM

Crash Test Results Weight (lb) Impact Speed (mph) Velocity Change (mph) Max. Acceleration Average Acceleration Energy Equivalent Speed Stiffness Impulse Duration Coefficient of Restitution e Separation Velocity Total Crush Energy Repair Cost

Nissan Micra 2195 19.91 11.2 415 164 15.5 24,000

Nissan/VW

VW New Beetle 2971 0 6.71 428 103 3.4 1002,806

100 0.29 3.98 8,703 12,258

Units lb mph mph ft/sec2 ft/sec2 ft/sec lb/ft ms mph lbft CHF

1,947

Table 1: Published Crash Data

Figure 4: Velocity and Acceleration Traces.

Impact Speed Speed After Impact Velocity Change Crush Energy Impulse

Crash Test Data

MARC1-W2 Crash Test-on-Paper

Nissan 13.91 2.74 11.2 8,703 1,119

Nissan 13.91 3.59 -10.32 7,466 1031

Beetle 0 6.71 6.71 907*

Beetle 0 7.62 7.62

mph mph mph lbft lbsec

*Calculated for P1 = (deltaV1)(m1) = (11.2)(1.466)(2195/32.2) = 1,119 lbsec)

Table 2: Crash Test Data and MARC1 –W2 Comparison

66

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 66

www.collisionpublishing.com

1/23/2017 2:01:46 PM

Inspection of the data comparison shows that the braking Nissan in the actual test does not push the VW forward as violently as the no-braking crash test-on-paper. Also, for the same reason, the after-impact speed of the Nissan is lower in the crash test with 2.74 mph versus 3.59 mph.

Crash Test Reconstruction For low-speed inline collision analysis see Section 33.0[8] of Limpert accident reconstruction book. For our reconstruction we do not know the impact speed of the Nissan.

However, we have inspected the damage of both vehicles and know that the Nissan driver was braking during impact. We first run all data as provided in the MARC1-W4 low-speed rear-end collision. As the reader will find out, the software cannot calculate the results. The reason is most likely a negative square root. The proper input shown in Table 1, while noticing the suggested input data ranges, are shown in the MARC1-W4 printout.

Data Comparison Crash Test and Reconstruction MARC1-W4 The data comparison is shown in Table 3. Which measured test data are important comparison measures and which are less critical may be a discussion topic by itself. However, based on my own experience, maximum accelerations for each vehicle, estimated crash damage and impulse time (Stosszeit in the German report) may be of importance. In this particular crash test no crush depth data were provided. The damage repair costs were published. Whether the Nissan repair cost of 12,258 CHF were due to an increase crush depth is not clear. I assumed a crush depth of 3.2 inches for the Nissan and 1.8 inches for the VW (1,947 CHF repair cost). I believe that acceleration, impulse time and crush energies should be used as an approximate guideline for input data analysis.

MARC1-W4 predicts impact speed reasonably accurately.

The user of the software must be careful not to over-extend

Impact Speed (mph) Maximum Acceleration (ft/sec2) Impulse Time (ms) Crush Energy (lbft) Departure Speed (mph) Delta-V (mph) Impulse (lbsec)

MARC1-W4 Crash Test” Reconstruction its capability. As a safety measure, the reader must always use MARC1-W2 as a truth maker tool.

Readers must remember, that nearly all tests involve to a large measure empirical data such as crush depth, crush energy, stiffness values, and others. While there are many high-speed crash test exceeding 35 mph, low speed crash tests are rarely conducted on a regular basis.

2015 ARC-CSI Low Speed Crash Test Mazda 3 and Land Rover At the 2015 ARC-CSI crash conference five low speed inline crash tests were conducted. Only some results were presented. A 2005 Mazda 3 was driven into the trailer

Crash Test Nissan VW 13.91 0 415 428 100 8,703 > Σ 6,711 2.74 6.71 11.2 6.71 1,119 907

MARC1 – W4 Nissan VW 13.72 0 427 426 105 89 3,758 2,953 2.97 6.94 -10.75 6.94 -1,075 939

Table 4. Crash Test and Reconstruction Comparison www.collisionpublishing.com

Volume 11 Issue 2.indd 67

Collision Magazine - Volume 11 Issue 2 67

1/23/2017 2:01:47 PM

hitch of a stationary 2001 Land Rover. The impact speeds ranged from 1.1 1to 6.1 mph. For each test the longitudinal delta-V and vehicle axis acceleration time histories were measured. The vehicle weights were not measured specifically. Published data from 4N6PRT SYSTEMS (R) were provided. The purpose of our analysis is to study the effects of various crash parameters on low speed inline crashes.

Impact Speed 1.1 mph The static contact damage shown in Figure 5 indicated within the blue marking.

Figure 7: Land Rover Data Traces.

Figure 5: Minor Damage in 1.1 mph Hitch Impact.

The data traces are shown in Figures 6 and 7. The maximum accelerations are approximately 0.45g for the Mazda and 0.38g for the Land Rover. The delta-Vs are approximately 1.18 and 0.12 mph, respectively.

The in-put and out-put data of our own crash-test-onpaper are shown in MARC1-W2 printout. Using a coefficient of restitution e = 0.70 yields reasonable data correlation, particularly for the Mazda. The crush energy is 36 lbft. The minor impact damage is reflected in Figure 5. MARC1-W2 does not consider any external tire forces.

For example, a tire rolling resistance coefficient of 0.015 indicates a “braking deceleration” of (0.015)(32.2) = 0.48 ft/sec2. In connection with a given impact speed of 1.1 mph (1.10 mph or 1.14 mph) and published vehicle weights (the test vehicles were apparently not measured), any further data analysis seems meaningless. However, we can conclude that the coefficient of restitution is approximately 0.7 and the crush energy 36 lbft.

Impact Speed 6.1 mph

Figure 6. Mazda Data Traces

68

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 68

The impact damage to the Mazda bumper is shown in Figures 8 and 9.

www.collisionpublishing.com

1/23/2017 2:01:47 PM

Figure 8: Mazda Impact Damage.

Figure 11: Data Traces for Land Rover.

As before, we run our own “crash test” using MARC1-W2.

The results are shown in the data output. For a coefficient of restitution e = 0.23 reasonable data correlation is achieved. The crush energy is 2,032 lbft. It appears that the influence factors such as tire rolling resistance are of little influence at an impact speed of 6.1 mph versus 1.1 mph.

A coefficient of restitution e = 0.23 is consistent with the initial and final velocities indicated in Figure 10 and 11 (Eq. 33-11 of Limpert book).

Figure 9: Mazda Impact Damage The acceleration and delta-V traces are shown in Figures 10 and 11.

Figure 10: Data Traces for Mazda.

The Mazda static crush depths shown in Figure 8 and 9 surprisingly were not measured by the Lantz Engineering, et al. crash test teams. Crush depth is a critical reconstruction parameter for any accident, particularly for a low speed crash. The reasons are very simple. In a high speed accident there typically are several other reconstruction parameters besides momentum such as vehicle linear and rotational runout dynamics, pre-crash dynamics, crash depth measurements and documented crash tests. In a low speed accident, if the static crush depth is not carefully

www.collisionpublishing.com

Volume 11 Issue 2.indd 69

Collision Magazine - Volume 11 Issue 2 69

1/23/2017 2:01:47 PM

measured and documented, we have practically nothing to work with.

2015 IPTM Low Speed Crash Test Analysis of Bus – Ford Car

Low Speed Crash Reconstruction

A city bus rear ends a stationary car at four different speeds The impact configuration is shown is in Figure 12. We will only analyze the 3.8 mph impact test. The measured output data are shown in Figure 13. Our own crash test data are shown in the MARC1‐W2 printout. As an inspection shows, calculated and measured data are in excellent agreement.

The purpose of the detailed 6.1 mph low speed crash test analysis is to show the reader what data analysis is available with the MARC1-W4 software.

In Section 3.2 we ran a crash-test-on-paper. Now we are trying to reconstruct the crash test based upon measured data and static crush depth estimates based upon vehicle photographs. We assume that the Land Rover hitch assembly is very stiff, the foam-filled bumper of the Mazda fairly soft. It would have been very helpful to measure the bumper stiffness in terms of force displacement versus deflection in a simple load cell test. However, it appears we made reasonable assumptions.

The MARC1-W2 output data show a total crush energy is 864 lbft. How it is ditributed among both vehicles is not known from our analysisi thus far. However, it is safe to say that the largest cruch energy is associated with the car. For a crash test with e = 0 (fully plastic impact) crush energies are distributed according to the respective deltaVs.

The calculated delta-Vs for the Mazda and Land Rover are 4.75 and 2.75 mph, respectively. The measured delta-Vs are 4.7 and 2.7 mph for the Mazda and Land Rover, respectively. The measured maximum accelerations are 6.3g (203 ft/sec2) and 5.1g (164 ft/sec2), respectively. Minor input data variation most likely will bring the acceleration for the Mazda (203 versus 239 ft/sec2) in better correlation.

The control parameters measured are longitudinal acceleration, delta-Vs and impact speed. Based upon the MARC1W2 data, we had some guide lines, namely coefficient of restitution, crush energy and impulse.

Figure 12: Bus-Ford Impact Configuration.

Figure 13: Impct Speeds and Measured Output Data.

The coefficient of restitution e = 0.44 is provided in Figure 13. It was calculated by (see Motor Vehicle Accident Reconstruction and Cause Analysis, Eq. 33‐11): e=(V22–V12)/(V11–V21)=[4.9–(3.8–0.57)/[3.8–0]=0.44

70

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 70

www.collisionpublishing.com

1/23/2017 2:01:48 PM

Figure 16: Car Acceleration Trace

Figure 14. Bus Acelleration Trace.

In Figure 14 the actual input screen of MARC1‐W4 is shown. In the printed version the decimal points are limited to three places behind the period. For example, the static crush depth for the bus bumper was assumed to have been 0.0001 ft or 0.0012 inches and not zero (0.000 as the reader might assume). When we first introduced MARC1‐ W4 we thought three digits behind the period was sufficient. The calculations are done with the four decimal places, however not printed. This sould probably be corrected.

Inspection of the measured and calculated data shows good agreement. W4 also shows that practically all crush energy is associated with the car (866 vs. 0.7 lbft).

Low Speed Reconstruction Help If any reader needs any assistance, please contact Rudy Limpert at 801‐201‐2532. I can also provide a large list of published papers including dissertaions, etc. on low speed crashes.

Figure 15: Actual Input Screen Data used for W4 Analysis. www.collisionpublishing.com

Volume 11 Issue 2.indd 71

Collision Magazine - Volume 11 Issue 2 71

1/23/2017 2:01:49 PM

ed w e i v e R Peer

Piston-Type Bumper Isolator Compression versus Delta-V in Bumper-to-Barrier and Bumper-to-Bumper Impacts Robert D. Anderson, Michelle R. Hoffman Russell L. Anderson, Michael Rosenfield

72

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 72

www.collisionpublishing.com

1/23/2017 2:01:51 PM

I

ntroduction In low speed collisions involving vehicles’ bumpers, the evidence of the amount of piston-type bumper isolator compression has been correlated with the severity of the collision. The extent of isolator compression is in effect a record of the magnitude of the collision force. For vehicles with isolators that have been involved in a low-speed collision, comparing the extent of isolator movement to that produced in collision tests of known impact severity is a reliable and accepted method for assessing impact severity. Since the mid-1990’s the use of relationships developed for each automobile manufacturer has been common.

Although many manufacturers have moved away from piston-type bumper isolators, BMW and Audi/ Volkswagen equipped some of their automobiles with isolators up to the 2006 and 2009 model years, respectively. Ford had isolators on some of their vehicles as recently as 2011, and Porsche had isolators on some of their vehicles as recently as 2013.

Starting in the 1990’s, low speed impact analyses commonly included evidence of the extent that isolators compressed during the impact. Isolator-equipped bumpers were common on 1990’s automobiles and much of the published data relating to isolator compression versus ∆V is from that era (Szabo and Welcher, 1992; King, et al., 1993; West, et al., 1993; Malmsbury, 1994; Siegmund, et al., 1994; Bailey, et al., 1995; Siegmund, et al., 1996; King, et al., 1997; Anderson, et al., 1998; and King, et al., 1999).

King, et al., (1993) concluded from static and dynamic bumper compression tests that isolator compression generally increased with speed change and that testing for each vehicle was the most accurate method to arrive at a ∆V for a particular amount of isolator compression.

The largest and most comprehensive data set of testing on isolator-equipped vehicles was published by Siegmund, et al., (1994). Specific average isolator compression-∆V trends for vehicle manufacturers were identified, which reduced the need for thorough testing of each specific vehicle. It was observed that struck vehicle braking, and/or additional weight from passengers and cargo produced over-estimates of the struck vehicle’s ∆V.

Both Siegmund et al., (1994) and Bailey et al., (1995) found that using isolator compression data from barrier impacts may over-estimate the ∆V in car-to-car impacts for some vehicles and that care must be exercised to ensure that the isolators are not bottomed out.

Siegmund, et al., (1996) found that the direct barrier and the momentum, energy and restitution methods provided reasonable predictions of the actual ∆V sustained by vehicles in car-to-car collisions, but that the quasi-static isolator force method did not.

While investigating more affordable methods to obtain isolator compression-∆V data, King, et al., (1997) found that vehicle characteristics, such as suspension properties, center of mass location, as well as the bumper mounting conditions including the bumper cover attachment may affect the isolator compression∆V relationship. King, et al., (1999) concluded that single isolator testing could be used to develop the relationship between isolator compression and vehicle speed change for some, but not all, types of isolators.

In practice, the manufacturer trends identified by Siegmund et al. (1994) have been commonly used in low speed impact analysis since the mid-1990’s. The majority of all the data cited above was obtained from 1980’s vintage automobiles. Recent crash test results have shown an apparent shift in the isolator compression versus ∆V trend for Ford automobiles (Ruberte, et al., 2014), suggesting that using the data from Siegmund et al., (1994) resulted in an underestimate of the vehicle ∆V. The purpose of this paper is to present data to help bridge the gap between the historic and newer data for Ford, BMW and Volkswagen, and to present a method for measuring real time isolator compression.

M

ethodology A series of low speed impacts were conducted on vehicles with piston-type bumper isolators. Both front and rear bumper impacts were conducted on 4 different vehicles (identified in Table 1) to determine the isolator stroke as a function of vehicle velocity change for vehicles manufactured during the 1990s.

Before testing, each vehicle was weighed with CAS Wheel Weigher scales, which have a 10,000 pound per wheel capacity in 5 pound increments. The test weights are shown in Table 1. It is noteworthy, that

www.collisionpublishing.com

Volume 11 Issue 2.indd 73

Collision Magazine - Volume 11 Issue 2 73

1/23/2017 2:01:53 PM

Vehicle Year 1995 1996 1997 1997 1999

Make *Buick BMW Mercury Lincoln Volkswagen

Model Le Sabre* 740iL Grand Marquis Town Car Passat

Weight (lbs) 3595 4155 3765 - 3815 4100 3310

Model Year Range 1992-1999 1995-2001 1995-1997 1995-1997 1998-2001

*The Buick was equipped with foam cushion bumpers and was used as a target for the vehicleto-vehicle testing with the Mercury.

