Richard S. Gallagher. Computer Visualization: Graphics Techniques for Engineering and Scientific Analysis

Подождите немного. Документ загружается.

Search Tips

Advanced Search

Computer Visualization: Graphics Techniques for Engineering and Scientific Analysis

by Richard S. Gallagher. Solomon Press

CRC Press, CRC Press LLC

ISBN: 0849390508 Pub Date: 12/22/94

Search this book:

Previous Table of Contents Next

Visualization allows us to explore our data in different ways and at different levels of abstraction (to find the

right level of detail). Data visualization provides new tools for exploratory data analysis. Such techniques are

most useful if they are highly interactive, permit direct manipulation, and include rapid response time. The

analyst must be able to navigate the data, change its grain (resolution), and alter its representation (change

symbols, colors, etc.). Graphic display of objects may use color differentiation (identity, type, attributes) or

special symbols (glyphs, stars, boids). The user can experiment with the image to find “the correct

representation.” The goal is to represent data so that its meaning (significance) is apparent, and we can see

patterns or relations in data.

In simulation experiments, data analysis, and display design, we want to incorporate the investigator into the

process to guide and steer computations, to evaluate output and make decisions, and to control the simulation

“on the fly.” This requires representations that are intuitive and easily interpretable so their significance may

be grasped quickly. The display must contain information at the correct level of detail (for the task) and

ideally focus attention on key information. Human perception, cognition, and response time will be the key

limitations of “person-in-the-loop” computation and simulation.

Visualization is essential for understanding the concept of multidimensional space. Clearly, the use of

graphics has enhanced our appreciation of 3-D space and enabled us to understand how 3-D objects map into

2-D views. Banchoff has explored graphic techniques for representing geometric phenomena in four (or more)

dimensions [3]. Shepard studied how people can construct meaningful representations of higher dimensional

space [35]. Just as first year engineering students come to interpret multiple 2-D views and construct a mental

image of a 3-D object, one can construct a 4-D model from a sequence of 2-D and 3-D images.

Special problems arise when we try to visualize multiple dependent variables.A variety of special symbols

have been invented to convey simultaneous variation on several measures taken on the same object [41]. In

two-way displays, these include Chernoff’s faces, glyphs, stars, and color mapping. Figure 9.3 shows a star

display applied to quality of life measures for various states [41 p. 228] Such displays may be readily used

with engineering data, and may even be incorporated into dynamic displays (creating “twinkling stars”!). The

spokes could be color coded to identify the variables. Three dimensional extensions of these schemes are

possible but the resulting objects are usually still mapped onto a 2-D screen.

A final problem concerns the display of variation over time. Multiple static representations can be used, but

accurate representation of dynamic behavior requires animation. Two solutions are discussed below in the

section on CFD: traveling icons (such as Kerlick’s boids) and responsive icons (such as Haber’s tensor

Title

-----------

glyph). The former go with the flow, while the latter respond to the wave as it passes.

Previous Table of Contents Next

reserved. Reproduction whole or in part in any form or medium without express written permission of

EarthWeb is prohibited. Read EarthWeb's privacy statement.

Search Tips

Advanced Search

Computer Visualization: Graphics Techniques for Engineering and Scientific Analysis

by Richard S. Gallagher. Solomon Press

CRC Press, CRC Press LLC

ISBN: 0849390508 Pub Date: 12/22/94

Search this book:

Previous Table of Contents Next

9.4.2. Finite Element Modeling and Analysis and Computational Fluid Dynamics

Both of these approaches are concerned with the spatial distribution of values of one or more variables over,

within, or around an object. They simulate the behavior of the object or system under specified conditions or

environments, or trace the effects of the object on the surrounding environment. It can be argued that

computer graphics is what made finite element analysis usable and comprehensible for most engineers. Until

there were post-processors which allowed the results of the analyses to be presented graphically, few

engineers were capable of interpreting the results of finite element analysis.

Finite Element Analysis (FEA) is an area in which visualization techniques developed in other domains have

been readily adopted. Since analysts are interested in the behavior within the object, volume visualization

methods are appropriate. Techniques from medical imaging, such as volume slicing, isosurfaces, and

translucent displays, are now common in FEA research, and are even available in commercial FEA software

(ANSYS 5.0). The application of these new techniques to FEA is a direct result of the kind of

crossfertilization the NSF visualization report was designed to encourage.

