Вы находитесь на странице: 1из 9

THE STRESS-STRAIN RELATIONSHIP FOR THE SKIN*

Pm TONGS
Transportation System Center, Department of Transportation, Cambridge. MA 02142. U.S.A.
and

YUAN-CHENG FUNG
Department of Applied Mechanics and Engineering Sciences/Bioengineering, University of California.
San Diego, La Jolla, CA 92093. U.S.A.

Abstract-The skin is an inelastic material. But in cyclic loading and unloading the stress-strain loop
is unique after preconditioning, and is insensitive to strain rate. We define a pseudo strain potential
for such a material so that the stresses can be derived from the expression Sij = ii(p,w)/deij, where
W is the strain energy per unit mass, p0 is the density of the material in the initial, undeformed
state. Sij is the Kirchhoff stress tensor, eij is the Green strain tensor. and the indexes i. j range over
1, 2, 3. For skin in a state of generalized plane stress, the pseudo strain potential is a function only
of e,,. ez2. ei2, where the index 1 refers to the longitudinal axis and 2 refers to the transverse axis
of the animal. It is shown in this article that the data collected by Lanir and Fung (1972) are best
represented by a pseudo strain potential of the form

P,W = f(o: e) + c exp CF(a, 41


where f(a, e) = ale:l + z&2 + 2a,e, 1e22
F(a, e) = ale:l + a2ez2 + 0sef2 + Zn,e,e, + yle: + yle: + y4e:ez + yse,e$
and a,. CQ. . .. a,, a2 . . . . y1 . ..ys and c are constants, and exp ( ) is the exponential function.

1. INTRODUCTION testing procedure for the skin, the requirement of pre-


conditioning, and the general features of the elasticity
The constitutive equation, i.e. the equation describing and hysteresis, are similar to those described by Fung
the stress-strain relationship of a material under a in 1967 and 1973.
general, three-dimensional stress field, must be known The original data were analog signals recorded
before any structural analysis can be made. In an either on magnetic tape or on x-y recorder. These
attempt to relate function with structure of organs, were digitized by Paul Sobin with an equipment
the determination of the constitutive equations of its designed by Professor Oreste Piccioni (UCSD) and
material components is the first order of business. organized into IBM card form by Conrad Fung. The
Yet, for all the biological materials, with the possible arduous task of programming and computing was
exception of the aorta, we do not know a single con- accomplished by Mrs. Ling Lin.
stitutive equation. This is partly due to the difficulties One of the principal observations is that the stress-
in testing and measuring biological tissues, and partly strain curves are practically independent of the strain
due to the difficulty in reducing the experimental data rate in loading-unloading cycles (within the range
to a mathematical expression. For this reason, the tested). It is therefore possible to express the stress-
study of the constitutive equation of the rabbit skin, strain relationship in loading (increasing strain) by a
to be reported below, is of considerable interest. unique equation. To describe unloading requires a dif-
This paper is based on the data collected by Yoram ferent equation. For either loading or unloading, a
Lanir in the senior author’s laboratory at UCSD, pseudo strain potential W(eij) can be defined so that
before he returned to Israel in January, 1973. For the stresses can be derived by differentiation:
the purpose of data collection a special-purpose test-
ing equipment was designed and constructed. The
Sij = g (i,j = 1,2,3). (1)
details of this equipment, which is nicknamed ‘Two- 1,
dimensional Biodyane” in our laboratory, have been
reported by Lanir and Fung (1974a). General obser- These pseudo strain potentials, however, do not have
vations of the elastic and viscoelastic properties of the thermodynamic meaning of the ‘strain energy
the rabbit abdominal skin, as revealed by these exper- function,’ because they depend not only on precondi-
iments, are reported in Lanir and Fung (1974b). The tioning, but also on whether the process is loading
or unloading. Nevertheless, the assumption of the
existence of a strain potential for either loading or
* Receiced 27 August 1975.
t This work was done while Dr. Tong was a Visiting unloading does simplify the mathematical problem of
Professor at UCSD. data reduction.
649
650 PIN TONG and YUAN-CHENGFUNG

