OCEANS'95 MTS/IEEESan Diego, California, U.S.A.

REAL-TIME MODELING OF MICRO CABLE DEPLOYMENTS

J. M. Andres, S. R. Jefferies, and G. D. Gillenwaters

Makai Ocean Engineering, Inc.Makai Research Pier

Waimanalo, HI 96795 USA

Abstract -Extensive at-sea experiments completedby the Naval Civil Engineering Lab. (NCEL)during the Underwater Deployment Experiment(UDX) have provided a unique set of data thatconfirms differences between deployment oftraditional large-diameter cables and the newgeneration of micro fiber optic cables (diam.<0.2").These data have been used to validate existing cabledynamic models, such as SEADYN and the Makaicable model, under different deployment scenarioswhich include slow and rapid deployment of microcables and deployment of micro cables with heavyin-line bodies. This paper discusses the results ofthese comparisons in terms of modeling accuracyand computational speed. The results demonstratethat the Makai cable model is capable of accuratelymodeling, faster than real-time, realisticdeployment scenarios of micro fiber optic cables.The paper goes on to briefly outline the adaptationof an existing real-time control system, successfullyused to deploy conventional cables, to automaticallycontrol deployment of micro cables with a highdegree of positional accuracy and slack control.

1. INTRODUCTION

With the movement towards micro fiber optic cablesfor military applications, the U.S. Navy through theNaval Civil Engineering Laboratory (NCEL,currently Naval Facilities Engineering ServiceCenter) initiated investigations into micro cabledeployment control systems for use in future Navycable deployment operations. As part of this effort,NCEL first needed to determine if existing cabledynamic models could be used to accurately modelthe performance of light weight micro cables. Sinceno appropriate test lay data were available for microcables to make this validation, NCEL initiated aseries of at-sea experiments, the UnderwaterDeployment Experiments (UDX), in an effort toprovide such data. Results from these experiments

have provided a unique set of data that confirmsdifferences between deployment of traditional largediameter cables and the new generation of smallfiber optic cables (d<0.2").

The cable models included in the validation exercisewere SEADYN, the premier cable dynamic modelused by the u.S. Navy, and the APC, the cable modelused in Makai's Integrated Control System (ICS).Makai's ICS is a semi-automated real-time cabledeployment control system used to accurately controlthe bottom placement and slack/tension of submarinecables. At that time, the Makai ICS had beensuccessfully used to deploy power cables withpositive bottom tensionl.2, and to lay communicationcables with an accurate control of cable placementand bottom slack3. The heart of the Makai ICS is athree dimensional cable model program (i.e. theAPC or Makai cable model) which is able to modelthe dynamics of a suspended cable using real-timedata measurements during a cable deployment orretrieval operations (i.e. ship position, cable payoutand ocean currents) 4. The Navy was interestedtolearn if the Makai cable model (and the ICS) couldperform as well on smaller fiber optic cables as ithad on larger power and communication cables.

In order to determine how well the Makai cablemodel could simulate the dynamic behavior of microfiber optic cable deployments, NCEL prepared a setof baseline deployment conditions from the UDXexperiment for input into the Makai model. Resultsfrom the Makai cable model were then compared byWestern Instrument Corporation to those ofSEADYN and to the real data collected from theUDX 5. The Makai cable model used for this studywas the same as that used during the SOAR IIproject (version 4.0)3. Only the formulation for thetangential and normal drag coefficientswere modifiedto agree with the formulation suggested by NCEL formicrocables.

Once existing dynamic cable models were validated,the Navy's next desired step is the development of asuitable semi-automated control system to assist inthe deployment of micro fiber optic cables. In thefuture this system could be extended to become afully automated cable deployment control system.

2. BASELINE CASES ANALYZED

During the UDX experiment, two strings of microfiber optic cable each with five cable packs and fourbodies were deployed. To have a meaningfulcomparison, two cable pack deployments (barecables) and two body deployments were selected as areasonable subset of data. A basic description ofthese cases is outlined in Table 1. Table 2 lists thebasic cable, body and environmental properties usedin the simulations.

