- an intelligent planning tool for evaluation of amphibious river crossing sites a218496

7/27/2019 - An Intelligent Planning Tool for Evaluation of Amphibious River Crossing Sites a218496

1/149

DTIC FL C3PY 2I00 A-N AN INTELLIGENT PLANNING TOOL

FOR EVALUATION OF AMPHIBIOUSRIVER CROSSING SITES

byMark W. Yenter

B.S., University of Nevada, 1981

DTICS DLECTE FEB 2 1 1990 D

A thesis submitted to theFaculty of the Graduate School of the

University of Colorado in partial fulfillmentof the requirements for the degree of

Master of ScienceDepartment of Civil, Environmental,

and Architectural Engineering1990

' MTRXTn0N SThTTk" A90 02 ,


2/149

REPORT DOCUMENTATION PAGE Fof 0704-018PuUOCwortmq aura". for t. aw of I of III:=ismulmd to evvuqe I ntow ow rowmea a i.' ism r rwv&"t innufm. semawi sq om sev qaftn" amlMnmw IllsOne nbsdd, land WfWA he ~ctni kOYdO.Sii iwnsqeqVI6di.nlnte-I alntam 1111 01mm~idu,,m. owsaq suqoummfor FINUaflMeu wi. o WinUWqM Hesndaualei Swtmw. okomm Ia.tdaemI (owuam amd mmU. 1215 .JlwUmi 0aft HlImy. S&d 1204. AMys.. VA aid to1144Oftli at lntwndnin AMd OVAUM Allan. Offle atManjiwuuU atto Budq@l. WWOM00. DC~ _______________1.AGENCY US E ONLY (Lam Bhaw) 2.REPORT DATE 3.REPORT TYPE AND DATES COVERED

9 Feb 90Fia4. TITLE AND SUBTITLE S. FUNDING NUMBERS

An Intelligent Planning Tool for Evaluation of AmphibiousRiver Crossing Sites

6.AUTHOR(S)CPT Mark William Yenter

7. PERFORMING ORGANIZATION NAME(S) ANiD ADDRESS(ES) 8. PERFORMING ORGANIZATIONHQDA, MILPERCEN (DAPC-OPB-D) REPORT NUMBER

9. SPONSORINGIMON'ITORtING AGENCY NAME(S) AND ADORESS(ES) 10. SPONSORINGMON ITORING AGENCYREPORT NUMBER

11. SUP PLEMENTARY NOTES

12a.DISTRIB IfAVAILABIUITY STATEMENT 12b. DISTRIBUTION CODEUnlimited

13. AB STRPACT (MaWenui 200 wco'&).'The use of two dimensional hydrodynamic finite elempnrt ni~els to evaluate

potential amphibious river crossing sites~jj

14. SUBJECT T~MS 4 15~~..-.i.UMBER OF PAGEStwo dimensional hydroYdlnarnic modeling, river crossing operations, ion PRICE CODEfinite element nbtworks, Bradley Fighting Vehicle, ' lf!. _________

17. SECURITY CLASSIFICATION I& . SECURITY CLASSIFICATION 19 . SECURITY CLASSIFICATION 20 . LIMITATION OF ABSTRACTOF REPORT OF THIS PAGE OF ABSTRACT A* UNCLASSIFIED UNCLASSIFIED UNCLASSIFIED SAR__aN7540-01-280-550 pliiiici y AN.SI id. MIS


3/149

This thesis for the Master of Science degree byMark W. Yenter

has been approved for theDepartment of

Civil, Environmental, and Architectural Engineeringby

Steven t." Chapra

Date


4/149

Yenter, Mark William (M.S. Civil Engineering)An Intelligent Planning Tool for Evaluation of

Amphibious River Crossing SitesThesis directed by Research Associate Professor Kenneth

M. Strzepek

In the past, two-dimensional hydrodynamicfinite element models have been used to successfullymodel water current velocity and surface elevation.Because they are mathematical models, they have are astatic, that is, initial boundary conditions are setand the model is run. The question is, can the resultsof static mathematical hydrodynamic modeling beextended to dynamic simulation?

In this thesis a decision support system isdeveloped that applies the static results of two-dimensional hydrodynamic modeling in a graphicsoriented, dynamic computer simulation of an amphibiousvehicle crossing a river. As a decision supportsystem, the program permits the user to build thesimulation scenario by selecting the crossing site,vehicle type that will attempt the crossing, watercurrent, and starting and finish poi:its. The decisionsupport system, called the River Crossing SiteSimulation and Evaluation Tool (RC-SET), assists the


5/149

ivuser in evaluating the feasibility of crossing a riverunder specific hydrologic conditions with a knownvehicle type.

IAccesiar For

DTIC EdJ' .... o,.By.......C . ....,..., (" "' "-ji


6/149

This thesis is dedicated to

Lisa,my best friend.


7/149

ACKNOWLEDGEMENTS

Many people have assisted me in my efforts. Mythanks to Dr. Kenneth M. Strzepek who helped me todevelop and focus this thesis, and whose imaginationand sense of humor have been an inspiration. My thanksalso go to the other members of my graduate committee,Dr. Steve Chapra, Dr. Earnest Flack, and Dr. TissaIllangasekare, all are academic professionals of thehighest caliber.

Many thanks to my friends and colleagues at theCenter for Advanced Decision Support for Water andEnvironmental Systems. Tom Over and Edith Zagona wherealways there when I needed advice about water resourcesissues or a shot in the arm of inspiration. TerriBetancourt introduced me to Geographical InformationSystems and was a dependable and never ending source ofcommon sense and good ideas. My deepest thanks toDavid Sieh, Brian Makare, and Steve Wehrend whoseinteresting combination of saintly patience and man-handling helped to transform me from a broken FORTRANIV programmer into a respectable FORTRAN 77 hacker. Myapologies to Jim Waterman, the computer systems guy,


8/149

viiwhose disk space I mercilessly poached during the lastyear.

A few people outside of the University ofColorado provided me with information and software,without which this thesis would not be possible.Thanks to Dr. D. Michael Gee from the Corps ofEngineers Hydrologic Engineering Center for providingthe source code and documentation of the twodimensional model, and the final report of theSacramento River study. My thanks to Captain BoUnderwood from the U.S. Corps of Engineers WaterwaysExperiment Station for taking the time to locate andsend to me the field testing data on a number oftracked vehicles.

And finally, thanks to my wife Lisa, and ourdarling daughter Lindsay Marie, for putting up with myabsence, both mental and physical, during the lasteighteen months.


9/149

CONTENTSCHAPTER

I. INTRODUCTION ................. 1The River Crossing Problem .... ......... 1

Current Doctrine .... .............. 1River Crossing Operations .... ........ 2River Crossing Engineer

Information Requirements .... ........ 6Focus of This Thesis ... ........... . 12

II. LITERATURE REVIEW .... ............. . 14Decision Support Systems .. ......... .. 14Simulation ...... ................ . 17

Modeling ...... ................ . 17Computer Simulation

As Experimentation ... .......... .. 19The Simulation Process .. ......... .. 21

Two Dimensional FiniteElement Models .... ............. . 28Finite Element ModelingSystem Operation .... ............ . 30

Data Collection ... ........... . 31Network Design .... ............ . 32Calibration .... ............. . 35Validation ..... .............. . 37Application .... ............. . 37


10/149

ixIII. RMA-2D . . . . . . . . . . . . . . . . . . . 38

Introduction ...... ............... .38Model Applicability ..... ............ .39

General Approximations............ 39Two Dimensional Elements .. ........ .. 41Equation Solution ..... ............ .42

Program Organization .... ........... .44IV. RIVER CROSSING SITE SIMULATION ANDEVALUATION TOOL (RC-SET) ... ......... .52

General ....... .................. .. 52External Data Preparation Tools ........ .54

GRASS ....... .................. .. 56MESH ....... .................. 58RMA-2D Modeling ..... ............. .60

Data Base Management System ... ........ .. 63Model Management System .... .......... .. 67Model Set Up ...... ............... .69

