3-D PARACHUTE DESCENT ANALYSIS USING COUPLED

CFD AND STRUCTURAL CODES

Jubaraj Sahu*
Gene coopert
U.S. Army Research Laboratory
Aberdeen Proving Ground, MD 2 1005-5066

and

Mr. Richard ~ e n n e y $
U.S. Army Natick Research, Development, and Engineering Center
Natick, MA 01760-5017

Abstract Introduction

A computational tool that models the terminal Parachutes have been an interest to man for over
descent characteristics of a single or a cluster of two thousand A brief but good review of the
parachutes is a technology that is needed by parachute history of parachutes can be found in ' ~ o c k r e l l , ~
designers and engineers. A joint effort between the U.S. ~ n a c k e ,and
~ the USA 1978 and 1963 Parachute
Army Natick Research Development and Engineering Design guide^.^.^ The consideration of parachutes as
Center (Natick) and the Army Research Laboratory high-performance Aerodynamic Decelerators did not
(ARL) to develop this computational tool is now take place until the middle of this century. One of the
underway. As a first, effort attempts are being made to features of parachutes that is being studied here is the
analyze both 2-D and 3-D flow fields around a parachute fluid flow field around one or more parachutes being
using a coupling procedure where the fluid dynamics is used as decelerators for heavy payloads dropped from
coupled to 2-D and 3-D structural dynamic (SD) codes. in-flight aircraft. In particular, this work investigate
This effort uses computational fluid dynamic (CFD) the coupling of the flow field around nonribbon
codes to calculate a pressure field, which is then used as parachutes with the dynamic structure of the
a input load for the SD codes. Specifically, this paper parachutes themselves.
presents the methods and results of the flow field plus
the structural characteristics of a single axisymmetric The flow around parachutes has been studied by
parachute and a 3-D gore configuration for the terminal many investigates's6,' using analytical methods for
descent velocity. Computed results have been obtained predicting canopy pressure distributions. cockrel12
using the payload weight and unstretched constructed presents a summary of potential flow solutions
geometry of the canopies as input. Significant progress describing steady-state canopy pressure fields, and
has been made in determining the terminal descent flow lbrahiml has found closed form analytical solutions
field along with the terminal shape of the parachute. A for the velocity potential around spherical cups. These
discussion of the fluid and structural dynamics codes, studies have been extended by ~ l i r n a s , who ~
coupling procedure, and the associated technical considered placing a vortex sheet to coincide with the
difficulties is presented. Examples of the codes' current location of a real canopy. He used an axisymmetric
capabilities are shown. Stoke's stream function to model viscous and
permeable material effects. The theoretical
predictions gave only fair numerical results, but the