TABLE 1: Case definition.

PROPERTIES S1 UNITSENGLISHUNIT S

0.077 inches7.18 Ib/1O00ft

5.11lb/1O00 ft3.810 ft

5.5 inch45lb

64.56 Ib/ft332.14 ft/s2

0.70

1.956 mm

0.105 N/m

0.0747 N/m

3.83.0m

0.140 m

200N

1035 kg/m39.80 m/s20.70

TABLE 2: Cable. body and environmental properties.

The following data were provided as input to the Makaimodel:

. Ship position (X,Y) as a function of timeLength of cable paid out as a function of timeCurrent velocity ('Ix, Vy) at different depths as afunction of time

Water depth along the cable routeInitial cable configuration at start of simulationCable, body and environmental properties as listedin Table 2

Cable and body normal and tangential dragcoefficient formulation as provided by NCEL

.

.

...

.With the exception of Case 1, Makai was neitherprovided with the UDX experimental results nor withthe SEADYN results (these results were only madeavailable after the Makai results were provided toNCEL). For Case 1, however,Makai wasprovidedwithpartial experimentalresults and with averageship speedand cablepayout rates during loweringof the cablebodythrough the water column. Case 1 was then used todetermine if the results predicted by the Makai modelusing all the mathematicalassumptions relevant to thisdeployment (e.g. new formulation for drag coefficient)wereconsistentwith the experimentalresults.

3. RESULTS AND COMPARISONS

Figure 1 presents plan, isometric and side views of atypicalcable lay showing the main parameters requiredas a part of the output for the simulation runs. Adescriptionof the conventionsused for the angles is alsoshown.

PLAN VIEW

Touchdown y

LxTransverse

OffsetT--

Barge Ship

ISOMETRIC VIEW

LTD

Depth! /0 76Z ~Offset

I

CASES SHIP PAYOUT FEATURES(ALL FROM SPEED SLACKUDX2) (kt) RATE(%)1. Body 1 2.0. 5% Controlled payout

rate.2. Pack 4 1.7 5% Controlled payout

rate.

3. Body 4 1.7 40% High slack bodydeployment.

4. Pack 5 3.8 5% High ship speed

SIDE VIEW

SURFACEBARGE-

Tbrep=AFT Cable Tension

Figure 1. Cable laying geometry showing the mainoutput parameters.

CASE 1: UDX2 Body 1 Deployment (5%Pavout Slack)

Results for the UDX2 Body 1 case are for a bodydeployedunder a controlledpayout rate with 5% slack.

Figure 2 depictsthe body sink rate as a functionof timeas predicted by the SEADYN and the Makai modelsand as measured during the UDX experiment. Thisfigure shows that, with the exceptionof the time whenthe body goes overboard,the body sink rates computedby the Makai model follow the same trend as thosemeasuredduring the UDX experiment.The initial highvalue of sink rate measured experimentally could berelated to a momentary change in cable payout rate asthe body is about to enter the water. This effect couldnot be modeled by the Makai cable model in thisspecificcase since only averagecable payout rates wereavailable.

1.5

.-...(f)

"-....

E 1.0' /

Exper.- SEADYN- Makai

(j)+-'0

0::: .5.:0<:

C

(/)

>- .0-00

CD

-.5I

15.8 15.9

Time (Decimal Hours)

Figure 2. Body sink rate as a function of time forUDX2Body 1 deployment.

Figure 3 shows the body pitch angle as a function oftime. The body pitch angle is measured such that a zerodegree angle corresponds to the case when the body ishorizontal (see Figure 1). The Makai cable modelshows values of pitch angle that decrease from 45° whenthe body enters the water to 20° at touchdown. Thesevalues are in close agreement with the values measuredduring the UDX experiment.

