An Evaluation of Windshield Glazing and Installation Methods for Passenger Cars

Charles J. Kahane, Ph.D.

Federal Motor Vehicle Safety Standard 205 sets requirements for the penetration resistance of windshields upon impact while Standard 212 regulates windshield retention in crashes. The High Penetration Resistant (HPR) windshield, developed during the 1960's, was designed to crumple and deform at speeds where earlier windshields would have been penetrated by head impacts. Adhesive bonding provided a tighter windshield installation method than earlier rubber gasket designs. The objectives of this agency staff evaluation are to determine if HPR windshields and adhesive bonding achieved their performance objectives in highway crashes and to measure their casualty-reducing benefits, side effects and costs. The study is based on statistical analyses of National Crash Severity Study, New York, Texas, mu1tidiscip1inary Accident Investigation and Fatal Accident Reporting System data.

It was found that:

• HPR glazing doubled the impact velocity needed for the occupant's head to penetrate the windshield, preventing 39,000 serious lacerations and 8,000 facial fractures per year.
• Adhesive bonding halved windshield bond separation and occupant ejection through the windshield portal, saving 105 lives per year.
• HPR glazing did not increase the risk of concussions and adhesive bonding did not increase the injuries of persons who were not ejected.
• HPR glazing added $6 (in 1982 dollars) to the lifetime cost of owning and operating a car, but adhesive bonding saved $15 per car.

Since 1960, there have been major changes in the design of windshields for passenger cars and in the techniques whereby windshields are installed in cars.

In 1965, the domestic manufacturers installed High Penetration Resistant (HPR) windshields, on an experimental basis, in a few models and in 1966, HPR became standard equipment in all domestic cars. Before HPR, the plastic interlayer of safety glazing used in windshields was easily torn by broken glass, permitting the occupant1s head to tear through and penetrate the windshield in low-speed crashes. Windshield penetration was believed to be the cause of most of the disfiguring or disabling head injuries associated with windshield contact. Rodloff, Patrick, Rieser and other researchers found techniques to obtain a looser glass-plastic bond in safety glazing, allowing the glass to crumple away rather than tear the plastic. The new manufacturing techniques, in combination with a thicker plastic layer, became known as the "HPR windshield", which was found to double the speed needed for the windshield to be penetrated in laboratory impact tests. Federal Motor Vehicle Safety Standard 205, which took effect on January 1, 1968, incorporated American National Standards Institute's safety codes which the motor vehicle industry had already imposed upon themselves to assure that all motor vehicles have windshields as penetration-resistant as HPR.

Before 1963, windshields were installed in a car by means of a rubber gasket. In 1963, butyl tape was used to adhesively bond the windshield to the frame on a small test fleet of General Motors cars. Adhesive bonding became standard on a few GM models in 1964. The domestic manufacturers gradually shifted from rubber gaskets to adhesive bonding (initially butyl tape and later, in some cases, polyurethane sealant) after 1964, but rubber gaskets remained on some domestic models until 1978. The objectives of adhesive bonding were not explicitly stated, but two may be inferred: to provide a tighter bond between windshield and car, preventing the windshield from becoming dislodged in a crash, denying occupants an avenue for ejection through the gap between windshield and frame; to reduce manufacturing cost by eliminating the rubber gasket. On January 1, 1970, in the middle of the transition from rubber gaskets to adhesive bonding, Federal Motor Vehicle Safety Standard 212 took effect for passenger cars. Standard 212 limits the amount of windshield bond separation allowed in a 30 mph barrier crash and has the explicit objective of preventing occupant ejection through the windshield portal. But the relationship of Standard 212 to adhesive bonding is not clear, since, as stated above, rubber gaskets continued to be used in some models well after 1970. It is possible that a 1976 modification in the temperature range for Standard 212 testing may also have accelerated the shift to adhesive bonding.

Foreign cars, as a matter or fact, continued to use mostly rubber gaskets throughout the 1970's. But Volkswagen, which had virtually a "pop-out" windshield before 1970, did install clips between the gasket and the frame in response to Standard 212. It is possible that other German manufacturers also implemented similar devices about that time.

Executive Order 12291 (February 1981) requires agencies to evaluate their existing major regulations, including any rule whose annual effect on the economy is $100 million or more. The objectives of an evaluation are to determine the actual benefits--lives saved, injuries prevented, damage avoided--and costs of safety equipment installed in production vehicles in connection with a standard and to assess cost-effectiveness.