Vehicle Movement .... ............ .70Model Limitations ..... ............ .71General Equations of Motion .. ....... .. 73Movement Simulation .... ........... .92Translational Motion ... .......... .. 92Rotational Motion ..... ............ .94

User-System Interface .... ........... .95


11/149

xV. THE CASE STUDY ...... ............... 105

Scope ....... .................. 105Reach Selection .... ............. 106Vehicle Selection .... ............ 111Calibration and Validation .. ........ .114

The Digital Elevation Model .. ...... 114RMA-2D ....... ................ 115Amphibious Vehicle Movement .. ...... 117

Simulation Experimentation .. ......... .120

Conclusions of the Case Study .. ....... 122Calibration and Validationof RC-SET ..... ............... 123Usefulness of RC-SET ... .......... 123

VI. RESULTS AND CONCLUSIONS ... .......... 124Results ...... ................. 124

Usefulness to the U.S. Army .. ...... 125Continuing Research ... .......... . 128

Amphibious Operations ... ......... 128Rafting Operations .... ........... 128Military Float Bridging .. ........ 129

Conclusion ....... ................. 129REFERENCES ........ .................... 132


12/149

xi

TABLES

Table1.1 River crossing information .... ........ 84.1 Data preparation tools output files .... 62


13/149

FIGURES

Figure3.1 RMA-2D program organization ... ........ ..453.2 RMA-1 finite element network .. ....... .. 473.3 RMA-2D VECTOR plot .... ............ .493.4 RMA-2D CONTUR plots ..... ............ .514.1 RC-SET program organization ... ........ .. 534.2 Data preparation tools ... .......... .. 554.3 RC-SET data base management system . . .. 644.4 RC-SET model management system ........ .. 684.5 Forces and moments onimmersed bodies ..... ............. .724.6 Rectilinear motion .... ............ .744.7 Rotational motion in a

horizontal plane .... ............ .764.8 Free body diagram of avehicle in water .... ............ .784.9 Rotational moments about

the mass center ..... ............. .854.10 Water resistance to rotation .. ....... .. 874.11 RC-SET title screen ..... ............ .964.12 RC-SET capabilities screen .. ........ .. 984.13 A successful river crossingusing RC-SET .... ............. . 1045.1 Sacramento river study area ........ .. 1085.2 DEM construction ...... ............ 110


14/149

xiii5.3 The Bradley Fighting Vehicle System . . . 1125.4 The Bradley Fighting Vehicle Systemwith swim curtain installed ....... . 1135.5 Study area finite element network . . . 1165.6 Acceleration test curve fitting ..... ... 1195.7 Three crossing methods .. ......... .. 1216.1 Crossing a river at a bend .. ...... 127


15/149

CHAPTER I

INTRODUCTION

The River Crossing Problem

Current doctrine. During the last decade the UnitedStates Army has recognized that the size of enemyformations and the lethalness of modern weaponsdictated a change in the way we would fight the nextwar. Our army is much smaller than our majoropponents, but is better trained and equipped. The newarmy doctrine seeks to overcome this disadvantage innumbers with aggressive operations that will allow usto maintain the initiative on the battlefield in orderto dictate how the battle is to be fought. This newdoctrine is known as the Airland Battle Doctrine.

The Airland Battle Doctrine identifies threeareas of combat where the battle will be waged. CloseOperations take place at the leading edge of thefriendly forces, where they engage and destroy theleading elements of the enemy. Deep Operations aredirected at the follow-on forces of the enemy and areconducted to destroy and disrupt reinforcing units toprevent them from supporting forward units. Rear


16/149

2prevent them from supporting forward units. RearOperations are those activities conducted in the reararea of friendly forces and are designed to ensurecontinuity of friendly operations (U.S.Army, 1989).

River crossing operations. The ability of the U.S.Army to cross rivers quickly and efficiently whileconducting Close Operations is essential to the successof the Airland Battle Doctrine (U.S.Army, 1984).Despite advances in high mobility weapon systems andextensive aviation assets, rivers remain majorobstacles. River crossings are among the mostcritical, complex, and vulnerable combined armsoperations (U.S.Army, 1984).

There are three types of river crossingoperations; hasty, deliberate, and retrograde. A hastyriver crossing is a decentralized operation that isconducted as a continuation of an attack by friendlyforces. It is used to cross rivers where the enemyforces are weak or disrupted, and in situations wherethe characteristics of the river are such that thestream is not a major obstacle. The methods used toconduct hasty river crossings include the use ofexisting civilian bridges and ferries, ford sites,amphibious vehicles, and military rafting and bridgingequipment.


17/149

3Deliberate river crossing operations are

conducted in situations where hasty crossing are notpractical or have failed. Deliberate river crossingoperations require detailed planning and the buildup ofassaulting forces, river crossing assets, and firesupport. It is a three-phased operation consisting ofthe assault phase, the rafting phase, and the bridgingphase. The selection of crossing, rafting, andbridging sites are critical to the success of theoperation (U.S.Army, 1984).

In the assault river crossing phase thecommander of the friendly forces attempts to crosssufficient forces to secure the far bank of the river.Assault river crossing sites must be located whereenemy forces are weak, there is concealment from enemyobservation, and friendly forces occupy the dominantterrain features. This insures security and supportingfire for the assaulting forces. Adequate routes to theriver, as well as from the river to follow-onobjectives are also required. The river currentvelocity must be small, less than five feet per second,and the river should be crossed at a narrow point tominimize the exposure of friendly forces to hostilefire.

When amphibious vehicles are used the ingressand egress bank conditions are very important. As a


18/149

4general rule bank slopes should have a slope of 30percent or less. Ideally the banks soil is firm andcapable of handling high traffic, allowing multiplepasses of ingressing and egressing vehicles.

If fording sites are available they must beshallow with a water depth not greater than 39 inchesfor dismounted soldiers and light armored vehicles, andnot greater than 42 inches for medium to heavy armoredvehicles. The ingress and egress banks must haveslopes of 30 percent or less if armored vehicles areused. The riverbed must be firm enough to withstandthe vehicle traffic.

During the rafting phase friendly forcesreinforce the assaulting forces by rafting armoredvehicles and anti-armor weapons. Rafting sites arelocated where they can provide the fastest access tothe far shore. Rafting sites require sufficient roadnetworks leading up to them on the near shore andadequate exit routes on the far shore to permit hastymovement. Both banks should be firm and have slopesnot greater than 20 percent. If possible the raftingsite should be located at a narrow part of the river,but must be free of sandbars or other obstacles thewould impede rafting operations. The draftrequirements of the four types of rafting equipmentcurrently in the Army inventory vary from 22 inches to


19/149

529 inches. Rafting operations can be conducted inwater current velocities as high as 10 feet per second,but the practical limit is less than or equal to 5 feetper second. Because there is a risk of rafts beingswept downstream, rafting sites are always locateddownstream of bridging sites.

During the bridging phase the bulk of theadvancing friendly forces cross the river. Organic(those normally found in the unit) and attachedbridging assets are used to construct temporaryfloating bridges that allow wheeled and remainingtracked vehicles to cross the river. Bridging sitesmust be located where adequate road networks alreadyexist or can be rapidly constructed. The bank slopes,either existing or constructed, cannot be greater than10 percent. The bank soil must be compacted andcapable of handling high traffic, either in it naturalstate or stabilized through standard constructionpractices or expedient methods such as temporarymatting. The velocity of the water cannot exceed 10feet per second, and the minimum draft requirements formilitary float bridging equipment range from 24inchesto 40inches. Floating bridges also require assemblysites along the water's edge where pontoons and othercomponents can be pre-assembled then maneuvered intoposition for anchoring. Obviously, the bridging site


20/149

6must be located at the narrowest point along the riverthat meets the above criteria.

Retrograde river crossing operations aredifferent from hasty and deliberate crossings becausethey are defensive in nature. They are conducted whenenemy forces threaten to overwhelm friendly forces.They are carefully planned operations that are designedto trade space for time. Retrograde river crossingoperations are planned to successfully extract friendlyforces from the enemy shore and force the enemy toconduct a deliberate crossing.

