Event Data Recorders in the Analysis of Frontal Impacts

Abstract

Evaluations of crash protection safety features require measures for quantifying impact severity. Velocity change (delta-V) is the major descriptor of collision severity used in most real-world crash databases. One of the limitations of delta-V is that it does not account for the time over which the crash pulse occurs (delta-t). Late model GM vehicles equipped with event data recorders capture the cumulative delta-V in 10 ms intervals over the crash pulse. Deceleration can be readily calculated from these data and provides a complementary measure of severity that has not previously been available for real world crashes. The relationship between maximum delta-V and deceleration was examined for different vehicle platforms involved in real world frontal impacts and frontal crash tests. Maximum deceleration was observed to be closely correlated to the maximum delta-V.

Many late model vehicles are equipped with an event data recorder (EDR) that records the time history of the forward or longitudinal delta-V during the crash pulse (NHTSA website). The recorded delta-V is available for most late model General Motors’ vehicles and can be downloaded using the Crash Data Retrieval System (Vetronix, Santa Barbara, CA). The on-board EDR continuously monitors the vehicle’s acceleration. Every 312 microseconds, the EDR samples the accelerometer and when two successive samples exceed about 2g of deceleration, the airbag deployment algorithm is enabled (AE). Four acceleration samples are averaged over each 1.25 ms period. The resulting values are then integrated to determine the vehicle’s cumulative delta-V. Values of the computed delta-V are stored by the EDR every 10 ms (Chidester et al., 1999).

The accelerometers in many of GM’s crash recorders are uni-axial and are oriented to capture acceleration along the longitudinal axis. In real-world crashes, the forces and accelerations may well be off axis, such that the EDR will only capture the longitudinal component of the total delta-V. More modern units have biaxial accelerometers and provide both the longitudinal and lateral components of ΔV. These latter systems may also provide a measure of the principal direction of force (PDOF) acting on the vehicle during the crash.

The delta-V is available for up to 300 milliseconds on some GM Sensing and Diagnostic Modules (SDM) in both deployment and non-deployment events. On other systems, the delta-V is recorded for up to 150 milliseconds during deployment and non-deployment events. In the deployment events, these systems typically record 100 ms after the deployment command is given, and up to 50 ms before. When deployment is commanded early, the system may only record 110 ms of data which may not capture the entire crash pulse.

Research has shown that General Motors’ EDRs generally produce delta-V values within the stated uncertainty tolerances. In a study conducted by Transport Canada and General Motors, Comeau et al. (2004) examined the accuracy of the delta-V versus time recorded by GM EDRs. The study compared the data from eight separate crash tests involving three vehicle models. EDR delta-V was found to be within ±10% of the delta-V as measured by the crash test instrumentation. The authors reported that under normal circumstances, the crash investigator should expect the EDR-based delta-V to be a reasonable approximation to the actual delta-V experienced by a vehicle in a frontal crash. The delta-V values were noted to be reasonably accurate and somewhat conservative.

Niehoff et al. evaluated the performance of EDRs in laboratory crash tests across a wide spectrum of impact conditions. The authors concluded that, if the EDR recorded the full crash pulse, the EDR average error in frontal crash pulses was just under 6% when compared with crash test accelerometers. The authors also reported an average error of about 6% in the longitudinal delta-V at 100 ms. In nearly all cases, the delta-V recorded by the EDR was found to be less than the true delta-V. They further noted that the majority of the EDRs did not record the entire event and in one-third (10 of 30) of the GM crash tests, 10% or more of the crash pulse duration was not recorded.

Chidester et al. (1999) examined the performance of EDRs from model year 1998 GM passenger vehicles. Accuracy was considered to be acceptable, however occasionally the EDRs would report slightly lower velocity changes than those determined from the crash test accelerometers. Lawrence et al. (2002) evaluated the performance of GM EDRs in low-speed collisions and found that the EDRs underestimated delta-V. Large errors were observed in collisions with a delta-V of 4 km/h although these errors declined at higher delta-Vs.

While EDRs provide tremendous opportunity for the analysis of real world collisions, one must be aware of their limitations. There is a small time lag between the impact that triggers the recording, the occurrence of AE, and the commencement of the actual recording process. Severe crashes or other adverse collision configurations may disrupt the electrical power supply to the EDR which can result in some data loss. In crashes involving multiple impacts the most severe impact may not be captured. Furthermore, in some cases current EDRs will not record the complete crash pulse and the delta-V is truncated.

