ht. J. Pm. Ves.

& Piping 62 (1995) 313-320

0 1995 Elsevier Science Limited
Printed in Northern Ireland. All rights reserved
ELSEVIER )308-0161(94)00026-3 030%0161/95/$09,SO

Destructive tests of 15CDV6 steel rocket

motor cases and their application to
lightweight design
A. P. Beena, M. K. Sundaresan & B. Nageswara Rao
Structural Design and Analysis Division, Structural Etlgineeritlg Group, Vikram Sarabhai Space Centre,
Trivandrum-695022, India

(Received 13 August 1993;accepted 15 April 1994)

Experimental stressinvestigationsare carried out on the burst testsof rocket

motor casesmadeof 15CDV6 steel.The biaxial ultimate strength of the motor
casesis evaluated. The feasibility of using the biaxial strength as the basisof
designis examined.

INTRODUCTION Technological excellence is required to develop

a highly reliable design of a lightweight solid
Two types of failure criteria’-3 are recognised by rocket motor case for the overall performance of
the rocket industry, yielding and fracture. Failure a launch vehicle or a missile system. The
due to yielding is applied to a criterion in which distortion energy theory’,” offers theoretical
some functional of the stress or strain is e:xceeded justification for using true stress in a biaxial stress
and fracture is applied to a criterion in which an field such as that which occurs in the cylindrical
already existing crack extends according to an portion of the rocket motor case. However, the
energy balance hypothesis. Experimentation with basic derivation of the theory is based on
a variety of materials would show that the theory Hooke’s Law, thus limiting the theoretical
works well for certain materials but not very well applications, up to the yield point of the material.
for others. Such attempts to devise failure The rational design of motor cases under internal
theories and experimentally verify them has led pressure requires the study of two modes of
to many failure theories,4m8viz., maximum normal failure. The first possibility of failure occurs when
stress theory, maximum shear stress theory, the deformation becomes excessive and there is
maximum normal strain theory, total strain a possibility of permanent deformation. The
energy theory, distortion energy theory and second type of failure occurs at a higher pressure
Mohr’s failure theory. and takes the form of bursting of the motor case.
Selecting the factor or margin of safety, The purpose of this note is to carry out
predicting the failure mode and determining the experimental stress investigations on burst tests
associated failure strength and calculating from of rocket motor cases made of a high strength,
these the design stress, are important steps in low alloy 15CDV6 steel. Strain data recorded
making design tools of combined stress failure during hydro-burst-tests of the motor cases are
theories. Substituting design stress for failure available in Ref. 10. The stresses in the motor
strength and using the equality in any applicable case under internal pressure are computed from
failure theory turns it into a design tool by which the measured strains and the biaxial strength of
critical dimensions may be established. Failure the motor case is evaluated. The applicability of
theories properly evaluated are, therefore, existing analytical expressions for burst pressure
among the most important links in the design evaluation of cylindrical vessels is examined. This
process. study also confirms that usage of the biaxial
313
314 A. P. Beena et al.

strength as a basis of design makes it possible to specified E, and E,. The value of E, is substituted
achieve the lightweight design of rocket motor in eqn (1) to get (T,. With these values, eqns (3)
cases. and (4) give the secant modulus (E,) and secant
Poisson’s ratio (YJ. Using this value of Y, in eqn
(2), E, will be determined. This iteration process
has to be continued until these two unknowns, vs
BIAXIAL ULTIMATE STRENGTH and E,, satisfy both eqns (2) and (3) with the
EVALUATION desired degree of accuracy.
With the secant modulus (E,) and secant
Biaxial ultimate strength of the material can Poisson’s ratio (YJ for the measured principal
strains E[ and E*, the respective principal stresses
generally be evaluated by burst-testing of the
subscale cylindrical vessels. For hydro-burst-tests, u1 and u2 are
the stresses in the inelastic range are determined
from the post-yield strain gauges. For the
calculation of stresses from the measured strains,
11fll

(T* = (1 EY:, [1: x:1 C5)

