Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

The Use of Inertia Relief to Estimate Impact Loads

Author(s): Mark F. Nelson and Joseph A. Wolf Jr.


Source: SAE Transactions , 1977, Vol. 86, Section 3: 770404–770718 (1977), pp. 2237-2243
Published by: SAE International

Stable URL: https://www.jstor.org/stable/44644543

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

SAE International is collaborating with JSTOR to digitize, preserve and extend access to SAE
Transactions

This content downloaded from


85.229.184.213 on Sun, 03 Oct 2021 19:33:04 UTC
All use subject to https://about.jstor.org/terms
770604

The Use of Inertia Relief

to Estimate Impact Loads


Mark F. Nelson and Joseph A. Wolf, Jr.
Research Labs, General Motors Corp.

THERE ARE SEVERAL REASONS why analysts are given above is violated; that is, when the
often reluctant to perform transient dynamic period of the loads is near the structure's
analyses for impact problems. First, the cost periods. Through simple examples this paper
may be prohibitive; and second, the results may will seek to give some insight into this and to
be questionable due to the influence of the suggest guidelines for using this technique.
high frequency modes. In real life much of the The next section examines the theory of inertia
high frequency response of these modes is relief, and the section following it looks at
suppressed by damping in the structure. Yet, the results of various examples.
for impact studies, these high frequency modes The final section illustrates how inertia
often yield low-frequency stress components, and relief can be used to determine the loads from
thus to suppress them in the analysis might lead a low-speed frontal barrier impact such as
to erroneous results. might be used for certifying bumpers (Federal
One way to partially overcome these prob- Motor Vehicle Safety Standard 215). The vehicle
lems is to use the method of inertia relief. in this illustration is a space frame composed
This is an approximate analytical technique, of welded tubular elements so as to represent
used by the aerospace industry for many years the geometry of a subcompact car. This tech-
[2,6]*, which will yield exact results if the nique has been used in the design of an experi-
periods of the dynamic loads are much greater mental vehicle [1].
than the periods of the modes of interest in
the structure. Under these conditions, one THEORETICAL CONSIDERATIONS
need only compute the inertia forces due to
these applied loads, superimpose them on the In order to perform an inertia relief
existing loads, and then solve the problem analysis, an analyst must first select single
statically. values to represent the force time histories
The question arises, however, as to how which are applied to the structure. These
good this technique is when the assumption forces are then applied to a partially or

&
Numbers in brackets designate References at
end of paper.

The method of inertia relief can provide a approximate technique is exact. However,
very inexpensive way of calculating dynamic should the periods of the loads be close to
forces in a structure. In order to perform an those of interest in the structure, then the
inertia relief analysis, an analyst must first results are not nearly as good.
select single values to represent the force Several examples are examined to determine
time histories which are applied to the struc- the accuracy of the inertia relief method.
ture. These forces are then applied to a The dynamic responses of simple series- and
partially or totally unrestrained structure parallel-connected spring-mass systems are
(i.e., free-free), and the resulting rigid body analyzed and compared with inertia relief
accelerations are calculated. From these calculations. An important conclusion is that
accelerations and the mass of the structure, for masses connected in series, the error in
the inertia forces can be calculated at all the inertia relief results increases as one
points in the structure and then applied along gets farther from the applied load. The final
with the original forces. Finally, the struc- example is a space frame structure composed of
ture is restrained from rigid body motion, and welded tubular elements and representing a
a conventional static analysis is performed. subcompact vehicle geometry. It is subjected
If the periods of the applied loads are to a load which is similar to that resulting
much greater than the periods of those modes from the 8 km/h (5 mph) bumper impact safety
that would be excited in the structure, this standard.

2237
0096-736X/78/8603-2237$02.50
Copyright © 1978 Society of Automotive Engineers, Inc.

This content downloaded from


85.229.184.213 on Sun, 03 Oct 2021 19:33:04 UTC
All use subject to https://about.jstor.org/terms
2238 M. F. NELSON AND J. A. WOLF, JR.

totally unrestrained structure (i.e., free-


free) , and the resulting rigid body accelera-
tions are then calculated. From these accelera- f m - »-Q
tions and the mass of the structure, the inertia
am K m
forces can then be calculated at all points in
the structure and then applied along with the Fig. 1 - Two degree of freedom example
original forces. Now the structure is restrain-
ed from rigid body motion, and a conventional
static analysis is performed. am K m
To express this procedure analytically,
consider the undamped equations of motion
Mx + Kx = P (1)
fo - ""O
which are partitioned as follows:
fo íTõt fo lis
Mff Mfr Xf + Kff Kfr xf (Pf Fig.
(2) 2 - Free body diagram showing inertia
Ml M x Kl K x IP forces
fr rr r fr rr r ' r

where primes (f) denote matrix transposition.