Delta-V is an important indicator of collision severity but does not account for the timing of the impact. Information on delta-t and deceleration is critical for comparison of crash tests to real world impacts. This paper compares delta-V and maximum deceleration in a sample of real world collisions to data from full-frontal high speed crash tests into rigid barriers. Maximum deceleration as determined from the CDR data is demonstrated to be a useful measure of crash severity particularly when the delta-V and delta-t have not been completely captured. Airbag deployment timing is also examined over a wide spectrum of frontal impact severity.

METHODS

Crash pulse data were obtained from the NHTSA website for 16 staged collisions with EDR-equipped vehicles. The full-frontal fixed barrier crashes were conducted at test speeds of 55 to 57 km/h on several different GM platforms. Reports on the recorded EDR data were available for each of the test crashes on the NHTSA website. A series of real-world motor vehicle collisions subject to conventional in-depth investigation and reconstruction techniques, were also included in this study. The individual cases were drawn from Transport Canada’s on-going collision investigation programme. They include investigations focused on airbag deployment crashes, moderately-severe side impacts, and a series of special investigations. A common element to the cases was the availability of cumulative delta-V data downloaded from event data recorders in GM vehicles. Maximum delta-V was determined from the recorded data. Depending on the SDM, the maximum delta-V was either recorded separately or had to be determined from the cumulative delta-V array. The average deceleration was calculated for each 10 ms interval. The maximum deceleration was determined and compared to the maximum delta-V. Airbag deployment timing was obtained from the recorded data when it was available and compared to the maximum delta-V. Real world collisions involved in this study include a sample of 58 frontal impacts involving 1996 to 2005 model Sunfires and Cavaliers over a wide severity spectrum. In addition, 35 frontal impacts with a minimum delta-V of 30 km/h were analyzed.

RESULTS

CRASH TESTS

Table 1 shows the maximum delta-V and the maximum 10 ms deceleration for the test crashes. The average deceleration was calculated for each 10 ms interval during the crash pulse and the maximum determined. The maximum decelerations ranged from 25.4g in test # 4472 to 39.5g in test # 4923. The average decelerations are shown in Table 2. In some cases the vehicle was still undergoing moderate deceleration in the last recording interval and there would have been additional delta-V.

Table 1.

Crash test summary

Test #	Case vehicle	Test speed (km/h)	Maximum delta-V (km/h)	Maximum 10 ms deceleration (g)	AE to deploy command (ms)
4923	2004 Cadillac SRX	56.7	−57.4	−39.5	5
4244	2002 Chev Trailblazer	56.5	−57.9	−35.3	5
4238	2002 Cadillac De Ville	56.8	−59.9	−32.5	17.5
4487	2003 Saturn Ion	56.1	−62.1	−32.0	10
4899	2004 Chevrolet Colorado	57.0	−58.4	−31.1	2.5
4198	2002 Saturn Vue	56.3	−61.6	−31.0	NA
4775	2004 Pontiac Grand Prix	55.9	−62.1	−30.0	10
4985	2005 Chevrolet Equinox	56.3	−59.3	−30.0	10
3952	2002 Buick Rendezvous	56.6	−66.6	−29.0	NA
4567	2003 Chevrolet Suburban	56.3	−60.3	−28.2	2.5
4549	2003 Chevrolet Tahoe	56.3	−58.4	−28.2	2.5
4464	2003 Chev Avalanche	56.6	−59.4	−26.8	5
3851	2003 Chev Avalanche	56.6	−62.8	−26.0	NA
4445	2003 Chevrolet Cavalier	55.9	−58.6	−26.0	5
4918	2004 GMC Envoy XUV	56.7	−54.4	−25.4	5
4472	2003 Chevrolet Silverado	55.9	−57.9	−25.4	7.5

Table 2.