Table 1: Vehicle information for the tested vehicles. the Mercury’s test weight varied by 50 pounds, which can be explained by different levels of fuel, at 6 pounds per gallon in the 20 gallon gas tank. Table 1 also lists the model year range that the test vehicles were unchanged as specified by the Mitchell Collision Estimating Guide.

Vehicle-to-vehicle testing was conducted on three vehicles.

In December of 2005, 50% offset and centered impacts were conducted between the front of the Mercury striking the Buick, and in January of 2006, tests were conducted with the front of the Mercury and the rear of the Lincoln.

impact speed by monitoring a digital speed readout from the VBOX3i. The vehicle was released just before impact and struck the barrier. For the Volkswagen tests, the engine was not running and vehicle was steered by a human through the driver’s window. For the BMW tests the engine was running and there was a human driver steering the vehicle into the barrier.

For the vehicle-to-vehicle tests, the bullet vehicle was driven and/or manually pushed into the target vehicle, which was in neutral. Both vehicles’ engines were running, and the impact speed was measured with a Racelogic VBOX3, which reported speed based upon the Doppler shift of GPS radio signals at 100 Hz with an accuracy of 0.06 mph. The ∆V of the striking vehicle was arrived at by integrating the vehicle acceleration measured with ICST/ MSI 3031 uniaxial, 100 g range accelerometers arranged in a triaxial configuration. A pressure sensitive tape switch was mounted to the bumper of the Mercury to establish time zero at contact and trigger data collection. The ∆V of the struck vehicle was measured with a Robic time trap, which measured the time between tape switch activations in one-thousandths of a second increments.

Vehicle-to-barrier testing was also conducted on three vehicles: the Mercury was tested in January of 2006, the Volkswagen was tested in September of 2013, and the BMW was tested in November and December of 2013.

In the Mercury barrier tests, the vehicle was manually propelled into a concrete loading dock, shown in Figure 1 (top), while it was unoccupied, its engine running, and its transmission in neutral. For the Volkswagen and BMW barrier tests, the vehicles were propelled into a 50,000 pound concrete block barrier shown in Figure 1 (bottom), by manually pushing the vehicle with the transmission in neutral. The vehicle was accelerated to the desired 74

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 74

Figure 1: Barriers - loading dock (top), and concrete block (bottom)

www.collisionpublishing.com

1/23/2017 2:01:53 PM

In all barrier tests, the vehicles were instrumented with a triaxial accelerometer comprised of ICST/MSI 3031 uniaxial accelerometers mounted at the vehicle’s approximate static center of gravity (cg). The vehicle’s barrier impact speed was monitored and recorded using a Racelogic VBOX3i. In some tests, the barrier impact speed was also measured with a FarmTek/Polaris time trap, which measured the time between tape switch activations in onethousandths of a second increments. A pressure sensitive tape switch was mounted to the bumpers of the vehicles to establish time zero at barrier contact and trigger collection of the data. The vehicles were inspected for damage after each test.

For the Volkswagen and BMW barrier tests, Celesco SP1 linear displacement transducers (string potentiometers) with a 50 inch range were also used to measure the piston stroke distance. See Figure 3.

For all isolator test vehicles, the isolator stroke was measured by observing the distance a loose-fitting cable tie was pushed along the exposed piston tube length by the piston housing, as shown in Figure 2.

Figure 3: Isolator piston and string pot before (top) and after (bottom) testing.

The accelerometer, linear displacement transducer and in some cases the speed data output from the VBOX were collected using a rack-mounted Diversified Technical System (DTS) TDAS Pro signal input module(s), each of which have a 16-bit resolution. All data was recorded at 10 kHz. Axis orientation and data filters complied with SAE J211 Recommended Practice.

Figure 2: Cable ties before (top) and after (bottom) impact tests.

Barrier impact speeds ranged from less than 1 mph to 7 mph and resulted in ∆V’s as high as about 8½ mph. The vehicle-to-vehicle impact speeds ranged from 1 to 8 mph producing ∆V’s as high as about 5½ mph. The test setup and results were documented with photographs and movies taken with various cameras. Although both isolators for any given impact often exhibited similar compression in the aligned impact, the two isolators’ compression was always averaged and then compared to the ∆V. For each vehicle’s average isolator versus ∆V data set, linear regression was used to obtain the best fit line and the coefficient of determination, R2.

www.collisionpublishing.com

Volume 11 Issue 2.indd 75

Collision Magazine - Volume 11 Issue 2 75

1/23/2017 2:01:54 PM

R

esults and Discussion: For the BMW, the front-to-barrier impacts were performed first. The data from the frontal impacts was then used to assess the relationship between barrier impact speed and vehicle ∆V to increase the accuracy of the expected ∆V from the rear-to-barrier impacts. The relationship between impact speed and vehicle ∆V for the frontal impacts is shown in Figure 4. For these frontal impacts, the coefficient of restitution ranged from 0.22 to 0.4. This barrier impact speed versus ∆V relationship allowed appropriate barrier impact speeds to be targeted for the subsequent rear impacts where a driver occupant would be exposed.

The average isolator compression vs. ∆V for all of the front and rear BMW-to-barrier impacts is shown in Figure 5.

The results of a number of rear-to-barrier impacts were excluded from the data set because the isolator compression values were unreliable due to the string potentiometer housing becoming dislodged. Compression of both isolators initiated at a ∆V of about 1.25 mph and steadily increased to 1.9 inches compression at a 7.7 mph ∆V. For the linear regression, the coefficient of determination (R2) was 0.9746, which demonstrates that the best fit line closely represents the data. Using the best fit line for the BMW data results in the following equation:

∆V = (IC + 0.2769) / 0.2788

Equation (1)

Where, ∆V is in mph and IC is the average isolator compression in inches.

The compression versus ∆V for the 1985 BMW 535i reported by Siegmund et al. (1994) showed compression of both isolators began at about a 1.5 mph ∆V and steadily increased to 1.1 inches at a 4.7 mph ∆V. Using Equation 1, with 1.1 inches for the average isolator compression, the corresponding ∆V would be 4.9 mph. Therefore, the results appear similar within one-quarter of 1 mph.

According to Mitchell Collision Estimating Guide, the model year range for the BMW 5 and 7 Series were essentially unchanged from 1982 to 1988 and again from 1995 to 2001, respectively. As such, there is not any evidence from the current test series that the BMW manufacturer trend has changed significantly from the early 1980’s through 2001. Therefore, it may be expected that appropriate results can be obtained using the previously identified manufacturer trend for low speed reconstruction involving bumper impacts.

The Volkswagen Passat was backed into the barrier at speeds ranging from 1.4 to 7 mph, which with coefficients of restitution ranging from 0.24 to 0.44 produced ∆V’s

Figure 4: Impact speed versus ∆V for the 1996 BMW 740 iL. 76

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 76

www.collisionpublishing.com

1/23/2017 2:01:54 PM

Figure 5: 1996 BMW 740iL front- and rear-to-barrier isolator compression versus ∆V.

Figure 6: 1999 Volkswagen Passat rear-to-barrier isolator compression versus ∆V. www.collisionpublishing.com

Volume 11 Issue 2.indd 77

Collision Magazine - Volume 11 Issue 2 77

1/23/2017 2:01:55 PM

within the 2 to 8½ mph range. The Volkswagen’s average isolator versus ∆V relationship is shown in Figure 6.

Again, with an R2 of 0.9959, the best fit line closely represents the data. Using the best fit line for the Volkswagen data results in the following equation: ∆V = (IC + 0.5416) / 0.3275

Equation (2)

Where, ∆V is in mph and IC is average isolator compression in inches.

The compression versus ∆V for the 1978 Volkswagen Rabbit reported by Siegmund, et al., (1994) showed compression of both isolators began with about a 1.5 mph ∆V and steadily increased to about 1.8 inches at a 6.7 mph ∆V before a modest apparent stiffening characterized by slightly greater ∆V increases for incremental increases in average isolator compression. Using Equation 2 with 1.8 inches for the average isolator compression, the corresponding ∆V would be 7.1 mph, which is within the bounds of the data for the Volkswagen Rabbit. Therefore, the results appear similar, within about one-third of 1 mph.

According to Mitchell Collision Estimating Guide, the model year ranges for the Volkswagen Rabbit and Passat were essentially unchanged from 1975 to 1979, and from 1998 to 2005, respectively. As such, there is not any

evidence from the current test series that the Volkswagen manufacturer trend has changed significantly from mid1970’s through 2005. Therefore, it may be expected that appropriate results can be obtained using the previously identified manufacturer trend for low speed reconstruction involving bumper impacts.

The Mercury Grand Marquis was rolled forward and backward into a loading dock at speeds within the approximate 0.7 to 5 mph range, which produced ∆V’s within the 1¼ to 6¼ mph range. As shown in Figure 7, the Mercury’s front bumper and rear bumper average isolator versus ∆V relationships were different.

The Mercury’s rear bumper exhibited a stiffer response characterized by initiation of isolator compression at a higher ∆V, and greater increases in ∆V for incremental increases in average isolator compression. It was noted that the rear bumper exposed piston tube length was 0.6 inches shorter than that of the front. With R2 of 0.9684 and 0.0933 for the front and rear bumper, respectively, the best fit lines closely represent the data. Using the best fit line for the Mercury data results in the following equations: Front ∆V = (IC + 0.4685) / 0.3644 Equation (3) Rear ∆V = (IC + 0.492) / 0.244 Equation (4)

Figure 7: 1997 Mercury Grand Marquis front- and rear-to-barrier isolator compression versus ∆V. 78

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 78

www.collisionpublishing.com

1/23/2017 2:01:55 PM

Where, ∆V is in mph and IC is average isolator compression in inches.

Due to the differences in the relationships for the Mercury’s front and rear bumper, combining the front and rear bumper data results in an R2 of only 0.7149. According to Mitchell Collision Estimating Guide, the Mercury Grand Marquis/Ford Crown Victoria were essentially unchanged over the 1992 to 1997 model years.

The combined front and rear average isolator compression versus ∆V for the 1979 Fairmont, 1981 Escort, 1989 Tempo, 1981 and 1989 Mustangs, 1982 Lynx, and 1985 Grand Marquis reported by Siegmund et al. (1994) showed compression of both isolators began at about 1.3 to 2.6 mph ∆Vs and steadily increased to about 2.2 inches at a 6.7 mph ∆V before the Escort’s isolator’s began to bottom out. This generally corresponds with the front bumper relationship observed on the 1997 Grand Marquis. However, the rear bumper relationship shown in Figure 7 more closely corresponds with the rear bumper data for the 2003 to 2011 Ford Crown Victoria observed by Ruberte, et al., (2014). The Mitchell Collision Estimating Guide indicates that the Crown Victoria/Grand Marquis from 2003

to 2011 were equipped with foam cushion inserts on the front bumpers, which is likely why Ruberte, et al., (2014) on reported on rear bumper isolator data.

Based both upon the current data for the Grand Marquis and Ruberte, et al.’s data for the Crown Victoria, it appears that the Grand Marquis/Crown Victoria were equipped with shorter and stiffer rear bumper piston-type isolators in the early 1990’s while the front bumper remained equipped with longer and softer isolators until changed to cushion insert type bumpers in 2003. Thus, it appears that there is evidence of a shift in the manufacturer trend for the Ford family of automobiles toward stiffer isolators on the rear. Therefore in the analysis of low speed bumper impacts involving Fords newer than the early 1990’s, the possibility that the manufacturer trend may be stiffer than previously published should be considered.

Finally, the Mercury was used to rear-end another a 1997 Lincoln Towncar which was also equipped with pistontype impact isolators and a 1995 Buick LeSabre, which was equipped with a foam cushion insert bumper. The Buick impacts were both one-sided and straight-on impact orientations. As shown in Figure 8, comparable results

Figure 8: 1997 Grand Marquis front isolator compression versus ∆V for barrier, Lincoln, and Buick impacts. www.collisionpublishing.com

Volume 11 Issue 2.indd 79

Collision Magazine - Volume 11 Issue 2 79

1/23/2017 2:01:55 PM

were obtained for the compression of the Mercury’s front isolators in the barrier, Lincoln and Buick impacts.

In addition, the Lincoln isolator versus ∆V relationship coincided with the above trend as also shown in Figure 8. This is consistent with the results of earlier test results, reported by Bailey, et al., (1995) that sometimes using isolator compression data obtained in barrier impacts can slightly over-estimate the ∆V in vehicle-to-vehicle impacts, and other times it does not.

Given the strong correlations between average isolator compression and ∆V, using the evidence of isolator compression in the reconstruction of low speed bumper impacts remains an appropriate and reliable method. For BMW and Volkswagen vehicles, using the previously identified manufacturer trend is appropriate for reconstructing low speed bumper impacts involving these vehicles.

However, for the Ford family of automobiles, a determination should be made whether it is appropriate to apply the original softer manufacturer trend or the newer stiffer manufacturer trend for reconstructing low speed bumper impacts involving these vehicles.

S

tring Potentiometer A string potentiometer is often referred to as a string pot, yo-yo pot, draw wire transducer, or cable extension transducer. They can be compact, economical and durable devices that use a flexible cable, a spring-loaded spool, and/or a rotational potentiometer coupled with the axle of the spool to measure the position separating two points.

By applying an excitation voltage (+in voltage), there is a voltage divider so that the output voltage is the percent-

age of the excitation voltage that coincides with percentage of the cable that is extended, as illustrated in Figure 9.

The output from a string pot is position, the derivative of which, with respect to time, is velocity. The output accurately represents the distance separating those two points, provided that the cable remains straight between the two attachment points. Sag (from gravity, physical interference or motion faster than the spool can retract the cable) reduces the accuracy.

One advantage of a string pot is the ease of installation, which simply consists of securing the cable end and the housing on the objects between which the relative distance is of interest. For the BMW and the Volkswagen barrier crashes, a string pot was mounted between the reinforcement bar and the front or rear body panels such that the cable was aligned with the isolator. The cable would retract or extend with the compression or expansion of each isolator, provided that the bumper did not compress faster than the string pot could retract.

For the BMW rear-to-barrier impacts, the string pot housings were affixed to the floor of the trunk and the cable end was attached to the reinforcement bar through holes drilled in the rear body panel. However, during the test series, the housings broke loose, allowing the cable to snag on the edge of the holes, thereby preventing retraction of the cable. As such, the data obtained from the string pots did not accurately represent the isolator compression in those runs.

Throughout the remainder of the Volkswagen and BMW testing, the maximum isolator compression measured via string pots coincided with isolator compression evidenced by the distance the isolator housing moved a loosely fit-

Figure 9: Voltage divider and string potentiometer output signal.