We would like to have a strain potential which 2. FINITE DEFORMATION OF THE SKIN
involves only a minimum number of material con-
stants. Biological specimens vary a great deal, not In this article the rabbit skin data collected by
only from animal to animal and specimen to speci- Lanir are analyzed. A skin specimen is a thin slab
men, but also for the same specimen as the strain which, when stretched in its own plane, produces a
history changes. An empirical expression that involves state of generalized plane stresses. Let the orthogonal
many material constants is likely to experience diffi- coordinates x1,x2 be chosen in the plane of the skin,
culty in determining these constants. For this reason with xi pointing in the longitudinal (head to tail) di-
a polynomial form of I+‘(e,) is rejected because, unlike rection, and let x3 be perpendicular to x,.x2. Then
the aorta (for which Pate1 and Vaishnav (1972) used the normal stress SjJ and the shear stresses
a second- or third-order polynomial), the skin would Ssi = Si3, SJ2 = SZj all vanish. Hence if equation
require a polynomial of high degree (cf. 18th degree (1) applied, the pseudo strain potential W will not
for the mesentery, 8th degree for the skeletal muscle, contain e33,e13,e31re32,e23 because S,, = 0 implies
in Blatz et al. (1969)). For an anisotropic material aW/ae,, = 0, etc. Therefore, for two-dimensional
such a polynomial of high degree would require specimens in a state of plane stress, W is a function
many coefficients. only of e11,e:2,e22. That ei2 must appear in even
In the selection of a proper form for the strain power is well known from the general theory of con-
potential, we were guided by the fact that many bio- tinuum mechanics.
logical materials are very flexible for quite a large In finite deformation it is necessary to distinguish
range of deformation (with stretch ratios up to the stresses and strains referred to the current, deformed
order of 1.5 or 2.0). and then they become very stiff. state or the initial, undeformed state. Thus, we have
Fung (1967,1972,1973) has shown that for the mesen- Referred to deformed state: Almansi’s strain tensor,
tery and muscles (in the passive state), the following Cauchy’s stress;
equation provides an excellent fit of the experimental Referred to undeformed state: Green’s strain tensor;
data of uni-axial tension tests, except for a small Lagrangian stress tensor (or Piola-Kirchhoff stress
region very near the state of zero stress: of the first kind); KirchholI’s stress tensor (or
Piola-Kirchhoff stress of the second kind).
dT
--aaT+/?,
de The simplest explanation of this complicated matter
is probably the one given in the senior author’s book
where T is the stress per unit original area, e is the (Fung, 1965, p. 91, and pp. 436439). It is shown that
strain relative to an initially undeformed state, and Cauchy’s stress tensor and Kirchhoff’s stress tensor
a, /I are constants determined experimentally. Inta are symmetric, whereas the Lagrangian stress tensor
gration of equation (2) yields a stress T as an exponen- is not. The strain tensors are always symmetric. For
tial function of the strain, e. Subsequently, he (1973) laboratory work on large deformations it is simplest
proposed a generalized constitutive equation for the to work with the Lagrangian stress tensor, & and
biological materials in the form of a strain potential Green’s strain tensor, ei) But since Tj is not sym-
function which can be written in tensor notation as metric it is inconvenient to use it in the stress-strain
relationship. For the latter purpose we use the Kirch-
hoff stress, S,. In Fung (1965) p. 439 et seq, it is
shown that (where E is used for our e here):

eii = *(A: - 1X e22 = +(A: - 1).

(-- >
where xi,hi, /&,,, vi,> Yijtlt and /?o are constants to I ax, ax,
e
be determined experimentally, and the Latin indices 12= Z au, aa ’
i, j, k,... range over 1, 2, 3. According to the rules
of tensor notation, repetition of an index means sum- si, = ah
- w (i,j = 1,2).
mation over the index; thus, Ctiei means aeij ’
aleI + a2e2 + a3ej, etc. The first term in equation
(3) is not logically needed; but it is introduced
T,, = &&xr T,2 = 12S22, i-12 = L2&2, L = hS21,

because the experimental data appear to be “bipha- where Li, A2 are the stretch ratios in the directions
sic.” In practice, we use the second term to express of xi and x2, W is the strain energy per unit mass
the behavior of the material at a high stress level, of the material, p. is the density of the material in
and use the first term to remedy the situation at a the initial, undeformed state, a,, a2 are the location
lower stress level. On further examination, Tong of a material particle in the undeformed state, xi,
found that a good fit with the experimental data of x2 are the location of the same particle in the
the skin can be obtained by expressing W as an deformed state, and finally, 1, = ax,/aa,,
exponential function of a polynominal of the second ~~= ax2iaa2.
and third degree, omitting the first degree terms. This In the skin experiment, T,, is the tension in the
is the form to be oresented
. below. x1 direction divided by the cross sectional area per-
Stress-strain relationship for the skin 651