Average 10 ms decelerations (g) for crash tests

Test #	10ms	20ms	30ms	40ms	50ms	60ms	70ms	80ms	90ms	100ms	110ms	120ms
3851	0.0	−12.0	−16.0	−20.0	−20.0	−18.0	−26.0	−26.0	−18.0	−12.0	−10.0
3952	−11.6	−11.0	−17.0	−19.0	−29.0	−27.0	−29.0	−25.0	−15.0	−5.0
4198	0.0	−0.6	−9.0	−13.0	−17.0	−23.0	−13.0	−31.0	−31.0	−23.0	−11.0	−3.0
4238	0.0	−7.1	−12.7	−7.1	−16.9	−22.6	−25.4	−18.4	−32.5	−18.4	−8.5
4244	−9.9	−24.0	−31.1	−35.3	−16.9	−16.9	−15.5	−9.9	−2.8	−1.4
4445	−2.0	−10.0	−10.0	−20.0	−22.0	−20.0	−24.0	−26.0	−20.0	−8.0	−4.0
4464	−8.5	−16.9	−16.9	−18.4	−15.5	−26.8	−26.8	−14.1	−14.1	−9.9
4472	−9.9	−19.8	−16.9	−14.1	−21.2	−25.4	−15.5	−14.1	−16.9	−9.9
4487	−8.0	−10.0	−14.0	−16.0	−22.0	−32.0	−30.0	−24.0	−14.0	−6.0
4549	−11.3	−16.9	−18.4	−18.4	−16.9	−28.2	−22.6	−12.7	−12.7	−7.1
4567	−9.9	−15.5	−16.9	−18.4	−16.9	−28.2	−24.0	−16.9	−15.5	−8.5
4775	−6.0	−14.0	−10.0	−12.0	−16.0	−30.0	−24.0	−20.0	−22.0	−16.0	−6.0
4899	−24.0	−25.4	−18.4	−28.2	−31.1	−12.7	−12.7	−9.9	−2.8
4918	−14.1	−18.4	−22.6	−25.4	−18.4	−19.8	−22.6	−11.3	−1.4
4923	−11.3	−16.9	−19.8	−25.4	−25.4	−39.5	−15.5	−1.4	−5.6	−1.4
4985	−6.0	−8.0	−14.0	−16.0	−14.0	−30.0	−28.0	−24.0	−16.0	−10.0	−2.0

In the test crashes maximum deformation occurs when the delta-V equals the impact speed and the vehicle’s centre of mass reaches a temporarily stopped condition. Review of the EDR data indicated that maximum deformation was reached in the 70 ms to 100 ms range for the test crashes. Once maximum deformation occurs the loading phase of the impact ends and the test vehicle enters the unloading or rebound phase of the impact. In all cases there was a significant reduction in the average deceleration in the intervals following maximum deformation.

FRONTAL IMPACTS SINGLE PLATFORM

There were 58 real world frontal impacts involving 1996 to 2005 model year Chevrolet Cavaliers and Pontiac Sunfires. These cases are summarized in Tables 3 and 4. This sample was restricted to frontal impacts involving airbag deployment where the time from AE to deployment command was recorded. The EDR recorded from 110 ms to 150 ms of data on the 2001 model and newer vehicles (N = 49) and up to 300 ms on the 1999 and earlier models (N = 9). There were no 2000 models in the sample as the timing of the deployment command was not recorded in these vehicles. The maximum delta-V ranged from 7.8 km/h to 90 km/h. In the majority of these deployment events the maximum delta-V was under 20 km/h. Firing times for airbag deployment varied considerably with much longer times between AE and initiation of the deployment command occurring in low delta-V impacts.

Table 3.

