The following are summaries of papers authored (or co-authored) by Wade Bartlett, at Mechanical Forensics. Click on the title to go to the summary. The SAE papers are copyrighted by them. MFES has have a limited number of hardcopies available free for whoever asks for them first, or they can be ordered through the SAE website at http://www.sae.org. The others are a little harder to find in the real world, but copies of the non-copyrighted ones can generally be sent via email upon request to the email contact listed below.
The brake system in modern passenger vehicles typically includes a vacuum booster. If the booster ceases to function properly, the pedal effort required to lock the wheels goes from a typical value of about 60 pounds (250 Newtons) to something closer to 200 pounds (900 N). This condition causes the brake pedal to get "high and hard," and is generally described by drivers as "brake failure." The pedal-depressing capabilities of men and women are examined, along with Federal regulations, and test results from several vehicles, in order to evaluate the possibility of estimating a person's deceleration rate during an accident involving a disabled booster.
The load-holding performance of polymeric clips exposed to elevated temperatures for 3 months was experimentally determined and compared to finite element predictions generated using two material models: one fairly simple one, and one which included an Arrhenius-type temperature term in the constitutive model. Parameters for both models were generated using least-squares curve fitting methods and standard tensile and creep test data available for ULTEM 1000. The variability of performance in the clips themselves was large enough to obviate any value in using the more complicated model in this case.
Large rectangular loads, such as lumber, are commonly strapped in place on flat bed trailers. Following a curved path and/or roadway superelevation can generate lateral forces high enough to cause the load to shift, stretching the straps, eventually to their breaking point. This paper presents a method of estimating the speed of such a vehicle based on the strength of the straps, including a sensitivity analysis. The result of scale model testing is presented, showing excellent agreement between predicted and experimental lateral force required to break the straps. A pdf-version of this paper is available by writing to me at the email listed in the MFES contact page. Further research on topics discussed in this paper, including friction between loads & decks, tension in straps, and effects of cargo movement on strap tension, has been published by the CCMTA and could at one time be downloaded for free from the US DOT Document Management Service search for Docket FMCSA-97-2289, and then look for the CCMTA reports - Warning, some of these files are 20Meg or more. [UPDATE: The main US document page was at http://www.regulations.gov/#!docketDetail;D=FMCSA-1997-2289 as of Jan 2015, including what appears to be the reports still available. 20Meg doesn't sound as daunting now as it did in 2000. -W]
During a sideswipe accident a rotating wheel of one vehicle may leave marks on another's side panels. This type of contact can occur during truck-to-automobile and motorcycle-to-automobile sideswipe accidents. The ratio of the speeds of the two vehicles involved can be computed once the marking tire's axle height and the period of the marks have been determined. This paper introduces the moving-frame-of-reference analysis which forms the basis of both the direct measurement method and a graphical comparison method. The former requires direct measurement of the mark period, while the latter offers a means for determining the mark period when the mark in not sufficiently complete to allow direct measurement. A previously presented vector-based method, founded on the principle of instant-centers-of-rotation, is also reviewed. This method can be used in cases where no mark period can be determined.
The most frequently cited papers on driver abilities are somewhat dated. This paper reports on the abilities of a large sample of drivers as they negotiated a closed cone-marked course using modern vehicles. The steering wheel position, brake line pressure, and throttle application were monitored, along with vehicle chassis accelerations.The objective of this paper is to report on the physical inputs utilized by operators, and compare gender-specific and vehicle-specific results. Willingness limits and g-g diagram results are presented. Results of this testing show that typical steering wheel rates in a modern vehicle are higher than earlier reported values, brake pedal forces utilized are significantly lower than those of which people are physically capable, only approximately half of all drivers utilized sufficient pedal force to lock the wheels, a significant portion of drivers never utilized wide-open-throttle, and the willingness limits for the two vehicles were virtually identical, despite variations in vehicle capabilities.There is a commonly held perception among law-enforcement drivers that traditional rear-wheel-drive cars are best suited to their performance-driving needs. However, the average and standard deviation of the front-wheel-drive group's elapsed time through the course were smaller than those of the rear-wheel-drive group, suggesting that traditional rear-wheel-drive vehicles require more skill to operate in near-limit conditions. This paper can be purchased at SAE's website, www.sae.org
When a motorcycle is involved in an accident, it is common to find a single skidmark leading up to the area of impact. It is sometimes averred that a wavy skidmark indicates the rear wheel was locked with no front brake use, while a straight skidmark indicates that the front brake was in use while the rear wheel was locked. This paper will discuss the interpretation of such skidmarks, and present the results of testing conducted with several motorcycles. It was determined that the skidmarks often became wavy at low speeds, but that the marks were often essentially straight at speeds above approximately 20 km/hr. A means of estimating the deceleration for any specific motorcycle during rear-wheel-only skidding is presented. The analysis is based on estimation or measurement of the vehicle's center of gravity and includes dynamic load transfer. Test results are compared to published data.
