Crash test dummy

Crash test dummies are full-scale anthropometric test devices (ATD) that simulate the dimensions, weight proportions and articulation of the human body, and are usually instrumented to record data about the dynamic behaviour of the ATD in simulated vehicle impacts. This data can include variables such as velocity of impact, crushing force, bending, folding, or torque of the body, and deceleration rates during a collision for use in crash tests. They remain indispensable in the development of and ergonomics in all types of vehicles, from automobiles to aircraft.

The need for testing

On August 31, 1869, Mary Ward became what is believed to be the first recorded victim of a steam powered automobile accident (Karl Benz only invented the gasoline powered automobile as we know it in 1886). Mary Ward was thrown out of a motor vehicle and killed in Parsonstown, Ireland. Some years later, on September 13, 1899, Henry Bliss entered the history books as North America's first motor vehicle fatality when he was hit stepping off a New York City trolley. Since that time, in excess of 20 million people worldwide have lost their lives to motor vehicle accidents.

The need for a means of analysing and mitigating the effects of motor vehicle accidents on human bodies was felt very soon after the commercial production of automobiles began in the late 1890s, and by the 1930s, with the automobile a common part of daily life, the number of motor vehicle deaths was becoming a serious issue. Death rates had surpassed 15.6 fatalities per 100 million vehicle-miles and were continuing to climb; vehicle designers saw this as a clear indication it was time to do some research on ways to make their products safer.

In 1930, the interior of a car was not a safe place even in a low-speed collision. Dashboards were made of rigid metal, steering columns were non-collapsible, and protruding knobs, buttons, and levers were ubiquitous. Seat belts were unheard-of, and in a frontal collision, passengers hurled through the windshield stood very little chance of avoiding serious injury or death. The vehicle body itself was rigid, and impact forces were transmitted directly to the vehicle occupants. As late as the 1950s, car manufacturers were on public record as saying vehicle accidents simply could not be made survivable; the forces in a crash were too great and the human body too frail.

Cadaver testing

Detroit's Wayne State University was the first to begin serious work on collecting data on the effects of high-speed collisions on the human body. In the late 1930s, there were no reliable data on the response of the human body to extreme physical injury, and no effective tools existed to measure such responses. Biomechanics was a field barely in its infancy. It was therefore necessary to employ two types of test subjects in order to develop initial data sets.

The first test subjects were human cadavers. They were used to obtain fundamental information about the human body's ability to withstand the crushing and tearing forces typically experienced in a high-speed accident. To such an end, steel ball bearings were dropped on skulls, and bodies were dumped down unused elevator shafts onto steel plates. Cadavers fitted with crude accelerometers were strapped into automobiles and subjected to head-on collisions and vehicle rollovers.

Albert King's 1995 Journal of Trauma article, "Humanitarian Benefits of Cadaver Research on Injury Prevention", clearly states the value in human lives saved as a result of cadaver research. King's calculations indicate that as a result of design changes implemented up to 1987, cadaver research has since saved 8500 lives annually. He notes that for every cadaver used, each year 61 people survive due to wearing seat belts, 147 live due to air bags, and 68 survive windshield impact.

However, work with cadavers presented almost as many problems as it resolved. Not only were there the moral and ethical issues related to working with the dead, but there were also research concerns. The majority of cadavers available were older European American adults who had died non-violent deaths; they did not represent a demographic cross-section of accident victims. Deceased accident victims could not be employed because any data that might be collected from such experimental subjects would be compromised by the cadaver's previous injuries. Since no two cadavers are the same, and since any specific part of a cadaver could only be used once, it was extremely difficult to achieve reliable comparison data. In addition, child cadavers were not only difficult to obtain, but both legal and public opinion made them effectively unusable. Moreover, as crash testing became more routine, suitable cadavers became increasingly scarce. As a result, biometric data were limited in extent and skewed toward the older white males.