River crossing engineer information requirements. Onthe battlefield the area commander determines thespecific area of operations along which he wants toconduct the river crossing operation. His decision isbased upon tactical considerations such as enemystrength, key terrain, and cover and concealment offriendly forces. The burden of determining thevalidity and practicality of the crossing sitesselected falls upon the Corps of Engineers' officerssupporting the maneuver commander. As discussed above,several characteristics of potential river crossingsites should be known to evaluate their feasibilityfrom an engineering standpoint. Table 1.1 lists the


21/149

7type of information that is required to evaluate aselected river crossing site.


22/149

-4

-HO

>0

F4Uw

clil0

0-44-4-00

0>41

0 0-O X >1-4

uu

r- ZO4~ -i-I V ~--4 .4

co0 41 14 04 4-4 0 -44 CU 0 aW (Z r-4u z >


23/149

9Table 1.1 dipicts the two general categories of

information are required, hydrologic measurements ofthe water velocity and depth, and geotechnicalqualification of the bank and riverbed soil conditions.Geotechnical issues are important, but beyond the scopeof this thesis. Army engineers have a variety ofsources of information to evaluate the hydrologiccharacteristics of the river.

Historical river data is the most dependablesource of hydrologic information. However, thisinformation may be limited or not available. When itis not available, or to update and augment thisinformation, engineer units will typically sendengineer soldiers to conduct a river reconnaissance.They will probe forward and gather hydrologic andgeotechnical information. To maintain cover andconcealment from enemy fire, their reconnaissance islimited to the bare essentials. River width is usuallyapproximated using a compass to determine the azimuthto a point on the far shore, adding 45 degrees to thisazimuth and siting the same point again on the newazimuth from a second position. The distance betweenthe two positions on the near bank is the approximatewidth of the river. When practical, river depth ismeasured using expedient rods or ropes tied to weights.


24/149

10Water velocity is approximated by timing a floatingobject, such as a piece of wood, as it travels over ameasured distance (U.S.Army, 1985). The hydrologicdata gathered by these soldiers is sketchy at best, butis extremely important when no historical records areavailable.

Topographic engineer units may provide evenless specific information about the river. On largerrivers it is sometimes possible to make a roughestimates of the water velocity from aerial photographs(U.S.Army, 1978). This is done by taking consecutivephotographs of an object floating in the river, such asa log, and noting the lapse time between them. Theappearance of the river can be an important indicatorof river velocity. Meandering rivers generally haveaverage water velocities of 4.5 feet per second orless.

In some cases, particularly where the river wasoriginally in friendly territory before the conflict,historical records of river stage and correspondingflow rates may be available in the form of ratingcurves or tabular data (Chow, 1988). Water velocitydata would be less likely to be available but would beuseful to evaluate specific locations, and could beused to approximate velocities at other sites in the


25/149

llriver if the reach geometry changes very little, andthe length of reach is fairly long.

Even though specific velocity information maybe available, rarely will it be in the quantity ordetail needed to indicate the velocity distributionthroughout the reach. Complex flow patterns can existwhere the channel flow is split by sandbars andislands, flows over and through submerged obstaclessuch as destroyed bridges and rock formations, andflows around sharp turns in the river. Engineerofficers are required to make a best estimate of theriver velocity distribution at different locationsthroughout the river reach, often from very littledata.

Needed is a method for determining the twodimensional flow patterns of the river at the existingflow rate and reach topography. This may not bepossible for those rivers located in enemy territorybefore the outbreak of hostilities, however, flow rateand corresponding velocity distribution data could becompiled for rivers in friendly territory. Thesevalues could be approximated using any of a number ofexisting two dimensional hydrodynamic models.


26/149

12With this information, engineer officers could

evaluate potential crossing sites with greateraccuracy. The magnitude and direction of the velocityvectors could be compared to the maximum valuesdiscussed above to identify amphibious crossing,fording, rafting, and bridging points. It would beeven more beneficial to apply the two dimensionalvelocity data in a computer simulation of the force ofthe river acting against vehicles and equipment as theyperform their river crossing tasks in order to test thefeasibility of a selected sites. A simulation of thissort would certainly reduce the risk of theseprecarious operations.

Focus of This ThesisThe purpose of this thesis is to determine the

feasibility of using a two dimensional finite elementhydrodynamic model as the basis of a decision supporttool that can be used to evaluate potential amphibiousassaulting sites.

The thesis will address hydrologic aspects ofthe amphibious river crossing problem. Thegeotechnical qualification of the bank slopes andtrafficability are important considerations, but theirstudy is beyond the scope of this thesis.


27/149

13This thesis is organized into six chapters.

The river crossing problem and current U.S. Armydoctrine has been addressed in Chapter I. The type ofengineering information required to evaluate potentialamphibious river crossing sites was also identified.Chapter II is a review of current literature aboutdecision support systems, simulation, two dimensionalfinite element models in general, and morespecifically, finite element modeling system operation.The hydrodynamic model RMA-2D is presented in ChapterIII. In Chapter IV the organization and methodology ofthe prototype decision support system used to simulateand evaluate potential amphibious river crossing sitesis discussed in detail. The prototype system is calledthe River Crossing Site Simulation and Evaluation Tool(RC-SET). The results of applying RC-SET in a casestudy are presented in Chapter V. In this case studyRC-SET is used to simulate an amphibious vehiclecrossing a river reach. Chapter VI summarizes thedevelopment and application of the prototype decisionsupport system. Continuing research and development ofRC-SET are also addressed.


28/149

CHAPTER II

LITERATURE REVIEW

Decision Suport SystemsBefore evaluating the feasibility of developing

a computer simulation of a potential river crossingsite we need to define terms and discuss the conceptsof decision support systems, modeling, and computersimulation. Today, computer programs are labeled asdecision support systems without much thought as towhether or not they actually are such systems. Bydefinition, a decision support system (DSS) is aninteractive computer-based system designed to helpdecision makers use data and models to solveunstructured problems (Hopple, 1988).

Although DSSs vary greatly in fields rangingfrom corporate planning to microbiology, they sharecommon characteristics. DSSs assist the user in makingdecisions, usually on semistructured problems. Theyare a tool used by humans to focus the decision makingprocess, and do not replace human judgment. Theyprocess, organize, and interpret voluminous quantitiesof information that would otherwise overwhelm the user,


29/149

15and by doing so enable the user to make decisions moreeffectively, and more efficiently. DSSs are designedto be used by people with little or no computerexperience. They should make the user feel comfortablewith the system and not intimidated by it (Hopple,1988). They also provide continuous interactiveproblem solving rather than off-line batch processing.DSSs should not impose or force a decision makingprocess on the user. Rather, the DSSs should providethe user with a selection of blocks to build thedecision making process. Additionally, DSSs should bedesigned in a modular system so that they can bereadily updated and improved as the designer's problemsolving methods change. This allows the designer tochange one portion of the DSS without redesigning theentire system.

DSSs consist of three subsystems: a user-systeminterface; a database management subsystem; and a modelbase management system (Hopple, 1988). The user-systeminterface allows the user to communicate with thecomponents of the DSS. The interface can be one ormore of a variety of input and output schemes generallycategorized as the action language, the display orpresentation language, and the knowledge base. Theuser uses the action language to communicate with theDSSs via keyboard, mouse, touch activated screen,


30/149

16joystick or even voice command. DSSs use display orpresentation language to communicate with the user viagraphics on the screen, audio output, line printers,and plotters. The knowledge base is that informationrequired by the user to use the DSS properly. If notalready known by the user, this information may beavailable through user's manuals, or on-line help keys.

The database management subsystem is a computerprogram designed to facilitate the management ofintegrated collection of data (Hopple, 1988). Databasesubsystems are particularly important when the DSS willbe used to evaluate unstructured or loosely structuredproblems where extensive and frequent datarestructuring will be required. Database managementprograms must also have the capability to load andmanage external files.

