SLIDE 1:

Recent NHTSA Compatibility Research 

Repeatability evaluation for recent compatibility test procedures

2005 SAE Government / Industry Meeting



SLIDE 2:
LC Barrier Repeatability

- 4 56 kph rigid barrier tests with comparable vehicles

- 2004 Honda Accord
  * Belted 50th into 8 x 16 LC barrier
  * Belted 5th into 8 x 16 LC barrier 
  * Belted 95th into NCAP 4 by 9 LC barrier 

- 2003 Accord
  * Belted 50th into MGA 2 by 3 LC barrier


SLIDE 3:
Test Comparison
[tables]
* the belted 95th test is considerably heavier than the other 3 tests

* Generally tests with the 2x3 barrier are not used for AHOF estimates


SLIDE 4:
Total Force (t)

* Only around 60 ms do the heavier and lighter vehicles differ



SLIDE 5:
Initial Stiffness
The slope for the Force-deflection profiles repeated well


SLIDE 6:
Acceleration (t)
Acceleration profiles are similar, but not as consistent as total force



SLIDE 7:
Height of Force (t)
  * 8 by 16 barriers are initially similar,but diverge after 50 ms
  * 4 by 9 barrier has different HOF(t)

MGA 2 Row barrier was not evaluated for HOF(t)



SLIDE 8:
Honda Accord – Row Forces
  * Test 5062, Note the strong impact on Row 1
  * Test 5145, Note force on Row 3 has increased and Row 1 has decreased



SLIDE 9:
Accord Pit Camera - 5062



SLIDE 10:
Accord - Impulse


SLIDE 11:
Deformation and Intrusion

Only TRC tests had matching measurements 
External Deformation matches well except for C6 (right side)

The right side IP measured additional intrusion in test 5062


SLIDE 12:
Repeatability - Observations
 - HR tests did not repeat AHOF as well as hoped
	 * Engine impact strongly influenced AHOF
	 * FWDB could reduce this effect

 - 4x9 row resolution affected HOF(t)	

 - Initial Stiffness repeated well
	* Total force was not affected by LC differences between 
	  4x9, 2x3, and 8x16 barriers.

 - Need to measure impact accuracy
	* European tests have shown impact location can vary up to 20 mm

	* Did impact accuracy affect the force distributions? 


SLIDE 13:
Impact Accuracy
- IHRA researchers has been monitoring the repeatability / accuracy of vehicle impact point
  * Pre test vehicle to barrier alignment is recorded using a sticker on barrier
  * Pin on vehicle puts a hole in sticker and the X and Y offset is recorded
  * Up to 35 mm deviation has been recorded

- NHTSA has just begun to record impact accuracy
  * Working to incorporate into standardized testing



SLIDE 14:
Impact Accuracy
* Camry Test,  Center of 75 mm target was aligned with the pin pre-test
* Accord Test
	[PHOTOS]

COMMENTS: Three inch target is shown, no measurements were taken


SLIDE 15:
Full Width Deformable Barrier Tests
Comparison of Plascore and Cellbond Barrier Faces



SLIDE 16:
Concorde Test Series

Plascore was low bidder
 Cellbond was manufacturer for FWDB development tests
 FWDB’s were 1000 mm high
 FWDB was mounted 125 mm above the ground
 AHOF was corrected by subtracting the vertical offset.
 Impact accuracy was not measured for the rigid barrier test

COMMENTS:
When Aloke purchased the FWDB faces, Plascore underbid Cellbond.  Since these were the first FWDB faces that we were aware of Plascore supplying (and since the vehicle handling group had surplussed 4 old Concorde vehicles)  we couldn’t pass up a chance for some comparison testing.

The AHOF correction wasn’t initially intended and we clearly took the simplest possible approach.



SLIDE 17:
Barrier Deformation
[post-crash photos: Cellbond & Plascore]

COMMENTS: Visually, the barrier deformation appear very similar.  