Volunteer testing

Some researchers took it upon themselves to serve as crash test dummies. Colonel John Paul Stapp USAF propelled himself over 1000 km/h on a rocket sled and stopped in 1.4 seconds. Lawrence Patrick, then a professor at Wayne State University, endured some 400 rides on a rocket sled in order to test the effects of rapid deceleration on the human body. He and his students allowed themselves to be smashed in the chest with heavy metal pendulums, impacted in the face by pneumatically-driven rotary hammers, and sprayed with shattered glass to simulate window implosion. While admitting that it made him "a little sore", Patrick has said that the research he and his students conducted was seminal in developing mathematical models against which further research could be compared. But while data from live testing was valuable, human subjects could not withstand tests which went past a certain degree of physical injury. To gather information about the causes and prevention of injuries and fatalities would require a different kind of subject.

Animal testing

By the mid-1950s, the bulk of the information cadaver testing could provide had been harvested. It was also necessary to collect data on accident survivability, research for which cadavers were woefully inadequate. In concert with the shortage of cadavers, this need forced researchers to seek other models. A description by Mary Roach of the Eighth Stapp Car Crash and Field Demonstration Conference shows the direction in which research had begun to move. "We saw chimpanzees riding rocket sleds, a bear on an impact swing...We observed a pig, anesthetized and placed in a sitting position on the swing in the harness, crashed into a deep-dish steering wheel at about 10 mph."

One important research objective which could not be achieved with either cadavers or live humans was a means of reducing the injuries caused by impalement on the steering column. By 1964, over a million fatalities resulting from steering wheel impact had been recorded, a significant percentage of all fatalities; the introduction by General Motors in the early 1960s of the collapsible steering column cut the risk of steering-wheel death by fifty percent. The most commonly used animal subjects in cabin-collision studies were pigs, primarily because their internal structure is similar to a human's. Pigs can also be placed in a vehicle in a good approximation of a seated human.

The ability to sit upright was an important requirement for test animals in order that another common fatal injury among human victims, decapitation, could be studied. As well, it was important for researchers to be able to determine to what extent cabin design needed to be modified to ensure optimal survival circumstances. For instance, a dashboard with too little padding or padding which was too stiff or too soft would not significantly reduce head injury over a dash with no padding at all. While knobs, levers, and buttons are essential in the operation of a vehicle, which design modifications would best ensure that these elements did not tear or puncture victims in a crash. Rear-view mirror impact is a significant occurrence in a frontal collision; how should a mirror be built so that it is both rigid enough to perform its task and yet of low injury risk if struck.

While work with cadavers had aroused some opposition, primarily from religious institutions, it was grudgingly accepted because the dead, being dead, felt no pain, and the indignity of their situations was directly related to easing the pain of the living. Animal research, on the other hand, aroused much greater passion. Animal rights groups such as the ASPCA were vehement in their protest, and while researchers such as Patrick supported animal testing because of its ability to produce reliable, applicable data, there was nonetheless a strong ethical unease about this process.

Although animal test data were still more easily obtained than cadaver data, the fact that animals were not people and the difficulty of employing adequate internal instrumentation limited their usefulness. Animal testing is no longer practiced by any of the major automobile makers; General Motors discontinued live testing in 1993 and other manufacturers followed suit shortly thereafter.

Dummy evolution

The information gleaned from cadaver research and animal studies had already been put to some use in the construction of human simulacra as early as 1949, when "Sierra Sam" was created by Samuel W. Alderson at his Alderson Research Labs (ARL) and Sierra Engineering Co. to test aircraft ejection seats and pilot restraint harnesses. This testing involved the use of high acceleration to 1000 km/h (600 mph) rocket sleds, beyond the capability of human volunteers to tolerate. In the early 1950s, Alderson and Grumman produced a dummy which was used to conduct crash tests in both motor vehicles and aircraft.