When performing calculations pertaining to the analysis of motor vehicle accidents, investigators often must select values for a number of parameters, such as drag factor, distances along roadways, crush depths, skid lengths, and yaw marks. The uncertainty of the final answers is a function of the variations of each parameter involved in the calculation. Though it has long been recognized that variations exist in all measured data, including those related to accident analysis, very little published information exists to assist the investigator in assigning appropriate values to the input parameter variations. This paper presents the results of recent tests conducted to obtain sample distributions of some common parameters, including measurements made with tapes, measurements made with roller-wheels, skidmark measurements, yawmark measurements, estimation of crush damage from photographs, measurement of crush damage, and measurement of drag factors, all of which may be used for uncertainty analyses. The paper also reviews the distributions of pertinent data reported by other researchers in the accident reconstruction field. Prudent application of proper distributions and ranges can lead to more accurate accident reconstructions, regardless of which range-evaluation method is selected. This paper can be purchased at SAE's website, www.sae.org
Bartlett, Wright, Brach, Baxter, Schmidt, Masory
The ability of stationary witnesses to estimate the speed of a passing vehicle was tested with a large group of participants. During one run, participants were not warned that this experiment was going to be conducted. During two others, they had foreknowledge of the experiment. The results indicate that stationary witnesses individually are not capable of accurate speed estimates, but the average estimate for groups of 25 or more may be very close to the true speed.
Information to help riders involved in accidents protect their own interests after an accident, including preserving scene and vehicle evidence, recording witness names, and taking useful photographs.
This paper discusses the conflicting published information on the use of conservation of linear momentum in motorcycle/automobile collisions. The proper methodology for both linear and angular momentum analyses in motorcycle collisions is reviewed and two case studies are included as examples of successful use of these techniques. The use of linear and angular momentum in collisions where significant weight disparities exist between the vehicles should always include a sensitivity analysis that evaluates the level of confidence of the speed estimates. Use of the sensitivity analysis will allow the reconstructionist to determine if the techniques should be applied to the given analysis or be abandoned in favor of other methods of speed analysis.
Monte Carlo analysis has been shown to be a powerful tool for evaluating confidence limits and probability distributions for values calculated during the analysis of vehicle accidents. Using this tool has generally required specialized software. This paper presents a method of using the tools provided with most simple spreadsheet programs to conduct Monte Carlo analysis with both evenly distributed and normally distributed variables for cases where the equations can be expressed in closed form. The accuracy one can expect given a particular number of trials is discussed. Example analyses using both even-probability and normal-probability variables are shown.
Various methods of assessing the overall uncertainty in a set of calculations based on the uncertainty of each variable have been promulgated. For most real-world accident analyses the available methods are simplistic, mathematically intractable, or highly computation-intensive. This paper presents a numeric approach to the partial differentiation technique that requires no high-level mathematical ability to apply, uses very little computation time, provides good results, and can be used with analysis packages of any complexity. Additionally, the method inherently incorporates a sensitivity analysis allowing evaluation of the effects of each variable. Three examples of uncertainty evaluation in accident analyses are presented.
A comparison of timed results through a cone-course obtained using a FWD cruiser and a RWD cruiser. Though the RWD car had a horsepower advantage that theoretically should make it faster, getting the most out of it required more skill than most drivers displayed. The FWD car was more forgiving, and despite a power deficiency generated overall times that were slightly lower than the RWD unit. Driver-selected limitations on cornering appear to be about the same for both vehicles.
