Skip to main content

Full text of "DTIC ADA619807: Progress in Finite Element Modeling of the Lower Extremities"

See other formats


ARL-MR-0890 • Jun 2015 


> 


ARL 

US Army Research Laboratory 


Progress in Finite Element Modeling of the 
Lower Extremities 

by Adam Sokolow 


Approved for public release; distribution unlimited. 


NOTICES 

Disclaimers 

The findings in this report are not to be construed as an official Department of the 
Army position unless so designated by other authorized documents. 

Citation of manufacturer’s or trade names does not constitute an official 
endorsement or approval of the use thereof. 

Destroy this report when it is no longer needed. Do not return it to the originator. 



ARL-MR-0890 • Jun 2015 


ARL 

US Army Research Laboratory 


Progress in Finite Element Modeling of the 
Lower Extremities 

by Adam Sokolow 

Weapons and Materials Research Directorate ; ARL 


Approved for public release; distribution unlimited. 



REPORT DOCUMENTATION PAGE 


Form Approved 
OMB No. 0704-0188 


Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the 
data needed, and completing and reviewing the collection information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing the 
burden, to Department of Defense, Washington Headquarters Services, Directorate for Information Operations and Reports (0704-0188), 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302. 
Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any penalty for failing to comply with a collection of information if it does not display a currently valid 
OMB control number. 

PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS. 

1. REPORT DATE (DD-MM-YYYY) 2. REPORT TYPE 3. DATES COVERED (From - To) 

June 2015 Memorandum Report October 2014-February 2015 

4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER 

Progress in Finite Element Modeling of the Lower Extremities 

5b. GRANT NUMBER 

5c. PROGRAM ELEMENT NUMBER 

6. AUTHOR(S) 5d. PROJECT NUMBER 

Adam Sokolow 

5e. TASK NUMBER 

5f. WORK UNIT NUMBER 

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION REPORT NUMBER 

US Army Research Laboratory 

ATTN: RDRL-WMP-B ARL-MR-0890 

Aberdeen Proving Ground, MD 21005 

9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR'S ACRONYM(S) 

11. SPONSOR/MONITOR'S REPORT NUMBER(S) 

12. DISTRIBUTION/AVAILABILITY STATEMENT 

Approved for public release; distribution unlimited. 

13. SUPPLEMENTARY NOTES 

14. ABSTRACT 

Human body modeling efforts for the purpose of Soldier protection need to address the current threats as well as have a vision 
for the future. Modeling the human body is a challenging endeavor due to its geometric complexity, numerous interacting 
layers, rich anisotropy, and wide variability. Developing a model for predictive injury capability, therefore, needs to be 
versatile and flexible to address different levels of modeling complexity. The vision presented here surrounds a flexible mix- 
and-match assembly approach. This assembly process has the capability to take a collection of source body part meshes that 
may have different resolutions, deform the meshes based on the individual to be simulated, and posture them into different 
positions so that the end result can be exported into multiple finite element solvers. The primary focus of the present effort is 
the mounted Soldier’s response to accelerative loading from underbody blast events. Many of the challenges in modeling the 
human body remain the same for applications such as the response of dismounted Soldiers. This report presents a progress 
report of our current efforts and documents some major improvements to the lower leg model with a vision of the future in 
mind. We also introduce significant details regarding an assembly architecture that is currently under development. _ 

15. SUBJECT TERMS 

lower extremities, FEM, accelerative loading, biological variability, solver independent 


16. SECURITY CLASSIFICATION OF: 

17. LIMITATION 

OF ABSTRACT 

18. NUMBER 

OF PAGES 

19a. NAME OF RESPONSIBLE PERSON 

Adam Sokolow 

a. REPORT 

b. ABSTRACT 

c. THIS PAGE 

uu 

22 

19b. TELEPHONE NUMBER (Include area code) 

