Paper RotCFD NASA

NASA Ames Research Center
RotCFD
Ovidio Montalvo Fernández
NASA Ames Research Center – Aeromechanics Branch,
Universidad Autónoma de Nuevo León (UANL).
ovidiomontalvo@hotmail.com.
Abstract: RotCFD is a software intended to ease the design of
NextGen rotorcraft. Since RotCFD is a new software still in the
development process, the results need to be validated to determine
the software’s accuracy. The purpose of the present document is to
explain one of the approaches to accomplish that goal.
Keywords: CFD, CAD Modeling, Rotorcraft, NextGen aircraft.
I. ACKNOWLEDGEMENTS
For all involved in this wonderful experience, and for all the
knowledge acquired during this amazing time, I give the main
credit to God who has blessed me with the opportunity of
being where I am and allowed me to enrich my life during all
this time; I know it had never been possible without you.
I owe special thanks to my family for all their unconditional
support in everything I needed and for being there with me
without hesitation helping me in every possible way.
I’m deeply grateful with my beloved partner Ana Lucía for
being the reason to force myself to persevere, and for being
there any moment, anywhere, anyhow. Thanks for all your
love, dedication and great patience. All my successes are
dedicated to you and my parents.
The success and final outcome of this project required a lot of
guidance and assistance from many people and I am extremely
fortunate to have got this all along the completion of my
project work. Whatever I have done is only due to such
guidance and assistance and I would not forget to thank them.
I especially want to thank Dr. William Warmbrodt for giving
me the opportunity to be at NASA Ames Research Center
working in the Aeromechanics Branch and for have trusted in
my skills and knowledge to perform as a member of such an
important team.
I am thankful to and fortunate enough to get constant
encouragement, support and guidance from Eduardo Solis
who was always looking for our welfare and for us to have the
best work performance in our own benefit.
I would not forget to remember - for their unlisted
encouragement and more over for their timely support and
guidance to Shirly M. Burek, Meridith Segall, Guillermo
Costa, Ashley Pete, and Carl Russell.
I owe a great many thanks to many people who helped and
supported me during the process preceding my internship at
NASA as well as those who were tracking the whole process;
thanks to my institution and my faculty members without
whom this project would have been a distant reality; thank you
to all you!
II. NOMENCLATURE
VTOL – Vertical Take-off and Landing.
CFD – Computational Fluid Dynamics.
Downwash – Air forced downward as a result of the
momentum provided by an airplane wing or a rotor blade.
Outwash – Air forced outward in a rotor.
ߩ – Fluid density
ܸሬԦ – Velocity field.
݂Ԧ – Body force.
‫ܫܫ‬௜௝ – Stress tensor.
ܵԦ – All terms not accounted in the equation i.e. Sources.
III. INTRODUCTION
NASA is aware of the high and constantly growing aircraft
congestion in the biggest airports all around the world, this is
why it has started to design new aircrafts which can easily be
accommodated by current and future airports. One of the
presented solutions is the design of VTOL civil aircraft. The
RotCFD Software Validation -
Computational and Experimental Data
Comparison.