This report is an evaluation of HPR windshields for passenger cars, adhesive bonding of the windshields of domestic cars and the changes in the installation of Volkswagen and other German windshields made in response to Standard 212. (HPR glazing and adhesive bonding were also implemented in vehicles other than passenger cars, but that will be evaluated at a later date.)

The report does not evaluate the effects of the shift from laminated to tempered side windows which took place in about 1960--there were far too few cases on NHTSA accident files of occupants who were injured by contact with side windows in cars of the 1960 era. It also does not evaluate glass-plastic glazing concepts such as Securiflex because they have not yet (October 1984) been implemented in large numbers on cars sold in the United States. NHTSA evaluations of existing safety devices, as stated above, are based on the actual operating experience of production vehicles: something not yet available in sufficient quantity for glass-plastic glazing. It should be noted however, that laboratory tests show that glass-plastic glazing may have great potential for reducing minor facial lacerations (a great many of which still remain, even after HPR) and occupant ejection through side windows. If the concept is implemented on a large number of production vehicles, NHTSA will certainly evaluate their on-the-road experience.

HPR windshields have already been informally evaluated. The dramatic reduction in the demand for facial plastic surgery following the introduction of HPR made it clear to the safety community that HPR has been, perhaps, more successful than any other standard. The effectiveness of HPR has been shown in a number of laboratory studies and statistical accident analyses. It remains for this evaluation to give specific estimates of the numbers and types of injuries prevented by HPR, to compare laboratory and highway accident performance, and to investigate the possibility of negative side effects such as blunt impact trauma or secondary benefits such as a reduction of minor injuries.

Windshield installation methods, by contrast, have been a controversial subject since the mid-1960's. On the one hand, no study to date appears to have shown whether or not tighter bonding methods achieved Standard 212's goal of reducing occupant ejection. On the other hand, studies by Fargo (accident analysis of pre-HPR cars) and Rodloff and Breitenbuerger (drop tests with HPR glazing) warn that tight bonding has serious negative side effects for persons who are not ejected: lacerations, blunt impact trauma and a reduced windshield penetration velocity. But Patrick's and Trosien's sled tests with dummies found little or no side effect. Who is right? Thus, the evaluation must analyze the effect of installation method on ejection and on persons who are not ejected. Both analyses must be performed separately for domestic cars (effect of adhesive bonding) and German cars (effect of Standard 212); the analysis of persons who are not ejected, separately for pre-HPR and post-HPR cars.

The strategy of this evaluation was to perform parallel statistical "injury" and "engineering" analyses of accident data. In the case of HPR, the "injury" analysis of the reduction of various types of head trauma was paralleled by an "engineering" analysis of the velocities at which heads penetrate windshields in highway accidents. The objectives were to give an engineering explanation of why injuries were reduced and to compare hardware performance in accidents to the laboratory. In the study of the effect of windshield installation method on ejection, the "injury" analysis of occupant ejection rates was accompanied by an "engineering" analysis of windshield retention in crashes. The analysis of the side effects of windshield installation method on occupants who were not ejected again compared types of head injuries and penetration velocities.

The "engineering" analyses were based on National Crash Severity Study (NCSS) data. The "injury" analyses, in each case, were based on at least 3 files: effect of HPR on injury rates--NCSS, New York State, Texas and Fatal Accident Reporting System (FARS); effect of installation method on ejection--NC55, Multidisciplinary Accident Investigation (MOAI) and FARS; effect of installation method on injury rates--New York State, NCSS and Texas. New York data were especially useful because they identified the body region and type of injury over a large sample of accidents. When large data files were used (FARS, New York, Texas, the analysis of German cars was limited to Volkswagen, where it is relatively clear that clips were installed very close to the beginning of the 1970 model year. For the smaller data files (NCSS, MDAI), the other German makes are included to increase the available sample size, even though it is not as well known when and if similar modifications were made. Thus, throughout the report, results on "German cars" are the ones based on NCSS and/or MDAI; those on "Volkswagen" are based on the other files. Practically speaking, though, the distinction is of minor importance since Volkswagen accounted for over 85 percent of the German cars sold here during 1965-74.

The cost of the vehicle modifications was estimated by analyzing the components of vehicles produced before and after the modification.