Unclassified 

Unclassified 

Unclassified 



410-306-2985 


Standard Form 298 (Rev. 8/98) 
Prescribed by ANSI Std. Z39.18 


it 



































Contents 


List of Figures iv 

1. Introduction 1 

2. Hexahedron Meshes 3 

3. Assembler Overview 6 

4. Contact, Material, Part Management, and Structural Element 

Generation 8 

5. Subdividing and Grading Materials 10 

6. Accounting for Biological Variability 11 

7. Conclusions 14 

Distribution List 15 


iii 




List of Figures 


Fig. 1 Coarse hexahedral meshes of the bones of the foot and lower leg.4 

Fig. 2 Coarse hexahedral meshes of the muscles and flesh of the lower leg.5 

Fig. 3 A coarse mesh of the calcaneus (left) is replaced with a refined mesh 
(right) by changing a single line of code. The 1-D structural elements 
are automatically updated during the assembly process in the model 
management software.9 

Fig. 4 Close-up of the cuneiforms and navicular. The left panel shows a set of 
1-D structural elements (purple) automatically determined from the 
model management software. The right panel shows a second set of 
elements determined from a different set of parameters.10 

Fig. 5 The calcaneus and talus are subdivided to account for a solid cortical 

layer (light gray) and trabecular interior (darker gray). The stair-stepped 
trabecular interior is due to the coarse mesh resolution.11 

Fig. 6 The relative length of the femur to tibia is taken as an input parameter 

and scaled independently from the overall model dimensions.12 

Fig. 7 The relative size of the thigh and calf are taken as input scaled 

independently from the overall model dimensions, and do not affect the 
underlying bone structure.13 


IV 











1. Introduction 


Developing a finite element model of the human body that has predictive injury 
capability presents many challenges, the first of which is identifying the types of 
conditions where injury prediction is needed. Our focus is on lower-leg injuries due 
to an underbody blast event beneath a vehicle. During such an event, the explosive 
gases and soil impart momentum to the vehicle hull. Defonnations of the hull affect 
the floor plate, which can be in direct contact with the vehicle occupants. The floor 
plate can also undergo rapid accelerations that result in occupant injury. 

Injuries that result from these events are often dubbed accelerative injuries, 
although this nomenclature is misleading in that it combines 2 types of injuries 
under the same terminology. The 2 cases separate into what can be considered a 
shock perspective and a structural perspective. To model these events properly 
requires 2 models to be developed by 2 classes of injury mechanisms: one model 
that can adequately resolve wave propagation and a second that is suitable for 
accumulated strain. While both are necessary in the long run, the current emphasis 
is placed on a model that is adequate to predict injuries due to accumulated strain. 
This type of model can be generated for the entire body and has a more immediate 
impact on improvements in Soldier protection. 

For the shock case, wave propagation within the tissues that make up the structure 
is incredibly important, as well as understanding the full constitutive response and 
the types of interfaces a propagating wave might encounter. This case would be 
most concerned with the initial impact of the floor plate with the occupant. After 
the initial strike, a shock wave, or a large and rapidly changing impulse, will 
propagate through the substructures and across the material interfaces. The 
complicated interplay that results from anisotropy, rate-dependence, fracture, and 
interface dynamics can all play a critical role in injury. Wave propagative models 
should be developed at the component level and take into consideration the 
numerous layers that are present. To account for the types of injuries seen in the 
calcaneus, such component models would likely require multiple elements through 
the flesh, fat, muscles, and tendons of the heel, and account for the grading of the 
cortical to trabecular bone of the calcaneus. While we currently do not have 
accurate geometries for all these layers, we may have them in the future. Thus, our 
modeling effort needs to be adaptable to allow for including refined meshes in 
future work. 

For the structural perspective, the model is most concerned with accumulated 
deformation imposed on the load-bearing structures. This accumulated deformation 
might result from the somewhat simple case of compression along the loading path, 


1 




or the more complicated case where inertial effects result in bending and 
subsequent injury, e.g., the distal tibia motion results in bending of the tibia rather 
than the tibia rotating about the knee joint. Although still an active research 
question, intuitively it seems that calcaneal fractures are more likely to result from 
wave propagative effects whereas tibial fractures result from accumulated strain. 
Namely, rapidly varying propagating waves may fracture the microstructure of the 
calcaneus, while the tibia fails as it bends like a beam or under compressive loads. 
Either or both of these cases might be necessary to model when developing a fully 
predictive model. However, the level of detail required and the refinement of the 
mesh is vastly different for the 2 cases. 