pendicular to xi in the initial, undeformed state. T,, (dS2/ae2)-’ the generalized Poisson’s ratios, and
is that in x2 direction. Ti2 is usually zero. We shall aSi2iaei2 the generalized shear modulus. They are
not consider Cauchy’s stress, oij, in this paper. It is functions of deformation.
related to the others by the equation
5. FITI-ING OF EXPERIMENTAL DATA
P axi p axi axjS,,,
oji = -- Tpi = --- (5)
p. da, p. aa, aa, There are many ways to determine the constants
from the experimental data. Unfortunately the com-
and is principally used in equations of equilibrium monly used method of least squares is not appro-
and motion. We shall write, for simplicity, e,, e2, S,, priate for the present problem because, at certain
S2 for eii, ez2, S,,, and Szz respectively. ranges of strains, stresses vary very rapidly. In the
present article we determine the constants by requir-
3. THE PSEUDO STRAIN ENERGY FUNCTION FOR THE ing that equations (4-13) fit exactly the experimental
SKIN data at some selected points. Since only tensile testing
We assume that the material is orthotropic. In an data are available, only those constants associated
abridged form for the two-dimensional case, we use with tensile strains can be determined. The shear
strain ei2 was zero in all experiments, hence as, and
W = i_(aIef + aze: + a3ef2 + fa,e,eJ a3 cannot be determined. Two sets of experimental
curves are needed, i.e. the stress-strain relations with
+ 3c exp(a,ef + aze: + a3efz + 2u,e,ez
e2 fixed, (called x-experiment in Tables l-2) as shown
+ he: + be: + y4e:ez + yse,ei), (6) by the solid curve in Fig. 1, and those with e, fixed,
(called y-experiments) as shown by the dotted curve
where a’s, a’s, y’s and c are constants, and e’s are
in Fig. 1. A pair of typical points A and C are located
strains referred to the undeformed smte. By replacing
on the curves in a region where stresses change
the terms fioflip,ei~~, in equation (2) by a single con-
rapidly, whereas the points B and D are located in
stant c, the stress-strain relations will be simplified
a region in which the strains are relatively small and
somewhat.
stresses increase very slowly as strains increase. The
terms associated with c and ai, a2, a4 in the stress-
4. STRESESTRAIN RELATIONS AND THE
strain relationship are very small at points B and D,
GENERALIZED ELASTIC COEFFICIENTS
hence the constants a,, a2 and a4 are determined
From equation (4) we obtain the stress-strain rela- essentially by the experimental data at B and D. Con-
tions versely, the constants c, a,, a2 and a4 are determined
essentially by the data at A and C.
S1 = g = aleI + a4e2 + cA,X
1 6. EVALUATION OF CONSTANTS

SZ = a; = aoel + azez + cAzX (7)


The number of experimental data points required
2
depends upon the number of constants to be deter-
mined. The location and the kind of data points are
S12 = F = a3e12 + ca,e,zX,
12 selected in an ad hoc fashion, with the only criterion
where that the constants should be easily determined, and
Ai = alei + a&2 + %?e: + Y&i+ + &es (8) that the resulting curves will hopefully fit all the ex-
perimental data. Two cases are considered:
A2 = a&i + a2ez + jr& + fv4e: + Ysele2 (9) (A) All the y’s are set to zero. In this case there
X = exp [a,ef + a2e: + a3ei2 + 2a4e,e2 are seven constants to be determined, namely: a,, a2,
and a,. Using subscripts A, B, etc. to
+ hei + Y24 + y44e2 + he,&. (10) a4, c, a,, a2*
denote experimental data points A, B, etc. we choose
The generalized elastic coefficients are defined as the following seven pieces of experimental informa-
tion:
aSi
c = a, + c[al + 3y,e, + y4e2 + 2A:]X (11) S1 = (S,), at A
1
(14)

as2
S2 = (S,), at A (15)
z = a2 + c[a2 + 3y2e2 + ysel + 2A$]X (12)
2
at A (16)
as,
-=- as2
ae, ae, as2/de2
- a2 ds2/ae2
- a2
at C (17)
= a4 + c[a4 + y4el + yse2 + 2A1A2]X. (13) S2 - a4el - a2e2 = ( S2 - a4e, - a2e2>c
S, = (S,JB at B (18)
One may call dSi/ae, and %,/de, the generalized
Young’s moduli (aS,/ae2) (aS,/aei)- l and (as,/&,) S2 = (S,), at B (19)
652 PIN TONG and YUAN-CHENGFUNG