The most important conclusions of this evaluation are that HPR glazing dramatically reduced the number and severity of facial lacerations and fractures while doubling the impact velocity needed to penetrate the windshield in crashes. Adhesive bonding saved lives because it halved windshield separation in crashes and occupant ejection through the windshield portal; the clips installed in the rubber gaskets of Volkswagens in response to Standard 212 had the same effect. In cars with HPR windshields, the installation method had little or no side effect on the injuries of persons who were not ejected. Because each of these conclusions is supported by analyses of multiple data files, which are remarkably consistent with one another and with the "engineering" analyses, they may be stated confidently.

In two areas, conclusions are drawn less firmly. One concerns the proportion of ejectees through the windshield portal who were killed by injuries sustained while they were still inside the passenger compartment. This proportion is estimated with reasonable precision from NCSS and MDAI data. It is then assumed to equal the proportion of persons, saved from ejection by adhesive bonding, who would have died anyway from interior contacts. The FARS data, unfortunately, were unsuited for an independent verification of this plausible assumption. Thus, in this case, "evidence from multiple data files" is lacking. The other area is the effect of adhesive bonding on injury risk in cars with pre-HPR windshields. New York data show significant negative effects but NCSS and Texas data show none. Thus, while the evaluation clearly shows no side effects with today's windshields, it is unable to resolve the controversy about adhesive bonding in cars with pre-HPR windshields--fortunately, the question has become moot because so few of them remain on the road.

The principal findings and conclusions of the study are the following:

Principal Findings

Effect of HPR glazing on windshield penetration by occupants

• When an occupant's head strikes a safety-glass windshield and tears and penetrates the plastic interlayer, the risk of serious lacerations or fractures to the face, scalp, eyes, nose or mouth is 3 times greater than when the impact merely breaks the glass but leaves the plastic layer intact.
• HPR glazing reduced the likelihood of an occupant penetrating the windshield in frontal crashes by 78 percent, relative to pre-HPR glazing.
• With pre-HPR glazing, there was a 50 percent probability that an unbelted occupant would penetrate the windshield in a frontal crash with a Delta V of 14 miles per hour. With HPR glazing, the likelihood of penetration does not reach 50 percent until Delta V is 31 miles per hour.

Injury-reducing effectiveness of HPR

• The reduction of serious head injuries involving windshield contact, by injury type, was:

	Reduction for HPR (%)
	Best Estimate	Confidence Bounds
AIS 2-4 lacerations	74	65 to 83
AIS 2-4 eye, nose or mouth injuries	72	58 to 86
AIS 2-4 fractures	56	27 to 85

Those are the types of injuries most characteristically associated with penetration of the windshield.

• HPR glazing also reduced minor lacerations due to contacting the windshield by 25 percent (confidence bounds: 5 to 45 percent).
• HPR windshields had little or no observed effect on injuries characteristic of blunt impact trauma: concussions, contusions and complaints of pain.
• Fatality risk in crashes was not significantly changed by HPR.

Effect of windshield installation method on windshield retention in crashes

• All American manufacturers shifted from installing windshields with a rubber gasket to an adhesive bonding process at some time between 1963 and 1979, depending on the car' s make and model. That resulted in an immediate 35 percent reduction and a long-term 50 percent reduction in the proportion of the windshield that became separated from the frame in a frontal crash. The long-term reduction is greater because some of the adhesive bondings initially used by General Motors were looser than their later practice.
• Volkswagen (and, possibly, other German manufacturers) responded to Standard 212 by clipping their rubber gasket to the frame, rather than shifting to adhesive bonding. The clips reduced windshield separation by 51 percent. Nevertheless, the statistics on windshield separation in accidents suggest that post-Standard 212 windshield installations in Volkswagens were looser than pre-Standard 212 rubber gaskets in American cars.
• In a frontal crash with a Delta V of 30 miles per hour, the average amount of windshield separation, by installation method, was:

	Separation (%)
American cars - rubber gaskets	22
American cars - adhesive bonding	15
German cars - pre-Standard 212 gaskets	59
German cars - post-Standard 212 gaskets	39

Standard 212 allows 25 percent bond separation in a staged 30 mph barrier impact.

• Polyurethane sealant and butyl tape--two alternative adhesive bonding methods--provided about the same windshield retention in crashes, for cars of the 1970's.