It has been shown that one can calculate the braking deceleration capabilities of an air-braked heavy truck given a modest amount of information about the components in the brake system and their adjustment level. The error introduced by ignoring the transient air pressure effects early in the event has been found to be negligible during stops from normal road speeds, but during stops from low speeds, the actual decelerations achieved can be expected to be lower than this overall average value. This paper presents an extension of the Heusser analysis technique to include the air pressure rise-time.
The critical speed model is a tool used by accident reconstructionists to determine vehicle speeds. One assumption implicit in the model is that when in a critical speed yaw, the vehicle's center of mass travels in a circular arc. The validity of this assumption was investigated by comparing the results obtained by manually measuring the tire marks, assuming them parallel to the center of mass path, and fitting a polynomial. The results indicate that the assumption of a circular path is reasonably accurate.
Bartlett, Baxter, Livesay, Schmidt, Stanard, Wright
Lightweight "drag sleds" have long been used by crash investigators to determine the "drag factor" at a crash scene. Despite this long history, no published work has ever shown a correlation between drag sled results and the skidding performance of vehicles on multiple "uncalibrated" surfaces. Indeed, some researchers have noted that their testing appeared to show a poor correlation between the two. It has become clear in recent years that the interaction between braking or skidding tires and pavement does not fit the simple weight- and speed-independent friction model that has been assumed, leaving the accuracy of drag sleds in doubt. This paper presents the results of several comparison tests at different locations, involving multiple skid-test vehicles, dozens of drag sleds of various designs, and more than a hundred "pullers," and attempts to correlate the results of the two methods. Drag sleds tended to read higher than even the peak value generated with skidding cars, though occasionally this was not the case, and no relationship between individual drag sled results and skidding vehicle results could be identified which applied across multiple surfaces.
Evaluating the deceleration rate of air-braked heavy trucks based on brake components and their adjustment has been well documented for vehicles not equipped with anti-lock brake systems (ABS). Though there are many non-ABS trucks still on the road, ABS is becoming more common, having been required by NHTSA on all heavy trucks manufactured since March 1998. It has been found that the stopping-capabilities of ABS-equipped trucks are often not well modeled by the traditional Heusser technique. This paper will present an expansion of that method to accommodate ABS-equipped trucks, a means of characterizing air brake chambers to obviate the need for lookup tables, and discuss some sources of uncertainty in the analysis. A comparison of predicted stopping characteristics to test results for one truck is presented.
Conducting brake testing using front, rear, and both brakes has been part of most motorcycle crash reconstruction classes conducted by IPTM for many years. The classes are run at various locations across the country, so though all tests are conducted on a hard roadway surface (pavement or concrete), the actual surfaces vary, as do the motorcycles and riders. The data from more than 200 such tests using 112 different motorcycles gathered during many classes conducted between 1987 and 2006 have been assembled and analyzed. This paper will present the results of those tests, breaking out the reported drag factor in a variety of ways.
Motorcycle drop-tests have been part of most motorcycle crash reconstruction classes conducted by the Institute for Police Technology and Management for many years. Some of that data was summarized by Badger (1992). The classes are run at various locations across the country, so though all tests are conducted on a hard roadway surface (pavement or concrete), the actual surfaces vary. The drop technique can vary, depending on the equipment available at the training facility, but generally includes pushing a motorcycle off the back of a trailer either sideways or on its wheels. The results are a chaotic mix of sliding and tumbling, not unlike real motorcycle crashes. The basic data from 237 such tests using 107 different motorcycles gathered during many classes conducted between 1987 and 2006 have been assembled and analyzed. This article presents the results, breaking out the reported drag factor in a variety of ways, along with some comparisons to earlier published data. All distances were measured from the first mark in a straight line to the bike’s CG.
The upcoming NHTSA Rule563 requires vehicle EDRs (if equipped) tolerate a variety of environmental conditions. A review of past performance suggests typical designs will be fine.
A summary of the technique's background in crash analysis, and a practical step-by-step demonstration of how it's done, including conditional sampling to accomodate data known through EDR or other means.
Passenger Vehicle airbag control modules with event data recording capability have provent to be quite durable. Occasionally, however, fire, water, or mechanical damage can make it impossible to complete a normal download, even if hte EEPROM memory chip which might contain recorded information is itself undamaged. In these cases, it may be possible to swap the memory chip from the damaged module into a good surrogate module to interrogate the chip. This article describes the technique as performed in one case involving a fire-damaged SDM, as well as review some information on EEPROM nomenclature.