Summary of real world crashes

Case #	Case vehicle	Object struck	Max. forward delta-V (km/h)	Max. 10 ms decel. (g)	AE to deploy command (ms)
ASF31312	1997 Sunfire	Truck	−90.0	−50.0	10.0
ASF41620	2003 Sunfire	Car	−83.3	−46.0	7.5
ACR51646	2002 Cavalier	Car	−48.0	−32.0	15.0
ACR61617	2002 Cavalier	Car	−47.3	−28.0	15.0
ACR51356	2002 Cavalier	Car	−48.7	−26.0	10.0
ACR71604	1999 Sunfire	Car	−65.7	−23.0	15.0
ASF39630	1998 Sunfire	Minivan	−47.3	−20.0	13.8
ACR51810	2002 Cavalier	Car	−38.1	−18.0	15.0
ACR61619	2002 Cavalier	Car	−42.4	−18.0	10.0
ACR61142	2003 Cavalier	Car	−36.7	−18.0	12.5
ACR61618	2002 Sunfire	Minivan	−33.2	−18.0	30.0
ACR51646	1996 Cavalier	Car	−28.6	−18.0	22.5
ACR61325	2003 Cavalier	Guardrail	−35.3	−16.0	35.0
ACR61206	2002 Cavalier	Tree	−28.9	−14.0	30.0
ACR51644	2001 Cavalier	Minivan	−17.6	−14.0	67.5
ACR61304	2002 Sunfire	Car	−28.2	−13.7	17.5
ACR61628	2003 Cavalier	SUV	−28.2	−12.0	20.0
ACR61347	2003 Cavalier	Pole	−26.1	−12.0	65.0
ACR61653	2004 Sunfire	Minivan	−25.4	−12.0	25.0
ACR51350	2002 Cavalier	Car	−25.4	−10.0	25.0
SID51210	1999 Cavalier	Car	−24.7	−10.0	28.8
ACR61368	2005 Cavalier	Guardrail	−25.4	−10.0	20.0
ACR61652	2004 Cavalier	Minivan	−19.8	−10.0	15.0
ACR51927	2002 Sunfire	Minivan	−18.4	−10.0	32.5
ACR61602	2002 Cavalier	Pole	−12.7	−10.0	70.0
ACR61217	2003 Cavalier	Pickup	−16.9	−10.0	72.5
ACR61364	2004 Cavalier	Tractor	−17.6	−10.0	35.0
ASF31619	1998 Sunfire	Minivan	−25.1	−9.0	55.0
ACR61631	2003 Sunfire	Car	−21.2	−8.0	25.0
ACR61624	2004 Sunfire	Car	−16.9	−8.0	27.5
ACR61806	2002 Sunfire	Car	−12.7	−8.0	27.5
ACR51655	2002 Cavalier	Pole	−25.4	−8.0	25.0
ACR51922	2002 Sunfire	Barrier	−24.7	−8.0	72.5
ACR61637	2004 Cavalier	Minivan	−19.8	−8.0	17.5
ACR51272	2001 Sunfire	Pickup	−19.8	−8.0	37.5
ACR51645	2001 Cavalier	Car	−18.4	−8.0	20.0
ACR61112	2003 Cavalier	Car	−17.6	−8.0	12.5
ACR51806	2002 Sunfire	Car	−18.4	−8.0	47.5
SID51810	1998 Sunfire	Car	−15.2	−7.0	55.0
ACR51348	2002 Cavalier	Car	−16.9	−6.0	40.0
ACR51926	2002 Cavalier	SUV	−12.7	−6.0	87.5
ACR61113	2002 Cavalier	Car	−16.9	−6.0	40.0
ASF31841	2001 Sunfire	Car	−12.0	−6.0	45.0
ACR61606	2003 Sunfire	Car	−10.6	−6.0	30.0
ACR61819	2002 Cavalier	Minivan	−14.8	−6.0	62.5
ACR61226	2002 Sunfire	Car	−14.8	−6.0	52.5
ACR61611	2002 Sunfire	Car	−10.6	−6.0	27.5

Table 4.