80

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 80

www.collisionpublishing.com

1/23/2017 2:01:56 PM

resented. A graph showing vehicle x-axis acceleration, ∆V and isolator compression is shown from a representative test in Figure 10.

The maximum isolator compression roughly corresponds with the time of the peak vehicle acceleration and the compression remains relatively constant for about 30 msec before dropping off.

Figure 10: Vehicle acceleration, ∆V and isolator compression from a representative test.

Figure 11: String pot retraction speed during a barrier impact. ting cable tie down the exposed piston tube length. Additionally, many times the final position of the cable tie also coincided with the end of the most recent scrape mark and material build up, an example of which is seen in Figure 3. In a few instances, when the cable tie was pinched during full isolator compression or entrapped between the piston tube and housing, the compression was obtained from scratch marks or the impact was redone.

Overall, the testing demonstrated that the string potentiometer was accurately capturing the extent of isolator compression. However, additional analysis was required to ensure that the dynamic isolator compression was also being repwww.collisionpublishing.com

Volume 11 Issue 2.indd 81

Figure 11 shows the speed derived from the string pot retraction in a 4.85 mph barrier impact that produced a 6.64 mph ∆V. The isolator speed approached 3 mph at 20 msec, at which time the vehicle had slowed by only 1 mph. Since the bumper cannot sink into the barrier this would infer that slack was developing in the string pot cable, since the bumper was only moving at about 3 mph when the vehicle is still moving toward the barrier at about 3.8 mph. However, there was an airgap between the reinforcement bar and the plastic bumper cover that undoubtedly compressed and there may have also been some dynamic flexion of the reinforcement bar.

Because the amount of dynamic bumper cover compression and/or reinforcement bar flexion had not been was not measured, the string pot retraction during one of the highest speed barrier impacts was compared to the retraction during the release of an unattached cable end, which produced an unrestricted string pot retraction. As shown in Figure 12, this demonstration shows that the string pot retracts faster than what was measured in the impact testing. As such, in addition to appropriately representing the maximum isolator compression, the string pot output also appears to appropriately represent the dynamic compression and extension of the isolator. Future work could

Collision Magazine - Volume 11 Issue 2 81

1/23/2017 2:01:56 PM

© 2016, Trimble Inc. All rights reserved. PN GEO-117 (11/16)

For crash and crime scenes.

For security management.

One powerful solution.

There’s now a simple, powerful end-to-end solution for collecting, documenting and analyzing the data at crime and crash scenes. And for designing reliable security systems that boost safety and minimize risk.

Trimble Forensics—formerly Visual Statement—is the new name for a complete, end-to-end hardware and software solution designed specifically to help law enforcement and private sector agencies handle the toughest information-collection challenges. And it’s not only backed by highly trained Trimble experts who are committed to the law-enforcement mission, it comes with unparalleled service, training and local support.

Visual Statement is now T TRANSFORMING THE WAY THE WORLD WORKS

GE0-120_Trimble_Forensics_Transition_Print_2pp_AD_USL_Collisions_1116_final.indd All Pages Volume 11 Issue 2.indd 82

1/23/2017 2:01:56 PM

CRASH RECONSTRUCTION Clear the crash. Then get the answers.

Trimble Forensics solutions quickly collect the data you need, enabling you to clear crash scenes quickly, reducing disruption, costs, and the risk to personnel. Officers can complete the important steps in accident reconstruction and analyze back in the office.

CRIME SCENE INVESTIGATION Fast help at the scene of the crime.

At the heart of Trimble Forensics are reliable, sophisticated total stations, scanners, and data collectors built to capture critical, detailed information in virtually any scene or environment. Combined with Trimble Forensics field and office software, our hardware streamlines the process of capturing accurate data in the field. The end result: information that will hold up under scrutiny in a court of law.

SECURITY PLANNING The tools to plan security. For any event.

When it comes to security of events, Trimble Forensics offers the tools to effectively plan for the unexpected.

From detailed floor plans of buildings to 3D tactical views of the event location, you can give your security personnel exactly what they need.

Learn more at forensics.trimble.com

w Trimble Forensics Volume 11 Issue 2.indd 83

18/11/2016 1:14:11 p.m.

1/23/2017 2:01:56 PM

Figure 12: String pot retraction in a barrier impact and in an unattached cable end release. include additional verification that no slack is developing in the string pot cable via high-speed video.

C

onclusions: • The strong correlation between impact speed and ∆V for the BMW front-to-barrier impacts allowed precise control of the subsequent rearto-barrier exposures for the human driver.

•

The Mercury’s front bumper isolator compression versus ∆V behavior in vehicle-to-vehicle impacts was indistinguishable from that in barrier impacts.

•

Similar front and rear bumper isolator compression∆V relationships were observed for the BMW and the Volkswagen.

•

There is not any evidence from the current test series that the BMW manufacturer trend has changed significantly from the early 1980’s through 2001 and the previously identified manufacturer trend is still appropriate to use in reconstruction of low speed bumper impacts.

•

There is not any evidence from the current test series that the Volkswagen manufacturer trend has changed significantly from mid-1970’s through 2005 and the previously identified manufacturer trend is still appropriate to use in reconstruction of low speed bumper impacts.

84

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 84

•

The Mercury’s front bumper exhibited a softer response that corresponds with published data of older Fords, while the Mercury’s rear bumper exhibited a stiffer response that corresponds with published data for newer Fords. Thus, as observed by other researchers, there is evidence of a possible shift in Ford’s manufacturer trend toward stiffer bumpers starting in the early 1990’s.

•

For low speed bumper impacts involving the Ford family of automobiles newer than the early 1990’s the possible shift in the manufacturer trend towards stiffer bumpers should be considered. That is, a determination should be made whether it is appropriate to apply the original softer manufacturer trend or the newer stiffer manufacturer trend for reconstructing low speed bumper impacts involving these vehicles.

•

Given the strong correlations between average isolator compression and ∆V, using the evidence of piston isolator compression in the reconstruction of low speed bumper impacts remains an appropriate and reliable low speed accident reconstruction method.

•

String potentiometers aligned with the bumper piston isolator appropriately represented the maximum isolator compression as well the dynamic isolator compression and expansion.

www.collisionpublishing.com

1/23/2017 2:01:57 PM

R

eferences: 1. Anderson, et al., “Effect of Braking on Human Occupant and Vehicle Kinematics in Low Speed Rear-End Collisions,” SAE Technical Paper Series No. 980298, February 1998.

2. Bailey et al., “Data and Methods for Estimating the Severity of Minor Impacts,” SAE Technical Paper Series No. 950352, February/March 1995.

3. King, et al., “Automobile Bumper Behavior in LowSpeed Impacts,” SAE Technical Paper Series No. 930211, March 1993.

4. King, et al., “Determination of Bumper Characteristics Using Prototype Moving Barriers,” SAE Technical Paper Series No. 970956, February 1997.

5. King, et al., “Comparison of Testing of Bumper Isolators,” SAE Technical Paper Series No. 1999-01-0096, March 1999.

6. Malmsbury, et al., “Damage and/or Impact Absorber (Isolator) Movements Observed in Low Speed Crash Tests Involving Ford Escorts,” SAE Technical Paper Series No. 940912, March 1994.

Volume 11 Issue 2.indd 85

7. Ruberte, et al., “Low Velocity Piston Isolator Testing of a Ford Crown Victoria,” Collision Magazine, Volume 9, Issue 2, Fall 2014.

8. Siegmund, et al., “Characteristics of Specific Automobile Bumpers in Low-Velocity Impacts,” SAE Technical Paper Series No. 940916, February/March 1994.

9. Siegmund, et al., “Using Barrier Impact Data to Determine Speed Change in Aligned, Low-Speed Vehicle-to-Vehicle Collisions,” SAE Technical Paper Series No. 960887, February 1996.

10. Szabo and Welcher, “Dynamics of Low Speed Crash Tests with Energy Absorbing Bumpers,” SAE Technical Paper Series No. 921573, August 1992.

11. West, et al., “Low Speed Rear-End Collision Testing Using Human Subjects,” Accident Reconstruction Journal, May/June 1993.

A

cknowledgements: The authors would like to acknowledge Biodynamics Engineering, Inc. for the use of their barrier, as well as Eric Miller, Matt Ivory, Alissa Tanner, Chris Furbish and Robert L. Anderson for their assistance.

1/23/2017 2:01:57 PM

ed w e i v e R Peer

The Longevity of Scene Evidence from a Rollover A Case Study

I

ntroduction

As with most any type of crash, reconstructing a rollover crash often begins with the development of a scaled diagram depicting physical evidence from the crash. This evidence will often include tire marks on the roadway, tire furrows in an off-road surface, gouges and scrapes, wheel landing marks on asphalt or in the dirt, vehicle-to- ground impact marks on an off-road surface, vehicle window glass deposits, and other debris from the vehicle. This evidence enables the reconstructionist to determine the location, speed, and timing of the loss of control, the trip, and the rollover. Analysis of the motion of occupants who were ejected from the vehicle can also be carried out. Locating evidence on a scaled diagram, of course, depends on the evidence being documented in some form (photographs or measurements) before that evidence disappears.

This evidence documentation may be completed either by investigators on scene immediately after the crash or by later investigators who may come to the site long after the crash occurred. Of course, some types of evidence deteriorate more quickly than others and, at some point, many pieces of evidence will become unavailable for direct documentation by later investigators.

86

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 86

www.collisionpublishing.com

1/23/2017 2:01:58 PM

The longevity of tire marks, off-road furrows, and impact marks on an off-road surface depends on many factors, including the initial prominence of the marks, the type of asphalt or off-road surface, the traffic volume through the area, the weather, and the frequency of roadway and off-roadway maintenance and modification. Gouges and scrapes in an asphalt surface can be long-lasting as long as a road is not repaved or modified in a way that removes them.

Vehicle window glass deposits can be a long lasting piece of evidence from a rollover. Przybyla [2015] studied the longevity of glass debris fields over a period of two years. He placed glass debris fields at three locations in the United States where they would be subjected to the influence of slope, weather, mowing operations, and other factors that could result in movement of the debris field. Each area of glass debris was periodically mapped to determine the degree to which the glass debris had moved or changed in size.

Nathan Rose and Neal Carter Kineticorp,LLC www.collisionpublishing.com

Volume 11 Issue 2.indd 87

Collision Magazine - Volume 11 Issue 2 87

1/23/2017 2:01:59 PM

Przybyla documented expansion and movement of each glass debris field over the study period, but these were small, and he ultimately concluded that “roadside glass debris fields remain generally static over time when subjected to natural weather patterns, normal landscape growth, and roadside mowing.”

Given their longevity, glass deposits are often valuable evidence marking the path of the vehicle during the roll phase and of roll orientations of the vehicle at discrete points along the roll path. In addition to aiding in the reconstruction of vehicle positions, the location of glass debris can also be essential in the analysis of when during the roll phase an occupant was ejected or from which seating position that occupant was ejected [Funk, 2007 and 2008].

In order for an occupant to be ejected out of a particular window of the vehicle, the glass of that window must first break out (assuming the window was initially rolled up).

Thus, knowing where along the roll path a particular window broke out will help place limits on when during the roll phase an occupant could be ejected out of that window opening.

To further explore and document, the longevity and deterioration of evidence from rollover crashes, this article examines the longevity of the physical evidence from one particular rollover crash by documenting the crash site six years and several months after the crash occurred. This particular rollover crash was a steering-induced crash test reported by Asay in 2010. This test was one in a series of 8 tests reported by Asay. These tests were conducted on a two-lane highway in the west desert of Utah (see Figure 1).

The roadway was a chip-seal coated surface with graded shoulders along both sides. The off-road terrain surrounding this highway was open desert with sagebrush and sparse grass. Each vehicle was shifted into neutral, towed to highway speed, then released. An automated steering controller

Figure 1: Rural Highway in Utah where Asay’s Tests Were Conducted 88

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 88

was used to input a sharp left steering input of approximately ¼ turn of the steering wheel, then one second later, this initial steering input was followed by a severe steering input to the right. This sequence of steering inputs led to a rollover occurring in 6 of the 8 tests. Asay documented the physical evidence from each of these tests using a Leica total station. He produced an evidence diagram for each test in a CAD package using the survey data. Asay provided us with his survey data and evidence diagrams.

T

est #6 – 2002 Ford Explorer This article examines the longevity of the physical evidence from one of Asay’s tests that resulted in a rollover – Test #6 from his 2010 article. This test, which was run on August 19, 2009, utilized a 2002 Ford Explorer. The vehicle was released at a speed of 70 mph (113 kph) and was steered first to the left approximately ¼ of a turn of the steering wheel and then aggressively back to the right until the rightward steering stop was reached.

The vehicle exited the left side of the roadway in a clockwise yaw with its driver’s side leading. After furrowing into the dirt and gravel and yawing past a 90-degree slip angle, the Ford began rolling over and completed 2 rolls in a distance of approximately 79 feet (24.1 meters).

Asay’s survey data was used in conjunction with an aerial photograph to create the evidence diagram included in Figure 2. This diagram shows tire marks, furrows, ground impact marks, and debris deposited by the vehicle during the loss of control, the trip, and the roll phases of this test.

Figure 3 again depicts the scene evidence from this test, but this diagram includes vehicle positions that show the vehicle motion evidenced by the yaw marks and furrows.

In this case, the vehicle deposited tire marks on the roadway while it was yawing in a counterclockwise fashion following the leftward steering input and then in a clockwise fashion following the subsequent severe rightward steering input. The vehicle then entered the dirt shoulder while yawing in a clockwise fashion with its driver’s side leading and deposited furrows in the dirt. The furrows in the dirt off the left side of the roadway are identified with orange flags in the photographs of Figures 4, 5, and 6. The rest position of the vehicle is also visible in these photographs.

The photographs of Figures 7 and 8, taken immediately after the test was run, depict ground impact marks and debris from the roll phase of this test. During the posttest documentation, these ground impact marks and debris were marked with white spray paint and this paint is visible in these photographs. These pieces of evidence, which are included in the diagram of Figure 2, would enable the reconstructionist to infer the path of the vehicle from the end of the trip phase to the rest position. Often though,

www.collisionpublishing.com

1/23/2017 2:01:59 PM

Figure 2: Evidence Diagram for Test #6 from Asay [2010]

Figure 3: Tire Mark and Furrow Diagram for Test #6 from Asay [2010]

Figure 4: Photograph of Tire Furrows from Ford Explorer

www.collisionpublishing.com

Volume 11 Issue 2.indd 89

Figure 5: Photograph of Tire Furrows from Ford Explorer

Collision Magazine - Volume 11 Issue 2 89

1/23/2017 2:01:59 PM

Figure 6: Photograph of Tire Furrows from Ford Explorer

Figure 7: Photograph of Ground Impact Mark and Debris from Ford Explorer

for real-world crashes, on-scene investigators do not document the locations of ground impact marks and debris.

Impact marks will often deteriorate before later investigators are able to document them and vehicle debris will often be moved or removed during cleanup of the scene.

