Accident reconstruction is not an exact science, yet we routinely see expert reports that present a single number as THE answer. This is lazy science. Engineers are often hired long after scene evidence has disappeared and vehicles have been crushed. Even when we personally inspect the physical evidence, there will always be some uncertainty associated with the variables that enter our analyses. This uncertainty must find its way into the answers we calculate.
For example, one of the most common assumptions made in the reconstruction of a motor vehicle crash is the friction between the tires and the road surface. It is rare that we can test the actual vehicle at the actual scene. Even if we could, are the tires at the same temperature? Is the roadway in the same condition? Clearly, a single friction value cannot capture the potential effects of these variables.
Our research into vehicle skid friction has shown that even repeated tests under the same conditions produces a range of friction values. Many of us are familiar with the concept of the bell curve, or normal distribution (Figure 1). This curve represents the distribution of data about some average value. There is a high probability that most results are in the central area of the bell and a low probability that results are at the tail ends. This same concept applies to values like tire friction in the reconstruction of a car crash.
The reconstruction engineer must assume a range of friction values that quantify the expected variability in this and other parameters. By quantifying the variability we expect in the numbers that enter our analyses, we can calculate the effect of this variability in our answers. This variability always yields a range of answers; it never yields a single number.
Figure 1. Bell curve or normal distribution.
If you receive an expert report that has an “exact” single answer, send it back for clarification. The expert’s confidence in their results will surely be tested during cross-examination and the court is much better served by directly acknowledging either the limitations of the analysis or the confidence range of the answer.
2011 was an active year for turbulence-related aviation injuries. Several major airlines operated flights that were affected by turbulence categorized as “moderate” or worse, with multiple events sending passengers and crew to the emergency room.
Turbulence is a disturbance in air movements that typically cannot be seen. Turbulence can occur as a result of any number of conditions, including proximity to the jet stream, mountain waves, or convective activity such as thunderstorms or frontal systems.
The culprit in most injury-causing encounters with turbulence is Clear Air Turbulence, (CAT). This type of turbulence is usually encountered at the higher Flight Levels utilized by airliners, (18,000 feet and above) and is especially dangerous because it is undetectable.
A fairly recent advancement is the “Turbulence Analysis Chart”, an example of which is shown in Figure 1. These charts show the areas with the highest probability of a turbulence encounter as well as the maximum forecast winds at various altitudes. Avoiding the colored areas depicted on the chart is recommended however doing so is no guarantee that the pilot will prevent the aircraft from encountering turbulence. The inability to accurately and precisely forecast CAT remains one of the biggest challenges in flight planning.
Figure 1. Turbulence Analysis Chart
White-knuckled flyers often cite turbulence as their greatest fear, wrongly believing a bumpy ride will make the wings fall off or cause the pilots to lose control. The fact is aircraft can withstand loads that far exceed those ever imposed by turbulence. During initial certification, airframe manufacturers are required to demonstrate a “limit load” test, with the final data point corresponding to the actual destruction of the aircraft wing. Engineers increase the wing loading at 10% increments up to 100% limit load, which is defined as the maximum in-service limit load the aircraft is ever expected to see. At 100% limit load, the test continues at 10% increments up to 130% limit load and beyond, to gather additional engineering data such as the actual load and location of structural failure. The Boeing 777 wing was pulled 24 feet above its normal position and remained intact until 154% of design limit load.
Passengers who later recount turbulence events often state that the aircraft hit an “air pocket” or that the plane “suddenly dropped thousands of feet.” In reality, air doesn’t disappear; however it does experience density changes and/or shear, which is a large change in speed or direction over a short distance. In response to these phenomena, the aircraft moves, usually up and down, causing the occupants to experience rapid changes in vertical acceleration. This is what causes the sensation of the aircraft “dropping”. Actual altitude changes during turbulence are typically less than 50 feet, except in rare cases of severe turbulence, where an altitude change might reach 200-300 feet. These changes pose no threat to a seated passenger whose seat belt is fastened.
The data acquired annually by the FAA demonstrates that turbulence-related injuries do not coincide with structural damage to the aircraft. Most injuries occur when standing occupants fall or when seated passengers not wearing seat belts are thrust up into the ceiling. According to the FAA, no airline pilot has been seriously injured by turbulence since 1962, when the regulation requiring pilots to fasten their seat belts while in their seats took effect. This is why the initial cockpit-to-cabin announcement will state something to the effect of, “…for your safety and the safety of those around you, please keep your seat belts fastened when you are in your seats as we do up here in the cockpit…”
Consider the following data from FAA.gov:
-
In nonfatal accidents, in-flight turbulence is the leading cause of injuries to airline passengers and flight attendants.