Keil and Benning” have described the analytical The secant modulus (E,) and the secant
method, for which Ref. 12 suggests a simplified Poisson’s ratio (Ye) in the proportional limit is
procedure which can be easily programmed on a equal to the Young’s modulus (E) and Poisson’s
digital computer. The analytical method for ratio (v) of the material. So the above-described
elasto-plastic analysis used is briefly described procedure can be used for the calculation of the
below. principal stresses in the proportional limit from
The stress(a)-strain(&) diagram of the mate- the measured principal strains. For this case, the
rial is represented by first iteration itself gives the solution.
The variation of the maximum principal stress
u = E&(1 + (&,F(Jy” (1) and the measured principal strain due to internal
pressure up to bursting of the cylindrical vessel,
where E is the Young’s modulus, a0 and y1are the represents the biaxial stress-strain curve of the
material constants. material. The maximum principal stress at the
Based on a biaxial stress field and the initiation of the burst pressure is the biaxial
distortion energy principle, an equivalent strain ultimate strength of the material.
(I,) is obtained for the measured principal strains
(E, and r,) as

E, = ((1 + Y, + Vf)(&f + &) BURST PRESSURE DETERMINATION

- (1 - 4Y, + Yf)Ele,}“*(l - Y:)-’ (2)
The pressure, Py, which is sufficient to initiate a
where the secant Poisson’s ratio, Y, is yield zone at the inner surface of a long
cylindrical vessel, is13
y, = $ - ($ - Y) 2ES (3) P&g l-f (6)
( )
the secant modulus, E,, is where K is the ratio of outer to inner diameter of
the shell and uys is the yield strength of the
Es=5 (4) material.
I” The formula derived by Marin and Sharma’”
for bursting pressure of a cylinder was
and the stress (u,) corresponding to the strain
(E,) is obtained from eqn (1). Since in eqns (2)
and (3), the values of Y, and E, are interdepen- Pb = 2.31(@577)“*[%)
I
dent, the following iterative process is necessary
to evaluate properly Y, and E, which will satisfy where the strain hardening exponent, ~1,=
both equations. Initially, Y, is assumed as Y and ln(1 f E,), E, is the strain corresponding to the
substituted in eqn (2) to obtain E, for the maximum stress, CT,of the tensile specimen at the
 Destructive tests of 15CDV6 steel rocket motor cases 315

initiation of failure, t is the thickness and Di is

the inner diameter. Svensson” developed the
same formula independently at about the same
time.
Reference 15 suggests the formula for bursting
pressure of a cylinder as

Following the work of Margetson,16 the burst

pressure of a cylindrical vessel is proposed as

where Y,, is the secant Poisson’s ratio at the

initiation of the failure of the tensile specimen.

DISCUSSION ON BURST TEST DATA OF

ROCKET MOTOR CASES

The rocket motor cases made of 15CDV6 steel

were subjected to hydro-burst-pressure test, the
details of which are available in Ref. 10. Stress
analysis of the motor case under internal ypressure Fig. 2. Motor cases after burst test. (a) Chamber 1; and
indicates that the cylindrical shell portion (b) chamber 2.
experiences the maximum stress. Hence, strain
gauges were mounted mostly in the cylindrical
portion, near the long seam weld joint. The which the first number is equal to 100 times the
details of strain gauge locations in the test concentration of carbon. The letters which follow
chamber are shown in Fig. 1. As expected, the indicate the other elements present, in the
failure of the chamber was initiated near the long decreasing order of concentration. The last digit
seam joint. A classical hoop failure was observed is equal to four times the concentration of
(Fig. 2). chromium. In the French notation, C stands for
Regarding the material of the test chamber, chromium, D for molybdenum, and V for
15CDV6 steel is a high strength, low-alloy steel. vanadium. It therefore follows that, in 15CDV6
The name 15CDV6 is a French designation in steel, the concentration of carbon is O-15%, while

1.063m

Fig. 1. Motor case and the strain gauge locations.