The third subsystem of a DSS is the model basemanagement system. A model is a simplifiedrepresentation of reality or some aspect of it. Amodel based management system is a computer programthat offers a wide range of models and allows forflexible access, update, and change to the model base(Hopple, 1988). The model base management systemprovides the user with the building blocks that areused to construct variations of the model system.


31/149

17Simulation

A decision support system is built around amodel that simulates the system to be evaluated. Incommon language, to simulate something means to assumethe appearance of that thing. However, a morecategorical meaning is required in engineering andscience. In these terms "a simulation is the formingof an abstract model from a real situation in order tounderstand the impact of modifications and the effectof introducing various strategies," (Negoita, 1987).The main objective in simulation is to aid the user indetermining what will happen in a given physicalsituation under certain simplifying assumptions. Themajor advantage of simulations is that they provide theuser with a means of experimenting without destroying,damaging or, in any way, modifying the real system.

Modeling. To simulate a series of actions, a modelmust be produced. Models have been defined in manyways by many scholars, but an adequate definition isthat a model is a formal, symbolic representation of asystem (Lewis and Smith, 1979). Models can be used toapproximate reality because both the model and the realsystem follow the same physical laws, however, modelsare simplifications of reality. The more closely themodel approximates reality, the more complex the model


32/149

18becomes. "Models should never be confused withreality. They just map some facets of reality(hopefully the ones we had in mind) into an abstractdescription," (Cellier, 1982).

Before the advent of computers, physical modelswere probably the most commonly used models. Examplesof physical models are wind tunnels used in aircraftand automobile design, and water tanks used inhydraulic studies and ship design. The benefits ofphysical models are that they are relatively easy toconstruct and can model complex physical phenomena,such as the formation of ice on submerged structures,that would otherwise be very difficult to model.Physical models do have disadvantages. Onedisadvantage is that they are highly specific. Aphysical model of one port facility will not be validfor another (Pidd, 1988). Another disadvantage is thatexperimentation with physical models usually requiresalteration of the model, and therefore, expensive andfrequent construction of variations of the model.

Scale can be a disadvantage in physicalmodeling. While it may be relatively easy to constructa scale replication of the real system, such as ahighway bride, it may not be possible to accuratelymodel very small systems. For example, consider ascale model of flow through a porous medium. In this


33/149

19case, the scale model is an enlargement of the realsystem. While it may be possible to model the physicalcavity space between typical grains of sand, but itwill not be possible to model the capillary forcesexerted by the water relative to the size of thecavity.

Since the mid-sixties computers have been usedsuccessfully to model mathematically engineering andapplied science systems. Mathematical models areeffective in classical engineering and physics wherephysical phenomena is understood and can be defined byequations and empirical coefficients. The initialsuccess of these models led many scholars to believethat they could be applied to other sciences likebiology, economy, sociology, and psychology (Cellier,1982). Such was not the case. In the mid-seventiesengineers studying water resources and environmentalsystems realized that they were difficult to describein precise mathematical terms. To compensate for this"softness" or "grayness," fuzzy logic methods weredeveloped. However, these fuzzy logic concepts proveto be inadequate for ill-defined systems (Cellier,1982).

Computer simulation as experimentation. Computersimulation is experimentation using a computer-based


34/149

20model of some real system. The model is used as thevehicle for experimentation, and is often a trial anderror way to test the effects of various policies whenapplied to the real system (Pidd, 1988). Just asmathematical models cannot model all systems, computersimulations are not a panacea. Realistic simulationsrequire long computer programs that can be quitecomplex. The more the model is made to approximatereality, the greater the overhead in time, effort, andhardware resources. However, in most cases computersimulations have distinct advantages over reale;,perimentation and mathematical modeling (Pidd, 1988):

Cost. Computer simulations can be more costeffective than direct experimentation. Even thoughcomputer simulation can be time consuming and expensivein terms of skilled manpower, real experiments can alsobe time consuming and expensive, particularly if thereal system must be modified.

Time. Computer simulations provide profitablereturn for the initial time investment of producingworking computer programs for simulation models. Thereturn on this initial investment is that computersimulations can simulate weeks, months or even years ina few seconds of computer time.


35/149

21Replication. Unlike physical experiments,

computer simulations are precisely repeatable.Physical experiments rarely yield the exact resultswhen repeated due to errors and physical variations inthe system. Computer simulations also allow theexperiment to be conducted repetitively withoutchanging the physical system.

Safety. Computer simulations allow us to testextreme conditions that would otherwise be hazardous.

Flexibility. Computer simulations are superiorto mathematical models because they can be designed tocope with transient and dynamic effects.

The simulation process. The simulation process can becategorized into five phases: (1) systems analysis, (2)program synthesis, (3) model verification, (4) modelvalidation, and (5) model analysis (Lewis and Smith,1979). In practical terms, these phases are notdistinct or separated from each other. The simulationprocess is iterative, starting with a general thesisand evolving into a final product as new requirementsand problem solving methods are introduced.

Systems analysis is the thorough and detailedexamination of the real system in order to decompose itinto understandable and manageable sized components.


36/149

22The purpose of the analysis is to establish theinteractions, dependencies, and rules governing thecomponents of the system so that a model of the realsystem can be developed.

Program synthesis is the most creative phaseand is subdivided into the modeling and the programmingstages. During modeling stage the developer mustdetermine what components are required to model thereal system. Components are simplified or even omittedif analysis suggests that their effects do not justifytheir inclusion (Lewis and Smith, 1979). The task isto keep the program simple, to reduce cost, yetdetailed enough to provide reasonable accuracy (Negoitaand Ralescu, 1987).

The developer should consider the nature of thesystem being evaluated, understand the purpose of thesimulation, and know what results are expected. Thedeveloper can use this criteria to determine the levelof accuracy and detail required (Pidd, 1988). There isno need to develop a highly accurate computersimulation if only crude estimates are needed.

During the modeling stage the developer mustdetermine how the program will handle the passage oftime, if the model is stochastic or deterministic, andif changes in the model are to be discrete orcontinuous (Pidd, 1988). As stated above, one of the


37/149

23advantages of simulation is that the speed at which theexperiment takes place can be faster than the realsystem. Time slicing and next event techniques can beused to handle time-flow within the simulation. Timeslicing is a simple technique that controls the flow oftime through the simulation in equal time intervals.The next-event technique is used when the system willinclude slack periods of inactivity. In this techniquethe model is examined or updated only when a state ofchange occurs.

The developer must determine if the simulationof the real system is stochastic or deterministic. "Astochastic system is one whose behavior cannot entirelypredicted, though some statement may be made about howlikely certain events are to occur," (Pidd, 1988).Probability distributions are used in stochasticsimulations. As the simulation proceeds, samples fromprobability distributions are used to determine thestochastic behavior of the model. Deterministicsystems are those whose behavior is completelypredictable. In deterministic simulations, each timewe start with the same initial conditions we get thesame results. Deterministic systems are usuallyexpressed as mathematical models.

During the modeling stage the developer mustalso determine how the variables of the model must


38/149

24change with respect to time to simulate the realsystem. In a discrete simulation the variables changeonly at known or predictable times. In models thatallow continuous change, the value of the variableschange continuously as the simulation proceeds. Thesecontinuous changes could be represented by differentialequations and, in theory, their value could bedetermined at any point in simulation. In reality,digital computers operate with discrete quantities.Changes in variable values cannot change continuously.Continuous change can be simulated by inspecting orchanging the value of the variables at a multitude offixed points in simulated time (Pidd, 1988).

During the programming stage the computerprogram is designed and the appropriate computerlanguage is selected. The program should be organizedwith a highly structured approach. Highly structuredprograms have well-defined subroutines, modules, orprocedures of manageable size (Pidd, 1988). Structuredprogramming is always good practice but is particularlyimportant in large and complex programs. Verificationof the program is much easier if each module can betested separately as the program is being built.Structured programming also expedites renovation of theprogram as the simulation evolves from a thesis to thefinal model.