Degrees of freedom of type !ff represent free or This equation can easily be solved for the
unrestrained motion while degrees of freedom of accelerations of the degrees of freedom of type
type Trł represent restrained motion. These łrł and by means of the second derivative of
latter freedoms represent the free body supports Eq. (4) the accelerations of all points in the
in the actual structure. Let us now perform a structure can be calculated.
condensation [7], or what NASTRAN [5,6] calls a Let us now introduce an inertia load vector
Guyan reduction, of these equations by assuming P^ which we will define as
that xf is a linear combination of xr. Specifi- P1 = -M x (8)
cally, let us assume that
where M is the mass matrix for the entire

Xf = -K¡P Kfr xr (3) structure. Since we know x , the acceleration


vector for all degrees of freedom in the struc-
which is the static relationship. Then ture, we can compute P* . Superimposing this
vector onto the original load vector, we can now

h * Kfr
restrain the structure and solve for the static

* * * r
* I (4) displacements by means of the equation

where I is an r x r identity matrix. With


Kff xf = Pf + PP (9)
this transformation, Eq. (2) reduces to Once Xf is known, the stresses in the individ-
ual elements can be calculated in the conven-
M* x + K* x = P* (5)
rr r rr r r tional manner.

where
The procedure just outlined is rigid
format 2 of the NASTRAN computer program.
M* = M - K' K~P - Ml K~* K_
rr rr fr ff fr fr ff fr

+frKI fr
ff ff K~p ff M„ ff K~* ff K, fr (6a)
ff fr

K* = ff
Kfr- K' fr K~P ff Kf fr (6b)
INERTIA RELIEF EXAMPLES
rr rr fr
As a first example, consider the structure
P*
r r=fr
P -
ffKl
f fr
v ' K~ļ
shownff
in Fig. 1 P. f (6c)
which consists v '
of two masses
joined by a spring. The system is unconstrained
If the supports are statically determinate, then and is loaded by a force at the left end which
K*r will be a null matrix. This is becausewe will choose to be

these supports are the minimum necessary to 1 2tt


prevent rigid body motion - any additional -r f (1 - cos pt), 0 < t < -
supports would only add redundancies to the f(t) = 2 ° - - p (10)
system. If a unit displacement were applied 0 , otherwise
to the structure in any one of the supported
(This particular forcing function was selected
directions, then rigid body motion would result,
due to its similarity to the impact loading for
and the force required to maintain the structure
an 8 km/h front end barrier test - see
in this configuration would be zero.
Fig. 10.) The reason for the 1/2 factor is
Thus, for a statically determinate struc-
ture, we may rewrite (5) as so that the maximum force will be f0. The exact
solution for the dynamic force in the spring
M*
rr
r r
x = P* (7) during the time the load is applied is

This content downloaded from


85.229.184.213 on Sun, 03 Oct 2021 19:33:04 UTC
All use subject to https://about.jstor.org/terms
THE USE OF INERTIA RELIEF 2239

clear that the inertia relief force, Fj, in the


spring will be
3.0 - c .. Jfi_
I 1 + a

fI"-īīī
2.5- Shown in Fig. 3 is the ratio of the maximum
dynamic force to the inertia relief force versus
am K m the ratio of the system natural frequency to the
f0-^CHwO forcing function frequency. Symbolically stated
2.0 -
T F(t)
R = (14)

^ 1.5 - / '
This ratio is plotted against m/p . From
this graph the reader should note the
following points.
LO - -4 (a) R is physically the dynamic overshot
or the degree to which the dynamic force exceeds
the inertia relief force.
(b) R is independent of the mass ratio a
although F and Fj are not.
(c) The maximum overshoot occurs when
a) = p which is the resonance condition.
0I
(d) Except for co/p < 0.33, R _> 1 which
0 1 2 3 4 5 6 means that the inertia relief results are

«/p generally unconservative.