316 A. P. Beena et al.

that of chromium is l-5%, and the concentrations locations away from the weldment is found to be
of molybdenum and vanadium are less than 1.5% more, compared to that at weld locations, which
each. Hence, it is a low-alloy steel, in which the may be due to higher thickness of the weld bead.
proportion by weight of all the alloying elements The variation in the stresses on either side of the
combined is less than 5 %. weld is mainly due to curvature change or
As per specification, the minimum guaranteed discontinuity, which causes bending. These
material properties are bending stresses are found to be diminishing as
Young’s modulus (E) 203557.50 X 10e6 N/m” the internal pressure increases and becomes
Poisson’s ratio (Y) O-3 negligibly small at the burst pressure. Another
Ultimate tensile observation is that, even though the peak stresses
strength (a,) 981 X 10d6 N/m* under internal pressure in the cylindrical shell
Yield strength (vJ 833435 X 10e6 N/m2 portion were observed in the parent metal,
Percentage elongation 8-12% failure was initiated near the long seam joint.
Density (p) 7850 kg/m3 This may be due to the weld efficiency (7,) being
less than 100%.
The dimensional details of the motor case The stresses in the cylindrical shell portion of
(cylindrical shell portion) are as follows:
the motor case under internal pressure are
Inner diameter (DJ 206.6 X lo-” m computed from the measured strains and the
Thickness (t) 2-6 X lop3 m stress-strain data of the material obtained from
The stress-strain data of the material obtained the tensile test coupons. The stress field in a
from the tensile test coupons corresponding to thin-walled motor case is biaxial. In the
the hardware are represented by the relationship cylindrical shell portion of the motor case, the
as given in eqn (1). 0.2% proof stress or yield hoop stress (a,) is higher than the meridional
stress (a,,) for this batch of the material is stress (a,) and the effective stress (a,,). The plot
915.27 X lop6 N/m2 and the material constants in of the hoop stress (a,) versus the hoop strain (cl)
eqn (1) are E = 203557.5 X lop6 N/m’, &o= O-005 up to the burst pressure of the motor case
and n = 3.2825. The data generated from the represents the biaxial stress-strain diagram. The
empirical relationship are plotted along with the plot of the effective stress (o,,=J versus the
actual stress-strain data in Fig. 3. effective strain (E& up to the burst pressure of
Based on the procedure explained in the the motor case follows the path of the uniaxial
previous section, stresses are computed from the stress-strain diagram of the tensile test coupon.
measured strains at various pressure levels. Table Figure 5 shows the comparison of uniaxial and
1 gives the stress analysis results at the maximum biaxial stress strain diagram for the motor case
stressed location (No. 3). near the long seam joint at which the failure of
Figures 4(a) and (b) shows the variation of the motor case was initiated. The biaxial ultimate
hoop stresses with internal pressure, at different strength of the motor case is worked out to be
locations identified in Fig. 1. The hoop stress at 1164.45 X 10e6N/m2. One interesting phenome-
non observed from these curves is the reduction
of ductility in the biaxially stressed cylinder as
compared to that obtained in the uniaxial
specimen. Because of loss of ductility in the
biaxial field, material discrepancies such as flaws,
- Eq.(1) local embrittlement and porosities will be more
DWPDT~S~
critical in causing premature failure in a vessel,
MATER!AL CONSTANTS
than in a uniaxial specimen. This may be the
E = 203557.5 x lo-' N/m'
reason that the maximum effective stress in the
" = 3.2825
motor case (a,) is 1010.43 X 10W6N/m’, which is
= 0.005
c.
less, compared to the ultimate strength
(1030.10 x 10-6-1079.10 X 10e6N/m’) of the ten-
sile test coupons. The weld efficiency (nW) is
worked out to be 93.5-98%. The strain E,
STRAIN, L ( Y lo-') corresponding to the maximum effective stress is
Fig. 3. Stress-strain curve for 15CDV6 steel. O-01623. If the motor case is not weld-free
 Destructive tests of 15CDV6 steel rocket motor cases 317

Table 1. Stresses and strains at the maximulm stressed location (No. 3) in the burst-tested motor case (see Fig. 1)

Int. Hoop Merid. Eff. Hoop Merid. Eff.

pressure, strain, strain, strain, stress, stress, stress,
&T UeE
(X 1ONim’) (x &) (X 3, (x 10-y (x 10-%,m2) (X 10-%/m’) (X 10m6N/m*)