39/149

25In the mid 1970s graphical interface with

computers began to replace batch mode operations.Today, users take for granted that data can be entereddirectly from the keyboard or using some help devicesuch as a mouse (Pidd, 1988). Well designed graphicsare the foundation of the user-system interface of thedecision support system. Many benefits come from usingdynamic graphical displays in simulations. Properlydesigned graphical displays give the user an idea ofthe logical behavior of the simulation program.Graphical displays can also aid the user whenexperimenting with the model. They can aid the user byidentifying when an experiment has gone aria and shouldbe terminated if there is no need to make a full run ofthe simulation. The conditions can be changed and thesimulation run again. Finally, graphical displays canbe designed to return messages as the program runs thattell the programer that the simulation logic is as itshould be. This is particularly important whenevaluating how the simulation reacts to state changesin the system.

The next phase of the simulation process is thesimulation model verification. Model verification isdefined as ensuring that the model behaves (runs) asintended (Cellier, 1982). In other words, each line ofprogram code should do what it is supposed to do.


40/149

26Structured programming will make the verification phasemuch easier. Modular, structured programming allowsthe programmer to test each subroutine as they arecoded. In this way each module tested and verifiedbefore it is combined with other modules that togethermake up the submodel.

After all modules have been tested and verifiedeach submodel and the overall model are tested(Cellier, 1982). At this level testing becomes moredifficult. Two approaches are commonly used to verifysubmodels and models. The first is static testingwhere the computer program is analyzed to determine ifit is correct by using correctness proofs, syntacticdecompositions, and examination of the structureproperties of the program (Cellier, 1982).

The other method is dynamic testing. Indynamic testing the computerized simulation is rununder different conditions and the values obtained areused to evaluate the correctness of the program.Dynamic testing techniques include traces,investigation of input-output relations, internalconsistency checks, and reprogramming of criticalcomponents to determine if the same results areobtained.

In the validation phase confidence in the modelis determined by comparing it to the real system to see


41/149

27how closely the model approximates reality. Thepurpose of program verification and validation is notto prove that the model runs as intended and is anadequate representation of reality under all sets ofconditions (Cellier, 1982). Models are verified andvalidated over the conditions that are applicable tothe realm of operation of the simulation. Designingmodels that exceed this requirement are uneconomical intime and resources. Many techniques can be used tovalidate the model. These include evaluation ofgraphical displays to see if the model appears to beacting correctly, face validity checks, whereknowledgeable people determine if the model isreasonable, internal validity, where several runs ofstochastic models are made to determine the stochasticvariability of the model, and subjecting a samplerecord from the model to statistical tests to determinecorrectness of the data.

The final phase of computer model simulation ismodel analysis. After the computer simulation modelhas been verified and validated it is ready forcomputer simulation experiments. Alternate inputvalues and conditions are applied to the model to studythe effects on the model output. Model analysis is theculmination the other phases of computer modelsimulation.


42/149

28Two Dimensional Finite Element Models

The finite element method was originallydeveloped to evaluate structures. Over the past twentyyears it has become an effective tool for evaluating awide variety of problems in the field of continuummechanics (Froehlich, 1988). Finite element methodshave been successfully used to analyze heat transferproblems, aircraft structural stability and behaviorunder vibration, friction modeling in structures, thedesign of warheads and ballistic penetrators (Tipnisand Patton, 1988), and a wide variety of fatigueanalysis studies in mechanical engineering. Only inrecent years have finite element methods been used tomodel surface-water flow problems. Severalmathematical finite element models have been designedto model two-dimensional surface water flow in ahorizontal plane.

The finite element method is a numericalprocedure for solving the differential equationsencountered in problems of physics and engineering(Froehlich, 1988). Application of this method producesa set of simultaneous nonlinear algebraic equationsthat must be solved by iteration. Typically there areseveral thousand equations with several thousandunknowns (Gee, 1985). Models vary in the numericalmethods they use to solve these equations.


43/149

29The common thread in finite element models is

that they all require rapid calculations and storageand retrieval of intermediate solution values.Computers are well suited to this task. In the pastonly large mainframe computers were available. Batchinput decks were the norm, and data pre- and post-processing was laborious. Smaller and more affordablecomputers have opened the door to the use of finiteelement methods by small companies and individuals whocould not afford large and expensive main framecomputers, or computing time. Many finite elementmodels designed for mainframe computers have be re-coded and are portable to personal computers (PCs).

Two dimensional surface-water models aresufficient for most practical hydraulic engineeringproblems, particularly where the horizontaldistribution of flow quantities are the main interest.They are best suited for modeling of shallow rivers,flood plains, estuaries, harbors, coastal areas, andalmost anywhere the depth-to-width ratio of the body ofwater is small. In an interesting study a twodimensional finite element model was used to simulatecurrent velocities created by the operation of anexisting dam and lock. The concern was that thestructure was causing current velocities that created ahazard for small boat operation, and effected the


44/149

30aquatic habitat downstream (Gee, 1985). They have alsobeen used effectively to model sediment transport andwater quality simulations. Probably the greateststrength of these models is that they can simulate flowaround and over irregular topography and geometry suchas islands and submerged obstacles. In some instances,such as flow over and through a submerged bridge, flowis modeled as a combination one and two- dimensionalflows.

Finite element based hydrodynamic modelssimulate both steady and time dependent two dimensionalsurface water flow. They solve vertically-integratedequations of motion and continuity, and use the finiteelement method of analysis to obtain depth-averaged ordepth-integrated velocities and flow depths (Froehlich,1988). All models include the effects of surfaceroughness of the geometry, as well as fluid stressescaused by turbulence. Most models have the ability toinclude the Coriolis force and surface wind-stress.

Finite Element Modeling System OperationThe steps taken to apply finite element two

dimensional hydrodynamic models are as follows: (1)data collection, (2) network design, (3) calibration,(4) validation, and (5) application (Froehlich, 1988).


45/149

31This is very similar to the steps in model developmentdiscussed earlier.

Data collection. The first step in a surface watermodeling is to define the problem system and gatherdata. Two types of data are collected. Topographicdata describes the geometry of the area under study andincludes an evaluation of surface roughness to be usedin estimating the bed friction coefficients.Hydrologic data consists of evaluation of stage andflow hydrographs, field spot measurements of stage,flow and velocity, rating curves, high-water marks andlimits of flooding, and wind measurements. Hydrologicdata are used to define the model boundary conditions,and are used later to calibrate and validate the model.

The type and amount of data required dependsupon the purpose of the model. It is difficult toestablish the minimum data requirements for aparticular application. Time, manpower requirements,funding, and the objective of the study will determinethe degree of detail in the finite element networkdesign. If the network provides a high level of detailthen the risk of not properly representing the systemwill be reduced. However, if only generalapproximations of the real system are needed, then itwould be uneconomical to collect voluminous data.


46/149

32Knowledge of the important physical processes thatgovern the response of a system under study is neededto evaluate the trade-offs between risk of notaccurately representing the system and the difficultyof obtaining a solution (Froehlich, 1988).

Network design. The next step is the design of thefinite element network. Network design is nothing morethan dividing the real system topography into severalfinite elements. The goal of the network design is tocreate a finite element network that adeg'iatelyapproximates the real system. Because every networkdesign problem is unique there are no set rules thatare applicable in every case. The following aregeneral guidelines that are applicable in most cases.

Design of the work requires deciding the

number, size, shape, and configuration of the elementsthat will approximate the real system. Smallerelements improve the accuracy of the solution but alsoincrease the number of computations. Elements need tobe small enough to provide reasonable accuracy, yetlarge enough to be computationally economical.

The first step is to draw the network on a mapof the study area. Because several iterations of thenetwork will probably be constructed it is advisable todraw the network on some type of overlay material, such


47/149

33as gridded mylar. The grid on the mylar will assist indetermining the coordinates of each node, while the mapis used to determine their elevation. The scale anddetail of the map will determine the accuracy of themodel.

The next step is to define the area to bemodeled. Model boundaries should be located where thewater-surface elevations and flows can be specifiedaccurately. If the exact locations of the water-elevations are unknown or vary, then the boundariesshould be placed far away from the areas of primaryinterest. This will minimize the influence of errorintroduced by the boundary placement.