(e) The unconservative error exceeds 25
Fig. 3 - Comparison of dynamic and inertia percent in the region 0.5 £ o)/p <_ 2 (this
relief results for two degree of freedom corresponds to a 33 percent overshoot) .
example
(f) Beyond co/p = 3, there is good agreement
between the dynamic and inertia relief results.
Kj * K K2 « K K3 ■ K This is as would be expected, for in this region
f - - MMł the period of the applied load is much greater
mmmm
than the period of the system.
(g) The value that R approaches as co/p
Fig. 4 - Four degree of freedom example increases depends on the single value chosen
to represent the time history of the load. For
this example choosing the peak value, f0, leads
to R 1. For other pulse shapes, the peak
value might not work as well, and possibly a
F (t)
root mean square value might be better.
2 (l+a) L l-(£)2
As a second example consider the four equal
1 27T (11) masses connected in series, as shown in Fig. 4.
- cos pt} , 0 < t < -
As before, the system is unconstrained and has
the load given by Eq. (10) applied to the left
where co is the natural frequency of the system end. The inertia relief forces in each of the
which is given by
springs are

«-J;r(i+b w m a (12)
w m a
Ft = - 0.75 f
11 o
The minus sign in (11) indicates compression. Ft = - 0.50 f (15)
1 2 o
Beyond t = 2fr/p , there is no load on the
system, and the problem reduces to a free vibra- Ft = - 0.25 f
tion one.
13 o
To obtain the inertia relief solution, the and the fundamental natural frequency is
forcing function must be applied statically
which means that a value representative of the
time history must be selected. For the time to = y 0.586 ^ (16)
history given in (10), we will use the peak Figure 5 shows the overshoot for each of the
value fQ. The rigid body acceleration is three springs. For each spring, the dynamic
f overshoot curve is very similar to that shown
a = ° . (13) in Fig. 3 for the two-degree-of-f reedom system.
m ( l+a . )
All curves show pronounced overshoot in the
and if the resulting inertia forces are applied region of w = p but small overshoot for
to the system, we will obtain the free body dia- w > 3p. The significant fact from this figure,
gram shown in Fig. 2. From this figure, it is however, is that the overshoot increases dra-

This content downloaded from


85.229.184.213 on Sun, 03 Oct 2021 19:33:04 UTC
All use subject to https://about.jstor.org/terms
2240 M.F.NELSON AND J. A. WOLF, JR.

ll o al a2 a3 a4

4 0 "
f -

nij fTi^ nrij


r' - B - R (3) F^ - -0.25 f0
Fig. 6 - Masses and springs connected in
series
30 I ' «-y 0.586 K/m
m. = m m . = 25m
VI mmmm

I j f - ►O-wO^OwO i- 1

2.5 -J ' Kj = K K2 = K K3 = K

2.0 -T ' ' K3 - K/2 m


- M/W

Fig. 7 - Four degree of freedom example


with parallel load paths
3.5 -

'i 0

2 0
A - o - R (3) , F. - -0.468 f
3.0 - ' , l3 o
/ ' -A - R (4) F. - -0.436 f
0'
/ ' ' - R (4) F. '4 -0.436 f 0
0 1 2 3 4 5 6

*/p
fj' ļ to - y 0.316 K/m
Fig. 5 - Comparison of dynamic and inertia
relief results for four degree of freedom
r ' ^ m.-m _ . .m m. = 25m
2.0-1 r V ' m.-m r - _ 2 . .m m. A-n =
example i ' ' - w/v-O 'vwyr-
4J V ' , K.-K/2 K? = K
" I V ' ' , fo~^ m3 = m
matically as one gets farther away from the I ' ' -A/WV-O-^VWV-
applied load. The first spring has an over-
15 'j '
shoot of around 1.3; the second has an over-
shoot of almost 2.0; and the last spring has
an overshoot of 3.9. ! o L/

If we examine the inertia relief forces,


we see that the force decreases as one gets
farther from the applied load. The force, for
example, in the last spring is one- third of
the force in the first spring. Thus we have
the situation that as we move farther away o

0 1 2 3 4 5 6
from the applied load, the inertia relief force
to/p
decreases, but the error (overshoot) increases.
To understand this phenomenon, let us Fig. 8 - Comparison of dynamic and inertia
examine the source of error. For a set of relief results for four degree of freedom
masses and springs connected in series, the example
forces in the springs can easily be found
once the inertia forces are known. Shown in Ft = f - m_a
Fig. 6 are a set of masses connected in series. Iļ o Ir
The accelerations of each of the masses as well F = Ft - m0a = f - m_a - m_a (19)
I2 Iļ 2 r o Ir 2 r
as the forces in each of the springs are shown.
Now from equilibrium considerations etc .