Alderson went on to produce what it called the VIP-50 series, built specifically for General Motors and Ford, but which was also adopted by the National Bureau of Standards. Sierra followed up with a competitor dummy, a model it called "Sierra Stan," but GM, who had taken over the impetus in developing a reliable and durable dummy, found neither model satisfied its needs. GM engineers decided to combine the best features of the VIP series and Sierra Stan, and so in 1971 Hybrid I was born. Hybrid I was what is known as a "50th percentile male" dummy. That is to say, it modeled an average male in height, mass, and proportion. The original "Sierra Sam" was a 95th percentile male dummy (heavier and taller than 95% of human males). In cooperation with the Society of Automotive Engineers (SAE), GM shared this design, and a subsequent 5th percentile female dummy, with its competitors.

Since then, considerable work has gone into creating more and more sophisticated dummies. Hybrid II was introduced in 1972, with improved shoulder, spine, and knee responses, and more rigorous documentation. Hybrid II became the first dummy to comply with the American Federal Motor Vehicle Safety Standard (FMVSS) for testing of automotive lap and shoulder belts. In 1973, a 50th percentile male dummy was released, and the National Highway Transportation Safety Administration (NHTSA) NHTSA undertook an agreement with General Motors to produce a model exceeding Hybrid II's performance in a number of specific areas.

Though a great improvement over cadavers for standardized testing purposes, Hybrid I and Hybrid II were still very crude, and their use was limited to developing and testing seat belt designs. A dummy was needed which would allow researchers to explore injury-reduction strategies. It was this need that pushed GM researchers to develop the current Hybrid line, the Hybrid III family of crash test dummies.

Hybrid III family

Hybrid III, the 50th percentile male dummy which made its first appearance in 1976, is the familiar crash test dummy, and he is now a family man. If he could stand upright, he would be 168 cm (5 '6 ") tall and would have a mass of 77 kg (170 lb). He occupies the driver's seat in all the Insurance Institute for Highway Safety (IIHS) 65 km/h (40 mph) offset frontal crash tests. He is joined by a "big brother", the 95th percentile Hybrid III, at 188 cm (6 ft 2 in) and 100 kg (223 lb). Ms. Hybrid III is a 5th percentile female dummy, at a diminutive 152 cm (5 ft) tall and 50 kg (110 lb). The three Hybrid III child dummies represent a ten year old, 21 kg (47 lb) six year old, and a 15 kg (33 lb) three year old. The child models are very recent additions to the crash test dummy family; because so little hard data are available on the effects of accidents on children, and such data are very difficult to obtain, these models are based in large part on estimates and approximations. The primary benefit provided by the Hybrid III is improved neck response in forward flexion and head rotation that better simulates the human.

Test process

Every Hybrid III undergoes calibration prior to a crash test. Its head is removed and is dropped from 40 centimetres to test calibrate the head instrumentation. Then the head and neck are reattached, set in motion, and stopped abruptly to check for proper neck flexure. Hybrids wear chamois leather skin; the knees are struck with a metal probe to check for proper puncture. Finally, the head and neck are attached to the body, which is attached to a test platform and struck violently in the chest by a heavy pendulum to ensure that the ribs bend and flex as they should.

When the dummy has been determined to be ready for testing, it is dressed entirely in yellow, marking paint is applied to the head and knees, and calibration marks are fastened to the side of the head to aid researchers when slow-motion films are reviewed later. The dummy is then placed inside the test vehicle. Forty-four data channels located in all parts of the Hybrid III, from the head to the ankle, record between 30 000 and 35 000 data items in a typical 100–150  millisecond crash. Recorded in a temporary data repository in the dummy's chest, these data are downloaded to computer once the test is complete.

Because the Hybrid is a standardized data collection device, any part of a particular Hybrid type is interchangeable with any other. Not only can one dummy be tested several times, but if a part should fail, it can be replaced with a new part. A fully-instrumented dummy is worth about €150 000.