Despite the significant alterations to the roadway and shoulder, we were able to locate three areas where vehicle window glass had been deposited on the ground (Figure 10), identify vegetation that had been pressed down by the vehicle, and determine the rest position of the vehicle based on the locations of individual bushes that were visible in the original photographs from the test and still visible during our inspection.

S

ite Inspection We visited the site of this test on December 9, 2015, six years and several months after the test was run.

Figure 9 depicts the site at the time of our inspection. As this photograph shows, the asphalt of the highway had been removed and only a dirt road remained. Clearly, this means that the tire marks on the roadway were no longer present at the time of this inspection. The grading of the shoulders appeared similar to what it was at the time of the test, but it was also apparent that the shoulders had been regraded since the test, likely during the removal of the asphalt. Chunks of asphalt from the original highway surface were present and mixed in with the soil on the shoulders. Given these modifications, the furrows in the off-road surface had also been obliterated and were no longer present at the time of our inspection.

These alterations to the test site were more obvious than many scene alterations that we have encountered in our accident reconstruction practice. Collisions investigators will often need to consider subtler changes to crash sites.

For instance, we have often encountered altered roadway delineations and changes to signage. Such changes will be relevant if a reconstructionist intends to use techniques of photogrammetry in the course their scene and evidence mapping, since such changes can prevent, or at least make more difficult, the use of stripping and signage as landmarks for evidence analysis.

90

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 90

The photograph of Figure 11 shows a close-up view of glass fragments located in the area designated with the Number 2 in Figure 10. This glass exhibited dark tinting and many of the fragments were painted red on one side. The glass had a thickness of approximately 0.185 inches (4.7 millimeters). The photographs of Figure 12 depict the test vehicle prior to the test. As these photographs show, prior to running this test, the left side windows of the test vehicle were painted green, the right side windows orange, and the rear hatch glass red. Thus, the red paint on the glass frag-

Figure 8: Photograph of Ground Impact Mark from Ford Explorer

www.collisionpublishing.com

1/23/2017 2:02:00 PM

Figure 9: Site of Test #6 from Asay [2010] 6 Years and Several Months After the Test

Figure 11: Close-Up of Glass Debris from Area 2 Identified in Figure 10, Site of Test #6 from Asay [2010], 6 Years and Several Months After the Test

Figure 10 : Glass Debris Fields Identified with Orange Flags, Site of Test #6 from Asay [2010], 6 Years and Several Months After the Test

Figure 12: Photographs of the Test Vehicle Prior to the Test with Left Side Windows Painted Green, Right Side Windows Painted Orange, and the Rear Hatch Painted Red [Photographs Provided by Alan Asay] www.collisionpublishing.com

Volume 11 Issue 2.indd 91

Collision Magazine - Volume 11 Issue 2 91

1/23/2017 2:02:00 PM

ments in Area 2 identify this glass as being from the rear hatch glass of the test vehicle.

and separated from the vehicle. These observations are consistent with the glass deposits we found at the site.

The photograph of Figure 13 depicts a glass sample we collected from the area identified with a Number 1 in Figure 10. This glass exhibited dark tinting, a thickness of approximately 0.150 inches (3.8 millimeters), and some of the fragments had green paint on them. This green paint is an indication that this glass was from a left side window of the test vehicle. Based on the thickness, this glass would be distinguishable from the rear hatch glass even without this paint. In real-world cases, where the windows clearly would not be painted different colors, other indications such as whether they are tinted or not or whether or not they have defogger elements embedded in the glass would help the investigator determine from which window the glass deposits came.

In addition to the glass deposits identified during our inspection, the locations of individual bushes that were visible in the original photographs from the test and that were still visible during this inspection made the rest position of the vehicle identifiable. Figure 16 is the same as the photograph of Figure 14. However, in this instance of this photograph, four specific bushes and an open dirt area around those bushes have been identified. These same four bushes and the open dirt area have also been identified in the image of Figure 17, which is one frame of a video take from an unmanned aerial vehicle (UAV) at the time our inspection.

The glass in the area designated with the Number 1 in Figure 10 was clear tempered glass. This glass did not fit with any of the windows on the Ford Explorer from this test, all of which were painted. This glass is likely from one of Asay’s 2009 tests, which were run in the same area, and in which the vehicle windows were not painted [Asay, 2009].

Figures 14 and 15 show the test vehicle at rest following the subject test. These photographs reveal that during this rollover, most of the window glass on the driver’s side of the vehicle broke (painted green) and separated from the vehicle while the passenger’s side window glass remained intact (painted orange). The rear window glass also broke

Figure 18 is the same photograph as Figure 15. However, on this instance of the photograph, bushes #3 and #4 have again been identified as have two additional bushes. All six of the bushes identified in Figures 16 and 18 have also been identified in Figure 19, which is an aerial photograph from our inspection, again taken with a camera attached to a UAV. These figures demonstrate that, using vegetation that was still in place 6 years and several months after the test, the rest position of the vehicle could be determined.

D

iscussion and Conclusions The alterations to the roadway and shoulder in this particular instance presented us with an interesting case study related to the longevity of particular pieces of scene evidence. The tire marks on the roadway

Figure 13: Close-Up of a Glass Sample Collected from Area 1 Identified in Figure 10, Site of Test #6 from Asay [2010], 6 Years and Several Months After the Test 92

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 92

www.collisionpublishing.com

1/23/2017 2:02:00 PM

Figure 14: Photograph Showing that Most of the Driver’s Side Window Glass on the Explorer Broke or Separated from the Vehicle During the Rollover

Figure 15: Photograph Showing that the Passenger’s Side Window Glass on the Explorer Remained Intact After the Rollover

Figure 16: Photograph Showing Bushes Surrounding the Vehicle Rest Position

Figure 18: Photograph Showing Bushes Surrounding the Vehicle Rest Position

Figure 17: One Frame of Video from UAV Footage Showing the Bushes of Figure 2.14, 6 Years and Several Months After the Test www.collisionpublishing.com

Volume 11 Issue 2.indd 93

Collision Magazine - Volume 11 Issue 2 93

1/23/2017 2:02:00 PM

Figure 19: One Frame of Video from UAV Footage Showing the Bushes of Figures 2.14 and 2.16, 6 Years and Several Months After the Test and furrows on the shoulder from this test were no longer present at the time of our inspection due to significant roadway modification. However, we were able to locate two deposits of window glass on the ground that could be tied to this crash, identify vegetation that had been pressed down by the vehicle, and to determine the rest position of the vehicle based on the locations of individual bushes that were visible in the original photographs from the test and still visible during our inspection.

These findings are likely, to some degree, unique to this particular location, with its low rain fall and minimal changes to the vegetation. Over the course of more than 6 years, it is of course possible that more significant changes would occur to vegetation in a less arid area. In areas of the United States with more significant annual rainfall and a warmer climate, it is possible that glass deposits could get buried or obscured by long grass. It is also possible that other sites would exhibit less significant roadway changes even over a 6-year time frame. Regardless, this case study shows that when investigating and reconstructing a rollover crash, a site inspection may yield useful evidence for the reconstruction, even more than 6 years after the crash.

94

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 94

R 2.

3.

4.

5.

eferences 1. Asay, A. and Woolley, R., “Rollover Testing on an Actual Highway,” SAE Int. J. Passeng.

Cars – Mech. Syst. 2(1):1237- 1267, 2009, doi:10.4271/2009-01-1544.

Asay, A. and Woolley, R., “Rollover Testing of Sport Utility Vehicles (SUVs) on an Actual Highway,” SAE Technical Paper 2010-01-0521, 2010, doi:10.4271/2010-01-0521.

Funk, J. and Luepke, P., “Trajectory Model of Occupants Ejected in Rollover Crashes,” SAE Technical Paper 2007-01- 0742, 2007, doi:10.4271/2007-010742.

Funk, J., Beauchamp, G., Rose, N., Fenton, S. et al., “Occupant Ejection Trajectories in Rollover Crashes: Full- Scale Testing and Real World Cases,” SAE Int. J. Passeng. Cars - Mech. Syst. 1(1):43-54, 2009, doi:10.4271/2008-01-0166.

Przybyla, J., Jupe, J., Rush, T., and Keller, R., “Glass Debris Field Longevity for Rollover Accident Reconstruction,” SAE Technical Paper 2015-01-1427, 2015, doi:10.4271/2015-01- 1427.

www.collisionpublishing.com

1/23/2017 2:02:01 PM

www.collisionpublishing.com

Volume 11 Issue 2.indd 95

Collision Magazine - Volume 11 Issue 2 95

1/23/2017 2:02:01 PM

ial r o t Edi

ed w e i Rev

Motorycycle Crash Testing: Advanced Boot Camp Was Born

Michael DiTallo, Thomas Green, Wes Grimes, Ken Salisbury, Brent Munyon, Tom Lawson, James Whelan, Eric Moody, Kevin Vosburgh, and Eric Paul 96

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 96

www.collisionpublishing.com

1/23/2017 2:02:02 PM

W

e would like to take this opportunity to introduce you to the 2016 ARC CSI Motorcycle Crash Team by providing an article that will briefly discuss the testing that has been conducted, member’s experiences and future plans. Rusty Haight and Scott Baker opened a door to allow Michael DiTallo to be a team leader conducting motorcycle crash testing at the 2016 ARC CSI conference. This was an incredible opportunity to form a team, secure motorcycles, procure equipment and develop and conduct crash testing. Rusty and Scott allowed Michael to pick his own team and create scientific motorcycle crash tests. We had one goal, quality first over quantity. Michael, working closely with Harley Davison, was able to secure 5 motorcycles for the ARC CSI, however this was only the beginning.

Photograph 1: Harley Davidson Inc., Donated Motorcycles The wish list of potential people joining the team was long as it had many experienced individuals covering many disciplines in our collective fields. The team was formed with 22 individuals consisting of the following members: Thomas Green, Kevin Vosburgh, Eric Paul, William Redfairn, Thomas Lawson, Wesley Grimes, James Whelan, Tim Reust, Jim Morgan, Kelley Adamson, Eric Moody, Rusty Haight, Brent Munyon, Billy Cox, Bob Anderson, David Hallman, Michael Rosenfield, Mike Merolli, Jim Dunlap, Ken Salisbury, Sean Haight and Team Leader Michael DiTallo.

Photograph 2: 2016 ARC CSI Motorcycle Crash Team Group Photograph www.collisionpublishing.com

Volume 11 Issue 2.indd 97

Collision Magazine - Volume 11 Issue 2 97

1/23/2017 2:02:04 PM

Editor's Note In the run-up to the 2016 ARC-CSI Conference, Mike Ditallo and his team from Dynamic Safety contacted the Crash Conference organizers and Boot Camp Team instructors to develop a plan to undertake the largest set of motorcycle crash tests and related research ever conducted. Ditallo arranged for 5 new Harley Davidson motorcycles to be donated to the effort as well as a variety of other bikes or various types and configurations for the testing. Crash tests into cars, SUVs and barriers, drop tests and more make up the total body of research. Some of the tests were part of the crash test day series at the 2016 ARCCSI crash conference and data from those is part of the data set distributed at the conference and included on the DVD with this issue of Collision.

The planning and coordination of the tests, instrumentation (for example the associated research done on the SlamStick accelerometer units as described in another article in this issue of Collision) and logistics of this undertaking evolved into a coordinated effort between the Dynamic Safety team and the ARC-CSI Boot Camp/Crash Team and is laid out in the accompanying article as the "advanced boot camp" concept.

The idea is to take experienced researchers with a research plan or concept and work with them, through the ARC-CSI Crash Conference team and provide an opportunity for that research project. of course, just conducting the testing isn't enough, there's also the idea that research is useful if published so others can benefit from it. The tie-in with Collision Magazine is a natural progression in terms of that larger concept of research and dissemination of the resulting information. In this issue, we find the first two of a larger series of articles which will flow from the largest motorcycle testing undertaking of its kind ever. The SlamStick research is an example of part of the larger project which goes beyond the motorcycle research. Future issues of Collision will include additional peer reviewed articles detailing the crash tests, drop tests and additional motorcycle data compiled. The accompanying article here describes the process and is the introduction to the larger project going forward.

98

Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 98

Now that the team was formed the real work began. Kevin Vosburgh and Michael DiTallo had been working on a new motorcycle delivery system. It was complicated by the fact that we wanted a safety control feature to be a part of any developed delivery system. The safety control ensured that the motorcycle would not be released from the delivery system unless one of the test conductors positively actuated the motorcycle release. Bill Redfairn, team supporter Jim Byrne and Michael DiTallo worked on locating additional motorcycles for the testing. We collected 26 motorcycles of various styles and makes. The team developed goals and the initial motorcycle crash test plans for ARC CSI. Members Tom Green, David Hallman, Tim Reust, Kevin Vosburgh, Billy Cox, Bob Anderson, Kelly Adamson, Bill Redfairn, team supporters Louis Peck, Jim Byrne and Michael DiTallo worked on equipment and preparation. One of the caveats for this opportunity was that the team would have to be self-sufficient with all needed equipment.

Then, in the days leading up to the event, the motorcycles were prepared, which included photographs, inspections, weights, center of gravity data collection and 3D scans. We arrived early in Las Vegas several days prior to the first scheduled crash test to do final preparations. We removed fuel, retrofitted a lateral bar onto the motorcycle frames that was used for the final delivery system, removed seats and batteries as needed, made sure the tires were at recommended air pressures and also replaced and fixed tires if they would not hold air. Table 1 contains a list of the motorcycles assembled in this effort.

Photograph 3: Motorcycle Preparation – Center of Mass Data Collection Example Here are some comments, in their own words, from a few members of the 2016 ARC CSI Motorcycle Crash Team.

Brent Munyon: While I had some crash testing experience; ARC-CSI Bootcamp and a dozen or so private tests, my personal view was that this limited experience was substantially less than my cohorts in this www.collisionpublishing.com

1/23/2017 2:02:05 PM

MC Color

Brand

Model Name

VIN

YEAR

Sportster

Model Number XL1200C

A

Red

B

Black

C

White

D

Black

E

Blue

F G H I J K L M N O P Q R S T U V W X Y Z

Black Red Black Black Maroon Red Yellow White/Red Red/Black Purple Black White/Red/Blue Black Blue Grey Black Black Black Silver Orange Red

Harley Davidson Harley Davidson Harley Davidson Harley Davidson Harley Davidson Buell Honda Kawasaki Suzuki Yamaha Honda Kawasaki Kawasaki Kawasaki Kawasaki Honda Honda Honda Yamaha Yamaha Yamaha Yamaha Suzuki Jonway Znen Kawasaki

1HD1CGP186K800057

2006

Fat Bob

FXDF

1HD1GY411AC300033

2010

Heritage Softail Road King

FLSTC

1HD1BW518BB010087

2011

FLHRC

SHD1FR4CXBB600079

2011

Electra Glide

FLHTCU

1HD1FC411BB600023

2011

Thunderbolt (Dirt Bike) Ninja 500R

S3

4MZRSllJOW3000257 JH2MD0105BM300463 EX500-D JKAEXVD171A065404 GZ250 VTINJ48A3A2100474 XV250 JYAVG04E79A009983 CBR600 JH2PC35081M203493 EX250-F JKAEXMF126DA21164 ZX900-B JKAZX2B11VA038708 VN1500-A JKBVNAA10KA003441 VN800-A JKBVNCA17TA031749 VT1100C2 1HFSC320XWA300006 XR250L JH2MD2217PM200732 CBR900RR JH2SC3307WM202405 YZF-R6 JYARJ04E2YA005812 JYARJ18EX9A002383 FZR600 JYA3UUC01PA009231 FXR600 JYA3HHE06NA040657 GS500E VTIGM51A352103042 YT50QT L8YTCKPM7CM401083 GTR 150 L5YTCKPA681228159 EN500-C JKAENVC194A185555

1998 1981 2001 2010 2009 2001 2006 1997 1989 1996 1998 1993 1998 2000 2009 1993 1993 2005 2012 2008 2004

Virago 250 Ninja 250R Ninja ZX-9R Vulcan Vulcan 800 Shadow Fireblade R6 FZ6R FXR FXR

Vulcan

Table 1: List of Motorcycles Assembled

adventure. With that in mind, my goal was to soak up anything and everything I could during this experience.

weighed, measured, photographed, gasoline removed from tanks, rigged for the delivery system and instrumented.