After the boundaries have been identified thestudy area is subdivided into relatively large areas ofsimilar topographic and surface roughnesscharacteristics. Subdivision lines between theregions should follow abrupt changes in topography andsurface cover.

Most two dimensional hydrodynamic models willaccept 6-node triangular, and 8-node and 9-nodequadrangular elements. The two kinds of quadrilateralelements are similar except the 9-node element has aninternal node. This additional node requires morecomputational time but provides greater accuracy thanthe 8-node element. For most networks a combination of


48/149

34the three element types will provide the bestrepresentation of the study area.

A network of uniformly sized elements is easyto construct but may not necessarily be the best way tomodel the system. One of the strengths of the finiteelement method is that the size and shape of theelements can be varied. Small elements should be usedin regions where the topography and surface varyconsiderably. These regions will cause large gradientsin the dependent variables of the model. The key is tokeep the elements as homogeneous as possible. Largevariations of these characteristics in the same elementwill cause model solution convergence problems.

Large elements can be used in areas where thetopography and surface roughness do not vary a greatdeal, and in areas where approximate solutions areadequate. Transition from large elements to smallelement should take place gradually. Large elementsshould not be placed immediately next to smallelements.

An important characteristic of element shape isthe aspect ratio. The aspect ratio is the ratio of thelongest element side length to the smallest. Theoptimum aspect ratio depends upon the local gradientsof the solution variables. The longest side of theelement should be aligned in the direction of the


49/149

35smallest gradient, and likewise, the shortest sideshould be aligned with the largest gradient of thesolution variables.

Calibration. A finite element model is a simplified,discrete representation of a complex and continuousphysical flow system (Froehlich, 1988). The threedimensional characteristic of the real system aremodeled by two dimensional elements and the flow isassumed to obey differential equations in which severalempirical coefficients are used. Model calibration isthe process of adjusting the dimensions of the finiteelements and the empirical coefficients so the valuescalculated by the model closely approximate valuesmeasured on the real system. The hydrodynamic model isready for calibration when it produces useful data.

The purpose of model calibration is to obtainan accurate mathematical representation of reality andnot to force-fit a poorly constructed model toapproximate these values (Froehlich, 1988). Sometimesparameters of poorly constructed models can be adjustedso the model produces results are close approximationsof the actual measurements. For example, the finiteelement network is probably a poor representation ofthe real system if a good fit of the data can only beaccomplished if the Manning roughness coefficient used


50/149

36is three or four times as large as the original valueassumed.

The model is calibrated by systematicallyadjusting parameters until computed and measured valuesagree as closely as possible. Measured calibrationdata consists of spot values of water-surfaceelevation, flow rates, and velocity components.Sensitivity of computed values to changes in model datashould be determined. Small changes in input data thatcause significant changes in model results indicatethat special care should be taken to obtain the inputdata.

Surface roughness coefficients have thegreatest effect on the model solution. The initialestimate of this empirical coefficient should not haveto be change much if the two dimensional network isproperly designed. Any changes in the roughnesscoefficient should be reasonable for the vegetativecover and topographic characteristics of the realreach.

Eddy viscosity coefficients usually do notaffect the model solution as much as roughnesscoefficients do. They have the greatest effect wherevelocity gradients are large. Increasing eddyviscosity coefficients will cause velocity gradients tobe reduced, and the horizontal velocity distribution


51/149


52/149

CHAPTER III

RMA-2D

IntroductionRMA-2D is the hydrodynamic numerical model

around which the river crossing site simulation andevaluation tool is assembled. RMA-2D is a finiteelement model used to solve two dimensional (in ahorizontal plane) depth averaged shallow water flowequations. The finite element technique produces a setof simultaneous nonlinear algebraic equations that mustbe solved by iteration. Typically there are severalthousand of these equations and several thousandunknowns. The unknowns are the two current velocitycomponents and the depth at each computational node.The model is designed to simulate both steady and non-steady state systems.

RMA-2D was designed for the U.S. Corps ofEngineers, Walla Walla District to study flow regimesfor the proposed Lower Granite Reservoir. It wasdeveloped under contract in 1973 by William P. Nortonand Ian P. King of Resources Management Associates.Originally designed for mainframe computers, the


53/149

39program has undergone considerable modifications sincethat time and is now portable to an IBM AT computer.Modifications to the model include use of curvedisoparametric elements, out of core equation solvers,sophisticated pre- and post-processing routines, andsimulation of alternately wet and dry elements during atidal cycle (King, 1988). The latest modificationallows a combination of one and two dimensionalelements to permit economical simulation of bays withtributary rivers and complex delta systems where only asmall area of the system is actually acting in twodimensional fashion.

Model Applicability

General approximations. RMA-2D simulates one and twodimensional depth averaged hydrodynamic systems withconstant water density over the entire area. Thedependent variables are water depth and the watervelocity in the horizontal plane. Reynolds assumptionsof shear stresses in turbulent flow are incorporated aseddy viscosity to approximate turbulent energy losses(Daugherty and Franzini, 1977). Either Chezy orManning equations can be used to approximate frictionlosses from the channel surface. Coriolis and surfacewind effects can also be approximated.


54/149

40The governing equations are as follows

(Froehlich, 1988):

Momentum:au at au a a t t E 2 L ItE d 2 uIt+01-+uh---gh +h - Sf +T=O (1)h t ax 3y- d x p ax 2 p dy 2

S U V a a h ItE y d 2 u hEYY d 2 UIt +Uh + vi-gt--gh ----- _-~-SfY+t


55/149

41Two dimensional elements. RMA-2D uses isoparametricquadrilateral and triangular elements to represent thegeometry of the reach being modeled. The boundaries ofthese elements can be either curved or straight. AGalerkin weighted residual approach is used to developthe finite element integrals and Gaussian quadrature isused to evaluate the final integral forms. The basisfunctions used are bi-quadratic for velocity componentsand bi-linear for water depth (King, 1988).

The Galerkin finite element method begins bysubdividing the physical region into a number ofsubregions, called elements. An element can be eithera triangle or quadrangle and is defined by node pointslocated along its boundary. Each node point is unique,and identified. The values of the dependable variables(velocity components and water depth) are approximatedwithin each element using values defined at theelement's node points and a set of interpolationfunctions. RMA-2D uses quadratic interpolationfunctions to interpolate depth-averaged velocities.Linear functions are used to interpolate flow depth.

Following this interpolation the method ofweighted residuals is applied to the governingdifferential equations for each element.Approximations of the dependent variables aresubstituted into the governing equations and usually a


56/149


57/149

43nonlinear inertia and bottom friction terms in thegoverning equation.

RMA-2D uses a frontal direct solution scheme inlieu of conventional finite element methods. Thesolution scheme is designed to minimize core-storagerequirements and the number of arithmetic operationsneeded to solve the system of nonlinear algebraicequations. The frontal solution scheme assembles andeliminates element equations at the same time. As soonas an equation is formed completely it is reduced andeliminated from the "active" coefficient matrix(Froehlich, 1988). It is written into a buffercontained in core memory. When the buffer is full, itis written to an auxiliary storage device. Thecoefficient matrix is usually never formed in itsentirety. At any time the active matrix contains onlypartially assembled equations or complete equationsthat have not yet been eliminated. The frontalsolution scheme requires rapid storage and retrieval ofintermediate values for the three unknowns, the twovelocity components and the depth at each node.Minicomputers are well adapted to this process bystoring intermediate solutions in large arrays ratherthan using programmed writes and reads (Gee, 1985).


58/149

44RMA-2D has a variety of options that can be

used to model rather complex systems. As they were notused in this thesis they will only be mentioned here.RMA-2D provides a choice of two element wetting anddrying routines can be used to simulate tidal marsheffects. One dimensional element networks can be usedto model a complex delta system with many individualchannels and junctions, large estuary systems where twodimensional modeling is uneconomical, and to modelsystems where two dimensional detail is required insome areas, but not over the entire study area. RMA-2Dallows control structures and locks to be included inthe model. This is essential in modeling not onlycontrol structures but also submerged obstacles such asbridges.