Hybrid's successors

Hybrid IIIs are designed to research the effects of frontal impacts, and are less valuable in assessing the effects of other sorts of impacts, such as side impacts, rear impacts, or rollovers. After head-on collisions, the most common severe injury accident is the side impact.

The SID (Side Impact Dummy) family of test dummies has been designed to measure rib, spine, and internal organ effects in side collisions. It also assesses spine and rib deceleration and compression of the chest cavity. SID is the US government testing standard, EuroSID is used in Europe to ensure compliance with safety standards, and SID II(s) represents a 5th percentile female. BioSID is a more sophisticated version of SID and EuroSID, but is not used in a regulatory capacity. The WorldSID is a project to develop a new generation of dummy under the International Organization for Standardization.

BioRID is a dummy designed to assess the effects of a rear impact. Its primary purpose is to research Whiplash, and to aid designers in developing effective head and neck restraints. BioRID is more sophisticated in its spinal construction than Hybrid; 24 vertebra simulators allow BioRID to assume a much more natural seating posture, and to demonstrate the neck movement and configuration seen in rear-end collisions.

CRABI is a child dummy used to evaluate the effectiveness of child restraint devices including seat belts and air bags. There are three models of the CRABI, representing 18-month, 12-month, and 6-month old children.

THOR is an advanced 50th percentile male dummy. The successor of Hybrid III, THOR has a more humanlike spine and pelvis, and its face contains a number of sensors which allow analysis of facial impacts to an accuracy currently unobtainable with other dummies. THOR's range of sensors is also greater in quantity and sensitivity than those of Hybrid III.

Further development is needed on dummies which can address the concern that, even though fewer lives are lost, there are still a hundred seriously injured passengers for every death, and crippling injuries to the legs and feet represent a great percentage of resultant physical impairments.

One important sector of the traveling public has yet to be represented in mainstream crash testing — pregnant women. The first prototype pregnant crash test dummy has been built by engineering researchers at Loughborough University UK with the aim of improving seat belt design. It has a fluid filled container above the pelvis to replicate the foetus and womb. Belts can be uncomfortable for pregnant women so some choose not to wear them, reducing their safety at a time when it should be increased. A second pregnant crash test dummy has been designed by a student at the University of Idaho.

Future of the dummy

Crash test dummies have provided invaluable data on how human bodies react in crashes and have contributed greatly to improved vehicle design. While they have saved millions of lives, like cadavers and animals, they have reached a point of reduced data return.

The largest problem with acquiring data from cadavers, other than their availability, was that an essential element of standardized testing, repeatability, was impossible. No matter how many elements from a previous test could be reused, the cadaver had to be different each time. While modern test dummies have overcome this problem, testers still face essentially the same problem when it comes to testing the vehicle. A vehicle can be crashed only once; no matter how carefully the test is done, it cannot be repeated exactly.

A second problem with dummies is that they are and will only ever be approximately human. Forty-four data channels on a Hybrid III is not even a remote representation of the number of data channels in a living person. The mimicking of internal organs is crude at best, a fact that means that even though cadavers and animals are no longer the primary sources of accident data, they must still be employed in the study of soft tissue injury.

The future of crash testing has begun at the same place it all started: Wayne State University. King H. Yang is one of Wayne State's researchers involved in creating detailed computer models of human systems. Currently, Wayne State's researchers do not have fast enough computers nor skilled programmer's to create full-body simulations, but injury analysis of individual body systems is beginning to produce reliable and encouraging results.

The advantage of the computer is that it is unbound by physical law. A virtual vehicle crashed once can be uncrashed and then crashed again in a slightly different manner. A virtual back broken can be unbroken, the seatbelt configuration changed, and the back re-broken. When every variable is controllable and every event is repeatable, the need for physical experimentation is greatly reduced.

At the beginning of the 21st century, legal certification of new car models is still required to be done using physical dummies in physical vehicles. The next generation of crash test dummies may perform their tasks entirely on a computer screen.