* Aerospace Engineer, Associate Fellow, AIAA

t Research Physicist
$ Aerospace Engineer, Member AIAA

This paper is declared a work of the U.S. Government and is

not subject to copyright protection in the United States.
internal flow behavior was reasonably good. weight and unstretched constructed geometry of the
canopy are given as input. This 3-D capability is
Continuing in the vain given previously, needed to allow design engineers to optimize a single
researchers have studied the flow around parachutes parachute design (for terminal descent characteristics)
using numerical methods to solve the governing on a computer. The effect on the parachute's
equations. A two-dimensional CFD code has been used terminal descent characteristics due to modifications
by the U.S. Army Natick Research, Development, and to a gore shape, suspension line length, etc., can be
Engineering Center. NATICK has applied this code to investigated without having to manufacture prototypes
flow problems that have axial symmetry and have and perform multiple drop tests. The 3-D CFDISD
calculated the flow and pressure fields around capability being developed will ultimately be applied
parachutes. One limitation of using a 2-D code is that to the parachute cluster problem to investigate the
the parachutes of interest are usually not axisymmetric. flow field surrounding various U.S. Army clustered
Therefore, this report is going to investigate the 3-D parachute configurations.
properties of flows around parachutes. The pressure
distributions calculated on the under and top sides of the This paper reports the progress made in the
parachutes are then used to couple the fluid dynamics to developmentVof the CFD and SD models. The
the structure codes used to model the structural approach used to couple the two codes is also
properties of the parachute itself. described. Examples of the codes' current capabilities
are presented.
The total drag of parachutes in a cluster has been
shown experimentally to be less than the sum of the Computational Fluid Dynamics Model
drags of the individual canopies. The ability to predict
the flow field characteristics, terminal descent The computational method used in the present
positionslshapes, and the drag on clusters of parachutes analysis solves the Incompressible Navier-Stokes
is needed to assist in the development of improved equations in 3D ( I N S ~ Dgeneralized
)~ coordinates for
parachute cluster design and should allow for low-speed flows. This technique is based on the
optimization studies to be performed numerically. A method of artificial compressibility and an upwind
joint effort between Natick and the U.S. Army Research differencing scheme. The pseudocompressibility
Laboratory (ARL) to develop this computational tool has algorithm couples the pressure and velocity fields at
begun with modeling flow about parachutes in three the same time level and produces a hyperbolic system
dimensions (3-D). This effort is utilizing 3-D of equations. The upwind differencing leads to a
computational fluid dynamic (CFD) codes in conjunction more diagonal system and does not require a user-
with 3-D structural dynamics (SD) codes. specified artificial dissipation. The viscous flux
derivatives are computed using central differencing.
The aerodynamic characteristics associated with a This code is capable of computing both steady-state
single or cluster of parachutes in the terminal descent and time accurate flow fields.
phase is extremely complex to model. The complexity
arises largely from the fact that the flow field is The governing equations are numerically
dependent on the canopy shape, which is itself represented and solved using a nonfactored Gauss-
dependent on the flow field. A correct model must Seidal line-relaxation scheme. This maintains the
include the coupled behavior of the parachute system's stability and allows a large pseudo-time-step to be
structural dynamics with the aerodynamics of the taken to obtain steady-state results. Details of the
surrounding flow field. The coupled model is being numerical method can be found in Rogers, Wiltberger,
developed to yield the terminal descent characteristics of and Kwak (1993) and Kwak (1989).9s'0 For
single or clustered parachutes including velocity, shape, computation of turbulent flows, a turbulence model
drag, pressure distribution, and the other flowfield must be specified. The present calculations use the
characteristics. one-equation turbulence model developed by Baldwin
and Barth." In this model a transport equation is
As a starting point, the terminal descent solved for the turbulent Reynolds number. This
characteristics of a half-scale C-9 solid flat circular equation is derived from a simplified form of the
parachute are determined. The goal for this step is to standard two-equation k-E turbulence model. The
predict the parachute's terminal descent shape, velocity, one-equation model has the advantage that it does not
pressure field, velocity field, etc,. when the payload require the turbulent length scale to be specified. It
is solved using the same Gauss-Seidal-type line- modeling capability. This section outlines the steps
relaxation scheme used to solve the mean flow that are taken to model either single or clusters of
equations. parachutes as the model progresses over the next year.
The process begins with the structural model, which
Current Structural Model requires the cut-out geometry of the canopy and
suspension lines as input. A guessed surface pressure
The current structural model being utilized for this distribution is utilized by the structural code to
modeling effort is the CALA (CAnopy Loads Analysis) determine a first guess deformed shape for the canopy
code.12 The next generation for the structural half of the or canopies.
terminal descent model of round parachutes will be a
three-dimensional membranelcable finite element based If the parachute system is a single canopy, the
code.13 CALA is a static code that predicts the steady- shape determined is supplied to the CFD code as is.
state shape and stresses for round parachutes. CALA A CFD mesh is generated around the surface shape