-
Each year, approximately 58 people in the United States are injured by turbulence while not wearing their seat belts.
-
From 1980 through 2008, U.S. air carriers had 234 turbulence accidents, resulting in 298 serious injuries and three fatalities.
-
At least two of the three fatalities involved passengers who were not wearing their seat belts while the seat belt sign was illuminated.
-
Generally, two-thirds of turbulence-related accidents occur at or above 30,000 feet.
In-flight turbulence is often unavoidable; your preparation for it and your adherence to flight crew instructions is key – not only to your enjoyment of the flight but, more importantly, to avoiding an in-flight injury.
Have a safe flight!!
Ten years ago, MEA bought the first publicly available system for downloading the electronic crash data stored in black boxes in cars. There was some nervous laughter when colleagues joked that accident reconstruction engineers would soon be out of business. We’re still going strong, and here’s why:
1. Crash data doesn’t answer all the questions. Fundamental concerns about the reliability and accuracy of the data need to be addressed by an expert; otherwise a judge may exclude the data from court. And even if reliability and accuracy can be guaranteed, crash data rarely answer all of the questions. For example, it might provide evidence of excessive speed, but it won’t tell you if speed contributed to the crash. Or, it might tell you that a driver was not wearing their seat belt, but not whether seat belt use would have prevented an injury. Like other sources of physical evidence, crash data needs to be interpreted by a properly qualified expert to have an impact in court.
2. The number of cars from which data can be downloaded is limited. A recent upgrade to our Bosch download equipment adds Toyota to the short list of vehicles (GM, Ford and Chrysler-Fiat) that we can get crash data from. But only newer cars and not all models are supported. While the addition of Toyota to our list is enough to get engineers excited, the bigger news is that in September 2012 the US government will require car makers to provide access to any crash data saved by their cars. The data will also have to conform to certain standards, increasing the quantity and quality of information available to accident investigators.
3. Crash data is volatile. In many cars, the on-board computer module which records crash data must be replaced after the airbags have deployed. Therefore, crash data can get thrown away during repairs. Crash data from less severe crashes can be overwritten by a subsequent crash. If it’s not downloaded soon after a crash, data disappears. Often we are retained after the data is gone. Specially trained police officers called Collision Analysts download data but generally attend only the most severe crashes. Insurance estimators and repair shops would also be good candidates to download data, but it is not currently part of their mandate. In short, crash data often slips through the cracks.
Photo of a damaged event data recorder
Electronic crash data is a new and powerful source of evidence for the accident reconstruction engineer, but like traditional pieces of evidence it needs to be properly documented and interpreted to be useful. The accuracy, reliability and limitations of electronic crash data have been the focus of recent MEA research resulting in seven peer-reviewed scientific papers. Our goal is to be in the best position to help our clients, and ultimately the court, understand electronic crash data.
Helicopters can land safely after the engine loses power because the rotor blades can continue to turn, or autorotate, and provide lift. Autorotation is enabled by inserting a freewheel between the engine and the rotor. The freewheel locks the engine to the rotor when the engine is applying power to the rotor, but disengages the rotor from the engine if the engine stops unexpectedly. In this way a stopped engine does not brake or slow the rotor.
There have been some helicopter crashes where the freewheel suddenly disengaged leading to an engine over-speed and subsequent shutdown. Freewheel damage was moderate so the rotors could still turn freely, but autorotation was unsuccessful because of an operational factor such as low altitude or high rate of ascent. In the Sikorsky accident shown in Figure 1, as well as in other accidents, the freewheel device meant to prevent an accident has instead caused one.
Figure 1. A Sikorsky helicopter engaged in a heli-logging crashed after a freewheel disengagement.
The freewheel of this Sikorsky helicopter consists of 12 rollers pinched between a flat cam on the engine side and a gear housing on the rotor side, as shown in Figure 2. When the engine applies torque, each roller is held in place by friction so long as the coefficient of friction (COF) is at least 0.06. If the COF is less than 0.06, the roller will slide off the cam flat. This is called a spit-out. A roller spit-out often leads to a sudden freewheel disengagement at high engine load.
Figure 2. Left: The engine turns the cam causing the cam flats to pinch the rollers against the inside of a gear. That gear drives the main gearbox and ultimately the rotor. Right: If the coefficient of friction is less than 0.06 then the roller will spit out and the freewheel can disengage.