Program OrQanizationRMA-2D is a modeling system not just a single

program. There are three components to the RMA-2Dmodeling system: a preprocessor (RMA-I); thehydrodynamic model (RMA-2D); and two postprocessingprograms (VECTOR and CONTUR), that assist ininterpreting and displaying the data. The data flowand linkage of the three components can be seen inFigure 3.1.


59/149

45

~0ub 0 0--

0*ao 4)0oo0 co -I

0

'.4


60/149

46The data preprocessor, called RMA-l, is an aid

in development and error checking of the finite elementnetwork. The purpose of this program is to generatetwo dimensional finite element networks that will beused by RMA-2D. RMA-I helps the user to develop theelement network by generating the quadrilateral andtriangular elements. It can also be used to modify andexisting network and develop an element order that willallow the most efficient run time for RMA-2D. RMA-Ihas the capability to read, edit and print geometricinput files from a standard input batch file, or from afile written by a previous run of RMA-l. In addition,RMA-l can produce a graphical plot of the entirenetwork, as shown in Figure 3.2.


61/149

47

OA A

Figure~~~~~~~ . /nt lm n ewr(Source:~~~~~~~~e,18,Ueo\\ToDmninlFo

Model~~~~~~ ouniyAu cHbtt .4


62/149

48The hydrodynamic model is the second component

of the system. Originally designed to use withmainframe computers, RMA-2D still retains some of thoseattributes. Batch files, in which each linecorresponds to a ADP (automated data processing) cardare used to input data. The sequence of cards isrigid, as is the spacing of characters and the use ofinteger and real number values. Output from RMA-2D isin the form of print files that are, unfortunately formost users, in a 132 character format. This too is ahold over from the mainframe computer days.

The postprocessing component of the modelconsists of two plotting routines that help the user tounderstand the numerical results of the model run. Thefirst routine, called VECTOR, produces plots of thecurrent velocity calculated in RMA-2D. VECTOR has thecapability of plotting either vertically averagedvelocity vectors, unit discharge, or near-surface ornear-bottom velocity vectors. If the data is from adynamic simulation, VECTOR can plot these values onestep at a time. Additionally, Vector plots can bedisplayed on the computer screen. Figure 3.3 is a plotof the velocity vectors corresponding to elementnetwork in Figure 3.2.


63/149

49

3. 00 Fps

,/

/ I

Figure 3.3 RMA-2D VECTOR plot(Source: Gee, 1986, Use of a Two-Dimensional FlowModel to uantify Agquaic Habitat, p. )


64/149

5oThe other postprocessing routine called CONTUR

is used to plot contours of data produced by RMA-l andVECTOR. Data from RMA-1 is used by CONTUR to plot thecontours of the bed elevation as approximated inelement geometry file. Contur can plot the watersurface elevation, isotachs, or bed shear stress from afile generated by the VECTOR routine. The contouringprogram can also plot the finite element networkoverlaid on any of the above plots. This is aparticularly useful tool in network development andmodification. Figure 3.4 is an example of an isotachplot and a pathline plot from the same reach as theprevious figures.


65/149

51

I /

/ /i1 /x / / !I%

.xamole Isocach ?Ioc Example Pathline Ploc

Figure 3.4 RMA-2D CONTUR plots(Source: Gee, 1986, Use of a Two-Dimensional FlowModel to Quantify Aquatic Habitat, p. 9)


66/149

CHAPTER IVRIVER CROSSING SITE SIMULATIONAND EVALUATION TOOL (RC-SET)

GeneralThe River Crossing Site Simulation and

Evaluation Tool (RC-SET) is a computer simulation baseddecision support system used to evaluate potentialamphibious river crossing sites. RC-SET allows theuser to specify a unique scenario by selecting thecrossing site, the vehicle type that will attempt tocross the river, the river flow rate, the vehiclestarting point, and the vehicle finish line. It has aninteractive graphical interface, prompting the user forneeded input, and animating the river crossing attemptin real time as the user actually operates the vehicle.

RC-SET assists the user in interpreting andapplying output from the numerical two dimensionalfinite element hydrodynamic model RMA-2D. Thevelocity components and water surface elevation valuesare used by RC-SET to determine the effect of the watercurrent velocity on an amphibious vehicle'stranslational and rotational movement. The velocity


67/149

53vectors of the selected flow rate are displayed to helpthe user select the best route across the reach.

As a true decision support system, RC-SETconsists of three major components: the data basemanagement system, the model management system, and theuser-system interface (Hopple, 1988). Before rawtopographic and hydrologic data can be used by RC-SET,it must be manipulated using a number of datapreprocessing tools. The data flow and logicalrelationship of these components can be seen in Figure4.1.

DataPreparation

Tools

Data BaseManaqement

System

Model User-ManaementystemSystem Interface

Figure 4.1 RC-SET program organization


68/149

54External Data Preparation Tools

The purpose of the external data preparationtools is to develop a topography file of the study areaand prepare output from RMA-2D for use in RC-SET. Themethodology is to first use a raster based geographicalinformation system (GIS) called GRASS to produce abinary image file of the reach topography. This fileis used initially by a finite element generatingprogram (MESH) to develop a finite element network ofthe reach. The same file is used later by RC-SET todisplay the study area. The output file from the MESHis combined with a batch file defining the study areahydrologic boundary conditions, and together theybecome the input to RMA-2D. Results from RMA-2D arecalibrated with real data from the field when possible.The same procedure is followed for each potential rivercrossing site to be evaluated using RC-SET. Figure 4.2shows the flow of data during the preparation of dataphase.


69/149

55

-41-al

'.44

04

lz

Im

0

,7,3I00E4U


70/149

56GRASS. The Geographical Resources Analysis SupportSystem (GRASS) is a general purpose, grid-cell based,geographic modeling and analysis program developed bythe Corps of Engineers Construction EngineeringResearch Laboratory (USA-CERL, 1987). It wasoriginally developed for environmental planners atmilitary installations to help them plan land use tominimize the environmental impacts of training andsiting of structures. GRASS is now a widely usedsystem, used not only by military planners, but also bymany universities, government agencies, and researchlaboratories. GRASS can be used in studies concernedwith environmental planning, facility siting andresource management applications.

GRASS is an interactive graphical program thatprovides tools for developing, analyzing, anddisplaying geographical information. It is a grid celloriented Geographical Information System (GIS). Ageographical information system is capable of doingproximity analysis, weighting overlays, andneighborhood processing of geographical data fromsatellite imagery, paper maps, and other sources (USA-CERL, 1987). GRASS represents the data base as araster image of continuous rectangles. The analogyused in the GRASS manual is that the map sheets andoverlays are checkerboards with each square being


71/149


72/149


73/149


74/149


75/149


76/149

62

TABLE 4.1 Data preparation tools output files

Source File Name File Type Use in RC-SET

GRASS topo.bin binary topo graphicsdisplayRMA-2D vector.dat ASCII vector graphicdisplayGRASS topo.head ASCII topographicdata baseGRASS topo.cell ASCII topographic

data baseGRASS wse.head ASCII hydrologicdata baseGRASS wse.cell ASCII hydrologicdata baseGRASS vx.head ASCII hydrologicdata baseGRASS vx.cell ASCII hydrologicdata baseGRASS vy.head ASCII hydrologicdata baseGRASS vy.cell ASCII hydrologicdata base


77/149

63Data Base ManaQement System

The data base management system is the first ofthe three components of the RC-SET decision supportsystem. The purpose of the data base management systemis to open and read files selected by the modelmanagement system, create topographic and hydrologicdata bases from these files, and when prompted, sendspecific topographic and hydrologic information back tothe model management system. The data base managementsystem also has direct file reading subroutines used bythe model management system to read topographic binarygraphic files and the vehicle data base files.

The data base management system is based upon asimilar system used in MESH. Figure 4.3 is a schematicof the data base management system and shows selectionand data flow relationships.