Final Report- RotCFD
main advantage of this designs is that they don’t necessarily
need a long runway for takeoff or landing.
A great deal of the new designs are aircraft that use rotors in
some way as a mean of lift or propulsion generation, and that
is why an efficient and easy-to-use CFD tool is needed to
generate all the analyses needed for the design.
Sukra Helitek Inc. and its newest software – RotCFD – were
selected to work in conjunction with NASA in the
development of this software, which will make the design of
NextGen rotorcraft much easier [1].
IV. THEORETICAL FRAMEWORK
III.1 Computational Fluid Dynamic Fundamentals
Computational Fluid Dynamics, also known by the acronym
CFD, is the branch of the fluid dynamics used to predict fluid
flow behaviors involving heat transfer, chemical reactions,
viscosity, etc. by means of computer-based simulations.
All of CFD in one form or another is based in the fundamental
governing equations of fluid dynamics: The continuity,
momentum, and energy equations which in turn, are the
mathematical representation of the fundamental physical
principles: mass conservation, momentum conservation
(Newton’s second law) and energy conservation (first law of
thermodynamics).
Having the fundamental physical principles is necessary to
apply them to a suitable model of flow; among these models
are the fixed finite control volume, moving finite volume,
fixed infinitesimally small volume or moving infinitesimally
small volume in order to obtain the mathematical equations
which embodies the aforementioned physical principles.
The fluid flow considered in CFD is assumed to be a
continuum medium in which the Navier-Stokes equations can
be applied; however these are nonlinear partial differential
equations and their solution is very difficult. At this time, the
computational resources are not capable enough to handle
such complex problems (at least in the commercial use). This
is the reason why the Navier-Stokes are usually simplified,
neglecting some terms.
The flow model used on RotCFD is the one in which an
infinitesimally small element of fluid is fixed in space having
a differential volume dV. The fluid is infinitesimal in the same
sense as differential calculus; however, it is large enough to
contain a huge number of molecules so that it can be viewed
as a continuous medium. Instead of applying the fundamental
principles to the whole flow they are applied to the
infinitesimally small fluid element itself [2].
The fluid flow in the present paper is governed by the
unsteady, laminar, incompressible Navier-Stokes equations.
For inviscid incompressible flow, conservation of mass and
momentum are sufficient conditions for solving the flow field,
thus making the energy equation redundant.
The conservation of mass law applied to a fluid passing
through the infinitesimally small fixed control volume
aforementioned yields to the following equation of continuity:
߲
߲‫ݐ‬
ሺߩሻ + ∇ ∙ ൫ߩܸሬԦ൯ = 0
Newton’s second law applied to a fluid passing through an
infinitesimal control volume could be expressed by the
momentum equation as shown below:
߲
߲‫ݐ‬
൫ߩܸሬԦ൯ + ∇൫ߩܸሬԦ ܸሬԦ൯ = ߩ݂Ԧ + ∇‫ܫܫ‬పఫ
෪ + ܵԦ
III.2 Discretization
The aforementioned equations assume that all dependent
variables change continuously through the domain. If the
equations are applied to a large domain, the variations in
midpoints would be unknown or inexact. This is the reason
why it is necessary to “break the domain up” into small
sections and apply the governing equations in those small
parts instead; in this way, an accurate estimation of the flow
behavior in all the domains can be obtained. This process is
known as discretization.
Formally defined, discretization is the process by which a
closed-form mathematical expression, such as a function or a
differential or integral equation involving functions is
approximated by analogous expressions which prescribe
values at only a finite number of discrete points or volumes in
the domain.
In contrast to an analytical solution of partial differential
equations, in which the variation of dependent variables are
given continuously throughout the domain, the numerical
solution can give answers only at discrete points in the
domain, called grid points.
The nature of the resulting algebraic system depends on the
character of the problem posed by the original partial
differential equation. Equilibrium problems usually result in a
system of algebraic equations that must be solved
simultaneously throughout the domain in conjunction with
specified boundary values. [3].
There are different kinds of discretization methods employed
in the CFD field; among the most common are the finite
difference, finite element, and finite volume. This last one is
the method employed in RotCFD.
One of the advantages of the finite volume approach resides in
the easy grid adaptability to the body surface and the
flexibility for the transition to the boundaries.
Computationally speaking, the finite volume approach is able
to work in a “physical” plane without the need of a
transformation from physical to a computational plane just as
finite differences approach does. The elimination of this step
allows the use of complex shaped grid elements [3].

III.3 Momentum source approach
The momentum source approach, first applied to vertical axis
wind turbines, simulates the rotor effect on the flow by means
of the momentum generated by the rotor blades’ geometry.
The rotor is replaced by distributed sources of momentum in
the flow. The direction and magnitude of the moment depends
on the rotor geometry and the local flow characteristics. The
main advantage of this approach is that it doesn’t need a body-
fitted rotor grid that requires a lot of computational resources
[4].
V. METHODOLOGY
The way chosen for the validation process is based in the
comparison between experimental and computational results
obtained for some rotorcraft; among these are the CH-53, CH-
47 Chinook, UH-60 BlackHawk, and the V-22 Osprey.
Specifically, the results will be compared using outwash &
downwash analyses for the previously mentioned rotorcraft
[5].
The first step was the CAD modeling of the four aircraft. The
technical drawings were obtained from the web and they are
of public domain.
The software used for the CAD modeling was
CreoParametric2®
. Front, side and top technical views were
placed in the three computational planes accordingly to their
location one respect to the others (Fig. 1). Special attention
has to be placed in the alignment and the dimensioning at this
step since any misalignment can create complications later on
the modeling (i.e. 3d curves may not intersect).
Fig. 1 3D Modeling - CreoParametric2.
Once the model was finished in CreoParametric2, it was
exported as an IGES file, which allowed the model to be
imported in a second CAD software called Rhinoceros®
. The
next step, called post-modeling, consisted of a series of
procedures in which the geometry was treated in order to
prepare it for RotCFD. The model has to be examined
meticulously; if any hole is found, even a tiny one, it has to be
patched. All the surfaces have to be joined together, creating a
kind of single-surfaced body. If all the previous steps are made
exactly as stated before, a lot of troubles will be avoided on
the grid generation. The final four aircraft models are shown
on fig. 2.
Fig. 2 3D Models - Rhinoceros (V-22, UH-60, CH-53, Ch-47).
RotCFD is capable of outputting two kind of files: STL and
P3D; therefore there are two different ways to follow before
RotCFD.
1. Export the geometry from Rhinoceros as an STL file
and then import it into the module RotCFD UNS.
2. Export the geometry from Rhinoceros as an STL file,
import it then into the RotCFD module called
ShapeGen, save it as a P3D file and finally import it
into RotCFD UNS.
The first option is the fastest one but the file size can be too
large. The second option is recommended since it has just an
extra and simple step and yields a more manageable file. The
extra step does not require a special treatment of the geometry
it is as simple as import the file and save it as a P3D file.
One important thing about the STL creation from Rhinoceros
is the surface mesh quality: the mean aspect ratio of the
triangles which conform the surface has to be kept small.
Their size also needs to be small enough to generate a smooth
surface instead of sharpened faces.
The following figure is an example of the geometry generated
using inappropriate mesh sizes. A well-parameterized mesh
would look at first glance as the original geometry.
Fig. 3 Unacceptable STL model.
The format manipulation process using different software
prior to RotCFD can be synthetized in the diagram shown on
Fig. 4.