Engineers often assume that the COF between two materials is constant across all loading conditions. This assumption, however, is not always valid. Figure 3 shows data from more than 500 laboratory tests of the friction between the rollers and cams of this freewheel. The data show the COF is not constant and is sometimes less than the minimum value (0.06) needed to prevent a roller spit-out.
Figure 3. Each data point represents the normal and tangential load on a roller that slid on a cam flat in the laboratory. The line represents the coefficient of friction of 0.06 necessary to prevent roller spit out. Data points below the line are where the COF is low enough for roller spit-out. (Click on chart to view larger PDF version)
A low COF at high engine torque is believed to have caused freewheel disengagements that led to a number of crashes of this aircraft type engaged in heli-logging. Heli-logging is a repetitive external lift (REL) activity that this aircraft was not originally designed for.
Seventy-eight percent of crashes are estimated to involve driver inattention. Of these crashes, many are related to drivers who distract themselves by attempting to multitask. The graph below shows the various multitasking activities that commonly distract drivers before a crash[1].
Multitasking generally impairs performance at each individual task. Drivers cannot look in two places at once, and thus shift their gaze and attention back and forth between one task and the other. As a consequence, a driver is more likely to misdial their cell phone and insufficiently scan the roadway for hazards when attempting both tasks simultaneously.
Where a driver looks, and thus, where a driver attends, can be recorded with an eye-tracker. Eye-trackers are special devices either worn by drivers or composed of cameras mounted across the dashboard. Human factors scientists use eye-trackers to record both a driver’s field of view and where in that field of view the driver is looking.
The figure below shows a series of frames from a movie recorded by an eye-tracker while a driver was adjusting the radio. The red cross-hair indicates where the driver was looking in each frame. In the first frame, on the left, the driver looks to a car in the distance. After looking back and forth between the radio and the road for 1.3 seconds, the oncoming car is now passing the driver (it can be seen on the extreme left side of the last frame on the right). If the oncoming vehicle had crossed the centerline while the driver was adjusting the radio, multi-tasking may have resulted in a fatally delayed response.
Some people claim themselves to be good multitaskers; however, a “good” multitasker may not be aware of what they are missing. Drivers directing their attention towards a non-driving task may be blissfully unaware of the collisions they narrowly avoided. Out of sight, out of mind.
For more information on distracted driving please see our recent article on Visual Attention in Vehicle Accidents.
-
[1] Dingus, T A, Klauer, S G, Neale, V L, Petersen, A, Lee, S E, Sudweeks, J D, Perez, M A, Hankey, J, Ramsey, D J, Gupta, S, Bucher, C, Doerzaph, Z R, Jermeland, J, Knipling, R R (2006) The 100-Car Naturalistic Driving Study, Phase II - Results of the 100-Car Field Experiment. (Contract No. DTNH22-00-C-07007). Washington, DC: National Highway Traffic Safety Administration.
Forensic engineers and scientists are supposed to bring engineering and science to the courtroom. Too often, however, they bring pseudo-science, or worse, pure opinion masquerading as science. Rulings, like Frye and Daubert in the US, have reduced the amount of junk science in courtrooms, but we still see some experts playing tricks to sidestep these rules.
One common trick is to create slack between the opinion and the science purported to support the opinion. This slack is achieved by not citing the science directly in the body of the report, but instead providing a list of relevant or supporting articles at the end of the report. Slack between the opinion and the science maximizes an expert’s wiggle room when the opinion is challenged. It also allows an expert to inflate the article list to make it appear as though there is considerable science supporting the opinion.
When publishing real science in real journals, references to the science must be cited directly within the article. This allows readers of the science to know exactly how, when and for what the authors are relying on the prior work of other scientists. Including references that are not cited in the body of the article is not permitted. Only in popular science magazines and books (which are generally not peer-reviewed) are lists of additional articles or suggested reading materials common.
So why are some experts citing science differently in the courtroom than in the scientific world? Because it is easier, benefits their client and they have been allowed to get away with it.
If you receive an expert report where the science is not directly cited in the body of the report, we suggest you return it and ask the expert to i) properly cite the articles relied upon within the body of the report, and ii) delete the articles that are not relied upon. This will improve the quality of the reports from your own experts, and simplify the analysis and rebuttal of the reports from the opposing experts. It will also help rid the courtroom of some residual junk science.