The second challenge in modeling the human is accounting for variability when 
necessary. In the context of underbody blast events, there is a large amount of 
variability in the loading that reaches the occupant. These events occur in theater, 
and there is a wide range of postures that the occupants can be positioned in during 
an event. There are also large variations between Soldiers, some of which are 
readily visible like height and weight, while others are more subtle, e.g., femur 
length, bone structure and density, or completely unknown, e.g., prior injury. The 
materials themselves also have variability, e.g., cortical bone from the skull is 
structurally different from the femur and therefore exhibits different material 
characteristics. Thus, a true statistical modeling approach needs to incorporate 
posture and structural differences, as well as variations in the materials. Often these 
are not independent characteristics and thus a larger understanding of covariates of 
the system is necessary. However, the degree to which any type of variability 
matters in the context of predicting injury is an open research question. 

These challenges and the unknowns introduce numerous practical concerns, 
including version control for the multitude of parts—whether it is material model, 
mesh resolution, contact, connectivity, or posturing. Practical issues exist like 
numbering parts uniquely, producing documented input files for the solvers, and 
facilitating presentation quality images. These challenges also introduce numerous 
research concerns such as parameterizing structural features like bone thic kn esses 
or bone dimensions, or material features like the defined directionality of tissue 
anisotropy. In this report, both the practical and the research concerns are addressed 
by centralizing the efforts into a single assembler. 

This report is organized as follows. First, a summary of the meshing efforts is 
presented, as it represents a substantial improvement from the previous version of 
the lower leg. The remaining sections outline the assembler program starting with 
an overview and then continuing into a description of its current capabilities that 
include model management, controlled generation of beam elements, optional 
subdivision of parts, and model morphing capabilities. 


2 



2. Hexahedron Meshes 


A typical criticism of previous modeling efforts has been that the model is built on 
tetrahedron elements. These types of elements behave too stiffly at large strain, and 
do not allow for accurate shock wave propagation. In an effort to improve both 
accuracy and simulation run time, a major effort has been made to develop 
hexahedral meshes. 

The source geometries that we are working with are from Zygote Media Group, 
Inc. in a sterolithographic (STL) format. This format is inherently different from 
the typical computer-aided design (CAD) fde because it is a surface mesh and not 
B-spline based geometry. The full capability of a meshing software is rarely 
available, e.g., CUBIT (Sandia National Laboratory) treats geometries in a CAD 
engine as a separate engine from STLs and they cannot be mixed. The stability of 
CUBIT is also highly questionable when using STL files and repeated crashes 
during mesh development are common. Fortunately, these structures can be meshed 
using a blocking scheme in meshing software that has this capability. We use ICEM 
CFD (ANSYS) meshing software to produce highly controlled and parameterized 
meshes of the bones, muscles, and flesh. 

Due to the complexity of the geometries, the general modeling approach is to treat 
the bones, muscles, and skin as separate components and mesh them individually. 
Meshes are then joined at a later stage through combinations of contact definitions 
and 1 -dimensional (1-D) structural elements, and where appropriate, node merging. 
These choices are all flexible and can be changed in later versions. 

Hexahedral meshes were developed for 23 bones in the right-lower extremity, 
including the pelvis, femur, patella, tibia, fibula, talus, calcaneus, cuboid, navicular, 
the cuneiforms, metatarsals, and phalanges. Figure 1 shows the lower leg below the 
knee to provide an idea of the mesh resolution of the coarse model. We note that 
each toe has been simplified so that the 2 or 3 small phalanges that form a single 
toe are merged into one. This results in a single phalange for each toe. The intent 
of the mesh resolution is to capture loading events that are on the order of 
milliseconds, i.e., accumulated strain of the bone structures and some kinematics. 