Some applications of 3-D visualization techniques to analysis and modeling were given in Gallagher and

Selker [18]. Figures 9.4 and 9.5 show traditional contour and isosurface displays of a rocket seal. Clearly the

isosurface representation in Figure 9.5 gives a clearer picture of the 3-D variation in the model’s interior. In

Figure 9.6, the interior thermal distribution of a hair dryer model is revealed using shading and translucency.

Volume slicing is used to show the state of stress in the interior of a model in Figure 9.7. These three

examples are courtesy of Swanson Analysis Systems, Inc.

Visualization sometimes helps detect errors in the model, code, or procedures. Not all results are what we

expect. An anomalous result is a puzzle. Are surprises “real” or artifacts? It is often difficult to tell. Da Rosa

conducted an aerodynamic study of a test vehicle using the FLOTRAN CFD software [11]. The vehicle was

an automobile bluff-body specified by Cray Research as a test problem for evaluating CFD software. Two

distinct models were analyzed: a 17,000 node coarse model, and a 114,000 node refined model. The final

solution from the coarse model provided the basis for creating the refined model. The details of the analysis

and results need not concern us here. In this project the visualization results revealed a problem with the

model vectors in the Ramp region as shown in Figure 9.8 (a) and (b). In the coarse model, the flow behaves as

expected, but in the refined model, the flow pattern violates our intuitive expectations. Unfortunately, it also

fails to match the results of windtunnel experiments. These anomalous results were readily apparent from the

Title

-----------

visualization. Of course, anomalous behavior in any part of the display causes us to question the overall

results. With purely numeric displays, this problem would have been more difficult to detect.

Figure 9.3 A Star display for data on seven quality measures for various states. (From H. Wainer and D.

Thissen, “Graphical Data Analysis.” (Reproduced with permission from the Annual Review of Psychology,

Figure 9.4 Contour plot display of a rocket seal. Ansys 5.0. (Courtesy Dedo Sistemi-Italcae SRL, Italy.)

Figure 9.5 Isosurface display of a rocket seal. Ansys 5.0. (Courtesy Dedo Sistemi-Italcae SRL, Italy.)

Figure 9.6 Interior thermal distribution for a hair dryer model. Ansys 5.0, Source: Swanson Analysis, Inc.

(See color section plate 9.6)

Figure 9.7 Volume slice display of stress states for the interior of a model. ansys 5.0, Source: Swanson

Analysis, Inc.

There is a critical distinction between expertise as an analyst versus competence as an engineer. The analyst

can tell if an analysis was done correctly, the engineer must know if the results are correct. Of course, with

any model, its accuracy must eventually be verified by empirical results.

Computational fluid dynamics (CFD), by definition, involves patterns of change over time and space. To

understand even simple phenomena often requires several types of representation. Truly informative displays

must be dynamic. Through animation, we can watch icons move and change over time. Techniques for the

display of flows include vectors, streamlines, streaklines, and particle paths [4]. Kerlick also discussed flow

ribbons and boids (including sphere, arrow and dart boids) [23]. Boids are bird-like icons in that they travel

with the flow. They change shape in flight in response to field variables. These icons are designed to represent

multiple properties of the flow simultaneously and indicate directionality. The star technique could also be

used in this context.

One of the most compelling demonstrations of the utility of many of these techniques is the simulation of the

evolution of a severe thunderstorm developed at the National Center for Supercomputing Applications at the

University of Illinois. This simulation is described in Kaufman and Smarr and is available on videotape from

ACM Siggraph [21].

9.4.3. Mechanical Dynamics

Several programs allow us to animate assemblies, mechanisms, machines, vehicles, and structures (ADAMS,

IMP, DADS), so we may study their behavior over time and visualize the interactions of their parts. These

dynamic visualizations often reveal operational characteristics of the assembly that do not fit our expectations

or intentions. They allow us to analyze multibody contact; mating/fit; interference; range of motion. Using

these programs, we can follow changes of state or spatial location over time. Thus we can model the activity

in a production cell complete with robots, machine tools, and material transport systems, evaluate its

performance, and redesign it as necessary.

Previous Table of Contents Next

reserved. Reproduction whole or in part in any form or medium without express written permission of

EarthWeb is prohibited. Read EarthWeb's privacy statement.