Fļ = f - mļaļ where f0 is a value representing the time


history f - say its peak value. The error
F2 = Fļ - m2a2 = f - iīiļaļ - m2a2 (17)
or difference between F^ and Fj is
etc .

Remember that the expressions in (17) are func- e. 1


1 = mf
t t1F. 1I - .
Ft I
tions of time. To get the inertia relief solu- 1 i (20)
tion, the rigid body acceleration is calculated
t ' °Tlf
= 2 J - a rt
f ' - f °
ar = (18) If the maximum of F-^(t) occurs at approxi-
mately the same time, t0 » for adjacent i
and the forces are computed from the equations then from (20) we can write that

This content downloaded from


85.229.184.213 on Sun, 03 Oct 2021 19:33:04 UTC
All use subject to https://about.jstor.org/terms
THE USE OF INERTIA RELIEF 224 1

INERTIA RELIEF CALCULATIONS APPLIED TO A


SIMPLIFIED AUTOMOTIVE STRUCTURE

The example structure used to illustrate


the method is the space frame illustrated in
Fig. 9. The space frame structure is composed
of tubular mild steel beams of rectangular cross
section, with welded joints. The structure has
been modeled using finite element beam segments
having six stiffness and three mass degrees of
freedom at each end. The joints between
elements are assumed to be rigid, except at the
Fig. 9 - Example space frame structure roof supports, where some joint flexibilities
are modeled using elastic springs. Additional
^8810 N concentrated masses representing the engine,
§ (1980 LB) front seat, and spare tire have been added for
z
dynamic studies, and are retained in the inertia
- / ' relief calculations. Due to structural and
o / V
loading symmetry about the center plane, only
§
one half of the frame has been modeled, with
í / N
g Û£ / /
Û£ / X
boundary conditions applied at the symmetry
plane to enforce symmetric structural deflec-
tions. This reduces the problem size by one
y[

f- -20 M
TIME IN MILLISECONDS
half, permitting computational cost savings.
The dynamics model described above has
Fig. 10 - Force-time history for 8 km/h 154 degrees of freedom.
frontal impact STRUCTURAL STIFFNESS AND MASS MODELING -
Structural modeling consists of two tasks:
stiffness modeling based on the elastic prop-
erties of the vehicle, and mass modeling based
on the inertial properties. For the study
described herein, the mathematical model is
e. - e. - + m. {a - a.(t )} (21)
i l-l - i r i o then generated by NASTRAN, based on input data
describing the vehicle geometry, section prop-
From the above equation and from the results erties, material thickness, elastic modulus,
we have seen thus far, it is clear that for a etc. Linearly elastic finite elements such as
set of masses in series, inertia relief is springs, beams, and plates are used to repre-
overestimating the accelerations and thus sent small homogenous regions of the structure,
underestimating the forces. This leads to the and the total structure is then represented by
error accumulation as one gets farther from the mathematically combining element properties to
applied load. give an overall stiffness and mass character-
The assumption made above that the maxima ization of the structure. Inertia properties
of the F-^(t) occur at approximately the same are represented by a lumped mass model.
time is not strictly true, for when a set of The independent displacements retained for
masses in series is impacted, there will be a inertia relief calculations include all trans-
propagating wave. However, comparing (21) to lational degrees of freedom of the stiffness
the results we have seen, this wave effect does and mass models. Free-free boundary conditions
not seem very important. are used, as inertia relief can only be calcu-
As a final example, consider a system lated for structures with one or more rigid
which contains masses connected both in series body degrees of freedom. The choice of struc-
and in parallel, as shown in Fig. 7. The tural model used is based on the input impact
applied load is as specified in (10) . This loading. For low-speed elastic impacts, the
system illustrates the accuracy of inertia entire structure, including representations of
relief for multiple load paths. frame, engine, and front sheet metal is
Shown in Fig. 8 are the inertia relief included.

results, the natural frequency of the system, APPLIED LOADS - The applied dynamic loads
and the overshoot in each of the four springs. can be obtained either from experimental barrier
We again notice the same behavior as before. force measurements or from a barrier impact
The inertia relief results are quite good for simulation program [3,4]. The peak load is
the front springs but are not very good for the used to determine stresses for inertia relief.
rear springs in the neighborhood of oo = p. This load is applied directly to the corre-
The observed phenomenon of obtaining worse sponding model grid point(s). For the low-
inertia relief results as one gets farther from speed impact of the example, the application
the applied load may not be very serious in a point is the bumper connection point. The
practical design situation. Elements far from applied loading for the example calculations
a load are probably not designed by that load has been adapted from an 8 km/h (5 mph) frontal
but rather by some other loading condition. barrier impact test, as shown in Fig. 10. A