3 





Fig. 1 Coarse hexahedral meshes of the bones of the foot and lower leg 

In addition to the 23 bones, there are 19 muscles that are currently meshed below 
the knee, these are shown in the left panel of Fig. 2a (red elements). There are also 
23 muscles between the pelvis and patella that have been meshed (Fig. 2b). Figure 
2c also shows the current flesh layer (green elements) that envelops the muscles 
and bones. This flesh layer is a homogenization of the multiple layers of skin, fat, 
and other connective tissues that surround the musculoskeletal systems. Figure 2 
also illustrates a number of 1-D structural elements (purple line segments) used to 
represent connective tissues such as tendons and ligaments. Where necessary, these 
1-D structural elements will be replaced with solid elements in future work. 


4 




























Fig. 2 Coarse hexahedral meshes of the muscles and flesh of the lower leg 


Using LS-Dyna (Livermore Software Technology Corporation), simulations 
without the flesh layer (corresponding to roughly 30,000 elements) can reach 50-ms 
simulation time on 16 cores in under an hour. Incorporation of the flesh layer into 
a running simulation is underway. The difficulty lies, in part, in the choice to mesh 
parts separately, but more specifically challenges are associated with errors in the 
Zygote Media Group, Inc. source geometries. The source geometries, as mentioned 
before, are STL format and have numerous surfaces that interpenetrate one another. 
This becomes an issue when the solver attempts to resolve a defined contact 
between 2 interpenetrating surfaces. In these cases, the softer material is moved in 
such a way that it can create a small or negative volume. Element erosion 
occasionally can fix this, but it can lead to a zero time step. Parallel efforts are 
currently underway to convert the meshed structures into CAD formats where 
Boolean operations can be performed to remove the interpenetrating surfaces as 
well as identify critical interpenetrating surfaces and attempt to resolve them at the 
blocking stage. Enabling contact definitions also may reduce the time step to 
maintain stability, thereby causing a reduction in performance. 


5 









3. Assembler Overview 


The challenges and questions raised in the introduction need to be answered for any 
model, and ideally, changing the specific configuration within a model should be 
fairly simple and require minimal work. The remainder of this report describes 
some of the features of an assembler program that is in development. 

The assembler program is divided into 4 main components: a collection of finite- 
element data structures called Gilgamesh, a model parser and assembler called 
Gargamel, a model defonner Gumby, and a model translator Galvatron. 

Gilgamesh handles the objects that are related to the model and the mesh, i.e., 
nodes, elements, parts, materials, boundary conditions, etc. Gargamel takes the 
input meshes and performs some practical operations including adding comments, 
material assignment, and colormap generation but also includes subdividing source 
meshes into multiple subparts, detennining contact pairs from sets, and creating 
additional connective elements such as 1-D beams. In this way, Gargamel takes 
specified inputs and operators to meld together a model that is represented as a 
Gilgamesh object. Gumby operators apply deformations to the nodes stored in 
Gilgamesh using the information specified by Gargamel. These deformations 
(discussed in Section 6) are used to accommodate biological variability in the 
structures and will be used in the future to accommodate some posturing. Galvatron 
is the last stage of the assembler code. Galvatron translates the model contained in 
Gilgamesh to a format that can be parsed by an external finite element method 
(FEM) solver. 

The program is written in C++ and heavily makes use of operator overloading, 
function pointers, and stated-structures. The end goal is to allow a highly 
specialized but simple interface for a human model developer. At this point, the 
interface is within the code itself and requires a recompile, but a parser could easily 
be added in a future revision. 