Summary of real world crashes

Case #	Case vehicle	Object struck	Max. forward delta-V (km/h)	Max. 10 ms decel. (g)	AE to deploy comm. (ms)
ACR61321	2002 Sunfire	Minivan	−14.1	−6.0	25.0
ACR61647	2005 Cavalier	Pickup	−7.8	−6.0	25.0
ASF31817	1999 Sunfire	Car	−9.5	−5.0	28.8
ASF31811	1999 Cavalier	Pickup	−13.1	−5.0	38.8
ACR61816	2003 Sunfire	Car	−11.3	−4.0	50.0
ASF31834	1996 Sunfire	Car	−9.5	−4.0	47.5
ACR51815	2002 Cavalier	Car	−8.5	−4.0	37.5
ACR51653	2001 Cavalier	Car	−7.8	−4.0	25.0
ACR61120	2003 Cavalier	Barrier	−12.7	−4.0	47.5
ACR61357	2004 Sunfire	Tree	−12.7	−4.0	27.5
ACR51638	2001 Sunfire	Wall	−7.8	−4.0	40.0
RR948	2005 Chev Malibu	Truck	−114.5	−65.9	14.0
ACR61202	1999 Chev S10 – V1	Pickup	−90.1	−43.0	7.5
ACR61605	2000 Pont Grand Am	Car	−85.6	−47.0	NA
ACR61236	2003 Chev Express	Pickup	−80.2	−25.4	NA
CFCP9611	1996 Saturn SL1	Car	−74.9	−29.0	12.5
ROPS9601	1999 Buick Century	Car	−69.5	−31.0	NA
ASF29610	1999 GMC Sierra	Bus	−66.1	−26.0	5
ACR61668	2005 Pontiac Pursuit	Tree	−65.4	−18.5	96.0
ACR61831	2004 Chev Silverado	Tree	−64.9	−31.1	NA
ASF41203	2000 Buick Century	Tree	−64.1	−31.0	NA
ASF41617	2003 Chev Avalanche	Truck	−61.8	−29.7	10.0
ACR51274	2001 Saturn SL1	SUV	−61.1	−38.0	10
ASF51101	2002 Pontiac Montana	Pole	−58.5	−25.0	NA
LTVS1293	1997 GMC Sierra	Truck	−57.9	−28.0	23.75
ACR51277	2001 Pontiac Aztek	SUV	−56.1	−23.0	NA
SID51202	1998 Saturn SW1	Car	−55.8	−35.0	11.25
ACR61219	2002 Pontiac Venture	Minivan	−54.8	−19.0	NA
CFCP9611	2000 Chevrolet Malibu	Car	−52.5	−27.0	NA
ACR51917	2001 Chevrolet Impala	Car	−51.0	−24.0	NA
SID61902	2003 Pont Grand Am	Car	−49.8	−22.0	7.5
ASF21819	1998 Chev Silverado	Car	−48.8	−27.0	10
ASF51606	2000 Pont Grand Prix	SUV	−48.3	−19.0	NA
RS030	1998 GMC 1500	Pickup	−46.9	−21.0	7.5
CFC21605	1996 Buick Skylark	Car	−45.2	−24.0	13.8
RQ376	1997 Buick Le Sabre	Pole	−40.9	−20.5	11.3
SID61601	2002 GMC Savana	Car	−40.9	−18.0	52.5
ACR61613	2000 Pont Grand Am	Car	−40.0	−19.0	NA
CFC21603	2000 Chevrolet Blazer	Car	−39.5	−24.0	10.0
ACR61645	2004 Pontiac Montana	SUV	−38.8	−20.0	47.5
ACR51652	2002 Saturn SL1	Car	−38.1	−14.0	12.5
ACR61615	2003 Pont Grand Prix	Car	−36.7	−16.0	12.5
ASF51605	2000 GMC Sierra	Pickup	−34.5	−15.0	NA
RS722	2005 GMC Envoy	Car	−32.8	−17.4	NA
ACR61609	2002 Saturn Vue	Pickup	−31.0	−11.0	NA
ACR71619	2006 Hummer H3	Pickup	−30.7	−11.6	20.0

FRONTAL IMPACTS MULTIPLE PLATFORMS

There were 35 real world frontal impacts involving multiple platforms other than Sunfires or Cavaliers. These cases were restricted to deployment events with a maximum delta-V above 30 km/h and are summarized in Table 4. The maximum delta-V ranged from 30.7 km/h to 114.5 km/h. Many of the frontal impacts were severe and in 19 cases (54%) the maximum delta-V exceeded 50 km/h. Recording duration varied from 100 ms to 300 ms depending on the characteristics of the SDM. In the 10 cases involving 1999 or earlier vehicles, the SDM was capable of recording 300 ms of delta-V after AE. Power loss, indicated by zero ignition cycles at the deployment event, occurred in 7 of theses 10 cases (70%) during the 110 ms to 200 ms time range. The cumulative delta-V suddenly went to zero and the driver's seatbelt status was recorded as unbuckled in all 7 cases. In 3 cases summarized below (ACR61668, ACR71619 and RR948) the late model vehicles had more advanced SDMs capable of capturing 300 ms of data including 230 ms after the deployment command and up to 70 ms before. The SDMs in the remaining 22 cases captured between 100 and 150 ms of data and in 18 cases it was 120 ms or less. The last 10 ms interval of data capture in these 22 cases had average decelerations ranging from 1g to 15.5g with an average of 6.1g. In some of these cases there may have been significant delta-V after data recording ended.