Table 1. The values of strains (Green’s) and stresses (Lagrangian) at the points A, 8, C, D selected for each case,
and the calculated constants nl. Q, etc. c, is assumed equal to z2. y’s are assumed zero. The cross sectional area
for specimen 36 were A, = 0.618. A, = 0.685 cm2. Those for specimen 32 were A, = 0.398, A, = 0.444 cm2

Fixed Vi),, (WA W/de),


transverse I::i I::i Vi), (TAB (d T/de),
1 (e2)c Vi), (UC (d T/d&
I:$ (e2)D Vi), (T,), (dT/d&
Exp. No. x exp Y exp ai a2 a4 c a, = a2 a4

1.000 0.974 1.1904 0.0 88.46 40.78 1416


Xl& & 0.5107 0.0 4.349 1.48
Yl(II) -0.0254 0.4726’ 6.53 60.13 1226
-0.0254 0.1045 0.6149 1.858
7.334 20.37 6.819 l&E-3 5.982 2.241
36 1.000 0.974 1.178 0.0 92.81 32.65 1882
X1(111) & 0.5296 0.0 4.742 1.202
Y l(II1) -0.0254 0.4931 9.559 80.95 1 5333
-0.0253 0.1670 0.648 2.109
9.498 68.10 6.573 7.3E-6 6.239 2.270
1.000 0.974 1.224 0.0 82.88 38.03 1753
Xl& & 0.5199 0.0 4.134 1.169
Y l(IV) -0.0254 0.494 8.24 76.28 2449
-0.0253 0.1635 0.468 2.106
9.662 31.92 9.0154 1.63E-6 5.568 2.248
36 1.039 1.057 1.1473 0.03997 80.277 40.89 1386
Xl&Y7 0.59 0.03997 4.151 0.5946
0.0595 0.4795 1.019 71.18 3726
0.0594 0.2080 1.031 2.062
7.794 53.58 6.069 7.78E-5 4.707 0.6342
36 1.1017 1.209 1.0293 0.1068 78.77 55.34 1620
X6& Y5 0.5239 0.1068 3.836 1.492
0.2319 0.4316 25.86 86.56 5230
0.2319 0.1838 1.604 2.356
9.975 66.29 5.877 1.2lE-5 4.782 1.595
32 1.361 1.273 0.8847 0.4271 78.57 88.99 1590
x7 & Y20 0.1507 0.4271 4.547 9.322
0.3113 0.6755 35.17 118.4 2918
0.3113 0.0713 3.593 3.487
5.105 9.798 22.30 3.42E-10 14.55 4.200
32 1.313 1.17 0.9832 0.3619 105.7 94.3 2030
X6 &c Y19 0.2216 0.3619 5.531 8.00
0.1851 0.7082 21.80 113.5 2596
0.1851 0.1053 2.628 3.799
8.099 14.64 11.57 5.99&8 14.46 3.865
32 1.189 1.170 1.238 0.2076 112.9 51.41 1242
x4 & Y19 0.3463 0.2076 6.451 5.355
0.1851 0.7082 21.80 113.5 2596
0.1851 0.1053 2.628 3.799
5.073 16.61 0.9141 9.26E-4 10.19 6.847
32 1.127 1.083 1.324 0.1351 141.0 57.96 1497
X3 & Y18 0.4804 0.1351 8.291 4.853
0.0868 0.7279 15.62 111.3 1981
0.0868 0.1328 1.631 3.884
4.343 12.25 2.019 1.89E-3 10.616 5.934
32 1.058 1.000 1.363 0.0597 140.9 39.36 1171
x2 & Y17 0.4801 0.0597 8.292 3.42
0.0 0.7489 10.71 98.96 1670
0.0 0.1618 0.8662 3.968
3.383 11.74 1.272 1.76E-2 11.54 5.182