Once the model has been imported into RotCFD UNS, the
location of the aircraft is simple and intuitive. It can be located
accordingly to global coordinates or using relative
frameworks; rotors or others aircraft can be added. RotCFD
also allows the user to change the pitch, roll and yaw relative
to the global coordinate axes (default: Pitch: X rotation, Roll:
Y rotation, Yaw: Z rotation).
The rotor generation is a very simple step on RotCFD and is
what makes it different from other CFD software. In a
conventional CFD software it is necessary to model the rotor
blade geometry and create a body-fitted mesh attached to the
rotating blade. This kind of mesh makes the calculations much
more difficult, and greatly increases the time required to
generate the grid and the solution.
RotCFD, on the other hand, calculates the flow through the
rotors using the momentum source approach, which simulates
the rotor influences on the flow by applying a series of
momentums equivalent to the ones generated by the original
rotor. There are a certain number of “sources” placed along
the radius of the rotor which represent the momentum
imparted to the flow. The momentum magnitude and direction
depends of the flow characteristics and the airfoils sections of
the blade.
Fig. 4 Process prior to RotCFD.
The rotor location is basically the same as the aircraft body
previously mentioned. The inputs for the rotor are the follows:
• Rotor radius: Total rotor radius.
• Cone angle: The angle between the blades respect to
the hub horizontal axis.
• Collective pitch: The pitch angle applied to all the
blades at the same time.
• Reference twist: Twist angle at 75% of the rotor
radius.
• Tip velocity: Rotor tip velocity.
• Cutout radius: ratio between the physical start of the
blade and the total rotor radius.
• Hinge radius: Ratio between actual hinge length and
rotor radius.
• Rotation direction: Clockwise or counter-clockwise.
• Source locations: Number of momentum sources
along the rotor radius.
• Cyclic pitch: Cyclic pitch coefficients.
• Flapping: Harmonics flap coefficients.
• Airfoil tables: Selects the airfoil used at different
fractions of the rotor radius.
• Chord/Radius: Chord/radius ratio at different
fractions of the rotor radius.
• Twist: Twist angles at different fractions of the rotor
radius.
• Outplane deflection: It allows the user to specify the
out of plane deflection at different fractions of the
rotor radius.
The next step once the model has been imported, located, and
the rotor has been created is to generate the boundaries that
will determine the analysis area. The distances and
specifications for each side of the walls can vary depending on
the kind of analysis. Specifically for this case in which the
helicopter is in hover flight far away from ground or any other
object, the boundaries are located at distances where the air is
barely influenced by the rotor and the body. In the case in
which the flow does interact with real walls (as the ground and
wind tunnel test section walls are) the boundary walls are
located at the same location that the real ones (fig. 5). The way
to make them “real” is by specifying conditions of
impenetrability, this is, make the normal component of the
velocity equal to zero.
Fig. 5 Boundaries and refinement boxes.
One of the most critical steps is the grid generation. The
greater the interest of the flow behavior in certain areas, the
smaller the grid size has to be. The same applies for high
pressure gradients or for complex geometries where high
definition is needed.
Some of the most important factors to take into account
specifically on RotCFD are the follows:
• Initial X (Y or Z) Cells: This number defines the
number of divisions of the total length between two
boundaries placed perpendicular to the direction
specified (X, Y, or Z). They are called “initial”
number of cells because if there is an object with a
specific refinement somewhere in the middle or near
enough to the trajectory, those cells will be divided in
even smaller elements (depending on the body
refinement).
• It is important to mention that the ratio total
length/number of divisions, has to be kept constant