0.00 0 0 0 o-00 0~00 o-00

197.18 283 61 287 67.40 32.64 58.38
396.32 583 124 590 138.70 66.85 120.14
601.35 911 194 923 216.59 104.47 187.61
774.99 1171 249 1186 278.03 134.11 240.83
978.06 1482 316 1502 351.02 169.66 304a4
1203.69 1845 393 1872 434.78 210.51 376.59
1425.39 2181 466 2217 51044 248.14 442.11
1479.35 2272 484 2310 530.39 257.81 459.39
1586.28 2465 524 2509 572.09 278.52 495-50
1678.49 2613 555 2662 603.34 294.25 522-56
1777.57 2748 585 2803 631.32 308.78 546.79
1847.22 2878 609 2937 657.36 321.53 569-34
1973.77 3063 648 3132 693.58 340.37 600.69
2059.12 3222 678 3298 723.37 35547 626-49
2155.26 3377 709 3463 75144 370.21 650.79
2250.41 3544 740 3640 780.31 385.20 675.79
2353.44 3735 768 3842 811.32 400.33 70264
2452.50 3996 784 4113 849.81 415-79 736-01
2473.10 4087 784 4207 862.22 419.90 746.79
2491.74 4165 786 4288 872.64 423.74 755.84
2502.53 4231 784 4356 881*08 426.23 763.17
2515.28 4318 785 4446 892.06 430.13 772.71
2530.98 4498 777 4631 913.25 436.08 791.17
2550.60 4658 777 4799 931.18 442.41 806.76
2570.22 4900 774 5054 956.06 450.93 828-42
2591.80 5257 767 5432 988.31 461.98 856.51
2602.59 5438 765 5626 1002-86 467.39 869.17
2607.50 5569 762 5767 1012.62 470.86 877.67
2626.14 5898 759 6123 1034.77 479.64 896.93
2648.70 6230 756 6485 1053.88 487.57 913.53
2667.34 6726 748 7030 1077-19 497.37 933.78
2689.90 7514 735 7904 1104.40 509.79 957.37
2698.73 7902 729 8337 1114.45 514.82 966.07
2707.56 8254 725 8732 1122.13 519.01 972.70
2729.14 9346 708 9963 1139.36 528.90 987.56
2747-78 10032 701 10741 1146-69 533.96 993.84
2765.44 11622 677 12548 1157.53 541.97 1003.12
2788.00 13506 652 14697 1164.38 548.41 1008-95

construction, the weld efficiency (7~~) has to be 1010.43 N/m2, E, = 0.01623 and vSv,,= 0.439, in
considered on the ultimate tensile strength (a”) eqns (6)-(g) are given in Table 2.
of the parent material while evaluating the burst In order to verify the reliability and perfor-
pressure. Otherwise, the prediction of the burst mance reproducibility of the 15CDV6 steel motor
pressure will be unconservative. cases, another hardware was subjected to
From Fig. 5, the biaxial gain (ratio of hydro-burst-pressure test. The failure of the
maximum principal stress in the cylindrical shell motor case was initiated in the long seam joint
to the uniaxial tensile stress) is found to be and a classical hoop failure was observed. The
17.9% at the 0.2% proof stress level, whlereas it recorded burst pressure of the motor case is
is 15.4% at the ultimate stress. 29.59 X 10e6N/m2. The closeness of the two
The pressure at which yield is initiated I(&) and hydro-burst-tests amply testifies the reliability of
the burst pressure (Pb) for the tested motor case the motor cases made of 15CDV6 steel.
obtained by using the data, Oi = 206.6 X 10v3 m, The biaxial strength of the hydro-burst-tested
t = 2.6 X low6 m, uYS= 915.27 X 1O-6 N/m”, u, = motor case (1164-45 X lop6 N/m*) evaluated from
 318 A. P. Beena et al.

64 Table 2. Comparison of analytical and test results on the

pressure at which yield is initiated (P,) and the burst
pressure (Pi) for the rocket motor case

(X lO~‘N/m’) (X lO’N/m’)

Analytical Test Analytical Test

ew (f-3)
eqn (7) eqn (8) eqn (9)
25.62 26.49 29.14 28.11 28.65 28.86

PRESSURE. P ( x lo-" N/m') calculated as 29.23 X lop6 N/m* which is in good

agreement with the two burst test results of the
rocket motor cases.
(b)
This study confirms that the formulae sug-
1.700 r
7
gested in eqns (6)-(9) for the calculation of yield
E
> 1000 and burst pressures can be used for the design of
0

b
pressure vessels made of 15CDV6 steel. How-
- 600 ever, the pressure value given in eqn (8) being
more conservative, it is preferred for use in
design.