RESULTS – CASE STUDIES

ACR71619

The front end of a 2006 Hummer H3 struck the front right side of a pickup truck at a rural intersection. The maximum averaged crush was 25 cm at the left front corner. The SDM in this vehicle was equipped with bi-axial accelerometers and the average force direction was 335 degrees. The maximum forward delta-V was 30.7 km/h (19.1 mph) at 240 ms. A subset of the cumulative delta-V from 50 to 180 ms is shown in Table 5. While the crash pulse was relatively long, the primary impact occurred between 50 and 150 ms and accounted for 87% of the forward delta-V

Table 5.

EDR Data

	Delta-V (km/h)	Accel. (g)
50	0	0
60	−2.0	−5.8
70	−6.1	−11.6
80	−9.2	−8.7
90	−13.3	−11.6
100	−16.4	−8.7
110	−20.5	−11.6
120	−22.5	−5.8
130	−25.6	−8.7
140	−26.6	−2.9
150	−26.6	0.0
160	−26.6	0.0
170	−28.7	−5.8
180	−28.7	0.0

ASF41617

A 2003 Chevrolet Avalanche sustained a severe full-frontal impact with extensive underride when it collided head-on with a heavy truck on a rural road. Maximum crush at the front bumper was 44 cm while maximum crush at the hood was 190 cm. A maximum delta-V of 62.4 km/h (38.8 mph) was recorded for the deployment event at 105 ms. The cumulative delta-V and deceleration are shown in Table 6. The vehicle may still have been in the loading phase of the impact at the end of the recording. However, the deceleration pulse suggests that the majority of the impact was captured.

Table 6.

EDR Data

Time (ms)	Delta-V (km/h)	Accel. (g)
10	−1.0	−2.8
20	−4.0	−8.5
30	−6.5	−7.1
40	−17.0	−29.7
50	−25.9	−25.4
60	−34.9	−25.4
70	−43.9	−25.4
80	−50.4	−18.4
90	−54.9	−12.7
100	−58.4	−9.9
110	−61.8	−9.9
120	NA	NA

ASF41620

A 2003 Pontiac Sunfire sustained a very severe offset- frontal impact when it collided head-on with a passenger car on a rural road. The maximum crush was 166 cm at the left corner of the front bumper, tapering to 68 cm at the right corner. A maximum delta-V of 83 km/h (51.8 mph) was recorded for the deployment event at 105 ms. The cumulative delta-V and deceleration are shown in Table 7. The magnitude of the deceleration had dropped considerably by the last interval. The vehicle was likely in the rebound phase of the impact at the end of the recording.

Table 7.

EDR Data

Time (ms)	Delta-V (km/h)	Accel. (g)
10	−2.1	−6.0
20	−5.6	−10.0
30	−11.7	−17.3
40	−23.6	−33.6
50	−39.5	−45.1
60	−49.5	−28.3
70	−64.2	−41.6
80	−69.9	−16.0
90	−75.5	−16.0
100	−79.8	−12.0
110	−82.3	−7.3
120	NA	NA

ASF51606

A 2000 Pontiac Grand Prix 4-Dr sustained a severe offset-frontal impact when it collided head-on with an SUV that was traveling in the wrong direction on a median-divided highway. The maximum crush was 110 cm at the left corner of the front bumper tapering to 10 cm at the right corner. The cumulative delta-V and deceleration are shown in Table 8. The delta-V reached 48.3 km/h (30.0 mph) at 120 ms and the decline in the magnitude of the deceleration suggests that the impact was nearly complete when the recording ended. The SDM in this vehicle did not record a maximum delta-V.

Table 8.

EDR Data

Time (ms)	Delta-V (km/h)	Accel. (g)
10	0.0	0.0
20	−3.1	−8.9
30	−5.6	−7.0
40	−10.9	−15.0
50	−17.6	−19.0
60	−23.6	−16.9
70	−29.6	−17.0
80	−33.5	−11.0
90	−38.8	−15.0
100	−42.7	−11.0
110	−45.9	−9.0
120	−48.3	−7.0

ACR61668

A 2005 Pontiac Pursuit travelled onto the roadside where it contacted several bushes before striking a tree stump. The maximum crush was 33 cm at the front bumper. The vehicle’s SDM was equipped with biaxial accelerometers and recorded both longitudinal and lateral accelerations. The delta-V reached 65.4 km/h (40.7 mph) at 300 ms and there was sill moderate deceleration at the end of the recording. The impact with the stump likely occurred around 180 ms and the vehicle had already undergone a 15.3 km/h forward delta-V. The long delta-t and moderate maximum deceleration provided added insight into crash severity.