Note: (1) e is Green’s strain, and is dimensionless; (2) T is Lagrangian stress, g/cm’; (3) Subscript 1 refers to x,
or longitudinal, midline of the rabbit abdomen; Subscript 2 refers to y, or transverse direction; (4) x-experiments are
‘those in which 1, = constant; y-experiments are those in which It, = constants.
Stress-strain relationship for the skin 653

Table 2. A list of constants for specimen No. 40. In all cases it is assumed that 01~= 2s. and y4 = ys. In some sets
of data all i.‘s are assumed zero. in others y4 is assumed finite while other 7’s are zero

Transverse
Exp. 1.
No. 40
case * x-exp y-exp a2 a4

x11 &y6 1.00 1.45 3.78 23.8 2.21 0.0202 15.6 8.63 0
x13 & )‘2 1.00 1.04 3.24 11.2 0.373 0.0244 10.8 2.9 1 0
x8 & r10 1.08 1.10 4.45 13.2 0.636 0.0063 1 9.53 2.14 0
x6 & i8 1.23 1.28 7.73 17.1 1.83 0.000134 10.3 2.81 0
x13 & yll 1.00 1.00 3.79 12.7 0.587 0.00794 10.4 2.59 0
3.79 18.4 0.587 0.00794 10.4 2.59 15.6
x5 & J5 1.31 1.52 9.36 27.2 0.0 0.0000446 10.5 3.13 0
9.32 27.2 0.0 0.0000469 9.60 4.24 9.13
x6 & ~6 1.23 1.45 8.20 23.1 0.794 0.000089 10.8 2.66 0
8.18 22.9 0.664 0.00101 9.62 4.99 6.24
xl & ?‘l 1.15 1.38 6.44 23.0 0.258 0.00044 10.8 2.45 0
6.49 25.2 0.107 O.OCUl 11.8 - 1.60 27.4
x8 & yll 1.08 1.04 4.12 11.5 0.821 0.0117 9.27 2.09 0
4.13 12.0 0.841 0.0114 9.56 0.80 14.9
x9 & J9 1.04 1.18 3.30 14.4 16.6 0.0511 8.87 5.27 0
3.30 14.4 16.6 0.0511 8.87 5.28 8.85

Sz = (S,),. at D. (20) 2a, = [X, - (e2 + a4eI/aJ’] [e2 + a4eI/a&’ at C.


With equations (11-20) (remembering that y’s are 0), (25)
we have

xl_i(~-%)/(& -ate, -abeAlA atA (21)


a4
-_=
a1 K X3e1

X = exp(a,ef
- ze2)/fe1 - X3e2)lA

+ aze: + 2a,e,e,)
(26)

(27)
c = [(S, - a,e, - a2e2Yb1e1 + w2YXL (28)
xz=[($f-Q)i(& -&de1 -B2ez)lcatC (22) SA= PI - c(alel+ w9.G (29)

X3 = C(& - a4e1 - wJ(S1 - alei - cc&IA at A &I = 1s~ - c(a4el + a2e2)Xlrr (30)
(23) SD = C& - c(a4el + a2ed-% (31)
20, = [X, - (el + a4e2/aI)-‘][eI + a4e2/u&’ at A a2 = CSdedB - Wlbllle2MeI)B - kQ&d~l
(24) (32)

60-

Tronsverae extenslon mtlo= I.000 I


L
cJ
_- X(length of bodyl I
--- Y (width d body) j
I
‘I
‘I
5 =IF;‘A&y!

0
1.0 1.2 I.3 1.4 1.5 1.6 I.7 IB 1.9

Fig. I. Force vs stretch ratio curves in x-experiments (1, fixed while 2, varied; solid curves) and
in ?:-experiments (i., fixed; dotted). The choice of the points A, B, C, D is illustrated. In this example
i., = 1 in x-exp. and I, = 1 in pexp. The difference of the curves is due to anisotropy.
654 PIN TONG and YUAN-CHENG FUNG