Current off-the-shelf software largely addresses either the meshing step or the 
solver step. Both sides have features that are similar to what is being described here, 
and a brute force approach could be used to accomplish these tasks. It is common 
for a modeler to have dozens of disjoint scripts to accomplish some of the tasks that 
are discussed in Sections 4-6. For example, LS-Dyna has an accompanying pre¬ 
processing software that allows both scripted and graphical interface interaction to 
manipulate the mesh and materials. Similarly, TrueGrid (XYZ Scientific 
Applications) is capable of outputting Dyna3D (Lawrence Livermore National 
Laboratory) input decks. However, in both cases, swapping out a part that is 
multiply connected by 1-D structural elements and inserting a finer mesh would 


6 




require manual manipulation and changes in multiple locations in the input deck. 
Similarly, deciding to subdivide a part into multiple parts according to a 
parameterized function requires external scripts or mesh manipulation and re¬ 
importing of the model. In the case of a functionally graded material, new parts to 
the mesh would have to be included, requiring new material definitions and new 
contacts to be defined manually. 

The mesh is often considered a single entity at the input stage of FEM solver codes, 
making version control of the model combinatorial. In the ideal case, the model 
should be the mesh, material models, and boundary conditions, and it should be 
somewhat solver independent. This procedure would also largely depend on the 
target solver. As one can see, it can become quite cumbersome for a modeler to 
toggle back and forth between solvers. However, by centralizing some of these 
capabilities and adding in very specific features that would otherwise appear in a 
tangled mess of scripts, the assembler organizes research projects into one program 
so that they can be used later on in a practical way. 

Another key strength of using a programming language like C++ is polymorphism 
and object-oriented programming. An object is said to be polymorphic if it behaves 
differently in different contexts. There are many different ways to accomplish the 
polymorphic design and our current approach is to utilize a state structure. For 
example, in the context of solver input decks, a material is output to a file in 
different formats depending on whether the assembler is in an LS-Dyna, Dyna3D, 
or SIERRA state (Sierra Solid Mechanics, Sandia National Laboratory). Thus, the 
modeler can work with the constitutive model and parameters instead of worrying 
about properly formatting the material for the code. This simple task can be prone 
to multiple errors. In the case of the lower leg, if each component has a unique set 
of material models and parameters, a few hundred material definitions would have 
to be translated from one solver to another. Since the material models and 
parameters are still being updated, this task would be very inefficient as it would 
have to be revisited numerous times. By utilizing polymorphic design, the model 
development is solver independent. In the current version of the code, the assembler 
outputs to LS-Dyna input decks, and is partially implemented to output Dyna3D 
input decks. Since this is done utilizing a single class structure, the Dyna3D portion 
can be omitted entirely and not affect the code. Similarly, future additions like a 
SIERRA output class could be added in the future, thus making it a flexible 
approach. What is important to note is that the polymorphic design lets the user 
interact with the assembler using the same commands, and it will produce different 
output depending on the solver fonnat that is desired. This allows the user to view 
the model as mesh entities, materials, boundaries, and contacts, and the code 
properly converts it to the syntax needed for the FEM parser. These translator 


7 



subroutines can be added as needed to accommodate the specific material models 
that are used, i.e., we are not attempting to recreate or accommodate for the entire 
material library. 

4. Contact, Material, Part Management, and Structural 
Element Generation 


Since we are using a higher-level assembler program that maintains an abstraction 
of the FEM model, the contacts defined, the materials, and the parts are all kept in 
data structures. The map data structure in C++ is utilized to associate common 
names of the part, e.g., “tibia” to class structures that have details about that part, 
e.g., the material of the tibia. The data structures allow for iteration and therefore 
simple renumbering when necessary (as is needed in Dyna3D). The part 
management also enables useful comments to be generated within the input decks 
that help isolate nodes, elements, parts to be associated with their “common name” 
as well as their internal solver identification numbers. These same keys are used to 
create a colormap customized to the model. 

Contacts are defined pairwise and have a similar data structure for their 
management. Similarly for materials, sets of data structures help organize the 
model. These are managed in a larger structure that assigns numbers to them 
properly and interacts with the elements and part objects to do similar assignments. 
Contacts can also be defined as one-to-many using sets or all combinations. For 
example, the tibia can be set to have contact with all the muscles below the knee, 
or bones of the foot can be assigned to have contact with each other. 