RR948

A 2005 Chevrolet Malibu Maxx travelled into the oncoming lane and collided head-on with a heavy truck.. The maximum delta-V was 114.5 km/h (71.2 mph) at 250 ms. Initial engagement occurred around the 50 ms interval and the delta-t was quite long at 200 ms. However, the vehicle underwent 90% of the delta-V within 140 ms after initial contact and by the 180 ms interval the deceleration had dropped off significantly from the maximum. The vehicle was driven rearwards and there was sustained contact and likely multiple impacts with the heavy truck. The cumulative delta-V and deceleration for ACR61668 and RR948 are shown in Table 9.

Table 9.

EDR Data

Time (ms)	DV (Km/h)	Acc.(g)	DV (Km/h)	Acc.(g)
	ACR61668		RR948
10	−2.2	−6.2	0.0	0.0
20	−2.2	0.0	0.0	0.0
30	−4.4	−6.2	0.0	0.0
40	−5.5	−3.1	0.0	0.0
50	−7.6	−6.2	0.0	0.0
60	−7.6	0.0	−2.2	−6.3
70	−9.8	−6.2	−5.5	−9.2
80	−9.8	0.0	−12.0	−18.6
90	−10.9	−3.1	−22.5	−29.7
100	−10.9	0.0	−45.8	−65.9
110	−12.0	−3.1	−54.1	−23.5
120	−12.0	0.0	−63.7	−27.2
130	−12.0	0.0	−75.2	−32.7
140	−13.1	−3.1	−82.9	−21.6
150	−13.1	0.0	−87.3	−12.3
160	−13.1	0.0	−92.7	−15.4
170	−14.2	−3.1	−98.2	−15.4
180	−15.3	−3.1	−101.4	−9.2
190	−16.4	−3.1	−104.7	−9.2
200	−20.7	−12.3	−106.9	−6.2
210	−25.1	−12.3	−108.0	−3.1
220	−31.6	−18.5	−109.1	−3.1
230	−38.2	−18.5	−110.1	−3.1
240	−44.7	−18.5	−112.0	−5.3
250	−50.2	−15.4	−114.5	−7.1
260	−54.5	−12.3	0.0	0.0
270	−58.9	−12.3	0.0	0.0
280	−62.2	−9.3	0.0	0.0
290	−63.2	−3.1	0.0	0.0
300	−65.4	−6.2	0.0	0.0

DISCUSSION

MAXIMUM DELTA-V

Incomplete capture of the crash pulse is an important consideration when determining maximum delta-V from EDR data. In the majority of the high severity real world frontal impacts where data capture was limited to 150 ms or less, a significant average deceleration was calculated in the last interval of the cumulative delta-V, indicating that the crash pulse was probably incomplete. Even when the SDM was capable of recording 300 ms of data, power loss was frequently encountered before the crash pulse was complete.

Establishing maximum delta-V and delta-t can be difficult in real world collisions as vehicles are often undergoing deceleration before and/or after an impact. For example, a vehicle that leaves the road may encounter a hazard shortly before or after the major impact. Multiple contacts can occur over a short time interval. The added delta-V that happens outside of the major impact can be significant as illustrated in cases ACR61668, ACR71619 and RR948. Maximum delta-V and delta-t depend on the criteria chosen to determine the start and end points of the impact which are not always readily apparent.

A dramatic decline in the magnitude of the deceleration was often observed at the end of the recorded crash pulse in both crash tests and real world crashes. In the crash tests, this decline represented a transition from the approach or loading phase of the impact into the rebound or unloading phase. While a vehicle may undergo considerable delta-V during the rebound phase of the impact, only in low severity impacts does it tend to be a large percentage of the total delta-V. In real world crashes, one often does not know for certain whether this decline in the magnitude of the deceleration represents a similar transition into the rebound phase. A sudden decline does not necessarily mean that the collision is almost over and could just represent a temporary lull before another violent loading.