requires the parachutes dimensions and a steady state provided. The CFD code solves for the flow field
pressure distribution along a radial (meridional arc characteristics with a prescribed inflow velocity. The
length) as input. A radial is defined as the continuation 3-D surface pressure distribution is extracted from the
of a suspension line from the skirt along a meridional CFD results and fed into the structural code. The
arc length to the apex of the canopy. The radial can also structural code is executed with the new surface
be visualized as the connection of two adjacent gores. pressure distribution and predicts a new deformed
shape. The total vertical drag is calculated based on
The pressure distribution across the surface of the using the shape and pressure distribution. The new
canopy is supplied by the CFD code as a function of the shape is fed back to a grid generator and a
three-dimensional deformed canopy shape. The CALA mesh is generated. This new mesh is used in the
code assumptions transform the pressure distribution into CFD code, and computations are performed to obtain
nodal forces that are tangential and normal to the radial the steady-state result with the same inflow velocity.
position. The CALA code assumes that the horizontal The process is continued manually until convergence
members (defined as the curved strip of a gore of shape and pressure distribution is achieved. The
connecting two adjacent radials) of a gore form sections solution may diverge for relatively poor initial guesses
of circular arcs and that the pressure distribution is of pressure distribution, in which case the process
uniform along the horizontal members. The horizontal may involve a more elaborate coupling procedure.
members lie in planes that are defined by the surface
normal vectors from two adjacent radials making up one For modeling of a cluster of canopies, the
gore. The CALA code defines the static force per unit predicted single parachute shape is rotated by an
radial length applied to a radial location. These force initial guessed angle about the confluence point. A
equations include the variable v, which is defined as the CFD mesh can be created around the surface shape
gore bulge angle. y is determined iteratively for the provided and utilize symmetry boundary conditions
current iteration's surface configuration based on the that are dependent on the number of parachutes in the
constructed gore shape. The forces include cluster. The CFD code solves for the flow field
approximations of the hoop force contribution based on characteristics with a prescribed inflow velocity along
the gore geometry. the vertical axis. The 3-D surface pressure distribution
is extracted from the CFD results, and the net radial
The CALA code iterates on the shape by assuming force on the canopy is determined. The canopy shape
as initial guess for the vent line tension. The code is then rotated in the direction that is consistent with
outputs the canopy shape, stresses, gore bulge angles, the predicted net force. The new canopy grid location
total drag, etc. The radial shape is extracted from the is supplied to the CFD code, a new mesh is created,
output and utilized in a FORTRAN code with the CALA and the resulting surface pressure distribution is used
assumptions to generate a set of 3-D data representing to determine the direction and extent of rotation
the shape of a gore. required to reach a force equilibrium in the "radial"
direction. Rerunning the structural code with the new
surface pressure distribution, once the radial forces
have been reduced to near zero, is then performed and
At this time, the codes are being manually coupled. the process continued. The total vertical drag is
The results section that follows describes the current predicted based on the shape and pressure distribution
of the single canopy multiplied by the total number of the streamwise direction, 19 points in the
canopies being modeled in the cluster. circumferential direction, and 80 points in the normal
direction away from the body surface. The grid
Results points are clustered near the body surface for viscous
turbulent flow computations. The outer, inflow, and
The coupled model was used to determine the 3-D downstream boundaries are placed sufficiently away
terminal descent characteristics of a half-scale C-9 from the body surface such that they do not interfere
canopy. The half-scale C-9 is a solid cloth, flat circular with the convergence and accuracy of the computed
canopy made up of 28 gores. The constructed diameter flow field results. The grid was obtained by the
is 14 feet and suspension lines are 12 feet in length. ~ a g l e v i e w ' ~grid generation program using an
The half-scale C-9 canopy has been used in a variety of algebraic method and elliptic smoothing procedure.
experiments and its opening behavior was predicted with An expanded view of the computational grid near the
an axisymmetric coupled model developed at ~ a t i c k . ' ~ parachute body surface is shown in Figure 2. This
This canopy was studied with the manually coupled figure shows the computational grid for both end
codes, and the terminal descent results are compared to planes in the circumferential direction. Also included
the results obtained from previous studies at Natick. is a shaded paiachute body surface. It shows the grid
clustering near the body surface more clearly for the
To start, an axisymmetric parachute shape was outer part of the body surface and the skirt. The grid
modeled to gain confidence in the three-dimensional was obtained for a circumferential plane and then
CFD code's predictive capability of the flow field rotated around to obtain the full grid containing 19
characteristics around parachute-like shapes. The planes in the circumferential direction.
terminal descent shape from the axisymmetric model
(case 1 in reference 14) was used to define the grid. A
computational mesh was generated around a quarter of
the canopy to utilize symmetry boundary conditions for
the axisymmetric test case. The surface grid along with
a longitudinal cross section of the full grid is shown in
Figure I. The full grid consists of 48 points in