7. RESULTS AND DISCUSSION


a4 = WE - a2e2UeJB (33)

al = CC& - wdleJ8. (34) Various sets of experimental data of the rabbit ab-
dominal skin (Lanir and Fung. 1974b) have been used
From these equations, an iteration scheme can be
to determine the constants. The coordinate axis x1
set up to determine the constants ai, a2, ah, c, a,,
or x is chosen to coincide with the midline in the
a2 and a.,. Starting with an initial guess of a’s and
longitudinal direction. The axis x2 or _ris perpendicu-
a’s, we evaluate the updated a’s from equations
lar to xi. The points A and C are chosen to be very
(21-26). The iteration process is continued with fixed
close to the maximum strains reached in the exper-
a’s until a set of convergent u’s is obtained. Then
iment, whereas the points B and D correspond to
we proceed to use equations (27-34) to evaluate the
stress levels at l/20-1/30 of the stresses of points A
updated c and a’s. If a’s converged, iteration stops;
and C. The iteration method was found to be ade-
otherwise, the updated a’s and the current u’s are used
quate to determine the constants up to 3 digits.
to start the iteration again from equations (21-26).
If the stresses (or strains) at points A and C are
70.00
much higher than those at points B and D, an excel-
lent guess of a’s can be obtained by evaluating equa-
tions (29-34) with a’s set to zero. If (ez)” < < (e,)A 60.w
and (e,)c < c (ez)c, the initial guess of a’s can be.
obtained by evaluating equations (21-26) by setting
(& = (e& = 0.
(B) yi and yz are set ro zero and y4 is set to equal
y5. In this case, there are eight constants to be deter-
mined (remember that eiz = 0). Then, in addition to
the conditions (N--20), we require

at A. (35)

The general iterative scheme is slightly more com-


10.00
plicated than that of equations (21-34). We shall
only consider the special case that the experimental
data (ez)A = (ez)a = 0. Then all the equations for 0.00
OM 1.00 1.20 1.60 I.60 200
iteration are the same as those of equations (21-34),
Stretch mtlo A
except that equations (25-27) are replaced by
70.00
I I I I I
(b) ’
2 = 2[r, (a;rT[ey) - (1 + 2c,e:)X,]A (36)

(37)

2~22 = {X2 - Cl + (Y4eh2llCe2 + a4@2 + y4

x he2 + te3/a21-LI

' (e2 + a4el/uz + y4(e,e2 + $ef)/a2jv1, at C


(38)
X = exp {ale! + ale: + e,e2C2a4 + y&i + e2111,

(39)
and equation (31) are replaced by
SD = [a2 - c(a.+ei + w2 + fr4d + y4eIe2)X1,.
W)

The iteration procedure is similar to that of case u.


It should be noted that, since (e2)” = (e2)* = 0, no Stretch mtio X
initial guess of a, and a4 is necessary. Once the initial Fig. 2. Comparison between experimental data and mathe-
guess of a’s is made, n, can be evaluated from equa- matical expression. The tensile forces F,,F, are given in
tion (24), y4 is then evaluated from equation (36X a, g. Lagrange stress TX is equal to F, divided by A, the
is evaluated from equation (37), and then uz is deter- cross sectional area I, axis. 7” (or TI 1) is related to Kirch-
hoff stress St (or S, ,) by Equation (4). S1 is given by equa-
mined from equation (38) by iteration. Actually, a2 tions (1) and (6). Squares: experimental data. Circles: From
can be solved from equation (38) by solving a quad- equations (1) and (6) with a, = az, all ys = 0. Crosses:
ratic equation. . From equation (6) with a, = a2, y1 = y2 = 0, y4 = y5 # 0.
Stress-strain relationship for the skin 655