SLIDE 18:
Total Force (t)

COMMENTS: Not surprisingly the rigid barrier had a higher, earlier peak for the total force.  The two FWDB tests compared well and have a very high correlation factor (0.998)


SLIDE 19:
Concorde CG Acceleration (t)
Acceleration profiles are noisy, which is common for older vehicles

COMMENTS:  The rigid barrier generated a very different acceleration profile with larger spikes.  The acceleration response of the two FWDB’s were not as identical as expected an on whole the accelerometer measurements were very oscillatory.  Since the test vehicles were over 8 years old, the test to test variations are probably not too surprising.



SLIDE 20:
Height of Force (t)
The HOF(t) are similar after correcting for impact accuracy

COMMENTS: Without correction, the cellbond HOF was clearly higher than measured for the Plascore.  However after subtracting the vertical misalignments from each HOF curve and the correlation clearly increases.  Between the two FWDB tests the correlation factor is 1.0



SLIDE 21:
Row Forces  (Uncorrected)
* Plascore Barrier

* Cellbond Barrier
(19 mm higher impact),
Increase force in row 4
Less in rows 1 and 2

COMMENTS: The force distribution seems to have changed between the rows 2 and 4.   How much of this is due to the 19 mm difference in impact alignment ?  Both tests show the peak response measured in row 3.

I cannot think of a reasonable way to correct row forces for any deviation in impact alignment.  If the entire difference between rows 2 and 3 are due to a 19 mm difference in impact deviation, it presents a significant consideration for homogeneity or CV computations



SLIDE 22:
Rigid Barrier Row Forces

The force distribution for the rigid barrier test is much different than either of the FWDB tests

COMMENTS: Almost all force is concentrated in row 3 with significantly less force measured in rows 2 and 4.  The force in row 5 remains consistent at aroung 50 k peak.


SLIDE 23:
Vehicle-to-Vehicle Repeatability

 * Full Frontal Collinear
 * Dodge Ram 1500 into
 * Honda Accord 

Since the Accord/ Ram 1500 test was repeated, this seemed like an ideal opportunity to look at the repeatability of the VTV tests.  No significant results came of this, but this was one of the items that we needed to review and use the data to get a better sense of test to test repeatability




SLIDE 24:
Test Matrix

NHTSA conducted a collinear frontal vehicle-to-vehicle test in which a driver air bag failure occurred in the Accord.

Honda repeated the Ram/Accord frontal test, providing an opportunity to examine the repeatability for this vehicle-to-vehicle test series

MGA ran both tests under identical conditions. 
Stationary Accord impacted full frontal collinear.
Both vehicle centerlines were aligned.

COMMENTS: The Accords measurements seem slightly different, but I did not get any photos or videos for test 5247, only the time series data, so I have to assume the vehicles were nominally the same.



SLIDE 25:
Ram 1500 Post Test
[post-crash photos] 
V 5247
v 5041


SLIDE 26:
Accord Post Test
[post-crash photos] 
V 5247
v 5041


SLIDE 27:
Acceleration 
Honda Accord	(CF = 0.96)
Dodge Ram 1500	(CF = 0.95)

COMMENTS:
The acceleration repeated very well for both vehicles.  There seems to be a slight time shift between the acceleration profiles for the Accords, the contact switch may not have been triggered exactly the same.  I did not compute a cross correlation between the signals, but I would expect a peak value for a few ms delay.



SLIDE 28:
Injury Measures
Honda Driver injury measures not compared

COMMENTS:
The injury measures Honda passenger repeated very well.   
The Ram driver also showed very consistent injury measured, but the Ram passenger demonstrated significantly different injury measures.  No video was available for the Ram tests conducted for Honda.



SLIDE 29:
Bullet Vehicle Deformation
<graphs>



SLIDE 30:
Target Vehicle Deformation
<graphs>


SLIDE 31:
Observations
- Vehicle measurements repeated very well

- Injury measures for Honda passenger and Ram driver repeated well

- Ram Passenger injury measures did not repeat
	No clear explanation for the difference in Ram 
	passenger femur or neck loads

- External deformation measurements repeated well
	The repeat test did not measure toe pan intrusions