Search Tips

Advanced Search

Computer Visualization: Graphics Techniques for Engineering and Scientific Analysis

by Richard S. Gallagher. Solomon Press

CRC Press, CRC Press LLC

ISBN: 0849390508 Pub Date: 12/22/94

Search this book:

Previous Table of Contents Next

9.4.4. Ergonomic Modeling

In one sense, modeling humans is a special case of “Mechanical Dynamics.” Anthropometric models are used

to test the fit, reach, and motions of people in vehicles and environments. They incorporate restrictions on

limb and joint movements, and may be scaled for different body types and sizes. They also permit assessment

of movement patterns of individuals and groups of people. Some programs include the ability to determine the

visual field for the model. Ergonomic models have been developed for a variety of purposes and vary greatly

in the type and amount of information they contain. Such models have been developed for automobile

companies, aircraft manufacturers, and the space program. They have been used for cockpit design, aircraft

crew cabin layout and instrumentation evaluation, aircraft maintainability studies, automobile seat and

passenger comfort studies, and space station construction planning.

Figure 9.8 Flow fields around an automobile like bluff body Using Flotran (a) Coarse model vectors in ramp

regions, (b) refined model vectors in ramp region. (From Da Rosa, 1989)

One early program SAMMIE (System for Aiding Man-Machine Interaction Evaluation) was developed at the

University of Nottingham and used by British Leyland to assess the fit of drivers and passengers in their

vehicles. It included variation in gender and somatotype: one could specify whether the model was male or

female; heavy, average, or slight in build. SAMMIE was used primarily for vehicle design and the interior

design and layout of rooms and office spaces. With such programs, we can assess how people fit into, and

function in, their environments before the design is finalized and the space actually built. Other early

anthropometric modeling programs were reviewed by Dooley [12].

At the University of Pennsylvania, Badler and his colleagues have developed human modeling at a much

Title

-----------

higher level of sophistication [2]. Their program JACK incorporates the major features of the earlier programs

(reach and visibility analysis), but has a more complete representation of the human body. In addition to 88

articulated joints in the body model, JACK has a 17 segment flexible torso, a hand model with automatic grip,

and strength guided motion. It also has a variety of features useful for animation and film (facial models,

walking, etc.). Models of the human body have been used in a range of studies from simulated crash test

dummies to the electronic cadaver for training surgeons or previewing an operation.

Usually visualization is helpful in the design process but sometimes too much detail or too literal a

representation may interfere with problem solving. Finding the right visualization may involve discarding

some of the information or freeing oneself from a model which is too compelling. Paul Chia and the author

analyzed the ergonomic design of a work area in a furniture factory [8]. A hot press machine was

systematically loaded by two workers and unloaded by a second pair. By mapping the motions and

movements of the workers we hoped to duplicate their tasks with one or more robots. We found we could, in

fact, replace these workers with machines which could repeat the same work patterns, but it would be

expensive, and there were serious questions about the reliable operation of robots in this environment. Forced

by reality to reconsider our solution, we found that a total redesign of the process and work area would

eliminate the need for either humans or robots. Focusing on the concrete details of the task as currently done

prevented us (initially) from finding a simpler more abstract solution. Visualization still guided our search for

a solution, but it was more abstract and schematic.

9.4.5 Factory Simulation

Discrete event simulation and continuous process simulation represent an alternative type of modeling and

analysis from those described so far. Here the goal is to predict the behavior of a system under various

conditions and constraints. These models involve probabilistic mechanisms. For example, in the discrete

event simulation of manufacturing systems, programs such as SIMAN/ Cinema allow us to model and

visualize the effects of different factory layouts,machine cells, material movement patterns, and inventory

buildup. With animation, we can control the pace and resolution of the model, and can explore what happens

as we change features of the system. An investigator can actually watch the behavior of the system evolve

over time.

CAD programs can be used to develop design alternatives (such as factory layouts or machine cells).

Simulation can then be used to evaluate the proposed designs to see if they achieve their intended benefits.

Keathley simulated production at a sheet metal production facility, identified operational problems, created

part families using group technology, and then proposed, modeled, and evaluated redesigns of the layout [22].

Simulation can provide a means of assessing the manufacturability of proposed designs and thus will be an

essential component of future DFM (design for manufacturability) programs. The designer could send his or

her concept to a “design critic” which would assess ease of production and flag problems. The design critic

would be based on a model of the production facility as it exists at the particular company.