Upon my arrival, the first thing that struck me was the sheer number of bodies that were needed to accomplish a testing of this magnitude. Twenty or so people were quickly divided into subgroups to accomplish the very nonglamourous pre-crash tasks necessary to accomplish a successful test. The motorcycles had to be cataloged, marked,

My position on the motorcycle crash test assembly line, working along with Kelley Adamson and Eric Moody, was to predetermine where and how the instrumentation package would be attached to each motorcycle. This singular instrumentation package included an accelerometer with six degrees of freedom, two Slam Sticks and a V-Box all

www.collisionpublishing.com

Volume 11 Issue 2.indd 99

Collision Magazine - Volume 11 Issue 2 99

1/23/2017 2:02:05 PM

housed in a steel box that would be mounted to the motorcycle frame. Following a crash test, this package would be removed, quickly downloaded and re-set, then installed in the next motorcycle in line for testing. With 27 different motorcycles, each presented a different set of challenges as to how the instrumentation package could be securely attached. Some of the motorcycle frames were constructed of steel while others were aluminum alloys. Frame angles also varied greatly with some being close to level while others were constructed with sharp angles. Ultimately in each case, various non critical components such as seats, batteries were removed and the instrumentation package was pre-mounted using substantial self-tapping screws directly into the motorcycles frame rails. With the location determined and marked the instrumentation package was removed until just prior to the actual crash test.

The team, a diverse group of twenty or so mostly TypeA personalities, worked well together. You can imagine with that number of people there was an equal number of ideas on how any particular task could be and should be accomplished. The task specific division of labor kept individuals busy and focused in a particular area. This, and the general knowledge that giving unsolicited advice to a guy with bloody knuckles from turning wrenches and sweat dripping in his eyes, may not be the best idea. When needed, advice was sought and given and the assembly line worked quite well.

Few test runs were completed on the day prior to the conference to work out the bugs in the system. The crash tests went quite smoothly. The one consideration that has to be taken into account when testing at the ARC-CSI conference is time. You have to be ready when it is your turn to crash a vehicle. The division of labor was modified from an assembly line model to a NASCAR pit crew mentality.

My particular assignment was to mount the instrumentation package on the motorcycle to be tested. Then following the crash, immediately remove the steel box and deliver it to the individuals assigned to download the data from each instrument. Upon completion of the download and re-set of the four instruments, the package would then be re-installed on the next motorcycle to be tested.

Ultimately the completion of a test day of this magnitude, 27 crash tests on a tight schedule in front of an audience, was accomplished by having enough qualified individuals with a well-managed plan and proper division of tasks.

Post event testing of motorcycle drag factors was accomplished in a similar fashion. We were divided into three teams as were the previously crashed motorcycles. Approximately one-third of the motorcycles were tested from an upright orientation, being pulled to speed trailing a vehicle 100 Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 100

that was holding the front wheel in an air operated clamp.

The second third were dropped laterally from the rear lift gate of a moving truck. The final third, and my assigned group, conducted tradition drag tests utilizing a tow truck and a load cell.

In total, 10 motorcycles were each drag tested four times, twice with the front wheel leading and twice with the motorcycle traveling laterally being pulled by both wheels.

James Whelan: The impacts of the motorcycles into barriers or other vehicles was documented with the use of a digital still camera and digital video camera. Both cameras were mounted on a single tripod via a custom fixture. The cameras were placed just inside the protective barriers on the crash pad and in plane with the impact surface. The camera operator stood behind the protective barrier outside of the crash pad.

The digital still camera was a model FH25 and F1 Casio cameras. It was set to a focal length of 135 mm and recorded 60 still frames per second (fps). The camera was controlled by a wired remote. The operator had to estimate when the motorcycle was less than 15 seconds from impact and take into account the time lag for the camera to operate after the button was depressed. Experience showed that this was a difficult task that required practice. The first approach was to identify known distances on the crash pad; calculate the expected distances and times that the motorcycle was from impact based on planned test speeds; and then depress the button on the wired remote with sufficient time such that the impact was captured or documented by the camera. Due to the numerous variables and dynamics involved in crashing an upright motorcycle into a stationary barrier or moving vehicle, the actual delivery speed of the motorcycle did not always match the planned test speed. The said approach did not always have the desired affect and the documentation of the impact could be missed. An experienced camera operator with a good eye-estimate of the motorcycles speed and feel of the camera equipment was able to effectively catch each impact by using intuition. This proved to be a reliable process.

The video camera was a Vixia Canon HFM50 and Casio FHS25 and F1 Cameras in video mode. It was set to a focal length of 128 mm at a resolution of 1080p and recorded at 60 fps. The video camera was controlled by its record button and required less skill than the digital still camera. The operator started the recording when the incident test motorcycle was traveling on the crash pad. The operator stopped recording the video after the incident test motorcycle came to final rest.

www.collisionpublishing.com

1/23/2017 2:02:05 PM

Ken Salisbury: So how do we ensure thorough understanding? With all the various forms of learning out there, how do we choose which learning style is most effective? I can learn to ride a bicycle by reading a book, but I would rather have the experience of “sitting in the saddle,” before hammering into traffic. Simply put, a hands-on approach to learning in collision reconstruction is incomparable.

I understand much can be learned from the work of others in an academic environment. However, I learned more participating in a recent field study than I have in many hours of classroom learning. Experiencing the data collection process was beyond measure of what I have read of past studies. The study comprised motorcycle fork deformation analysis, motorcycle sliding drag factors, and mechanical braking dynamics.

Our group analyzed the largest known sample group of motorcycles ever tested. The sample group included 26 motorcycles with various forms of make, model, and design. 22 of the motorcycles were used in impact testing with fixed barriers, stopped vehicles, and moving vehicles.

Measurements, inspections, and scans were recorded before and after the impact testing. The predominate focus area of the testing was speed estimates from motorcycle relative to fork deformation.

This type of testing has been completed and analyzed in the past, but there have been several changes in motorcycle engineering and design since the previous study. One of the most significant changes in motorcycle engineering is the design in forks. Most modern motorcycles are equipped with inverted forks. Inverted forks have an increased rigidity because they are affixed by the slider. The slider is the larger diameter portion of the fork compared to the stanchion and typically comprises more of the overall length of an inverted fork. Conventional forks are affixed by the stanchion. There is typically more overlap in the portion of stanchion that protrudes into the slider when compared to conventional forks. The sample group of our study consisted of motorcycles with inverted forks and conventional forks. The results yielded in our study include the significant change in motorcycle engineering.

Our group used three different methods to analyze the deceleration values of motorcycles sliding on their sides.

One method was dragging a sample group of motorcycles on their sides with a load cell to measure the horizontal force required for the slide. The horizontal force was recorded and divided by the respective motorcycle weight to obtain the deceleration value (drag factor).

Another method used was a “drop test” of a sample group of motorcycles from the lift gate on the back of a moving

truck at a known speed. Several recording devices were used for redundancy in data collection.

The final method used was new in comparison to past motorcycle testing. A sample group of motorcycles was towed upright behind a moving truck with the use of a clamp on the front wheel of the motorcycle. A pneumatic valve released the pressure to the clamp allowing the motorcycle to overturn and slide parallel to the velocity vector of its center of mass. As with the previous method, several recording devices were used for redundancy in data collection.

There are many unique and specific areas of the collision reconstruction field. As a Detective with the Las Vegas Metropolitan Police Department Fatal Traffic Collision Detail, most of my case work involves pedestrians and motorcyclists. The compilation of data from this study was directly relative to a large portion of the work I am assigned on a yearly basis.

Almost all of the academic experience I have gained in collision reconstruction has been through Northwestern University Center for Public Safety. One course in particular was the Motorcycle Crash Reconstruction course. After attending the course, my ability to explain the application of drag factors in motorcycle collisions was proficient. After participating in this study, I am able to explain the application in greater detail and attest to the knowledge with first-hand experience.

In criminal cases, the ability to convey information to a jury with credibility is crucial. Being able to say you developed your opinion of the collision dynamics based on principles of physics accepted in the crash reconstruction community with scientific certainty demonstrates credibility. Being able to say you were a participant in the collection of the data used in the foundation of those principles adds superior credibility.

One of the predominate aspects of motorcycle crash reconstruction involves the deceleration value the motorcycle achieved before and after impact. Analyzing the speed of a motorcycle through the sequence of events can have varying degrees of difficulty due to variables unknown to the investigator. For example, “Did the rider apply the front brake, rear brake, or a combination of the two?” If the brakes were applied, “What percentage of braking was used?” If the motorcycle overturned, “What deceleration value did the motorcycle achieve?” Answers to these questions, within a degree of scientific certainty, can help the investigator analyze the speed of the motorcycle more accurately. In almost every incident, the speed of a motorcycle is imperative because the speed on approach to the collision area may relinquish the rider’s right-of-way to

www.collisionpublishing.com

Volume 11 Issue 2.indd 101

Collision Magazine - Volume 11 Issue 2 101

1/23/2017 2:02:05 PM

the occupied lane or diminish the possibility of a collision avoidance technique. In many cases I have investigated the motorcyclist could have avoided a collision by simply applying the brakes and not overturning before impact.

The closing speed is commonly so great, the rider implements “panic braking” resulting in overturn and / or longer deceleration distances.

One of the key things that we learned is that the surface upon which the run-up for the motorcycle takes place on is a critical factor for determining the path and stability of the motorcycle post-launch. Whenever possible, I would recommend performing some test run-ups at the actual crash tests location prior to any crash tests being performed.

While most facets of collision reconstruction are objective, a common misnomer is the subjective application of input variables within mathematical formulas. For instance, an unlikely deceleration value used in a basic motion equation has the potential for yielding an incorrect velocity for a motorcycle.

Tom Lawson: I was excited and honored to be asked to join the 2016 Motorcycle Crash Team for the ARC-CSI Crash Conference. I had attended the conference three times over the past seven years and felt the live crash testing and data review reinforced the principles taught in the various investigation and reconstruction classes I had taken during my seventeen years as a law enforcement reconstructionist. During that time, I had the opportunity to experience other live testing while taking The C.R.U.S.H. and Auto-Ped courses with Rusty (Haight) while he was still with Texas A&M and had taken Northwestern University’s Motorcycle and Heavy Vehicle Courses. All four of these gave me practical application and data collection experience; but as students, the tests were defined for us and instrumentation was extremely limited due to the era of the testing and/or the availability of instruments. Here was an opportunity to participate in the full testing cycle; from planning, preparation, execution, documentation, analysis, and reporting.

In almost every case, the most accurate method of analyzing deceleration rates would be testing and analysis with the respective units at the scene. However, testing at the scene may be impractical, if not impossible, in many instances. The unit involved may be too damaged for testing, resources may not be available, the testing may be too time extensive to justify the roadway closures. Due to these circumstances, law enforcement personnel are heavily reliant on data generated from these studies.

In conclusion, my knowledge and experience as a collision Reconstructionist was significantly enhanced with my participation in this study. I found the experience of handson data collection and analysis was superior to other methods of learning. The results of the data compiled in this study provide information based on quantifiable sampling with up-to-date conditions.

I have an in-depth understanding of this important study because I was a participant. The level of understanding I have has ensured I am producing accurate analyses. Seeing the development and implementation of a study of this magnitude gave me experience that cannot be duplicated in a classroom or at the scene of an actual collision.

Thomas Green: The task of performing multiple moving motorcycle crash tests has not been performed nearly as often as moving vehicle crash tests. One reason may be that the task of keeping an unmanned moving motorcycle upright and on target is not as easy as it may seem.

We used a cradle mounted to the front of a pickup truck as our propulsion and launch system. This launch system worked well during our initial launch system prove out tests on one surface (which was fairly flat and uniform).

The actual location of the multiple tests had a different surface, with pitch, grade and surface irregularities that were unknown and unforeseen. These surface irregularities forced some changes to the launch system.

102 Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 102

As with any planned event, plans rarely survive first contact with the enemy. In our case, the enemy was the delivery system used to accelerate the motorcycle from rest during the crash test. Kevin Vosburgh and Mike DiTallo had designed a jig that was attached to the front of a Dodge truck. Each motorcycle had a metal pipe welded to the frame. That pipe contacted the jig to both balance the motorcycle at rest and to accelerate the motorcycle as the truck accelerated. The weakness of the system was the flex experienced as the motorcycle and truck independently navigated the high points, low points, and surface joints of the test surface. Added to the initial ‘wobble’ experienced in the front tire of the motorcycle as it initially accelerated, the motorcycles routinely came out of the jig and fell before release. The problem was discussed over dinner at Macaroni Grill. The solution was literally sketched in crayon on the paper table cloth during the meal.

By extending the support welded to the motorcycle and modifying the contact with the jig, the delivery problem was solved and the crash testing was an incredible success.

Not only were we able to repeatedly deliver a moving motorcycle to contact with a fixed barrier, but we successfully conducted two tests of a driverless motorcycle into the side of a moving target vehicle.

www.collisionpublishing.com

1/23/2017 2:02:05 PM

For me, the data analysis I found most interesting was the comparison of the crash pulse to the high speed video. In the videos, stages of the motorcycle’s engagement with the barrier were easily defined; i.e., tire contacts barrier, wheel deforms, forks deform, forks and/or wheel contact frame, etc. Once the video timeline and the crash pulse timeline were synchronized, the peaks and valleys of the crash pulse matched to the stages of engagement between the motorcycle and the barrier. This was demonstrated during the Motorcycle Testing Summary presentation at the conference.

Wes Grimes: First off, it was fantastic to work with such a great group of people. This was a lot of work, for everyone, but I personally heard no one complaining about it.