- Exper.- SEADYN- Maka;

15.9

Time (Decimal Hours)

Figure 3. Body pitch angle as a function of time forUDX2Body 1 deployment.

Figure 4 and Figure 5 depict the time variations of cabletension computed aft (below) and forward (above) of thebody, respectively. In both cases, the agreementbetween the measured and computed values by theMakai cable model is excellent considering that theinput data used in the Makai cable model were not exact(only average values were available).

40

~

D

c 300(f)c(j)

I- 20

(j)D0

U 10+-'

:q:

0

- Exper.- SEADYN- Maka;

15.8 15.9

Time (Decimal Hours)

Figure 4. Tension in cable aft of body as a function oftimefor UDX2Body 1 deployment.

60.-...

50Q"\(j)

40-0'-.../

30(j)

20Q"\C 10

<r:0

L

-2 -10

0... -20

>- -30-00 -40CD

-50

-60

15.8

As in the case of the body sink rate, the initial highvalue of aft cable tension measured experimentallycould be related to a momentai)' decrease in cablepayout rate as the body enters the water. This effectcould not be modeled by the Makai model since onlyaverage cable payout rates were available. However, asimilar effect is observedand properly modeled by theMakai modelfor the UDXl Body4 case discussedlater.

60- Exper.- SEADYN- Maka;

~

-D

50c0

~ 40QJ

f-

QJ .30-D0

U 20

-0L

0 10~L0

LL 0

158 15.9

Time (Decimal Hours)

Figure 5. Tension in cable forward of body as afunction of time for UDX2 Body 1 deployment.

CASE 2: UDXl Pack4Deployment (5%PayoutSlack)

Resultsfor the UDXl Pack 4 are for a bare cablebeingpaid out with 5% slackand a ship speedof 1.7knots.

Figure 6 shows the results predicted by the Makai andSEADYN models for the cable angle e versus time (seeFigure 1 for definition of angles). The results are almostidentical. The angle starts at a high value caused by theprevious body deployment. Once the body touchesdown, the large body weight is removed from thesuspended cable and the cable angle decreases fromapproximately 33° to an almost "steady state" value of14.3° in about 30 minutes. The final steady state cableangle measured during the UDX experiment was 14.4°.

Figure 7 shows the values of the offset angle ~ as afunctionof time as predictedbyboth cable models. Thisfigure shows differences between the two programs.The results fromthe Makai model contain an oscillatorypattern not shown in the SEADYN results. This is theresult of Makai's input for ship course, which includedslight oscillationsabout the mean path. The SEADYNsimulation approximated the ship path by a straightline, so the oscillatory effect present during the realcable deployment does not comes into play. However,both results showthe same general trend.,an angle with

magnitude increasing until 18.7 hours, then slowlydecreasing through the rest of the deployment. Theexperimentaloffsetangle ~could not be measured as afunction of time during the experiment, but wasestimatedfrom the measurements made after the cablelay. For this deployment, the final offset angle wascalculatedto be -2.5 °.

~ry 40QJ

0' /

CD .30

QJryc

<C 20

>-.0

---1

QJ

-D0

U

- SEADYN- Maka;

10

0

18.5 19.5"19.0

Time (Decimal Hours)

Figure 6. Cable lay angle (J as a function of time forUDX2 Pack 4 deployment.

ryQJ -1

0

-e-

~ -3ryc

<C -4

a:; -5(/)'+-

0 -6

0

- SEADYN- Make;

-2

-7

185 19.0

Time (Decimal Hours)

19.5

Figure 7. Cable offiet angle <pas a function of time for

UDX2 Pack 4 deployment.

Figure 8 compares the values of cable top tension as afunction of time predictedby the Makai and SEADYNmodelswith the experimentalmeasurements. Althoughdifferent in detail, both models show that they are ableto simulate the very low tensions characteristic of thismicro fiber optic cable deployment (primary tensionvalue varies between 4 and 5 pounds). The Makairesults show more spread because the model used theraw cable payout rate input data (unsmoothed data),while the SEADYN simulation was done with thepayoutrate averagedeveryminute.