Figure 2. An Expanded View of the Grid Near the

Parachute.

The input velocity used is 18 feet per second, which

Figure 1 . Computational Grid for Axisymmetric was the terminal descent velocity determined from the
Parachute. axisymmetric model. All numerical computations
were performed at this inflow velocity and at a = 0'.
Results are now presented for the axisymmetric model.
The computed results were obtained using the INS3D
code. Figure 3 shows a velocity vectors field in the
vicinity of the axisymmetric parachute shape. The
incoming flow stagnates at the body surface, and a
separated flow region is formed in front. It also shows
flow separating at the skirt and then forming a large
region of recirculatory flow behind the body. This
separated flow region in the wake gives rise to lower
surface pressure on the outer body surface.
0 Outer Surface
X Inner Surface

8
-o1 *.O80.0 1.0 2.0 3.0 4.0

Meridional Length (feet)

5.0 6.0 7.0

Figure 4. Comparison of Pressure Distributions

(Axisymmetric Case).

of computed results agree very well for both outer

and inner surface pressures. The pressure
distributions are very close especially noting the fact
that the axisymmetric coupled model had not
completely damped out at the time for which the
shape was extracted. Small differences can be
observed between these results near the skirt of the
axisymmetric parachute. As expected, the inner
surface pressure is quite uniform and is a lot higher
than the predicted pressure on the outer surface. This
gives rise to drag force, which is consistent with the
Figure 3. Velocity Vectors for Axisymmetric Parachute, payload for this parachute at this terminal velocity.
a = oO..
The next step in this modeling effort involved
The pressure distribution on the inner and outer surface extracting a three-dimensional grid of a single gore
of the canopy from both codes is compared in Figure 4. from the axisymmetric shape. This was done by
In this figure nondimensional pressure coefficient is applying the CALA assumptions of the gore shape to
plotted as a function of the meridional length. Here, the the radial node points. The axisymmetric shape is
meridional length is the distance which is measured from assumed to coincide with a radial. The three-
the apex (vent) of the parachute to the outer section dimensional gore surface was obtained using this
(skirt) along the canopy surface. The total meridional procedure and further smoothed especially near the
length thus corresponds to half of the diameter of the vent region to eliminate discontinuities in the surface
parachute in the unstretched position (14 feet diameter). definition. The smoothed surface grid is shown in
The computed pressure distributions obtained using the Figure 5. This surface grid contains 24 grid points
INS3D code are compared with the previous computed in the meridional direction and 22 points in the
pressure distributions using a 2-D code. The two sets circumferential direction.
 were generated algebraically15 and then appended
together to form the full grid. The parachute gore
surface is a part of an interior grid line, and thus the
no-slip boundary condition is applied along this
interior boundary. For viscous flow computation,
which is of interest here, the grid points are clustered
near the parachute gore surface in the normal
direction to resolve the flow gradients in the boundary
layers.

Computed results have been obtained for the

three-dimensional gore configuration and are now
presented. Figure 7 shows the pressure contours for
a circumferential end plane. For the incoming flow
the pressure is uniform (free-stream pressure). As the
flow approaches the inner surface, pressure builds up
and forms a large region of high pressure in the
vicinity of the inner surface. The pressure in the
wake region is a low pressure region. Although
there is some variation of pressure in the low pressure
Figure 5. Single Gore Surface Grid. region, the outer surface pressure is a la lower than
the inner surface pressure. The flow expands around
This surface grid was used to generate a three- the skirt region of the gore, and sharp changes in the
dimensional computational grid (see Figure 6) for the pressure field can be observed in going from the inner
calculation of fluid flow over the gore configuration. to the outer surface.
The 3D gore grid consists of 53 x 22 x 84 in the axial,
normal, and circumferential directions, respectively. The
section containing the gore surface was generated using
an 0-topology. Another section of the grid was obtained
using rectangular mesh topology. Both grid components