Everyone was pleasant to work with and seemed to anxiously be involved. I saw several occasions when someone would have nothing immediate to complete on their assigned tasks and they would quickly step in to assist someone else that need help at that time. The expertise of those involved was also impressive, we could have conversations about complex concepts in a matter-of-fact fashion. Most of us in the field of accident reconstruction would probably have difficulty having such sharp, focused, conversations in our day-to-day interactions. It was challenging to keep up sometimes, but was a joy to participate in. Complex challenges were discussed and problems laid out, then decisions made that might have taken longer to arrive out without such a well-informed, technically savvy group.

The instrumentation was a challenge. One of the main requirements was to move instrumentation from one motorcycle to another in an organized and efficient manner.

Unfortunately, this may have produced situations where the easier and quicker path of mounting the instrumentation was taken, this contributed to some instrumentation becoming loose.

Overall we found that the Slam Sticks produce very noisy data, but once it is properly filtered and analyzed it seems to correlate very closely to the IST instrumentation Eric Paul: It was a great experience being able to participate in the motorcycle testing at the 2016 ARC/CSI Conference. I’ve had the opportunity to be part of other types of testing in the past, but not to the degree I had with this testing. Seeing the process from beginning to end was a great learning experience. Best of all, I was able to meet some other wonderful experts and learn from them.

Kevin Vosburgh: Months of preparation, countless hours sweat out and dollar signs flying out the door. All culminating in a medieval looking apparatus attached to the front of an 8,000 pound partially dismantle truck. Every-

The task at hand is to crash 27 or more motorcycles in a span of just a few days. The motorcycle delivery system has to operate flawlessly so that we can run dozens of crash tests in 2 days. thing that could be imagined to go wrong discussed and contemplated and agonized over. “How can we get this motorcycle up righted and traveling 40+ mph into an object… safely? ... and then do it 26 more times!”

The task at hand is to crash 27 or more motorcycles in a span of just a few days. The motorcycle delivery system has to operate flawlessly so that we can run dozens of crash tests in 2 days. I am spearheading this delivery system.

Our team looked at multiple versions of systems used by other parties and applied our ideas to create a no-compromise system.

So the day comes to load up our system and send it across the country to the testing facility. We are headed to Las Vegas Motor Speedway. With one successful low speed test-run conducted on a flat parking lot, we are confident we can run in Vegas, so off goes the truck with our load of equipment and 5 brand new Harley motorcycles.

Agony, fear, anxiety, tension and hope fill the air as we set up to run our first test in Vegas… and then it happens. The first motorcycle results in controllability issues.

Two dozen people working toward a solution. We have to

www.collisionpublishing.com

Volume 11 Issue 2.indd 103

Collision Magazine - Volume 11 Issue 2 103

1/23/2017 2:02:05 PM

Photograph 4: Instrumentation Mounting Example overcome this set back. Modifications are made, more like band aids masking the true underlying root-cause of the issue. And yet we struggle on through the first day, crashing fewer motorcycles than planned, but at least we are running.

Dejected and frustrated we call it a night and head out for dinner. At the table sit 8 worn down experts. But then something magical occurs. The Italian restaurant covers the tables in white paper sheets. Discussion were had and a pen or two come out. The white paper is now our design canvas. We discuss the physics and failures. Removed from the unrelenting sun, our minds all start to work like a finely tuned instrument. A solution is born and plans made to be constructed in the early morning before the show goes on for hundreds to watch.

We make changes to the system and our first tests runs nearly perfect. A huge sigh of relief falls over the entire team. We did it. And now, we are ready. In fact, we are so well prepared and running smoothly that we are able to keep the event running smoothly in the face of other teams set-backs and miss-timing. We keep the show going ahead of schedule!

104 Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 104

I am thankful for the opportunity to have worked with such a great group.

On Saturday May 21, 2016, the day before the main ARCCSI crash day, we started the motorcycle testing. We instrumented the first motorcycle (which is what I believed would be the hardest motorcycle for our testing system, as it had extended handle bars). We used a variety of testing instrumentation, with redundancy. We installed IST and TDAS accelerometers in target vehicles, VBOX sport and two Slam Stick accelerometers on the motorcycles (For a more detailed description of the instrumentation, see the article “Evaluation of the Slam Stick X as a Low-Cost Accelerometer and Data Acquisition System for Vehicle Crash Testing” in Collision Magazine). We used high speed cameras, HD cameras and digital video cameras mounted around the crash scene, on the motorcycles and on the test delivery vehicle. In addition, we set timing traps and utilized an aerial drone with a camera.

The crash scenes and the motorcycles were both scanned and photographed post collision. The motorcycles were inspected again post collision with notes, documentation and a specific inspection of forks and wheelbase deforma-

www.collisionpublishing.com

1/23/2017 2:02:07 PM

Scanner Image 1: Post Collision Data Collection with a Faro Scanner

Photograph 5: Barrier Crash Test With Motorcycle M

www.collisionpublishing.com

Volume 11 Issue 2.indd 105

Collision Magazine - Volume 11 Issue 2 105

1/23/2017 2:02:07 PM

Photograph 6: Barrier Crash Test With Motorcycle N

Photograph 7: 2016 ARC CSI Motorcycle Crash Team Delivery System 106 Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 106

www.collisionpublishing.com

1/23/2017 2:02:09 PM

Photograph 9: Downed Motorcycle Testing Drop Test Example

Photograph 8: Downed Motorcycle Testing Tow Rig Example tion. The same preparation and instrumentation continued on the next day, Sunday May 22nd.

Our overall test plan included a collection of frontal impacts for the motorcycles into the sides of stopped vehicles, fixed barriers and into the side of a moving vehicle. We conducted a few lower speed crash tests in Illinois to prove our delivery system before traveling to Las Vegas. The delivery system tests were conducted on flat, smooth, pavement with positive results. We knew that our Las Vegas crash test environment was not flat and entirely smooth. Thus, we encountered some issues with controllability of the test motorcycles on Saturday, day one, in Las Vegas. Our team worked hard and collectively tried to work through the issues. We made some real time changes to our mounting and delivery system on day one that helped, but still was not as good as we desired. Some of our tests resulted in better motorcycle controllability than others. Our team worked hard that day with lots of thought, effort and collaboration. When we concluded that night, we were uncertain as to how Sunday’s testing would go. A group of us went to dinner, and continued to collaborate. We used the paper tablecloth at the restaurant to sketch out our design changes and we devised

Photograph 10: Downed Motorcycle Testing Load Cell Pulley Example a small change to our delivery system that we believed would add the controllability we desired. Early Sunday morning we were off to the hardware store for additional parts. We made a change to a test motorcycle and conducted a crash test before the attendees arrived. Success!

We completed a test that was as we had hoped. Our team worked hard to implement our newly devised changes, instrument the vehicles and be ready for the ambitious test schedule that was planned for attendees. Our testing went well and we crashed the 5 Harley Davidsons motorcycles. We stayed long after the main event was over to continue conducting crash tests on the remaining motorcycles until late in the evening.

Collectively we conducted 27 instrumented crash tests, which is the largest number of motorcycle crash tests ever conducted at one time. We also returned to Las Vegas in July, 2016 and used our crash tested motorcycles to conduct drag factor testing of downed motorcycles.

www.collisionpublishing.com

Volume 11 Issue 2.indd 107

Collision Magazine - Volume 11 Issue 2 107

1/23/2017 2:02:12 PM

In future papers and/or articles we will discuss the results of these crash tests. This overview and introductive article is meant to give the reader a foundation. Within this article of Collision, we are publishing our first main article on an evaluation of the Slam Stick accelerometer. Our goals and future expected publications including papers and/or articles are as follows: 1. A comparison of testing data collected with the Slam Stick accelerometer and Vericom to determine coefficient of friction.

2. A study of motorcycle brake lag time. We utilized some of Harley Davidson motorcycles for testing and data collection during our ARC CSI testing for this study.

3. A study of the Slam Stick accelerometer on low speed motorcycle impacts. We utilized some of our testing motorcycle for data collection during our ARC CSI testing for this study.

4. A study and comparison of down motorcycle testing and resulting drag factors.

5. A publication for our testing sample of motorcycles for center of mass data.

6. A publication for our testing sample of resulting fork deformation.

7. A study and comparison of our testing sample of ways to use motorcycle and/or vehicle deformation to calculate impact speed, delta v and/or barrier equivalent velocity. Our data will also be compared to other tests conducted in the past.

108 Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 108

8. A publication on the use, results and validation of 3D Scanners at the 2016 ARC CSI Motorcycle Crash Testing.

F

inal Thoughts The ARC-CSI conference is well known for the crash testing that occurs each year and the expert team leaders that accomplish that testing. The Advanced Boot Camp team was comprised of experts in the field of accident reconstruction and as a team had extensive collective experience in all aspects of crash testing. In addition, this team developed a unique testing procedure and set of goals that had not been attempted before. The Advanced Boot Camp team concept allows experienced individuals to form teams and conduct unique scientific testing.

We have given you, the reader, a quick view inside the effort to build a crash test team from scratch, develop an aggressive test matrix, and accomplish some meaningful testing. There is no doubt that this was time consuming, but it was worth it to those involved. The foundation has been laid with ARC-CSI to have other groups assemble independent testing teams and complete unique testing that would not easily be accomplished otherwise. There are so many intelligent people in the accident reconstruction community whom have great ideas for testing to generate reference data, to explore concepts, or other areas we have not thought of. The opportunities presented by the ARC-CSI conferences which now includes the Advanced Boot Camp concept will hopefully encourage testing by individuals and teams to pursue scientific research in the private sector. Come on, what are you waiting for?

www.collisionpublishing.com

1/23/2017 2:02:13 PM

www.collisionpublishing.com

Volume 11 Issue 2.indd 109

Collision Magazine - Volume 11 Issue 2 109

1/23/2017 2:02:13 PM

dy

tu e S

Cas

Case Study Solution:

Evaluating Crash Data from late model GM vehicles w. R. Rusty Haight

Collision Safety Institute

T

here may be different ways to look at a given case where different people would reasonably come to the same ultimate conclusions and the solutions to the case problems offered here should not be taken as the only way to address the use of CDR Tool retrieved crash data in a more general sense. The solutions offered here are specific to the data presented in these cases keeping in mind that the theme for the case problem set is: “this is a reading intensive technology.”

that is needed to record a Non-Deployment Event, is five MPH [8 km/h]. …”

The Verano Case Problem

-Battery Cut-Off Deployment …”

The question posed for the Verano data set was: is this data set, the crash record retrieved using the CDR Tool (from the Verano), really representative of a “Deployment” or a “Non-Deployment” Event?

While this focuses on a seemingly narrow topic, in a larger sense, one should see the question leading toward evaluating whether or not the data is or can be associated with the crash under investigation. But more specifically to this first case problem, it pushes one toward really reading the CDR Tool report in totality together with a background of related references and influential documents rather than picking and choosing one or two elements and “running with it.”

A large part of really understanding what’s going on here is, of course, to look at the Data Limitations text for this report as a starting place. The Data Limitations text for the Verano data set includes, among others, these two passages: “… There are two types of recorded crash events for Front, Side, and Rear (FSR) Events. The first is the Non-Deployment Event. A Non-Deployment Event records data but does not deploy the air bag(s). The minimum SDM Recorded Vehicle Velocity Change,

110 110 Collision Collision Magazine Magazine - Volume - Volume 11 Issue 10 Issue 2 2

Volume 11 Issue 2.indd 110

“…Also, a Non-Deployment event can be recorded if one of the following occurs without the Deployment of any of the frontal air bags, side air bags, or roll bars: -Pretensioner(s) only Deployment -Head Rest Deployment

In the first passage, perhaps the part which should stand out to the CDR Tool data user in this example is: “... A Non-Deployment Event records data but does not deploy the air bag(s). ...” In the context of the idea that “this is a reading intensive technology,” another axiom to apply here might be: “what it doesn’t say is often just as important as what it does say” (with respect to this technology and in particular in the Data Limitations or in the body of a report). The passage tells us, for certain types of events, what a NonDeployment Event is and specifies that it’s where the airbags are not deployed but it does not include or reference seat belt pretensioners.

In the second paragraph - in what might be seen as amplification or maybe something of a clarification of the first entry - we find that an event where there were there was a “Pretensioner(s) (only) Deployment” without the deployment of any airbags, that would be recorded and labeled a “Non-Deployment.”

It would seem to be clear that the Data Limitations text is telling us (1) a Non-Deployment Event record is a set of data where there were no airbags deployed and, at the same time, (2) a Non-Deployment Event is one without airbags but with a pretensioner deployment. With the use of the phrase “pretensioner (only) deployment,” the apparent disconnect seems to ask the

www.collisionpublishing.com www.collisionpublishing.com

1/23/2017 2:02:14 PM

question: “so when is a ‘deployment’ a “Non-Deployment’?” (One clue here might be found in the select capitalization of the words.)

One might think it is almost like a riddle but if we start the analysis by looking to virtually any common dictionary definition of deployment we find the word used to mean “to use or employ something in an effective way.” Next we might look to SAE J1538 the “Glossary of Automotive Inflatable Restraint Systems” where we find this definition: “...3.78 DEPLOYMENT ... The activation of an inflatable restraint device.”

Notably there is no use of the word “Event” as part of this definition. There is no definition in that document for a “non-deployment;” however, the definition for a deployment uses the phrase “inflatable restraint device” and it would seem, given other references found in J1538, that a seat belt (safety belt) pretensioner is not considered “inflatable.” That is, of course, consistent with the way we think of how a belt pretensioner normally works; it’s not something we think of as being “inflated.”

identifier “02011_RCM-AB10P”) includes this passage: “... Some RCM may also categorize Non airbag deployment event. This type is an event in which non airbag devices such as pretentioners, knee bolster etc… have deployed. Note that such event can be overwritten given a subsequent '”deployment” event. ...” (sic) We’re back to the idea that maybe pretensioner deployments aren’t, well, “Deployments...”

In that context then, we might next consider 49CFR563. There we see (among others) these definitions which include the word “deployment:” “...Deployment time ... means ... the elapsed time from crash time zero to the deployment command, or for multi-staged air bag systems, the deployment command for the first stage. ...” “... Time to deploy, pretensioner means the elapsed time from crash time zero to the deployment command for the safety belt pretensioner (for both driver and right front passenger). ...”

In the Bosch CDR Tool Help file glossary we find these definitions:

“... Event means a crash or other physical occurrence that causes the trigger threshold to be met or exceeded, or any non-reversible deployable restraint to be deployed, whichever occurs first. ...”

“... Deployment (Event) ... Acceleration observed along one of the car’s axes sufficient to cause the control module’s crash sensing algorithm to “enable” or “wake up” and which is sufficient to warrant a commanded deployment. ...” “... Non Deployment (Event) ... Acceleration observed along one of the car’s axes sufficient to cause the module’s crash sensing algorithm to “enable” or “wake up” but which does not warrant a commanded deployment. ...”