'D9

'--" 8

§ 7

~ 6

~ 5

Q. 40

f- 3Q)

:D 20

U 118.5

- Exper.- SEADYN- Maka;

18.75 19.0 19.25

Time (Decimal Hours)

19.5

Figure 8. Cable top tension as a function of timeforUDX2Pack 4 deployment.

Figure 9 showsa comparisonof the installedcableslackas computed by the Makai and SEADYN models withthe measured values of slack. Although there is areasonable correlation, there are significant differencesin detail in this graph. As in the case of Figure 8, theinstalled slack computed by the Makai model showsmore spread as a result of using directly the measuredcablepayout rate. For example,at time t=18.603hours,the raw data for cablelengthjumped +9 m overa periodof 1 second. This high value of cable payout rategenerated a rapid decrease in the value of top cabletension (see Figure 8) and increased momentarily thecable bottom slack to a value of 137% (see Figure 9).Although the case just described at time 18.603 isobviouslyan exampleof erroneousdata input, there areother cases where it is difficult to differentiatebetween

noise and real measurements. The results computedbythe Makai cable model using noisy data show that themodel is able to predict the behavior of the cable underabrupt changes in cablepayout rates.

120,.--....~'--" 100

-YU 800

(/) 60

"D 40Q)

0 20~(f)

c 0

140

Exper.- SEADYN- Makai

-20

0 2000 40001000 3000

Cable laid on Bottom (m)

Figure 9. Installed slack as a function of cable laid onthe bottomfor UDX2 Pack 4 deployment.

CASE 3: UDX2 Body Deplovrnent (40% Pavout Slack)

Results for the UDX2 Body 4 are for a cable with an in-line body being deployed at a controlled payout rate of40% slack and the ship moving at 1.7 knots.

Table 3 summarizes the results provided by both cablemodels and the experimental measurements for threedifferent body deployment properties. The propertiescompared are: the advance distance, the transverseoffset, and the average sink rate. The final values forthe body advance and body offset are given by thelongitudinaland transverseadvance distances from thelaunch location to the body touchdown respectively.The averagebody sink rate is computed as the averageverticalvelocityof the bodyfrom launch to touchdown.

All percentages are percent of water depth.

TABLE 3: Comparison of body deployment properties.

The target accuracy for the body advance distance andbody transverse offset was ::tl6% of the water depth6.As seen in Table 3, both models predict these propertieswithin the desired accuracy. In the case of thetransverse offset, the Makai model comes closer to the

measured values than the SEADYN prediction. This isalso true for the average sink rate, which is within thetarget value of :iO.03 m/s of the measured value6. Thebody sink rate as a function of time during the bodydeployment is shown in Figure 10. This figure showsthat there is a very good correlation between the resultsof both cable models with the experiment.

";:;) 1.5'-----E

'--" 1.0

Q)~0

0::: .5

-'C.S 0.0(/) - Exper.- SEAOYN- MakeiA"D -.50

m 19.7 19,8

Time (Decimal Hours)

19.9

Figure 10. Body sink rate as a function of time forUDX2Body 4 deployment.

PROPERTY SEADYN MAKAI EXPERIMENTAdvance -10.8% -12% -0.6%Distance

Transverse -6.7% -3% -4.3%Offset

AverageSink 0.67 m/s 0.81 m/s 0.83 m/sRate

Figure 11 depicts the values of the body pitch angle asthe body sinks in the water. Although the Makai andSEADYN models yield comparable results, with theMakai model showing somewhat better correlation withthe measured data shortly after the body launch, none ofthe models are able to predict accurately the largeoscillatory behavior of the body pitch angle. Since inthe at-sea experiments the pitch angle has not beenshown to be of great importance in predicting body sinkrate or final body touchdown location, no attempts weremade to improve the modeling capabilities of thisparameter. It is interesting to note that this oscillatorybehavior was not observed for Case 1 where the bodywas deployed with only 5% slack For Case 1, the cablemodels and in particular the Makai model, were able topredict the measured values accurately.