Effect of windshield installation method on occupant ejection

• In American cars whose windshields were installed by rubber gaskets, 15 percent of the occupant ejections (complete or partial) took place through the windshield portal: in pre-Standard 212 Volkswagens, 17 percent. (Persons who merely penetrate the windshield's plastic interlayer with part of their heads are not normally coded as "partially ejected" by NHTSA accident investigators.)
• Adhesive bonding reduced the risk of ejection through the windshield portal by 50 percent in American cars (confidence bounds: 34 to 66 percent). The clips installed in Volkswagens in response to Standard 212 had the same effect. Reductions of complete and partial ejection were similar.
• On the other hand, only 30 percent of the persons who were ejected through the windshield portal received their most serious injuries as a consequence of the ejection--i.e., from objects exterior to the passenger compartment. As a result, adhesive bonding saves 15 percent (50% of 30%) of the deaths and serious injuries of windshield ejectees (confidence bounds: 7 to 22 percent). Standard 212 had the same effect in Volkswagens.

Effect of windshield installation method on windshield penetration by occupants

• There was no evidence that tighter bonding increased the risk of an occupant penetrating the windshield. In fact, the following nonsignificant reductions were observed:

Adhesive bonding vs. rubber gasket in American cars:	1 percent
Adhesive bonding vs. rubber gasket in American pre-HPR cars:	7 percent
Post-Standard 212 gasket vs. pre-Standard in German HPR cars:	19 percent

Effect of windshield installation method on the injuries of persons who are not ejected

• In American cars with HPR windshields1 the installation method (rubber gaskets or adhesive bonding) had little or no effect on the likelihood of any type of head injury.
• In Volkswagens with HPR windshields, Standard 212 likewise had little or no effect.
• In American cars with pre-HPR windshields, the following statistically significant increases of head injury risk were associated with adhesive bonding in New York State data:

	Increase for Adhesive Bonding (%) (Cars with Pre-HPR Windshields)
"Severe bleeding" (i.e., nonminor lacerations)	20
Concussions	50
Contusions and complaints of pain	20

The Texas data do not show any increase, however, in overall injury rates.

Cost of HPR glazing

• The incremental costs per car (in 1982 dollars) for HPR glazing, relative to pre-HPR, based on analyses of vehicle components, are the following:

Initial purchase price reduction	$ 4.45
Lifetime fuel savings due to 1.05 pound weight reduction	1.05
TOTAL SAVINGS PER CAR	$ 5.50

• The annual cost of HPR glazing in the United States (based on 10 million cars sold) is $55 million.

Cost savings due to adhesive bonding

• Adhesive bonding was a less costly way to install a windshield than rubber gaskets. The cost savings per car (in 1982 dollars), based on analyses of vehicle components, are the following:

Initial purchase price reduction	$11.50
Lifetime fuel savings due to 3.98 pound weight reduction	3.98
TOTAL SAVES PER CAR	$15.48

• The annual savings due to adhesive bonding in the United States (based on 7.5 million domestic cars sold) is $116 million.

Annual benefits of HPR glazing

• The annual benefits, when all cars in the United States have HPR glazing, will be:

Reduction of Head Injuries with	Best Estimate	Confidence Bounds
AIS 2-4 laceration or avulsion	39,000	25,000 - 53,000
AIS 2-4 fracture	8,000	1,000 - 18,000
AIS 2-4 (any type)	47,000	31,000 - 62,000
AIS 2-4 eye, nose or mouth injury	19,000	9,700 - 29,000
AIS 1 laceration	142,000	22,000 - 315,000

Annual benefits of adhesively bonded windshields in American cars

• When all domestic cars in the United States will have windshields installed by adhesive bonding, it will save 105 lives per year (confidence bounds: 35 to 175) and 160 nonfatal AIS 3-5 (serious) casualties.

Annual benefits of Standard 212 for Volkswagens

• When all Volkswagens registered in the United States meet Standard 212, it will save 7 lives and 11 AIS 3-5 casualties per year.

Cost-effectiveness

• Since HPR windshields save 47,000 AIS 2-4 injuries and cost $55 million, they eliminate 850 AIS 2-4 injuries per million dollars of cost (confidence bounds: 560 to 1130).
• Adhesive bonding saves lives while reducing the cost of a car.