APPLICATION TO LIGHTWEIGHT
DESIGN
0

PRESSURE, P ( x lo-' N/m') Some thoughts are given to achieve lightweight

Fig. 4. Hoop stress versus pressure at (a) locations 1, 2, 3 design of large diameter motor cases using biaxial
and 5; and (b) locations 4, 6 and 7. strength as the basis of design.
In the fabrication of a large motor case, welded
joints are inevitable. In the case of large diameter
the present experimental stress analysis is motor cases, controlled welding is very difficult to
equated to the hoop stress (PD/2t) of its achieve, leading to mismatches at the welded
cylindrical shell portion and the burst pressure is joints. These mismatches causes stress concentra-
tions near the weldments. Following Ref. 17, the
principal stresses u1 and g2 near the long seam
-0--- weld joint of a cylindrical pressure vessel under
,_-- HOOP
,
/ internal pressure (P), with a mismatch (S), in the
/
/ EFFECTIVE
proportional limit can be written as
/

where K, = (1 + 38/t) and K, = (5 + 3Slt).

The effective stress (T,~ in the presence of a
weld mismatch is
0:
0 4000 8000 12000 16000 20000

STRAIN, E ( Y lo-')

Fig. 5. Comparison of hoop stress and effective stress at the

maximum stressed location (No. 3).
 Destructive tests of 15CDV6 steel rocket motor cases 319

The efficetive stress ueff in the weld-free diameter rocket motor case. Reducing propor-
cylindrical shell is tionally, the evaluated biaxial strength of the
burst-tested 15CDV6 steel motor case, and
tiPD
u eff =-- (13) equating to the hoop stress ((PD/2t)K,) in the
4 t long seam joint of the cylindrical portion of the
From eqns (12) and (13), the stress colncentra- large diameter (2.8m) rocket motor case, the
tion factor K, in the long seam weld joint of the minimum burst pressure is worked out to be
cylindrical vessel due to weld mismatch of 6 is 8.05 x lop6 N/m2. Usage of biaxial strength to
assess the structural integrity of a motor case is
K, = 2
K: + K; - K1K2 “’
3 1 (14) thus simple and accurate.
The factor of safety (minimum) over proof
This stress concentration factor is diminished pressure is expected to be l-22.
in magnitude by local plastic flow. The plastic
stress concentration factor, K, for any type of CONCLUDING REMARKS
geometrical discontinuity, suggested in Ref. 18 is
A rocket motor case is used only for a short
K, = 1 + (K, - l)E,/E (15) duration during flight. For effectiveness of
To illustrate the use of these ideas, the vertical flight, the rocket motor case used in the
problem of designing a 2.8 m diameter rocket advanced solid propulsion system is essentially a
motor case made of 15CDV6 steel for a ;specified lightweight shell, acted upon by static internal
proof pressure of 6.48 X 10P6N/m2 is considered. pressure and dynamic and thermal loads during
Maximum permissible long seam weld mismatch flight. For practical purposes, consideration of
is assumed as 5% of thickness. Weld efficiency qw internal pressure is all that is necessary in
on the minimum guaranteed properties is investigating the structural integrity of the rocket
assumed as 94%. Using eqn (14), the elastic motor case.
stress concentration factor K,, for S = 0.03, is Experimental stress investigations are carried
obtained as 1.1505. The plastic stress concentra- out on burst tests of rocket motor cases made of
tion factor K, at the yield stress level is found to 15CDV6 steel. The biaxial strength of the motor
be 1.12. case is evaluated from the recorded strains in the
Allowing the material to yield at proof cylindrical shell portion and the applicability of
pressure, the thickness of the motor case, using the existing analytical expressions for the
eqn (6), is obtained as estimation of burst pressure is examined. It is
demonstrated that usage of biaxial strength in the
design of a large diameter motor case is quite
simple and accurate.
= 11.35 X 10M3m As a rule of thumb, the distortion energy
theory can be used for isotropic ductile materials.
The stress concentration factor K, at the While there is a well-established basis for the
ultimate strength is 1.0507, the ultimate effective effectiveness of proof testing for brittle materials,
stress and strain are the logic is not straightforward for tough ductile
CT
eff= qwa,lK, = 877.6 X 1O-6N/m2 materials which exhibit stable flaw growth.
Where possible, a fracture mechanics analysis
and E, = O-01459.
should be performed to establish the proof
Equating this ultimate effective stress for the pressure factor required to determine the
ultimate strength (T, in eqn (8), the burst pressure maximum possible flaw size, for verification of
of the motor case for the thickness of service life with respect to cyclic and sustained
11.35 X 10e3m is obtained as 7.91 X 10P6:N/m2. load history. To examine these aspects, fracture
The ultimate tensile strength (a,) of the tensile toughness and crack growth data of the material
test coupons (which were made from the motor are essential, in addition to the tensile strength
case after the burst test) was found to be! 5-10% properties.
higher than its specified value. The specified Burst tests involve enormously high cost for
minimum guaranteed strength properties of the large diameter rocket motor cases. Qualifying a
material are considered in the design of a 2.8 m new design of a rocket motor case generally
320 A. P. Beena et al.