Simulation can be a design tool—we can explore the behavior of systems that do not yet exist. Animation and

graphics are essential to understanding and communicating the results of a simulation. Visual representations

and especially animation can convey what is happening in a complex system, so that even those unskilled in

the methods of simulation can understand the results. Such applications of visualization are fairly

straightforward, but developing meaningful icons is important. Visualization is essential to convey the

probabilistic nature of simulation. It reveals the significance of material handling, queues, inventory

accumulation, and bottlenecks. Simulation can be extremely valuable for uncovering problems and assessing

their probable causes, and then for assessing the adequacy of our proposed solutions.

Previous Table of Contents Next

reserved. Reproduction whole or in part in any form or medium without express written permission of

EarthWeb is prohibited. Read EarthWeb's privacy statement.

Search Tips

Advanced Search

Computer Visualization: Graphics Techniques for Engineering and Scientific Analysis

by Richard S. Gallagher. Solomon Press

CRC Press, CRC Press LLC

ISBN: 0849390508 Pub Date: 12/22/94

Search this book:

Previous Table of Contents Next

9.5 Educating Engineers in Design and Analysis

9.5.1. Using Visualization to Improve Design Education

At the University of Virginia we have developed a unique approach to CAE education which emphasizes 3-D

thinking, modeling, and visualization. Our approach emphasizes the central role of geometric modeling. Our

students learn to do solids modeling from the start of their CAD experience. They use the CAD system to

create 3-D representations of objects and assemblies. In our approach, the students learn to describe and

represent objects in the simplest and most informative ways. They learn how to model, not just how to make

pictures. To us, CAD means “Computer Aided Design,” not drafting.

We start with simple 3-D models: extrusions and volumes of revolution. Students learn how to sweep profiles

through space to define objects by linear translation. The basic ideas of modeling are introduced: the

distinction between profile lines and information lines; the functions used to define solids, holes and depth

points; and the difference between models and views (pictures of the model). Then, they learn about volumes

of revolution: how to construct the profile, place center lines, and inform the system about the functions

necessary to generate the model. With these two basic modeling techniques, students are able to generate a

surprising range of simple models. They can also model several objects on the same display and relate these

models to each other.

When they have mastered both types of models, they are introduced to Boolean Operations, and learn how to

make complex parts from simple operations. This greatly expands the domain of possible models they can

create. They can also start to appreciate the concept of feature-based design and the relation of design features

to production operations. Other modeling techniques (slides, ruled surfaces, pipe and wire line models, and

surface networks) are also covered.

After mastering all these various modeling techniques, our students learn how to control views of the model.

This reinforces the idea that views (pictures) are distinct from the model. The model can be viewed from any

direction or perspective we choose, and we may generate as many views of the model as we wish.

The final topic in modeling is assemblies—creating submodels (parts) which can be stored independently and

then be brought together into a final product. By this time, the students are used to thinking about individual

models. They must now solve the problem of linking separate coordinate systems—the parts (components)

with the assembly sheet. After mastering these concepts with simple problems, the students are given a major

Title

-----------

project. They must select an assembly of 8 to 10 distinct parts and model it. The assembly should be

inherently interesting to the student, and may involve a novel concept or design.

After creating a model for the geometry of a solid object, mechanism, assembly, or structure, most students

proceed to analyze their model using either finite element methods or mechanical dynamics. There is a natural

transition from creating a solid model to specifying the properties of the object represented by that geometry.

The students lead us to advanced topics. When they want to assign material properties to their model and

study its behavior, we introduce them to FEA. A design for a mechanism or vehicle leads to animation and

mechanical dynamics programs. If they want to fabricate their object, they learn about manufacturing

processes, operations, and strategies, and simulation software. Self-directed learning is the most efficient

kind: when students are ready and motivated they quickly learn the material and retain it well. The

visualization capabilities of all these software products facilitate ease of use and ease of learning, and are

highly motivational.

9.5.2 Education for Visualization

A major potential benefit of visualization is its more general use in education. Developing expertise requires

having experiences: doing things, seeing things, developing correct intuitions. With visualization, we can

provide controlled experience, replicate situations, highlight selected aspects of a problem, control event

timing, control grain (detail or resolution), control focus of attention, and isolate features of special interest.

All of the applications discussed in this book are potential teaching aids as well as analysis tools.