If we look next at CDR Tool retrieved crash data for a historical reference we find, for example, in Ford data as far back as model year 2001 that an event where there was an airbag commanded for deployment with or without belt pretensioners or where there was a belt pretensioner but no airbag commanded both types of events were referred to as “Deployment Events.” Ford reports change; however, corresponding to about the time of the implementation of 49CFR563, adopting a naming convention where a recorded event without an airbag deployment - but where only the belt pretensioners were “deployed” - would be identified as a “Non-Deployment Event.” For example, a report from a 2013 Ford Escape (Data Limitations text

“... Time zero means whichever of the following occurs first: ... (3) Deployment of a non-reversible deployable restraint. ...”

All of these from 49CFR563.5 would seem to give us clues but no real specific definition of what a “deployment” or “non-deployment” is (or isn’t). The first associates the word “deployment” with a function of an airbag system but makes no reference to belt pretensioner. The second definition, like the first regarding a time from “time zero,” includes the word deployment and references belt pretensioners. The last two bring a new term - “non-reversible deployable restraint” into play but provide no direct insight into what that means (or doesn’t mean).

We find at least what might be considered an interpretation of this concept in Chrysler report Data Limitations where we see these passages:

www.collisionpublishing.com

Volume 11 Issue 2.indd 111

“... Activation of only the (Active Head Restraints), if stored, will be a nondeployment event. ...”

Collision Magazine - Volume 11 Issue 2 111

1/23/2017 2:02:15 PM

“... For non-NAFTA ACMs that control pedestrian protection devices, a non-deployment event will be stored when the pedestrian protection devices are activated. ...”

It would seem that, at least for Chrysler systems, the “activation” of Active Head Restraint systems and Pedestrian Protection devices (such as “hood actuators”) would fall outside the notion of a “deployment,” they are “activations.”

Looking then to other references associated with 49CFR563, in the Federal Register, Vol. 77, No. 154, Thursday, August 9, 2012, “Rules and Regulations” we find a discussion of the notion of a “non-reversible deployable restraint” in these passages: “... E. Non-Reversible Deployable Restraints ... The Alliance and ASC petitioned the agency to clarify the requirements for data capture by EDRs in events utilizing non-reversible deployable restraints other than air bags.

They stated their belief that the agency intended to allow capture and recording of deployments of devices such as pretensioners at the option of the manufacturer, ...” “... E. Non-Reversible Deployment of Restraints ... We are denying the Alliance and ASC petitions to clarify how the deployment of non-reversible restraints other than air bags are recorded. We believe that section 563.9(b) is clear that when a memory buffer is available, EDRs must capture and record current event data that does not involve deployment of an air bag. ... In this manner, manufacturers may prioritize the capture of events that do not involve deployments of air bags, but do deploy other nonreversible restraints regardless of type. ...”

Here, of course, we see a reference to “non-reversible restraints other than air bags” and it contextually ties that term to belt pretensioners. So why is all this important and how does this apply to this analysis?

Ultimately, as 49CFR563 was implemented the manufacturers had to make choices such that their systems were “563 compliant,” part of compliance goes to the idea of “when is something recorded and when can or should it be overwritten by a later occurring event?” The regulation calls for “Deployment Events” to be prioritized as it relates to recording and preventing them from being overwritten by later occurring events and that “Non-Deployment Events” are a lower priority and generally may be overwritten. This appears to be where the decision has been made, at least by some manufacturers, to categorize events where there were no air bags deployed but the belt pretensioners have been deployed (perhaps a “better way” of saying

112 Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 112

it might be to describe the action along the lines of what Chrysler has: “activated”) as “Non-Deployment Events.”

For example, in Data Limitations for Toyota crash data, we find this passage: “... If any of the front passenger seat airbags, side airbags, or Curtain Shield Airbags have deployed, data will not be overwritten or deleted by the airbag ECU following that event. If none of the airbags have deployed, the data of that event may be overwritten by a following event even if other airbags (pretensioner, rear seat airbag, etc.) have deployed. ...” (From Data Limitations associated with the text code 05012_ToyotaS02std) Although it would seem this all comes down to a matter of semantics, the implication of the naming convention and the use of the word “deploy” or deployment” connected to the function of the belt pretensioner is what would at least appear to play the larger role in the way crash data is recorded as the manufacturers work to be “563 compliant;” at least in terms of an understanding, application or interpretation of 49CFR563.

All of this of course, brings us back to the Verano case problem and the notion that the event recovered is designated as a “Non-Deployment Event” although, using the vernacular of the GM Data Limitations, the pretensioners were...deployed. (By extension, this discussion will also play into some aspects of the Corvette case problem in this issue.)

It doesn’t take much imagination to foresee a situation where one is testifying to a set of data as part of a larger analysis and the cross examining attorney begins a line of questions which lead us back to the riddle: “when is a deployment a Non-Deployment (Event)?”

The answer might ultimately call for us to refocus on the verbiage in the two noted paragraphs from the Data Limitations for the Verano; however, the verbiage still seems to mix, on its face, the words or terms, if not the ideas, of a “deployment” and a “non-deployment” in the same application.

It is here where this case problem was designed to lead the reader to contemplate how they would respond to the question or line of questions on the stand which could recall the historical naming convention of what is and is not a deployment and definitions from the various resources (among other potential resources).

With all this in mind, one might suggest that one way to think about a response to such a line of questions would be to contemplate this in two parts where, on the one hand

www.collisionpublishing.com

1/23/2017 2:02:15 PM

the word “deployment” can be used to describe a specific type of event and, on the other hand, the word “deployment” can be used to describe a function commanded by an airbag control module. In that way, there really isn’t so much a disconnect and the application(s) are consistent with the various descriptions or definitions previously discussed.

Keep in mind this is about language and reading. In that respect, if we take that approach seriously, we can see where one use of the word makes it a noun, perhaps almost a proper name as seen in: “Deployment Event.” Another application makes it a verb (a transitive verb) as in “commanded to deploy.” And, in a third application, it would be an adjective as we’d find it in the phrase “non-reversible deployable restraint.”

The suggestion for this case problem was there appears to be an inconsistency in the use of the word “deployment;” however, thinking about it in terms of the grammatical applications of the word, thinking about the use of the word in context, it’s really not so much of an inconsistency. This is particularly true where we might look at the context of how the word is used (and maybe even whether or not it is capitalized setting aside the potential for a typographical error).

If we’re saying “Deployment Event,” That’s a name assigned to a type of event. It might be shortened to “Deployment” and perceived to have the same meaning. If we talking about “the deployment of (a device)” that’s a description of a function and, there, the word is used to show an action or describe a function. In that context, there really isn’t a riddle but a better understanding of what is being described or identified.

Taking this analysis back to the underlying crash, we’re now in a position to evaluate and apply the remainder of the data in the context of what we know about the human, vehicle and environmental information. We can say that the event is, by definition a “Non-Deployment” event although the belt pretensioners were deployed by the airbag control module.

The Corvette Case Problem There are many similarities between the data sets found in this issue of Collision for the Verano case problem and the Corvette case problem. The airbag control modules for each provide substantially similar data (in both format and content); however, they are not the exact same module. The Data Limitations text code for the Verano identifies that vehicle’s module with the Data Limitations code: “01042_SDM 10P-autoliv.” The Corvette module/data is

identified by the Data Limitations code: “01050_SDM30delphi.”

With the previously stated background for the Corvette crash (the video and diagram), we are left to evaluate the data retrieved from the Corvette to try to see if that data reflects the crash we’re investigating. One of the first steps in that analysis is to look at the three recorded events and establish their order/sequence such that it might be compared to what we know from the crash scene. The first two questions posed in this issue’s case problem were: 1. What is the chronological order of the events recorded? Meaning, the data is presented as “Event Record 1 (then 2 and 3) but is that the chronological order in which they occurred?

And 2. In the data table “System Status At Time of Retrieval” we see these data elements: “Deployment Event Counter,” “Multi-Event, Number of Events (Event Counter)” and “OnStar Notification Event Counter.” How do these data elements relate to the events under investigation? Meaning, it would appear that there have been 4 lifetime events observed by the airbag control module (Multi-Event, Number of Events (Event Counter) = 4) although there are only three recorded. You should be ready to have to explain which events are recorded and, since only three are recorded, what happened to the 4th event?

These two questions can effectively be answered together toward a more complete analysis of the data.

We know that there are three recorded events: a Non-Deployment (Event), another Non-Deployment (Event) and a Deployment (Event) represented here in record space order. A careful reading of the Data Limitations suggests how the events are recorded and when a previously recorded event might be overwritten. A review of the data gives us clues or perhaps a confirmation of the recording order (for example, the ignition cycle count for an event record).

Toward this end, in the Data Limitations, we find these passages: “... The oldest Non-Deployment event can be overwritten by a Deployment Event, if all three records are full and the Non- Deployment Event is not locked. ...” “... A Non-Deployment Event can be overwritten by a more recent Non-Deployment Event if all three records are full and the Non-Deployment is older than approximately 250 ignition cycles. ...” “... The SDM can store up to three Events. ...”

www.collisionpublishing.com

Volume 11 Issue 2.indd 113

Collision Magazine - Volume 11 Issue 2 113

1/23/2017 2:02:15 PM

A more complete explanation of what we might call the “overwrite or protect logic,” the recording function of this module, might be laid out like this: This airbag control module can record data for up to three events at a time. The data associated with a qualifying event will be recorded to what will be identified as an “Event Record.”

The first qualifying event “experienced” by the airbag control module will be recorded to Event Record space 1. The module’s second qualifying event will be recorded to Event Record space 2. The module’s third qualifying will be recorded to Event Record space 3.

Maintaining the described Event Record recording order, after three event records have been recorded, the next occurring event which would qualify for recording would be recorded in Event Record 1 space overwriting the event previously recorded there if the previously recorded event was a Non-Deployment Event and (1) that previous Non-Deployment Event is more than 250 ignition cycles old or (2) the newly occurring event is a Deployment Event or (3) the newly occurring event is a “pretensioner(s) only deployment,” “head rest deployment” or “battery cutoff deployment” (those three being what we might call “special case” or “priority” Non-Deployment Events).

If the data already recorded to Event Record #1 is a Deployment Event, then it cannot be overwritten and the module will attempt to write the newly occurring data to Event Record #2 and the same recording logic/ restrictions previously described apply.

The airbag control module can record three Non-Deployment Event records at any one time, three Deployment Event records or a combination of Deployment and Non-Deployment Event records. When there are three Deployments Events recorded, the module cannot record any more data and has to be replaced.

With this information in mind, as we may look to the data as reported and see that there are three events recorded and, looking at the “System Status At Event” data table for each record, we find that the data element “Ignition Cycle, Crash (Ignition Cycles at Event)” for each is 3810 (cycles).

Now we know they all occurred on the same ignition cycle but what we don’t know so far, what we still haven’t established or confirmed yet, is the actual order.

To do that, we could start by comparing the data element “Multi-Event, Number of Events (Event Counter)” for 114 Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 114

each. Although there is no specific definition for this element provided in the Data Limitations, we can infer from context and other definitions in the Data Limitations that this element represents, for each individual event, the number of events (Deployment Events and Non-Deployment Events) that the module has “seen” up to the currently recorded event (as seen in an individual event record).

Since we have data tables for three events to compare, one possible analytical method would be to take the pages from a printed report and lay them out side-by-side to compare the data tables. Another analytical approach could be to (1) export the data to a Comma Separated Values (CSV) file using that function from within the CDR Tool software and then (2) open that file in a spreadsheet program like Excel. There, one might (3) cut-and-paste the data for the various tables together so they can be viewed sideby-side in a spreadsheet format. Table 1 was made using the CSV export function within the Bosch CDR Tool program and then cutting-and-pasting some of the data from the three Event Records side-by side for comparison. [Editor’s note: In the interest print space and to focus narrowly on the analysis here for the case problem questions, only select element are shown in Table 1 and the reader should refer to the complete data tables displayed earlier in this issue of Collision for other types of analysis or additional information.]

In Table 1, we first see the data found in the “System Status At Time of Retrieval” data table associated with the three system-based global elements: “Dynamic Deployment Event Counter,” “Multi-Event, Number of Events (Dynamic Event Counter),” and “Dynamic OnStar Notification Event Counter.” These are defined in the Data Limitations text (seen earlier in this issue of Collision) band are effectively module lifetime counters rather than event specific or “to the time of an event” counters. What we find there is that this module has “seen,” in its lifetime, a total of 4 events (Multi-Event, Number of Events (Dynamic Event Counter) = 4).

Further down in Table 1, under the heading “System Status at Event,” we find the ignition cycle count previously described. We also find the data element “Multi-Event, Number of Events (Event Counter).” For that element, it tells us that in Event Record 2 space, we find the 2nd event this module recorded in its lifetime. In Event Record 3 space, we find the 3nd event this module recorded in its lifetime. In Event Record 1 space, we find the 4th event this module recorded in its lifetime. Thinking back to the more detailed explanation of the recording process previously described in this solution discussion, the event counter order is consistent with the notion that the 2nd

www.collisionpublishing.com

1/23/2017 2:02:15 PM

qualifying lifetime event would be recorded in Event Record 2 space, the 3rd in Event Record 3 space and then the 4th would be, if possible, recorded in Event Record 1 space. Here, that was apparently the case.

Also in Table 1, we find the data element “Time From Event 1 to 2 (Time Between Events) (seconds).” The time from the previously occurring event to the event where the time is recorded is reported for this data element. As described in the Data Limitations, this element means: “... Time between events is recorded in 10 msec intervals and is displayed in seconds for a maximum time of 655.33 seconds. The counter measures the time from the start of one event to the start of the next event if both events occur within the same ignition cycle. ...”

What isn’t described in the Data Limitations is why, as in this case, instead of a numerical value for “Time Between Events” for Event Record 2 we may find: “Data Not Available.” Without saying it in so many words in the Data Limitations and based on this review and others, we may conclude that where there is no event to keep track of time from to the event which is recorded, there would be no time data to record, the proper reported value would then be “Data Not Available.” Another way of saying it would be: on ignition cycle 3810, if there was no event occurring before what would be recorded here in Event Record 2 space, there’s nothing to keep track of time from on ignition cycle 3810 to the start of what was ultimately recorded in Event Record 2 space; data for that time element in Event Record 2 space is therefore “not available.”

Using this information, we can conclude that on ignition cycle number 3810, the first occurring event is represented by what is recorded in Event Record 2 space, it is module’s 2nd observed lifetime event. The next occurring event, which began 0.53 sec after the start of that recorded in Event Record 2 space is recorded in Event Record 3 space.

The start of that event precedes the start of the 4th lifetime event which is then recorded in Event Record 1 space. The time from the start of the event recorded in Event Record 3 space to the start of the event recorded in Event Record 1 space is reported as 1.67 sec.

One might ask: why wasn’t the first event recorded in Event Record 1 space? The answer is: it would have IF there wasn’t an event already recorded there: the module’s 1st lifetime event.