for all three directions; in other words, each of the
elements formed in the grid has to be a perfect cube.
The mathematical behavior of the governing
equations is very complicated, and if this ratio is
maintained, the computational solution will be much
easier to obtain.
• Body Refinement: Allows the grid to be smaller in
the surrounding of the body. 6 or 7 are common
values used.
• Rotor Refinement: Allow the grid to be smaller in the
rotor surroundings. 6 or 7 are common values used.
• Fit bodies by default: This box enables an algorithm
that makes the rectangular grid adapt to the body
creating a tetrahedral grid near the body. If the box is
not checked the geometry will be simplified too
much and at the end it can give results different to the
expected ones.
Besides all these results there is another tool that increases the
grid refinement in specified areas, this is called Refinement
Box. As its name implies this tool generates a simple box that
increases the refinement of the grid in all the volume
contained on it. Usually they are used in zones of special
interest, i.e. rotor surroundings, body surroundings, on the
ground for Outwash & Downwash analyses (fig. 6).
We have shown up to this point all the steps and the inputs
needed for the grid generation.
Once all is ready the grid is generated by pushing the “Run
Grid Generator” button. The results looks like the follow
screenshot.
Fig. 6 Grid.
In some cases the geometry have small holes, in such cases the
grid would be generated even into the body. One way to verify
that everything is working well up to the present step is by
hiding the body as shown in fig. 7. In this specific case the
grid was generated properly what means that the geometry
was well-modeled.
Fig. 7 Grid inspection.
RotCFD has a helpful tool called “Body Surface Tester” that
enables the visualization of the geometry that actually is going
to be analyzed (The original geometry is just a medium for the
last one.
As previously mentioned, the definition of the final body
depends on the grid but also on the original geometry.
In figure 8 two different V-22 geometries are shown, the only
difference between both is the grid refinement. In the first one
the geometry looks almost as the original geometry, the only
differences are in thin parts such as the vertical stabilizer or
the wing trailing edge. Moreover the fig. 8 (b) looks a little
coarser with a surface not as smooth as the previous one.
The geometry refinement is an important factor buy it is not
the only one to take into account; reduction of time is
something desirable on CFD analyses and an analysis using a
grid as refined as the one used in fig.8 (a) will take much more
time than the second one and sometimes the results are almost
the same. This is why there is not just a single right solution.
Trade-off between this two factors is essential in CFD.
The physical flow properties in RotCFD are essentially the
same that the ones at any CFD software and all of them
depend on the atmospheric conditions at which the analysis is
taking place i.e. density, temperature, gas constant, specific
heats, viscosity, pressure, etc. It is assumed that the reader
knows all this properties therefore extra explanation is not
included.
Fig. 8 (a,b) Body Surfaces comparison.
One of the tools RotCFD has specialized for aircraft is the
“Flight Condition” module. This section makes the
characterization of the flow direction and velocity very simple
with minimal user interference needed. Five different flight
conditions are available:
• General: It allows the user to specify the magnitude
for all three velocity components.
• Hover: As this is a flight characteristic in which the
velocity relative to the rotorcraft is cero not input is
needed here.
• Forward Flight: Only the velocity for the direction
of the flight is needed.
• Climb: Only the climb velocity is needed and it is
not necessary to specify directions.
• Descent: Only the descent velocity is needed and it is
not necessary to specify directions.
The specification of the grid times and iterations is the last
step before running the simulation. The options available are:
• Time Length: This is the physical real time that the
software is going to calculate, the analysis start
calculating the flow at time cero with the boundary
conditions and flow properties as initial conditions