The use of visualization in design and analysis, and the educational uses of visualization to teach a variety of

subjects, are now well established. Even in our first year Engineering Concepts course, we teach our students

to model and to visualize. They learn to sketch and draw. This helps them learn to see: they come to

understand those features and aspects of the world which artists capture in pictures. They also study the

psychology of visual perception. When we introduce CAD, they learn to model (create 3-D representations of

objects) using the Silverscreen software package. The emphasis on 3-D modeling and visualization captures

their interest and increases their understanding of both engineering design and the artifacts they are designing.

Students can develop a sensitivity to visualization through systematic exposure to it. Greater emphasis on

visual representation in our courses can only enhance our student’s ability to visualize and extract information

from visual displays. Students must also learn to be critics of visual displays. Visualization can be effectively

used to misrepresent information, as well as to accurately convey it. “How to lie with visualization” is now a

popular topic at professional meetings. Graphics packages and visualization software allow one to generate

really terrible displays as easily as producing good ones. Thus we must educate our students about how to

display data, and how to avoid common mistakes and artifacts.

How do we teach visualization itself? Scientific visualization does not fit neatly into existing academic

curricula or departments. The necessary skills and knowledge are found in various domains: art, psychology,

statistics, and the subject matters of particular scientific and engineering disciplines. Currently, each domain

teaches those concepts and procedures of visualization relevant to its own problems and concerns. But this

approach fails to provide students an appreciation of the full range of techniques available. The collaborative

research teams found at NCSA [10] may be the model for future teaching teams who will develop and teach

the courses and curricula on visualization methods. Such crossfertilization has already yielded results for CAE

research and commercial software development.

Previous Table of Contents Next

reserved. Reproduction whole or in part in any form or medium without express written permission of

EarthWeb is prohibited. Read EarthWeb's privacy statement.

Search Tips

Advanced Search

Computer Visualization: Graphics Techniques for Engineering and Scientific Analysis

by Richard S. Gallagher. Solomon Press

CRC Press, CRC Press LLC

ISBN: 0849390508 Pub Date: 12/22/94

Search this book:

Previous Table of Contents Next

9.6 References

1. Bachi, R., Graphical Statistical Methodology in the Automation Era. Graphic Presentation of Stat.

Inf.: Presented at 136th Ann. Meet. Am. Stat. Assoc., Soc. Stat. Sect. Sess. Graphical Meth. Stat. Data,

Boston, 1976, Tech. Rep. 43, Washington, DC: Census Bureau, 1978.

2. Badler, Norman I., “Human Modeling in Visualization” Chapter 15 in Thalman, N.D. and Thalman,

D. (eds), cited below, pp. 210-228.

3. Banchoff, Thomas F., “Visualizing Two-Dimensional Phenomena in Four-Dimensional Space: A

Computer Graphics Approach” in Wegman, E.J. and DePriest, D.J. (eds.) Statistical Image Processing

and Graphics, New York: Marcel Dekker, Inc., 1986.

4. Bryson, S. and Levit, C., “The Virtual Windtunnel; an Environment for the Exploration of

Three-Dimensional Unsteady Flows” reprinted in Introduction to Scientific Visualization Tools and

Techniques, Course 02, Siggraph ’93, Anaheim, CA, 1993.

5. Chasen, S.H., Geometric Principles and Procedures for Computer Graphics Applications,

Englewood Cliffs, N.J., Prentice-Hall, 1978.

6. Chasen, S.H., “Historical Highlights of Interactive Computer Graphics,” Mechanical Engineering,

Nov. 1981, pp. 32-41.

7. Chambers, J. M., Cleveland, W. S., Kleiner, B., and Tukey, P. A., Graphical Methods for Data

Analysis, Belmont, CA, Wadsworth International Group, 1983.

8. Chia, Moon-Hsiang, Robot Selection and Design Based on Ergonomic Modeling, M.S. Thesis,

University of Virginia, Department of Mechanical and Aerospace Engineering, January, 1991.

9. Cleveland, W., The Elements of Graphing Data, Pacific Grove, CA, Wadsworth & Brooks/Cole,

1985.

10. Cox, Donna J., “Scientific Visualization: Collaborating to Predict the Future” EDUCOM Review,

Winter, 1990, 38-42.

11. Da Rosa, W.A., Simulation of Aerodynamic Flow Over Automobile-Like Bluff Bodies Using

Flotran Software, Senior Thesis, Aerospace Engineering, University of Virginia, April, 1991.

12. Dooley, M., “Anthropometric Modeling Programs—A Survey,” IEEE Computer Graphics and

Applications, 2, November, 1982, pp. 17-25.

Title

-----------