The mesh used for a part is included in its internal information, so making a 
substitution from a coarse mesh to a fine mesh is a simple file name substitution 
(see Fig. 3). The data structures that manage the parts also communicate with the 
contact and 1-D structural element generation. This allows the user to simply 
comment out a part on a single line and the contacts, 1-D structures, material files, 
elements, and nodes will all be renumbered and accounted for automatically. Both 
of these cases are particularly useful in Dyna3D since any change in element count 
directly affects the nodes, elements, parts, in header files and also impacts the 
number of “cards” the solver is expecting. Similar interdependencies exist in LS- 
Dyna as well, although they are not as pervasive. 


8 





Fig. 3 A coarse mesh of the calcaneus (left) is replaced with a refined mesh (right) by 
changing a single line of code. The 1-D structural elements are automatically updated during 
the assembly process in the model management software. 

A major drawback to the Zygote Media Group, Inc. geometries that we currently 
have is a lack of detail in the connective tissues. This results in gaps between bones, 
which is clearly problematic for simulating wave propagation. Figure 4 shows a 
close-up of the cuneiforms of the foot. In the body, these bones are tightly coupled 
to one another and have cartilage filling the gaps between bones. Due to the lack of 
detail in the source geometries, the choices on the modeling side are either to 
manually create cartilage structures or to fill these gaps with a simpler 
approximation (background grid approaches are also being explored). The auto 
1-D-element calculators are the simplest approach to connect adjacent bones. In 
Gargamel, one can request a connection between 2 objects. This connection can be 
parameterized so that the connected nodes between 2 objects can be adjusted. 
Figure 4 shows an example where the number of connections and their desired 
minimum and maximum lengths are changed. One can also keep the same 
parameters, and make a mesh substitution and the beam connectors will be 
recalculated (Fig. 3). 


9 









Fig. 4 Close-up of the cuneiforms and navicular. The left panel shows a set of 1-D structural 
elements (purple) automatically determined from the model management software. The right 
panel shows a second set of elements determined from a different set of parameters. 


5. Subdividing and Grading Materials 

There are numerous transitions of materials within the body. Some of these 
transitions can be considered as disjoint layers while others are more appropriately 
modeled as functionally graded materials, e.g., trabecular bone. However, this is 
not a simple material swap in the case of switching between a functionally graded 
material and 2 distinct layers. 

The source geometries correspond to a major anatomical component. In the case of 
the bones, it is only the surface of a particular bone. Therefore, any detail within 
that volume needs to be generated by the modeler. Specifically, the thickness of the 
cortical shell is an issue. This outer region of the bone varies in thickness from one 
bone to the next and within a bone itself. Some researchers choose to represent this 
layer with shell elements, while others assume a unifonn thic kn ess of solid 
elements, while others take into account some of the varying thic kn esses. A shell 
representation may allow for a considerably larger time step as well as fewer 
elements. The current version of the assembler program lets the user switch 
between representing the surface by shell elements or a uniform thickness by 
changing one line of code. This is done within Gargamel by taking the input mesh 
of the entire part and subdividing it into multiple parts according to a thic kn ess 
parameter. Figure 5 shows an example result of this operation for the calcaneus and 
talus. This procedure is limited by the resolution of the mesh and therefore can 


10 














































produce a stair-stepped interior, as seen in Fig. 5. This procedure can be generalized 
in the future to account for graded materials or alternate methods of subdividing the 
part. 



Fig. 5 The calcaneus and talus are subdivided to account for a solid cortical layer (light 
gray) and trabecular interior (darker gray). The stair-stepped trabecular interior is due to the 
coarse mesh resolution. 

6. Accounting for Biological Variability 

Biological variability is another issue to be accounted for when modeling the 
human. The goal of the assembler is to have a single set of source geometries and 
their associated meshes that can be distorted to represent other individuals. Often a 
simple uniform scaling can be used to match a single length, e.g., femur length. 
Unfortunately, femur length and other measurements do not correlate to all other 
possible anthropometric measurements. Here, we briefly describe a capability of 
the assembler that enables a single FEM model to undergo simple modifications in 
the code to output a different model for different Soldiers. 