MAXIMUM DECELERATION

Figure 7 show the relationship between maximum 10 ms deceleration and delta-V for the crash tests and all the real world collisions. The maximum deceleration was observed to be closely related to the maximum delta-V and in some cases may prove useful for estimating delta-V. However, for a given delta-V there can be considerable range in the maximum deceleration. For example, in ACR61668 the maximum delta-V was 65.4 km/h yet the maximum 10 ms deceleration was only 18.5 g. This was much smaller than in the crash tests and was indicative of the very long delta-t.

Figure 7.

Comparison of deceleration and delta-V

Delta-V is of fundamental importance in crash reconstruction and as a measure of crash severity. Maximum deceleration averaged over a suitable time period is a complementary severity indicator with some important advantages due to its ease of calculation and relationship with delta-t. In some non-GM vehicles, the EDR captures less than 100 ms of data making it difficult to establish the maximum delta-V from the EDR data although maximum deceleration can be readily established. Incomplete data capture is less important because the maximum deceleration generally occurs in the middle of the pulse.

Figure 8 compares the forward deceleration during the first 100 ms of the crash pulse for the crash tests and the real world impacts. The forward decelerations are the average values in the time intervals for each category shown. Thus, the crash test deceleration pulse is the average of all the crash tests under study. The average deceleration pulse for the severe real world frontal impacts appeared quite similar to that for the crash tests. Full frontal crash tests into rigid barriers appear to provide delta-t and maximum decelerations reasonably similar to real world frontal impacts. When the time from AE to deployment command exceeded 40 ms the crashes involved low decelerations.

Figure 8.

Average decelerations during crash pulse

AIRBAG DEPLOYMENT

Figure 9 shows the time when airbag deployment was commanded for all frontal impacts in the study where these data were available. While airbag deployment was commanded extremely early for the crash tests, typically in less than 10 ms, much later deployments were common in the real world crashes. Figure 10 compares the maximum deceleration to the time when airbag deployment was commanded. A low maximum deceleration was often associated with long airbag firing times.

Figure 9.

Time from AE to deployment command

Figure 10.

Maximum deceleration and time to deploy (ms)

CONCLUSIONS

Delta-V is the primary description of severity used in most crash databases and has traditionally been estimated using standard collision reconstruction techniques. EDRs provide a new source of objective data, directly related to a vehicle’s crash situation. In particular, specific data are provided on the crash pulse experienced by a vehicle equipped with an EDR, yielding measurements of delta-V as a function of time and, frequently include a supplementary measure of the vehicle’s maximum delta-V.

The present study has shown that, for General Motor’s vehicles, post-processing of the data obtained from EDRs can provide considerable insight into various aspects of real world collisions, including the timing of crash pulses, measures of vehicle decelerations, and airbag firing times. Indeed EDR data are the only means by which crash severity and the collision performance of advanced restraint systems, including devices such as seat belt pre-tensioners and airbag systems, can be directly monitored in the field.

Crash pulse data in the form of delta-V as a function of time can be readily converted into a time history of the vehicle acceleration. In particular, we have seen that the maximum acceleration that occurs during the crash phase of the collision is closely correlated to the vehicle’s ultimate change in velocity. Current EDRs are frequently subject to memory limitations whereby a completed recording of the crash pulse is not obtained, and hence the vehicle’s delta-V may well be underestimated. There is normally no such limitation on computing the maximum acceleration of the vehicle since this occurs well before the end of the crash pulse. Consequently, for vehicles equipped with EDRs, the computed acceleration may become a useful measure of crash severity for automotive safety researchers wishing, for example, to explore the relationship between collision severity and occupant injury.

EDRs also give valuable insight into the collision performance of airbag systems by providing data elements on deployment characteristics. For example, we have seen a disconnect between airbag firing times occurring in staged collisions, i.e. regulatory tests and staged collisions nominally intended to provide consumer information with respect to vehicle safety, and those achieved in the real world. The considerably longer firing times seen in the latter raises a note of caution for regulators and may indicate a need to consider alternatives to collisions with rigid barriers in order to more fully exercise airbag sensing and control systems and minimize the potential for late deployment.

Figure 1.

2006 Hummer H3

Figure 2.

2003 Chevrolet Avalanche

Figure 3.

2003 Pontiac Sunfire

Figure 4.

2000 Pontiac Grand Prix

Figure 5.

2005 Pontiac Pursuit

Figure 6.

2003 Chev Malibu Maxx