See 2009 article from ITAI for same stuff with some additional data
Some crash analysis practitioners maintain that the Critical Speed Formula (CSF) cannot be used in the presence of braking or acceleration, nor can it be used on surfaces other than dry pavement. Some recent research (and some not-so-recent) refutes these positions. This paper will review some of the literature on the matter, and will present results of a variety of yaw tests conducted under those conditions. The results reported here bolster earlier reports that the CSF model is tolerant of braking, regardless of the source, as long as wheels are not locked, though some literature suggests that even that is not a problem. The CSF model has been shown to work well under full throttle application on dry pavement. Results for front wheel drive cars tend to be less conservative than results for rear wheel drive cars. The actions of electronic stability control systems do not appear to reduce a vehicle’s lateral force capability to such an extent as to meaningfully affect application of the CSF model, either on gravel or dry pavement. The CSF model has been shown to work reasonably well on nominally flat unpaved gravel and grass surfaces. Best results were obtained by using the ABS-to-stop friction values. The CSF model produced slightly conservative results more often than not. Bellion’s 1/3 + 2/3 friction recommendation produced very reasonable results for split-friction events, when using the ABS-to-stop friction values for the two surfaces.
It has been observed that locked-wheel skidding friction values are essentially vehicle- and tire-independent. It has been tacitly assumed by most crash reconstructionists that any ABS-equipped vehicle would also decelerate at nearly the same rate as any other ABS-equipped vehicle. This paper will review literature with relevant straight-line test results on paved roadways and gravel, and present additional results from recent tests generated with four modern vehicles built by three manufacturers. It was found that the stopping distance reduction with ABS on dry paved surfaces varied from surface to surface for each vehicle. The ABS-improvement on a particular surface (over locked wheel stops) varied from vehicle to vehicle, and even between similar vehicles from the same manufacturer. The average ABS improvement on dry pavement with ABS was found to line up well with Lambourn's 12% value.
The accuracy of the EDR function of the Airbag Control Module (ACM) was tested on 2010 and 2011 Toyota Camry sedans during straight line operation. During steady state operation, and maximum ABS-braking runs starting from approximately 80 km/h (50 mph), and 113 km/h (70 mph), non-deployment events were artificially induced to store event data. Following each run, the EDR was imaged using the Bosch Crash Data Retrieval (CDR) system. The CDR reported speed values were compared to Racelogic VBox differential GPS speed records. Data recorders were also used to monitor the vehicle Controller Area Network (CAN) bus traffic, including the indicated speed, brake pressure, engine RPM, and accelerator pedal position. The speed and RPM reporting algorithms stated in CDR Data Limitations were confirmed. Exemplar graphs of EDR-reported speed/brake/RPM/accel pedal data versus GPS speed and CAN bus data are presented and discussed. The timing of the reported data with respect to the event is also discussed. The difference between vehicle speed recorded by the EDR and the GPS speed during steady state operation varied from +0.4 to -2.3 km/h, with the EDR typically reporting lower than the GPS. During heavy braking the difference in speed was observed to be from -7 to +15 km/h, with wheel slip causing negative differences, and time delay since the last CAN bus update causing positive differences.
The evasive capabilities of motorcycles and riders are often an important consideration when analyzing a motorcycle crash. Specifically, the longitudinal distance or time required for a motorcycle to move laterally some distance can be of critical interest. Previous publications on this topic have not all measured the same thing and have often included limited test data so their results can be difficult to compare or apply. In addition to reviewing some of the literature on the topic, this paper will present the results of a series of tests conducted with four riders on four motorcycles swerving 2 m (6.5 ft) to their left after passing through a gate at speeds of 40 to 88 km/h (25 to 55 mi/h). The most recent testing involved relatively skilled riders who had faster transitions and greater willingness to lean than the “average” rider generally described in the literature. Separating the perception-reaction time from the evaluation of the turnaway maneuver itself simplifies the analysis, though wide individual performance variation still exists.
Motorcycles have followed passenger cars once again: this time by installing EDRs that record some information about the situation leading up to the time when a crash occurs. This paper reveiws what is known about the newest Kawasaki EDRs.
Mechanical Forensics Engineering Services, LLC.
This page last modified 09-JAN-2017