i.e. Analyze the rotor wake the first 20 seconds once
the rotor star to spin.
• Time Step: This is the number of divisions that the
Time Length will be spitted for the calculus. The grid
solution will be calculated at small time increments
of ∆t= Time Length/ Time Step. The larger the Time
Step, the larger the results are for each grid discrete
point. The results are also more accurate but the
simulation time will take longer.
• Iterations: Number of iterations for each ∆t at each
grid discrete point.
• Relaxation numbers: They prevent large increments
in the equation dependent variables from one
iteration to other. The value 0.01 means that only 1%
of the present variable value is been used for the
calculus the remaining 99% depends of the previous
iteration.
Up to this point everything needed for the simulation is ready
but is recommended to enables the “Save restart history”
option because is very common to have the analysis stopped
for any reason, in that case all the information previously
calculated would be lost.
The simulation time can vary from hours up to weeks or
months depending on computer capabilities and grid
refinements. It is recommended to run some simple
simulations at first using coarse grid refinement and verify that
everything works as expected, so then increase refinements
and run loner simulations.
VI. RESULTS
As was suggested, some preliminary analyses were run at first
and the grid was inspected to verify that everything was
working properly. Fig 9 shows some of these preliminary
analyses.
Fig. 9 Preliminary results
Having these results is possible to assume that the grid is
properly working, unfortunately these were not the only
results but also there were some results in which some errors
were presented giving some extremely high pressure gradients
in very small areas.
In general CFD does not offer straightforward solutions that’s
why a lot of Time Length, Time Step, Iterations and
Relaxations number analyses combinations were run in order
to find the best combinations. Some of the most stable results
were found using relaxation numbers of 0.01, Time Length
higher to 10, Time Steps in a range from 1000 to 10000 and
Iterations around 20.
In the fig. 10 are shown the screenshots in where is possible to
see some of the different results using different combinations.
Fig. 10 Results
All the previous analyses were run at hover flight which
means that the aircrafts are in a supposed single point without
relative wind velocity.
In fig. 10 a) the CH-53 was analyzed at cero collective pitch,
with a blade without twist and using generic airfoils NACA
0012. Basically the rotor is not generating lift force and this is
why is not possible to see the wake under the rotor.
The V-22 analysis was run at 10 seconds with small
increments of ∆t=0.1 and using collective pitch of 10°. In this
analysis is possible to see how the wakes does develops under
the rotor. The different colors on the surface are the pressure
variations on the body surface due strictly to the flow. The
highest pressure is presented in the upper wind which makes
sense because is where the flow has the higher velocity normal
to the body surface.
The helicopter CH-47 opposite to the others is analyzed near
the ground and this is why the flow develops in the outward
direction near the lowest boundary condition.
The last picture shows the UH-60 the one is using the same
specifications than the V-22. Basically the difference is the
interaction between the flow and the fuselage. In the V-22 the
body pressure is higher in the upper part of the fuselage due to
the high velocity flow under the tip of the rotors but in the
UH-60 the highest velocities of the flow under the tip rotors
are not above the upper part of the fuselage.
Essentially this is the whole procedure used for the validation
of RotCFD results, due to data confidentiality, all the present
analyses are not using the real rotor specifications neither the
real airfoils. All resting is to change the above mention rotor
characteristics, run the analyses and finally compare the
experimental data whit RotCFD results.

VII. CONCLUSIONS
All the modeling processes and all previous procedure to the
usage of a CFD tool have to be performed carefully and
inspected over and over. Sometimes a simple error in the
geometry can delay the project too much time if not detected
on time.
The Project consummation will take place once all the rotor
specification are changed for the same than the Outwash &
Downwash ones and the analyses are run; having all these
same specifications, both results are directly comparable.
VIII. REFERENCIAS
[1] Sukra Helitek, Inc. “RotCFD Solver Theory” February 5 2013.
[2] Anderson John Jr. “Fundamentals of Aerodynamics” 5th Edition 2011.
[3] Anderson John Jr. “Computational Fluid Dynamics The basics With
Applications” 1995.
[4] Guntupalli Kanchan “Development, validation and verification of the
Momentum Source Model for discrete rotor blades” owa State
University 2011.
[5] Smith Robert D. “Heliport/Vertiport Design Deliberations 1997-2000.
IX. ROJECT MEMBERS
William G. Warmbrodt1
, Carl R. Russell1
, José Luis
Arciniega Jiménez2
, Zulema Guadalupe García Lozano2
,
Erik Alberto Márquez Angeles2
, Eliud Israel Meza
Escamilla3
, José Alberto Ramírez Juarez2
1
NASA Ames Research Center (NASA ARC).
Rotorcraft Aeromechanics Brach, ZIP. 94035.
2
Universidad Politécnica Metropolitana de Hidalgo
(UPMH).
Departamento de Ingeniería en Aeronáutica, CP. 42040.
3
Universidad Autónoma de Nuevo León (UANL).
Facultad de Ingeniería Mecánica y Eléctrica, CP. 66450.

Paper RotCFD NASA

More Related Content

Similar to Paper RotCFD NASA (20)

Paper RotCFD NASA