Figure 6. 3-D Computational Grid for Gore. Figure 7. Pressure Contours for Gore, a = 0'.
The computed pressure map on the gore surface itself is assumptions to generate a new 3-D gore grid. This
shown in Figure 8 for both inner and outer surfaces. new grid was used, computations were performed, and
Although computations were performed on one gore CFD results of the resulting flow field and pressure
shape due to symmetry, the canopy consists of 28 gores, distributions were obtained. This coupling procedure
which are all shown in this figure. Again, it shows high between the fluid dynamics and structural dynamics
pressure on the inside surface, which is rather uniform codes were repeated three times. The three different
except for the skirt region. The computed pressure on gore shapes which resulted from this analysis are
the outer surface is lower and is not as uniform. shown in Figure 9. The changes in the gore shape
from the first iteration to the second was observed to
be bigger than that from the second to the third
iteration of this coupling procedure. The change
between the shapes between the second and the third
iteration was rather small, which indicates the near
convergence of the body shape to its final terminal
descent configuration.

Figure 8. Computed Surface Pressure for Gore, a = 0,

[Inner and Outer).

The pressure distribution over the gore surface was

used by the CALA code to predict a new shape for a
gore. The pressure distribution required by CALA is
assumed constant over the horizontal members of the Figure 9. Comparison of Gore Shapes for Different
gore. Therefore, the CFD predicted surface pressure Coupling Iterations, a = 0.
distribution was averaged over the gore surface to obtain
an average pressure distribution versus meridional arc The outer surface pressure distribution over the
length set of data. This was accomplished by averaging gore (first iteration) is shown as a function of
the surface pressure values at the radial location and the meridional gore location in Figure 10. The total
corresponding gore mid-point values. This pressure meridional length again corresponds to half diameter
distribution was used as input by the CALA code which of the parachute in the unstretched position. The
output the new predicted shape. The shape of the radial computed pressure distributions are obtained at
predicted to the CALA code was used as an input to a different locations in the circumferential direction.
FORTRAN code that utilizes the CALA gore shape The first plane in this figure refers to the edge of the
gore whereas the half-plane corresponds to the mid- this figure is the computed pressure distribution for
section (center) of the gore. The difference in the outer the axisymmetric parachute shape. The pressure
surface pressures is rather small over a large portion of distribution over the 3-D gore is nearly constant near
the gore except near the skirt region. Although not the apex and deviates across the gore as the skirt is
shown here the same is true of the inner surface pressure approached. This is expected because the shape is
distributions. For the subsequent figures on surface nearly axisymmetric near the apex and becomes less
pressure comparisons, the 3-D gore surface pressure is axisymmetric towards the skirt. The CFD code was
used at the half-plane or the mid-section. run with the same inflow velocity, and the three-
dimensional parachute gore surface pressure
distribution was extracted from the CFD results. The
change in inner surface pressure is small between the
different iterations or gore shapes and is especially
small between the second and third iterations. The
difference between the predicted pressure for the gore
shapes and the axisymmetric configuration is larger,
which indicates the necessity of three-dimensional
modeling of the fluid and structural dynamics. Figure
12 shows the outer gore surface pressure comparisons
at a = 0'. The three sets of gore shapes yield similar
pressure distributions. The difference between the
second and the third iteration again *is smaller
-0.2 compared to that between the first and the second
0 1 2 3 4 5 6 7 8 iteration. The pressure distributions for the different
Meridional Lengt h (feet) gore shapes do not show as large a variation in
pressure as is observed for the axisymmetric
Figure 10. 3-D Gore Pressure Distributions at Different parachute shape. The computed surface pressures for
Circumferential Locations. both 3-D gore and axisymmetric parachute agree
rather well with each other only near the apex (vent).
The computed pressure distributions are all As expected, the three-dimensional effect is felt more
obtained using the INS3D code and are now compared near the skirt of the canopy. The pressure differential
between the different iterations of the coupling between the inner and the outer surface is what
procedure described earlier. The inner surface pressure determines the drag force for these parachute shapes
distribution over the 3-D gore is shown as a function of for a given terminal descent velocity.
meridional gore location in Figure 11. Also included in

0.0 - 0 . 6 0 1
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
Meridional Length (feet) Meridional Length (feet)