This content downloaded from


85.229.184.213 on Sun, 03 Oct 2021 19:33:04 UTC
All use subject to https://about.jstor.org/terms
994?
M. F. NELSON AND J. A. WOLF, JR.

peak loading be multiplied by a factor of 1.25


-180 ^ to get internal loads for an 8 km/h (5 mph)
barrier impact.
vtfs 20 y' Transient response results - a transient
~í 555 _ y>755^-^, response analysis was also carried out for the
1120 V' / « ^SS - _ .^Li20Ł - -frame
- ļ ^ structure using NASTRAN . Input data
140 94a ' / « 235 s yy ' xT-^ži3 I
' ho $/As' s yy ' xT-^ži3 VvP^Sl
included the structural model described earlier,
^mRìV
and a digitized representation of the force time
-7Ź50 I ' "143°
I "282° NUMBERS REPRESENT history of Fig. 10 sampled at intervals of one
-2835 AXIAL FORCES IN millisecond. Forces at other time stations
were provided for the numerical integrations by
NEWTONS

Fig. 11 - Element axial forces for 8 km/h


linear interpolation. The response included
frontal impact
all vibration modes up to 2500 Hz. This choice
was based on preliminary studies which indicated
that a large contribution to the low frequency
peak load of 53,800 N (12,100 lb) was obtained portion of the forced response was due to the
for a vehicle with a mass of 1256 kg (weight higher frequency modes. One percent of critical
= 2768 lb) . This force is reduced to 8810 N damping was added to each mode to ensure
(1980 lb) for the 205 kg (weight = 453 lb) numerical stability. A time step size of 50 ys
half-structure to give equal decelerations. was used, giving 8 steps per cycle at 2500 Hz.
INERTIA RELIEF RESULTS - For these calcu-
A typical transient response computer run was
lations, the frame structure is supported at about five or six times as expensive as the
grid points 25 and 40, and loaded at grid point companion inertia relief calculation. Program
8, as shown in Fig. 9. For each of the beam output included grid point acceleration and
elements used in the frame, the force outputs element forces at selected output time steps.
include bending, shear, axial force, and torque. Results are presented in Table 1 in terms of an
For example, axial forces are shown in Fig. 11 overshoot factor, equal to the peak transient
for elements of the front and side frame por- response force divided by the inertia relief
tions of the structure. The load paths are force for element bending moment and axial load.
clearly represented, with the forces generally This factor is given in terms of a weighted
decreasing as one moves aft away from the impact average and a range for various locations in
point. This is especially noticeable in the the structure.
front structure, where the engine inertia is
Let
responsible for a decrease from 7250 N to 2835 N
T
in the lower beams. Further discussion of the
computed results will be deferred to a later
F^ - magnitude of peak transient response
force for the ith element,
section where comparison will be made with
transient response calculations. F± = magnitude of inertia relief force for
ESTIMATION OF INTERNAL IMPACT LOADS - The the i element,
last step of the method is to estimate the T I
internal impact loads, based on the inertia and Ri = F¿ /F , the trans
relief results. Because of the dynamic over- Then an average transient overshoot ratio is
shoot phenomenon observed in any transient given by
problem, the inertia relief calculations are
generally unconservative. For an 8 km/h
barrier impact, the force input is shown in
R=2>iV£Fi (22)
Fig. 10, and has a duration t = 0.156 s. For
a single degree of freedom model responding
to the input of Eq. (10), the dynamic over- Table 1 - Transient Overshoot Ratios for 8 km/h Fro

shoot will be 1.25 or less if o)t/2tt Bending Moment Overshoot Ratio Axial Force Overshoot Ratio