In Gumby, various functional nodal mappings can be applied to and taken relative 
to specific structures. Figure 6 shows a particularly extreme example where the 
relative length of the femur to the tibia is taken as an input parameter. In Gumby, a 


11 




source object like the tibia or femur is used to define a coordinate system upon 
which a smoothly varying function defines the nodal mapping. In this case, the 
nodes of the leg are either compressed or stretched to translate the relative location 
of the knee up or down. Aside from shortening or lengthening the surrounding 
muscles, this mapping does not change the other dimensions of the leg. Although 
this method is still under development and can be improved to minimize knee 
distortion, it shows potential for accommodating multiple Soldiers from one FEM 
model. The 9 legs shown were generated in seconds by adjusting a single input 
parameter. 












Fig. 6 The relative length of the femur to tibia is taken as an input parameter and scaled 
independently from the overall model dimensions 


Another Gumby-related distortion is shown in Fig. 7. Here the relative thigh and 
calf thicknesses are adjusted by an input parameter. Using the surface location of 
the long bones, a functional mapping can be defined to only alter the nodal locations 
of the flesh. This type of morphing provides the capability of modeling individuals 
with a larger soft tissue mass relative to their bone structure. 


12 



Fig. 7 The relative size of the thigh and calf are taken as input scaled independently from 
the overall model dimensions, and do not affect the underlying bone structure 

Both of these approaches can be combined together with an overall scaling of the 
model. All of these features in combination would enable FEM simulations of 
individuals not represented by the original geometries. This procedure can easily 
be extended to the other regions of the body and enables the vision of a having an 
anny of human body models to subject to the same threat. 


13 







7. Conclusions 


We have briefly presented some of the progress on modeling the lower leg. A major 
effort has been put into mesh development. The coarser resolution and the update 
to hexahedral elements have tremendously reduced computational run time 
compared to the earlier version of the model. The model still requires a lot of work 
due to interpenetrating volumes that are present in the source geometries. A 
framework is being developed to manage and assemble the human body model to 
allow for rapid prototyping and incorporating enhancements that are specific to the 
human body. The goal of the framework is to be solver independent, to organize 
and streamline the human body modeling effort, and to mix-and-match component 
models from different resolutions and configurations to account for numerous 
Soldiers and postures. 


14 




1 DEFENSE TECHNICAL 
(PDF) INFORMATION CTR 

DTIC OCA 

2 DIRECTOR 

(PDF) US ARMY RESEARCH LAB 
RDRL CIO LL 

IMAL HRA MAIL & RECORDS 
MGMT 

1 GOVT PRINTG OFC 
(PDF) AMALHOTRA 

43 DIR USARL 
(PDF) RDRL WM 
SKARNA 
RDRL WML C 
T PIEHLER 
RDRL WML H 
B SCHUSTER 
RDRL WMM B 
BLOVE 
RDRL WMM G 
L PIEHLER 
N ZANDER 
RDRL WMP 
S SCHOENFELD 
RDRL WMP B 
A DAGRO 
A DILEONARDI 
W EVANS 
C GUNNARSSON 
C HOPPEL 
Y HUANG 
M KLEINBERGER 
MLYNCH 
J MCDONALD 
P MCKEE 
S SATAPATHY 
A SOKOLOW 
C WEAVER 
T WEERASOORIYA 
S WOZNIAK 
T ZHANG 
K ZIEGLER 
RDRL WMP C 
R BECKER 
S BILYK 
T BJERKE 
D CASEM 
J CLAYTON 
D DANDEKAR 
M GREENFIELD 
B LEAVY 
RDRL WMP D 
R DONEY 


J RUNYEON 
RDRL WMP E 
P SWOBODA 
RDRL WMP F 
D FOX 

E FIORAVANTE 
N GNIAZDOWSKI 
R GUPTA 
R KARGUS 
RDRL WMP G 
R BANTON 
N ELDREDGE 
S KUKUCK 


15 


Intentionally left blank. 


16