Figure 11. Comparison of Pressure Distributions (3-D Figure 12. Comparison of Pressure Distributions (3-D
Gore, Inner Surface). Gore, Outer Surface).
 Conclusions
4. Ewing, E.G., Bixby, H.W., and Knacke, T.W.,
The complexity of modeling parachute "Recovery Systems Design Guide,"
characteristics and phenomena in terminal descent and AFFDL-TR-78- 151, December 1978.
during the opening process stems from the coupling
between the structural dynamics of the canopy, lines plus 5. Performance of and Design Criteria for
payload, and the aerodynamics of the surrounding fluid Deployable Aerodynamic Decelerators, USAF
medium. ASD-TR-6 1-597, December 1963.

This paper has described on going research being 6. AIAA 1st Aerodynamic Deceleration Systems
conducted at both Natick and ARL to predict the three- Conference, Houston, Texas, September 1966.
dimensional terminal descent characteristics of single
and clusters of round parachutes. This involves the 7. Klimas, P.C., "Internal Parachute Flows," Journal
coupling of a 2-D and/or 3-D CFD code with 2-D and/or of Aircraft, Vol. 9, No. 4, April 1972.
3-D parachute structural codes. The solution to the
coupled problem is expected to assist in the development 8. Rogers, S.E:; Kwak, D., and Kiris, C., "Steady and
of future U.S. Army airdrop systems and other round Unsteady Solutions of the Incompressible Navier-
parachute systems. The capability of accurately Stokes Equations," AIAA Journal, Vol. 29, No. 4, pp.
predicting the behavior of parachute systems will 603-610, April 1991.
significantly reduce the amount of testing currently
required. The prediction of the terminal descent 9. Rogers, S.E., Wiltberger, N.L., and Kwak, D.,
characteristics of a half-scale C-9 parachute has been "Efficient Simulation of Incompressible Viscous Flow
demonstrated by manually coupling a 3-D CFD code to over Single and Multielement Airfoils," Journal of
the CALA code. A 3-D membranelcable finite element Aircraft, Vol. 30, No. 5, pp. 736-743, September-
code will be used as the next-generation structur?l October 1993.
model. Time will also be spent combining the two
codes into a more user-friendly environment and 10. Kwak, D., "Computation of Viscous
enhancing the pre- and post-processing codes to shorten Incompressible Flows," NASA Technical
the turnaround time. Memorandum 101090, March 1989.

This paper has presented the current status of the 11. Baldwin, B.S. and Barth, T.J., "A One-Equation
modeling effort and outlined the direction being pursued Turbulence Transport Model for High Reynolds
to address the complexity of the parachute characteristics Number Wall-Bounded Flows," AIAA Paper No. 91-
and the associated flow fields. Future computational 0610, 1991.
models are expected to provide better understanding of
the physics that govern the parachute and flow field 12. Sundberg, W.D., "New Solution Method for
interaction. The codes will allow the user to examine Steady-State Canopy Structural Loads," Journal of
the effect of vent size, vent location, reefed Aircraft, Vol. 25, No. 11., November 1988.
configurations, varying suspension line lengths, etc.
13. Benney, R.J., and Leonard, J., "A 3-D Finite
References Element Structural Parachute Model," PAPER # 95-
1563 13th AIAA Aerodynamic Decelerator
1. Maydew, R.C., Peterson, C.W., and Conference May 1995.
Orlik-Ruckemann, K.J., "Design and Testing of
High-Performance Parachutes," AGARDograph No. 3 19, 14. Stein, K.R., and Benney, R.J., "Parachute
November 1991. Inflation: A Problem in Aeroelasticity," U.S. Army
Natick Research, Development, and Engineering
2. Cockrell, D.J., "The Aerodynamics of Parachutes," Center. Natick Technical Report No. NATICWTR-
AGARDograph No.295, July 1987. 941015, August 1994.

3. Knacke, T.W., "Parachute Recovery Systems Design 15. Thompson, J.F., "A Composite Grid Generation
Manual," NWCTP 6575, Naval Weapons Center, China Code for 3-D Region - The EAGLE Code," AIAA
Lake, California, June 1987. Journal, Vol. 26, No. 3, pp. 271-272, March 1988.