= o)/p >2.2 (Fig. 3). To satisfy this cri- Location Weighted Average Range* Weighted Average Range*
terion, the lowest vibration frequency should Front 1-09 1.01 - 1.24 1.07 0.97 - 1.28

exceed 14.1 Hz. This condition is met for the Cowl and Torque Box 1.10 1.00-1.29 1.08 0.97-1.12
space frame of the sample problem. Engine Supports 1.02 - 0.70
For the actual multi-degree of freedom Floor, Tunnel,

system, the total overshoot cannot be specified and Rocker 1.15 1.01-1.54 1.06 0.82-1.21

beforehand, as it will involve the response of Roof and Pillars 1.22 - 1.51

more than one mode. However, in problems in KickuP 1.19 1.15 - 1.21 1.19 1.07 - 1.24

aircraft and building dynamics, only a few Rear End 1.28 1.15 - 1.35 1.19 1.16 - 1.22

modes, usually those with the lowest fre-


quencies, respond enough to add significantly
to the inertia relief stresses. For vehicle
impact, a few of the higher frequency modes
contribute as well, but the overshoot is not
much more, as will be seen in the example of * Range based on element moments and forces larger than
respectively.

the next section. Therefore, it is recom- Indicates insufficient data, fewer than 5 "large" forces in location.
mended that the inertia relief results for

This content downloaded from


85.229.184.213 on Sun, 03 Oct 2021 19:33:04 UTC
All use subject to https://about.jstor.org/terms
THE USE OF INERTIA RELIEF 2243

T I similar results would be anticipated for other


Where F¿ = or F¿, arid the summation is single load problems.
carried out over a given local area or sub-
structure. Theoretically, the transient ACKNOWLEDGMENT

response forces F¿ should be used as the weight-


ing factors Fi in this equation, because they The authors wish to thank their General
represent the actual dynamic response. However, Motors colleagues Vernon F. Fishtahler, John F.
the calculation takes an especially simple form Harris, Richard W. Marks, and Knut S. Skattum
if the inertia relief forces are chosen as the for providing data for the loads and frame model
weighting factors. With this choice, (22) may used as the example structure in this paper.
be rewritten as
REFERENCES

Ř = Līļ /EF* (23) 1. J. A. Augus titus, M. M. Kamal, and


Equation (23) is used to calculate the results L. J. Howell, "Design Through Analysis of an
of Table 1. Experimental Automobile Structure," SAE Confer-
COMPARISON OF RESULTS - The weighted aver- ence Proceedings, Second International Confer-
ages for elements in the front, cowl, and torque ence on Vehicle Structural Mechanics, South-
box substructures, near the loading point, are field, Michigan, 1977.
between 1.07 and 1.10, as shown in the second 2. R. L. Bisplinghof f , H. Ashley, and
and fourth columns of Table 1. The range for R. L. Halfman, "Aeroelasticity , " Addison-Wesley
individual elements in these substructures is Publishing Company, Inc. , Reading, Massachu-
between 1.00 and 1.29 for bending moment and setts, 1955.
between 0.97 and 1.28 for axial force, as shown 3. M. M. Kamal, "Analysis and Simulation
in the third and fifth columns of Table 1. The of Vehicle to Barrier Impact," SAE Transactions,
overshoot ratio increases with increasing dis- Paper No. 700414, Volume 79, pp. 1498-1503,
tance from the front input load location. 19 70.
However, the actual force values in the roof, 4. K. H. Lin, J. A. Augustitus, and
kickup, and rear end are generally smaller M. M. Kamal, "Computer Simulation of Vehicle- to-
than in the front, as shown in Fig. 11. Thus, Barrier Impact - A User's Guide," General
these structures, remote from the frontal Motors Research Publication No. 1943, Warren,
loading, would be designed to resist loading Michigan, August 8, 1975.
applied nearer their locations. For these 5. C. W. McCormick, ed., "The NASTRAN
loads the element forces in the roof, kickup, User's Manual," National Aeronautics and
and rear end would be larger than for frontal Space Administration, Special Publication
loading, and the overshoot ratios for these 222(01), Washington, D.C., May 1973.
loads would probably be equal to or below the 6. R. H. MacNeal, ed., "The NASTRAN
proposed inertia relief multiple of 1.25. Theoretical Manual," National Aeronautics and
It should be emphasized that the ability Space Administration, Special Publication
of the inertia relief method to provide a good 221(01), Washington, D.C., December 1972.
estimator for low-speed barrier impacts is 7. M. F. Nelson, "The Use of Condensation
fundamentally an empirical finding, and cannot Techniques for Solving Dynamics Problems," SAE
be deduced solely on theoretical grounds. Transactions, Paper No. 740330, Volume 83,
However, based on these preliminary studies, pp. 1435-1444, 1974.

This content downloaded from


85.229.184.213 on Sun, 03 Oct 2021 19:33:04 UTC
All use subject to https://about.jstor.org/terms

You might also like