The exact locations of the points A, B, C and D


are not very important. The value of the constants
changes very little by choosing different (reasonable)
points on any two sets of data curves. For any given
two sets of data, the theoretical curves of both the
cases a and b give very good fit to the entire range
of these two sets of experimental curves.
A typical comparison between the experimental
data and the mathematical expression is shown in
Fig. 2. In Fig. 2(a) the longitudinal force F, is plotted
against the longitudinal stretch ratio 1, while the
transverse direction is kept at the natural length so
that I, = 1, whereas in the other direction F, is plot-
ted against 1, while li, is kept at 1. The experimental
data points are shown in squares. Data computed
from our theoretical formulas (l-7) are plotted as cir-
cles for case A, (all y’s = 0), and as crosses for case
B (y4 = y5). The fit is good for the entire curve. In
Fig. 2(b) the stress in the transverse direction is plot-
ted against the longitudinal stretch ratio while the
longitudinal strain is kept at zero, and vice versa.
Again the fit between the mathematical formula and
the experimental data is good. Although these good
fittings are not surprising because the constants a,,
az, etc. are determined from these two sets of exper-
20.w 40.00 60.w smo 1w.w
imental data, the ability of the mathematical formulas
to fit the entire set of data when data at only four Tension 6 or5
points on the curves are used is nontrivial. Consider-
ing the tremendous nonlinearity of these curves, one Fig. 3. Comparison of the derivatives of tensions with re-
might say that the fitting is remarkably good. spect to stretches from experimental data with those calcu-
lated from the mathematical expression under the assump-
Additional comparison is shown in Fig. 3, in which tion z1 = tlz, all y’s = 0.
the derivatives dFJde, aF@e, etc. are plotted
against the forces F,, F,. It can be seen that the deri-
vatives aF/ae become very small when the stresses bilities: either the mathematical formula and the nu-
tend to zero. At higher stresses, dF/de vs F curves merical accuracy of the experiments and the curve-fit-
become essentially straight lines. The strain potentials ting are inadequate, or the material is not elastic. By
given in equations (6) and (7) are formulated on the convention, a body is defined as elastic if the stress
basis of these observations. and strain in the body are related by a unique, single-
However, if we use the constants determined in one valued relationship which is unaffected by strain rate
experiment with a specific protocol and precondition- or strain history. Applying this definition to our
ing to compute the stress-strain relationship in other results and considering the satisfactory accuracy of
experiments on the same specimen, but with a different the mathematical fitting for individual experiments,
protocol and a second round of preconditioning, the we must say that the constants changed from one ex-
success is not uniform. Figure 4(a) shows the compari- periment to another, and therefore the material is not
son of the transverse force F, vs longitudinal stretch elastic.
i, when the transverse direction was subjected to a It is perhaps disappointing that a program to deter-
moderate stretch 1, = 1.23, and a large stretch, to mine the elastic constants would have to end by de-
1, = 1.41. Figure 4(b) shows the corresponding cross claring that the material is inelastic. But to us this
plot: F, vs 1, in the same experiment. The agreement is indeed expected. Biological tissues are inelastic. We
between the mathematical formula and the exper- have not seen any tissue that is truly elastic. We
imental data is still reasonable when L, = 1.23, but emphasized at the beginning of this article that a pro-
in the case of 1, = 1.41 there is no correlation at all. cess of preconditioning is necessary for each exper-
Finally, in Fig. 4(c) results are shown for the same iment in order to obtain repeatable results. Precondi-
specimen subjected to initial longitudinal stretches of tioning is a repetition of the same procedure a suffi-
I, = 1.12 and 1.48, and the fit is seen to be reasonable cient number of times until a state of homeostasis
again. is obtained. Preconditioning a material to be tested
If a set of constants determined by one sequence in a new procedure is to create a new material. We
of experiments is used to predict the outcome of other have also emphasized earlier that all soft tissues show
experiments on the same specimen, and the result is features of stress relaxation, creep, and hysteresis. The
not a uniform success, then there are only two possi- apparent independence of the stress-strain curve from
656 PIN TONG and YIJAN-CHENGFUNG

the strain rate in a loading process is an empirical


fact attributable to a very broad relaxation spectrum
of the material (Fung, 1972). Taking all this into con-
sideration, we conclude that the elastic constants and
the strain potential describe only the preconditioned
(homeostatic) property of the tissue. Change of strain
procedure (protocol) alters the constants. Neverthe-
less, the empirical, simplified, and restrictive (or
generalized, depending on the point of view) stress-
strain relationships of soft-tissues are as useful to
physiology and medicine as the corresponding rela-
tionships for aluminium and steel are for the design
of airplanes and bridges.
With this understanding, we present some typical
results below.

70.00-
(0) ’
I 1 I I I 0.5 I .w I 20 Lso 2w

0
BOOO- 0
Fig. 4(c).
OF, “I x,
D X,=1.23 Fig. 4. Agreement and disagreement between the exper-
50.00- imental data and mathematical formula when the con-
0 stants determined by one experiment were used to calcu-
late stresses in another experiment. game specimen, differ-
ent transverse stretch, different preconditioning.