78/149

64

r0

41)I U2

4J

tz

L* 0

E4O1410ow 41 C42

>qi


79/149

65While Figure 4.3 may look complex, the data

base management system is really quite simple. Dashedlines in the figure correspond to topography, vehicle,and flow rate selections from the model managementsystem that indicate which data base files are to bebuilt. Solid lines indicate data flow, either ascoordinates from the model management system, ortopographic and hydrologic data corresponding to thosecoordinates being passed back to it.

RC-SET allows the user to build the simulationby selecting the river crossing site, the vehicle typethat will attempt to cross the river, and the flow rateof the river. As these selections are made, the database management system opens and reads the appropriatefiles and builds the topographic and hydrologic database. The model management system opens and reads thevehicle data base, and the topography and velocityvector graphics files directly. As the simulation isrun the model management system sends the coordinatesof the vehicle to the data base management system,which returns corresponding topographic and hydrologicdata. The model management system uses this data tocalculate the translational and rotational motion ofthe vehicle, and its new position. The new coordinatesare passed to the data base management syst.em and theprocess continues.


80/149

66The data base management system uses two

computer subroutines to build and access the data base.They are modifications of similar routines used inMESH. The first, called UTIL, is a file utility thatreads the header and cell file information from theGRASS ASCII map layer files. The header file is usedto dimension a matrix into which the cell file data isread. UTIL calculates x and y vectors that correspondto the world coordinates of the center of each cell.The vector entries correspond to the location where theclass information from that cell is stored. The classinformation is either the ground surface elevation,water surface elevation, water velocity component inthe x direction, or the water velocity component in they direction.

The other computer software subroutine iscalled AVE is used to access the data base matrices andlinearly interpolates the class value at the currentvehicle coordinates. UTIL translate world coordinatesinto vectors that are used to identify a point inspace. It reads the class values of the center ofsurrounding cells and calculates a two-way linearinterpolation to determine the class value at thecurrent coordinates. This value is passed back to themodel management tool.


81/149

67Not all topography and hydrologic files are

opened and read into the data base. To minimize memoryrequirements while running RC-SET, only the-. filescorresponding to selections made in the modelmanagement system are opened and read into the database.

Model Management SystemThe model management system consists of several

subroutines that assist the user to set up the modelproblem and simulate an amphibious vehicle's movementover land and in the water. The user communicates withthe model management system via a user-system interfaceof graphic displays, messages, menus and other devices.The function and design of the user-system interfacewill be covered in detail later in the thesis. Figure4.4 shows the logic network of the model managementsystem.


82/149

68

.1o YesWge-DT>Elev?

Ground AMphibiousTrans Trans

tout-zle5 .outines

Roan

Figure.4 odeta naetntSse


83/149

69Model set up. The model management system allows theuser to select the building blocks that make up thesimulation model: the crossing site, the vehicle type,the water flow rate, the vehicle starting point, andthe vehicle finish line. After the amphibious crossingsite is selected from a menu, a raster image of thecrossing site is displayed and the topography cell fileis open and read into the topographic data base by thedata base management system. The user then selects thetype vehicle that will attempt to swim across theriver. A vehicle data base file is opened andcharacteristics of the vehicle such as mass, width,height, length, and draft are assigned as globalvariables.

The user is then prompted to select the flowrate of the water that will flow through the modelreach. The velocity vector file is displayedindicating the direction and magnitude of the currentvelocity at the finite element node coordinates. Thedata base management system opens and reads thecorresponding water surface elevation, and watervelocity component map sheet cell files into thehydrology data base.

With the topographic and hydrologic data basesset up, the next step is for the user to select thevehicle starting point. Using the mouse, the user can


84/149


85/149

71rotational motion of the vehicle to determine theangular acceleration, velocity, and angle of rotation.In the fourth step the model management system sendsthe world coordinates and angle of rotation to theuser-system interface for graphic display. The fifthand final step is to determine if the mass center ofthe vehicle is in the finish line polygon. If it is,the simulation is terminated, if not, the movementroutine returns to step 1 and continues as before.

Model limitations. The primary purpose of RC-SET is tosimulate amphibious vehicle movement in water. RC-SETalso approximates vehicle ground movement, but thisfeature has not been validated by field data and shouldbe used only as an aid in learning to operate thevehicle or in some other way to enhance the simulation.

Equations determining movement of the vehiclethrough water have been simplified but still retain alevel of accuracy required to accurately approximatethe real system. Any body immersed in water willexperience forces and moments from the flcw. Theforces acting on the body in three dimensions are drag,lift and side force, and the corresponding moments areroll, yaw, pitch (White, 1986). Figure 4.5 shows thisrelationship.


86/149

72

4J)

0 4) 000000 - -

4 .,4

044 W

.H-I

00)(0 40

0 J4

0 0-ri~ 0Hn

>V 4 C-~ 4 J

>~


87/149

73Amphibious vehicle movement in this model is

simplified from this three dimensional system to twodimensional system by ignoring the effects of lift, androlling and pitching moments on the vehicle. Movementof the vehicle in water is simplified to translationaland rotational motion in a horizontal plane. In otherwords, RC-SET determines the vehicles orientationrelative to che z axis. This is sufficient toapproximate movement of the vehicle as it crosses theriver. However, it does not address the danger of thevehicle swamping due to excessive pitching or rolling.

General equations of motion. Specifying the positionof a rigid body in the horizontal plane of motionrequires the definition of three scalar values, the twocoordinates of the mass center and the angular positionof the vehicle about the mass center (Meriam, 1978).Three independent scalar equations are required todescribe planar motion, one for each scalar value.General plane motion of a rigid body is a combinationof translational and rotational motion. Thecoordinates of the mass center are determined by theequations of translational motion, and the orientationabout the mass center is determined by the rotationequation.


88/149

74Translation is defined as any motion in which

every line in the body remains parallel to its originalposition at all times (Meriam, 1978). In translationalmotion there is no rotation of the body. In rectilineartranslation all points in the body move in parallelstraight lines. Figure 4.6 is an example ofrectilinear motion in the x and y directions.

// A' 1 // / i/ W / I/i I/ / /

S A'i / / /

*- - - 3- - , /A

Figure 4.6b

Figure 4.6a

Figure 4.6 Rectilinear motion(Adapted from: Meriam, 1978, Dynamics, p. 264)


89/149

75Translational motion is defined by Newton's

second law of motion of a mass system. The law statesthat the resultant of external forces on any system ofmass equals the total mass of the system times theacceleration of the system mass:

Fx = m(dx) (4)Fy = m(y) (5)where:

Fx, Fy = the sum of the forces in the xand y directions, respectivelym = the mass of the body

dx, dy = instantaneous linear accelerationof the body in the x and ydirections, respectively

Rotational motion is defined as the orientationof the rigid body about some point of reference.Rotational motion can be calculated using the momentumequation (Meriam, 1978):

ZM = Ia (6)where:

ZM = the sum of the moments aboutthe mass center due to external forcesonly

S= the mass moment of inertia of thevehicle about the z axis

a= the angular acceleration of the vehicleabout the mass center, relative to thex axis


90/149

76Figure 4.7 represents rotational motion about

the mass center in a horizontal plane.

y

&K

B10

Figure 4.7 Rotational motion in a horizontal plane(Adapted from: Sears, Zemansky, and Young, 1976,University Physics, p. 160)


91/149

77The angle G is the original orientation of the

body relative to the x axis, 0' is the new orientationof the body after rotation. The mass moment of inertiais a constant property of the body and is a measure ofthe distribution of mass around the z axis and throughthe mass center, and is analogous to mass intranslational motion.

Figure 4.8 is a free body diagram describingplane motion (x-y) of an amphibious vehicle movingthrough water.


92/149

78

00

W Ul

4J .0

100

0-- 4

P44

a4 w110 4 -4

coJ 44

>1 4 - 4-40 Q0 m >

t N-4 "4

- an intelligent planning tool for evaluation of amphibious river crossing sites a218496

Documents