HPR Windshields

• HPR glazing greatly reduced the risk of serious lacerations of the face, scalp and mouth, fractures of the facial bones and nose and occular avulsions.
• The HPR windshield achieved its objective of steeply increasing the impact velocity needed for an occupant's head to tear and penetrate through the windshield's plastic interlayer. That explains HPR's success in mitigating the types of serious injuries listed above, because all of them are characteristically associated with penetrated windshields.
• The penetration velocities of windshields, both HPR and pre-HPR, in highway accidents were almost identical to those observed in laboratory tests. In short, HPR fully delivered in real crashes what it promised in the laboratory.
• Also, as predicted from laboratory testing, HPR had no negative side effects, such as increasing the risk of injuries associated with blunt impact trauma (concussions, contusions and complaints of pain). It had little or no effect on fatalities.
• The accident data indicate that HPR has also reduced minor lacerations significantly. Those injuries are typically associated with windshields that are cracked but penetrated. Their reduction confirms heretofore anecdotal evidence that the interlayer, in addition to resisting penetration, causes glass to crack into smaller, less injurious pieces. Nevertheless, the majority of minor lacerations still remains even after HPR.
• About half of the much smaller number of serious injuries which still occur after HPR are concussions--blunt impact trauma. The other half are lacerations and fractures: two-thirds of them did not involve windshield penetration and only the remaining third occurred at speeds too high for HPR to prevent penetration. In other words, what was once the most characteristic windshield-related serious injury has been largely eliminated.
• HPR glazing is a highly cost-effective safety device. HPR eliminated about 80 percent of penetration-related serious lacerations. No other safety device evaluated by NHTSA to date (October 1984) has come that close to eliminating the injury mechanism it was targeted to mitigate.

Windshield installation methods

• The risk or (complete or partial) occupant ejection through the windshield portal was significantly reduced when domestic manufacturers began to install windshields by adhesive bonding rather than rubber gaskets. A similar reduction was accomplished when Volkswagen began to clip its rubber gaskets to the car's windshield frame, in response to Standard 212.
• Unlike the situation that prevails with other ejection portals, the majority of persons ejected through the windshield portal received their most serious injuries before they left the passenger compartment--and would still have received them even if their ejection had been prevented. This attenuates the life-saving potential of tighter windshield installation methods; nevertheless adhesive bonding and Volkswagen's clipping of the gasket significantly reduced fatalities and serious injuries.
• Virtually all ejections through the windshield portal occur after the windshield has been partially or completely dislodged from its frame. Adhesive bonding and Volkswagen's clipping prevented ejection because they reduced the amount of windshield separation from the frame--in fact, the reductions in ejection and bond separation were nearly identical.
• Butyl tape and polyurethane sealant--two alternative adhesive bonding techniques--provide approximately equal windshield retention in crashes.
• The types of rubber gasket installations found in American cars could have or did pass Standard 212. The domestic manufacturer's shift from rubber gaskets to adhesive bonding was not necessitated by Standard 212 but was prompted by other factors, such additional safety or cost savings. By contrast, the rubber gaskets of pre-Standard 212 Volkswagens were much closer to a "pop-out" design; the gaskets were clipped to the frame at the time that Standard 212 took effect. Even the post-standard Volkswagen windshields were more loosely installed than those of pre-standard American cars.
• In cars with HPR windshields, the shift to adhesive bonding clearly did not have any negative side effects such as reducing windshield penetration velocity or increasing the risk of any type of injury to occupants who are not ejected. Neither did the shift to clipped rubber gaskets in Volkswagens.
• The accident data strongly confirm Patrick's and Trosien's sled tests with dummies and full windshield assemblies, which also showed no side effects. They do not support Rodloff and Breitenbuerger's drop tests of headforms onto less-than-fullsize glazing samples, which showed that tight bonding defeated a substantial proportion of HPR's gain in penetration velocity. It must be concluded that the drop tests simulated the interaction of the windshield and the frame in crashes less realistically than the sled tests.
• The accident data of this report show that Fargo's analysis of ACIR accident data, which was based on pre-HPR windshields and found significant negative side effects for adhesive bonding, cannot be carried over to HPR windshields.
• In cars with pre-HPR windshields, one of the accident files analyzed in this report associated significant increases in lacerations, concussions and minor blunt-impact trauma with adhesive bonding, supporting Fargo's results. But analyses of two other accident files did not confirm that association. The question or whether adhesive bonding had negative side effects in pre-HPR cars remains unresolved; it has, however, become moot because few pre-1966 cars remain on the road.
• The shift from rubber gaskets to adhesive bonding reduced the cost of purchasing and operating a car.
• Since adhesive bonding (in cars with HPR windshields) provided significant benefits without negative side effects while reducing cost, it is certainly a cost-effective safety improvement.