Table 1 lists the results of 10 experiments on speci-


mens 36 and 32. As an example of the choice of the
points A, B, C and D named above, the values of
e,, e2, T,, T2, dT,/de, and dT21de2 at these four points,
as well as the values of the constants a,, a2, a4, c,
a,, a2 and a4 are presented in a format listed at the
top of the table. Lagrangian stresses are listed here
because the raw data were obtained in the form of
tensile forces, which, when divided by the initial cross
A..
sectional areas, yield the T’s. These are then con-
Fig. 4(a). verted to the Kirchhoff stresses Si, S2, etc. in order
to evaluate the constants a,, a2, etc. From this table
70.00 the individual variations of different experiments can
L I I I I
lb) ’
be seen very clearly.
Table 2 lists the values of the constants a,, a2, etc.
for a specimen subjected to many different tests. In
some cases we assumed all y’s are zero, in others we
also evaluated Yo = Ys while assuming
yi = y2 = y3 = 0. We found it is adequate to assume
a2 = a,.
When we first attempted to reduce the mass of ex-
perimental data three years ago, we had an ambitious
plan to obtain the statistical mean values of the con-
stants that appear in the constitutive equation and
then to correlate them with such physiological par-
ameters as age, sex, etc. We thought it would be sim-
ple to determine these constants by a minimization
program comparing all experimental data of one ani-
mal against a mathematical expression. This, unfor-
tunately, cannot be done. The stresses rise very fast
at large strain, but exactly at what strain do the
stresses really shoot up depends on the test procedure
Fig. 4(b). (protocol, preconditioning). A reasonable fit-as
Stress-strain relationship for the skin 657

shown in Figs. 2(a) and 2(b) can be obtained only Patltuccl for their contributions to data reduction. and of
course. to Dr. Yoram Lanir, for having done these exper-
in a more modest approach.
iments in the first place. Without Mrs. Line. Lin’s extremely
careful and patieni work. this research could not have been
8. CONCLUSION completed.

For any set of two experiments on the rabbit skin-


one with a fixed specified strain in the J! direction REFERENCES
while the strain e, is varied and the stresses S,, S, Blatz. P. J.. Chu. B. M. and Wayland. H. (1969) On the
are measured continuously; another with a specified mechanical behavior of elastic animal tissue. Truns. Sot.
e, while e, is varied-the exponential type of strain Rheol. 13, 83-102.
potential given in equation (6) can fit the experimental Fung, Y. C. (1967) Elasticity of soft tissues in simple elon-
data very well. The skin, however, is not a truly elastic gation. Am. J. Physiol.. 213, 1532-1544.
Fung. Y. C. (1971) Stress-strain history relations of soft
material. Use of the elastic constants determined by tissues in simple elongation. Ch. 7 of Biomechanics: [rs
one set of experiments in the mathematical formula Fourldatiorls und O/+x-tires, (Edited by Fung. Y. C.).
to predict the stress-strain relationship of other sets Prentice-Hall. NJ 18l-208.
of experiments has not always been successful. The Fung. Y. C. (1973) Biorheology of soft tissues. Biorheolog?
10, 139-155.
mechanical property of the skin is influenced by the
Fung. Y. C. (1965) Foundations qf Solid Mechunics. Pren-
experimental loading procedure (preconditioning pro- tice-Hall. Englewood Cliffs. NJ.
cess). The strain Dotential is useful, but must be used Lanir, Y. and Fung. Y. C. (1974a) Two-dimensional
with understanding, properly qualified, and without mechanical properties of rabbit skin-I. Experimental
exaggeration. system. J. Biomechunics 7, 29-34.
Lanir. Y. and Fung. Y. C. (1974b) Two-dimensional
mechanical properties of rabbit skin--II. Experimental
Acknowledgements-The support of this work by USPHS results. J. Biomechnnics 7, 171-182.
NIH Grant No. HL 12494 from NHLI and the National Patel, D. J. and Vaishnav, R. N. (1972) The rheology of
Science Foundation through Grant No. 74-12664 is grate- large blood vessels. Ch. 2 in Cardiovascular Fluid D.wa-
fully acknowledged. We thank Mrs. Ling Lin and Annie rnics (edited by D. H. Bergel) Vol. 2. pp. l-64.
Ng Chen and Messrs. Paul Sobin, Conrad Fung, and Paul Academic Press, New York.

Вам также может понравиться