The follow up question might be: if there was an event in Event Record 1 space, how is it that Event #4 over wrote it? Going back to the “overwrite or protect logic” laid out previously, the 4th occurring event, since it is identified

as a Non-Deployment (Event), overwrote whatever was recorded in Event Record space 1 because what was there was a Non-Deployment Event which was more than 250 ignition cycles old. (1) What was there could not have been a Deployment Event, that would have been locked and couldn’t be overwritten. (2) What was there could not have been a Non-Deployment Event less than 250 ignition cycles old or it would have been protected and the Non-Deployment Event we see which was recorded there wouldn’t have been able to overwrite it. So, what’s left is that what was there was a Non-Deployment Event older than 250 ignition cycles old at cycle 3810 when the module went to record the three events we see here and it was overwritten.

Ultimately then, the answer to the questions together is, although the events are displayed in the order NonDeployment, Non-Deployment and then Deployment (Events) in the report based on Event Record space order, they occurred in the chronological order: Non-Deployment (lifetime event #2), Deployment (lifetime event #3), then Non-Deployment (lifetime #4). The 4th event in the module’s lifetime had been recorded in Record Space 1 as lifetime event #1 and was overwritten by what we see there now.

3. Taken in light of the previous case problem involving the Verano, the data in Event Record 3 for the Corvette is described as a “Deployment;” however, like the data in the Verano case, the only devices commanded for a deployment appear to be the driver and passenger belt pretensioners. One should expect to have to address that apparent conflict or at least “difference” in terms of the event naming convention. Like the Verano data set: is this data set, the crash record retrieved using the CDR Tool, really representative of a "Deployment" or a "Non-Deployment Event?”

In the Verano example, earlier in this issue’s Case Problem, we looked in detail at the naming convention logic for newer vintage GM vehicles as it relates to what is a “Non-Deployment Event.” We established that, when a seat belt pretensioner is the only device commanded for a deployment (lower case “d”), the airbag control module records an event which is identified as a Non-Deployment Event; however, in this example, we have a set of data which identifies the event type as a “Deployment” while the only devices commanded for deployment were the seat belt pretensioners. Given that the descriptors we found in the Verano case example are the same as those found in the Corvette case example, did we make a mistake? Is this a contradiction?

www.collisionpublishing.com

Volume 11 Issue 2.indd 115

Collision Magazine - Volume 11 Issue 2 115

1/23/2017 2:02:15 PM

SYSTEM STATUS AT TIME OF RETRIEVAL Dynamic Deployment Event Counter Multi-Event, Number of Events (Dynamic Event Counter) Dynamic OnStar Notification Event Counter SYSTEM STATUS AT EVENT Event Record Type OnStar Deployment Status Data Sent Complete file recorded (Event Recording Complete) Crash Record Locked OnStar SDM Recorded Vehicle Velocity Change Data Sent Deployment Event Counter Multi-Event, Number of Events (Event Counter) OnStar Notification Event Counter Time From Event 1 to 2 (Time Between Events) (seconds)

1 4 2 (EVENT RECORD 2) Non-Deployment No Yes No No 0 2 0 Data Not Available 3810 Yes No Yes Yes No No No No No No No No 2

Ignition Cycle, Crash (Ignition Cycles at Event) Algorithm Active: Frontal Algorithm Active: Side Algorithm Active: Rollover Algorithm Active: Rear Concurrent Event Flag Set Event Severity Status: Frontal Pretensioner Event Severity Status: Frontal Stage 1 Event Severity Status: Frontal Stage 2 Event Severity Status: Left Side Event Severity Status: Right Side Event Severity Status: Rear Event Severity Status: Rollover Maximum Delta-V, Longitudinal (Maximum Longitudinal SDM Recorded Vehicle Velocity Change for FSR Event) MPH Time, Maximum Delta-V (Time From FSR Time Zero to Maximum 300 Longitudinal SDM Recorded Vehicle Velocity Change)(msec) Maximum Delta-V, Lateral (Maximum Lateral SDM Recorded Ve-1 hicle Velocity Change for FSR Event) MPH Time Maximum Delta-V, Lateral (Time From FSR Time Zero to 158 Maximum Lateral SDM Recorded Vehicle Velocity Change)(msec)

(EVENT (EVENT RECORD 3) RECORD 1) Deployment Non-Deployment Yes Yes Yes Yes Yes No Yes Yes 1 1 3 4 1 2 0.53 1.67 3810 Yes No Yes Yes Yes No No No No No No Yes Data Not Available Data Not Available Data Not Available Data Not Available

3810 Yes No Yes Yes No No No No No No No No -5 204 11 204

Table 1: Data from the three Event Records side-by side for comparison (continued on following page)

116 Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 116

www.collisionpublishing.com

1/23/2017 2:02:15 PM

EVENT DATA Driver 1st Stage Deployment Loop Commanded Passenger 1st Stage Deployment Loop Commanded Driver 2nd Stage Deployment Loop Commanded Passenger 2nd Stage Deployment Loop Commanded Driver Pretensioner Deployment Loop #1 Commanded Passenger Pretensioner Deployment Loop #1 Commanded Driver Pretensioner Deployment Loop #2 Commanded Passenger Pretensioner Deployment Loop #2 Commanded Frontal Air Bag Deployment, Time to 1st Stage Deployment, Driver (Driver 1st Stage Time From Time Zero to Deployment Command Criteria Met) (msec) Frontal Air Bag Deployment, Time to 1st Stage Deployment, Right Front Passenger (Passenger 1st Stage Time From Time Zero to Deployment Command Criteria Met) (msec) Side air bag deployment, time to deploy, driver (Driver Thorax/ Curtain Time From Time Zero to Deployment Command Criteria Met) (msec) Side air bag deployment, time to deploy, right front passenger (Passenger Thorax/Curtain Time From Time Zero to Deployment Command Criteria Met) (msec) Pretensioner Deployment, Time to Fire, Driver (Driver Pretensioner Time From Time Zero to Deployment Loop #1 or Loop #2 Command Criteria Met) (msec) Pretensioner Deployment, Time to Fire, Right Front Passenger (Passenger Pretensioner Time From Time Zero to Deployment Loop #1 or Loop #2 Command Criteria Met) (msec)

(EVENT RECORD 2) No No No No No No No No Data Not Available

(EVENT RECORD 3) No No No No Yes Yes Yes Yes Data Not Available

(EVENT RECORD 1) No No No No No No No No Data Not Available

Data Not Available

Data Not Available

Data Not Available

Data Not Available

Data Not Available

Data Not Available

Data Not Available

Data Not Available

Data Not Available

Data Not Available

253

Data Not Available

Data Not Available

253

Data Not Available

Table 1: Data from the three Event Records side-by side for comparison (continued from previous page)

Keeping in mind this is a reading intensive technology, we should turn our attention back to the Data Limitations text where we previously found the description of what is a Deployment Event and what is a Non-Deployment Event and we find these passages: “... There are two types of recorded crash events for Front, Side, and Rear (FSR) Events. The first is the Non-Deployment Event. A Non-Deployment Event records data but does not deploy the air bag(s). The minimum SDM Recorded Vehicle Velocity Change, that is needed to record a Non-Deployment Event, is five MPH [8 km/h]. ...” “... For FSR Events, SDM Recorded Vehicle Velocity Change reflects the change in velocity that the sensing system experienced during the recorded portion of the event. ...”

“... For Rollover Events, the SDM may record Lateral Acceleration, Vertical Acceleration, and Roll Rate data, if the SDM is rollover capable. ...”

The Data Limitations text is essentially the same as what we saw earlier in the Verano case problem. So again why does there appear to be a difference in the naming convention used in the Verano data compared to that seen in the Corvette data? In the Verano case problem, when the airbag control module commanded the deployment of the belt pretensioners without airbags, the event was identified as a Non-Deployment Event. Here, when the airbag control module commanded the deployment of the belt pretensioners without airbags in the sequence for the Corvette, the event is identified as a Deployment Event.

What’s different? Given what we saw in the Verano example, shouldn’t this have been a Non-Deployment Event?

www.collisionpublishing.com

Volume 11 Issue 2.indd 117

Collision Magazine - Volume 11 Issue 2 117

1/23/2017 2:02:15 PM

To answer these questions, we have to look back at a phrase from the Data Limitations text (which appears in that text for both the Verano and Corvette): “... for Front, Side, and Rear (FSR) Events ...” Next there has to be a recognition that the event recorded in Event Record space 3 - and identified as a “Deployment Event” - for the Corvette is not a “Front, Side, (or) Rear Event,” it’s something else.

we might return to the passage from the Data limitations text shown immediately above which includes this phrase: “For Rollover Events, the SDM may record ...” Note it reads “may record” not “will record.” It did record two of the rollover related elements, not recording one of those listed isn’t necessarily unanticipated given the information provided in the Data Limitations text.

It is, dynamically, described as a “rollover” and, therefore, the rules associated with the naming convention which lead us to understand why the Verano belt pretensioner deployment is labeled as a Non-Deployment Event don’t apply and this event for the Corvette is labeled a “Deployment Event.”

In this instance we had benefit of a home security camera view of the crash. Still frames from that video were provided earlier in this issue. In the video, as seen in the still frames, the Corvette never actually “rolled over” such that the “painted side contracted the roadway.” The idea that this is a “rollover” is as a descriptor of dynamics and, as we can see in the still screen shots, after striking the curb, the Corvette rotated around its longitudinal axis moving toward the left side impact with the pole. The rotation of the Corvette body around the longitudinal axis wasn’t a frontal impact, it wasn’t a rear impact and it wasn’t (until it hit the pole) a side impact. Calling this a “rollover” describes the dynamic motion of the vehicle and is appropriate in this instance.

So, how do we know that the data found in Event Record space 3 corresponds to a “rollover?” There are several clues in the data. First among them is found in Table 1 where we see the data element “Event Severity Status” and we find that the translated value for the event found in Event Record space 3 is “Rollover.” Compare that to the similar element for the other similarly named elements for front, side or rear events.

Another indicator that the data in Event Record space 3 is representative of a “rollover” is found, in part, in the Data Limitations where we find these passages: “... For FSR Events, SDM Recorded Vehicle Velocity Change reflects the change in velocity that the sensing system experienced during the recorded portion of the event. ...” “... For Rollover Events, the SDM may record Lateral Acceleration, Vertical Acceleration, and Roll Rate data, if the SDM is rollover capable. ...”

That is taken then with Table 1 where, we find for the data recorded in Event Record space 3, for the “Maximum Delta-V, Longitudinal” and “Maximum Delta-V, Lateral” it reads “Data Not Available.” It’s not available because, as indicated in the Data Limitations text above, those data elements are associated with FSR events and this event is a “rollover.” In Table 1, we see values for those data elements for the data recorded in Event Record spaces 1 and 2.

Those are FSR events and there are values for “Maximum Delta-V, Longitudinal” and “Maximum Delta-V, Lateral.”

In summary, it becomes clear that the event names seen in the Verano data set compared to that seen in the Corvette data set aren’t at odds after all. The two data sets reflect two entirely different dynamic events and a careful read of the Data Limitations text - keeping in mind the notion that what they specify, what they say, is as important as what they don’t say.

This set of case problems should reenforce another idea: that being that one should look at each individual CDR Tool report in preparation of a collision analysis with “fresh eyes;” as though it were the first CDR Tool report they’ve ever seen. As soon as we become complacent, as soon as we start working with this type of data as though “if you’ve seen one, you’ve seen them all” you are guaranteed to make a mistake in the handling, in the application, of the data in a given case.

Looking back to the problem statement portion of the case problem found earlier in this issue of Collision, we see the condensed CDR Tool reports include, for Event Record 3, “Rollover Crash Pulse” and “Lateral Acceleration for Rollover Event” data. One might point out that the record does not include data for the “Vehicle Vertical Acceleration for Rollover Event.” True enough; however, 118 Collision Magazine - Volume 11 Issue 2

Volume 11 Issue 2.indd 118

www.collisionpublishing.com

1/23/2017 2:02:15 PM

www.collisionpublishing.com

Volume 11 Issue 2.indd 119

Collision Magazine - Volume 11 Issue 2 119

1/23/2017 2:02:16 PM

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:

www.4n6xprt.com/4n6as.htm

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/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 Enclosed is:

- 8.50%

TOTAL (U.S. Dollars)

= 0.00 = ______

= ________

= ________

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.

Volume 11 Issue 2.indd 120

1/23/2017 2:02:17 PM

We have the STIFFNESS data you need for your calculations!!

Now includes a FORCE BALANCE calculation module!!

4N6XPRT StifCalcs®

puts the NHTSA Crash Test database on YOUR 32 or 64 bit MS-Windows computer, automatically calculates A-B-G values Front, Side, and Rear - and Test Specific Crush Factors based on the NHTSA test data, allows the creation of “CLASS” vehicles and determine the “CLASS” stiffness values, PLUS conduct FORCE BALANCE calculations when needed, all of this with no need for you to get on the internet, and all for the LOW price of $650.00*!

Can you do this yourself “for free”?? Only if YOUR TIME = $0.00!!

For more information, program order form, and sample printouts please visit our web site at http://www.4n6xprt.com.

4N6XPRT StifCalcs® is provided by 4N6XPRT Systems

8387 University Avenue - La M esa, CA 91942 - USA Phone: 800-266-9778 / (619) 464-3478 Fax: (619) 464-2206

Individual Vehicle Stiffness Data So, you want Stiffness data for a vehicle, QUICKLY & CONVENIENTLY, but you don’t want to buy a program? Order your data on a vehicle-by-vehicle basis from us at only $40.00 per test!

Year: _________ Make: _____________________ Model:____________________________________ Impact on (circle one): Front - Side - Rear Impact speed (circle one): High - Low YOUR CONTACT INFORMATION

Name: ________________________________________ Company: _____________________________________ Address: ______________________________________ City/State/Zip: _________________________________ Phone: _______________________________________ Fax: __________________________________________ Email: ________________________________________ Visa/MC/AmEx: ________________________________ Exp.:____/____ Case Reference: ___________________ Send Order to:

4N6XPRT Systems

Fax: 619-464-2206 Ph: 800-266-9778 8387 University Avenue / La Mesa, CA 91942 / USA E-mail - 4n6@4n6xprt.com

BackCover.indd 2

1/23/2017 2:11:26 PM

© 2016. Trimble Inc. All rights reserved. PN GEO-117 (11/16)

For nearly 20 years, Visual Statement has been the leading software provider for crash and crime scene analysis and reconstruction. Now, as Trimble Forensics, it’s part of something even better—a powerful end-to-end solution for collecting, documenting and studying the data law enforcement personnel need to do their jobs effectively. With exceptional hardware, software, support, training and service, Trimble Forensics isn’t simply a name change. It’s a whole new way to work.

Learn more. Visit forensics.trimble.com

Visual Statement is now Trimble Forensics.

TRANSFORMING THE WAY THE WORLD WORKS

GE0-117_Trimble_Forensics_Collisions_Print_AD_USL_1116.indd 1 BackCover.indd 1

27/10/2016 1:59:13 p.m. PM 1/23/2017 2:11:25