requires, in addition to the proof tests, bursting 10. Anon., Test Report on the Qualification Test on
PSLV-Retro Chambers RSL and RS2 (Report No.
of one or more prototype motor cases. VSSC-SEG-STL-LT-79-89). Structural Testing Labora-
tory, Structural Engineering Group, Vikram Sarabhai
SpaceCentre, Trivandrum, India, 1989.
REFERENCES 11. Keil, S. & Benning, O., On the evaluation of
elasto-plastic strain measured with strain gauges.
1. Williams, F. A., Barrere, M. & Huang, N. C., Exper. Mechanics, 19 (8) (1979) 265-70.
Fundamental Aspects of Solid Propellant Rockets 12. NageswaraRao, B., A simplified procedure for stress
(AGARD No. 116). The Advisory Group for analysisin the elasto-plasticrange with the aid of strain
Aerospace Research and Development, Technical gauges.Exper. Tech., 15 (3) (1991) 38-9.
Services,Slough, UK, 1969. 13. Svensson,N. L., Bursting pressure of cylindrical and
2. James, B. H., Structural Integrity Analysis of Solid sphericalvessels.J. Appl. Mech., 25 (1956) 89-96.
Rocket Motors. Conference on Stress and Strain in 14. Marin, J. & Sharma, M. G., Design of a thin walled
Engineering, Brisbane. National Committee on Applied cylindrical vessel based upon plastic range and
Mechanics, The Institution of Engineers, Australia, consideringanisotropy. Welding Res. Council Bull., 40
1973. (1958).
3. NageswaraRao, B., Fracture of solid rocket propellant 15. Bohm, G. J., Cloud, R. L., Hsu, L. C., Pai, D. H. &
grains.Engng Fract. Mech., 43 (3) (1992)45.5-9. Reedy, R. F., Pressure Vessel and Piping-Design and
4. Marin, J., Mechanical Behavior of Engineering Mate- Analysis-A Decade of Progress (Vol. 1, Analysis for
rials. Prentice-Hall of India (Private) Ltd, New Delhi, Design). The American Society of Mechanical Engi-
India, 1962,pp. 104-69. neers,New York, USA, 1972,pp. 23-32.
5. McClintock, F. A. & Argon, A. S., Mechanical Behavior 16. Margetson, J., Burst pressure predictions of rocket
of Materials. Addison-Wesley, Canada, 1966, pp. motors. AIAA Paper No. 78-1569, AIAA/SAE 14th
273-308. Joint Propulsion Conference, Las Vegas, NV, USA,
6. Polakowski, N. H. & Ripling, E. J., Strength and 22-27 July, 1978.
Structure of Engineering Materials. Prentice-Hall, Inc., 17. Johns, R. H., Theoretical Elastic Mismatch Stresses
Englewood Cliffs, NJ, USA, 1966,pp. 99-132. (NASA TND-3254). Lewis Research Center, National
Sines,G., Elasticity and Strength. Allyn and Bacon, Inc., Aeronautical and Space Administration, Cleveland,
Boston, MA, USA, 1969,pp. 54-6. OH, USA, 1966.
Collins,J. A., Failure of Materials in Mechanical Design, 18. Hardrath, H. F. & Ohman, L., A Study of Elastic and
Analysis, Prediction,‘ Prevention. John Wiley & Sons, Plastic Stress Concentration Factors due to Notches and
Inc., New York, USA, 1981,pp. 126-63. Fillets in Flat Plates (NACA TN-2566). Langley
Dieter, G. E., Mechanical Metallurgy. McGraw-Hill Aeronautical Laboratory, National Advisory Com-
Book Company, New York, USA, 1988,pp. 69-99. mittee for Aeronautics, Langley Field, VA, 1951.