CJ1 60Q)

0'-../ 40

Q)

g' 20<{

L 00U--'

Q.. -20

:>,v -400m 197 19.8

Time (Decimal Hours)

19.9

Figure 11. Body pitch angle as a function of time forUDX2 Body 4 deployment.

Figures 12 and 13 show the results of the cable tensionjust aft and foreward of the body, respectively. Thesetensions are closely related to the body pitch angle, ascan be seen in the experimental data for theseparameters. Since the simulations could not predict theoscillations of pitch angle for a cable payout slack of40%, then it is expected that they can not predict theoscillations observed in the tension measurements.

D 30

c

Q 20(f)cQ)

~

Q) 10D0

U

--'

~0

Exper.- SEADYN- Maka;

19.7 19.8 19.9

Time (Decimal Hours)

Figure 12. Tensionin cable aft of body as a function oftimefor UDX2Body 4 deployment.

- Exper.- SEADYN- Maka;

19.8 19.9

Time (Decimal Hours)

Figure 13. Tension in cable forward of body as a

function of time for UDX2 Body 4 deployment

CASE 4: UDXl Pack 5 Deplovrnent(High Ship Speed)

Resultsfor the UDXl Pack 5 are for a bare cable beingdeployedwith 5% slack and a ship speedof3.8 knots.

Experimental results for cable lay angle, e, and offsetangle, ~, are not available for Pack 5 because themeasurementtechniques could not be used at this shipspeed. However,the values predictedby the Makai andSEADYN models showed agreement with each other,similar to thoseof Figures6 and 7 for Pack 4 (graphicalresultsfor Pack 5 not shown).

Values of cable top tension as a function of time andinstalledslack as a function of cable laid on the bottomare presentedin Figures 14 and 15 respectively. Figure14shows goodcorrelationbetweenthe predictionsfromSEADYN and the Makai cable model, although thepredicted values are higher than those measured.However, it is important to note that values of toptension were difficult to measure during thise.\:periment5.

D15- Exper.- SEADYN- Make;

c0(f) 10C(lJ

~

g- 5~

ClJ

-g 0

U 19.75 20.0 20.25 20.5

Time (Decimal Hours)

Figure 14. Cable top tension as a function of time forUDX2Pack 5 deployment.

/""'0. 50D'-../

c: 400(f)c:Q) 30

Q)

D 200

U

D 10'--0

'--00

u..19.7

Figures 15 shows that although none of the models canpredict accurately the installed slack, both models reflectthe general trend revealed by the experimentalmeasurements. That is,an increase in installed slack

just in front of an in-line body, with slack valuesdecreasing to the nominal slack as the distance from thebody decreases.

250

~

~ 200'---"

- Exper.- SEADYN- Maka;

-Y 150()0

(j) 100-0QJ

0..--UJC 0

-50

0 200 400 600

Coble Laid on Bottom (m)

800

Figure 15. Installed slack as a function of cable laid onthe bottomfor UDX2 Pack 5 deployment

Table 4 summarizes the cases run and thecorresponding execution times. Exact values ofexecutiontimes for the SEADYNmodel were not madeavailable, but they were considerablyslower than realtime for all the casesanalyzed.

TABLE 4: Execution timesfor the deployment casesrun.

4. RECENT DEVELOPMENTS

Since the completion of this project, the AdvancedResearch Projects Agency (ARPA) through Hawaii'sCenter of Excellence for Ocean Science (CEROS) hasfunded an R&D project to develop a real-time,automated system able to control the deployment ofmicro cables under realistic scenarios. Preliminaryresults of this ongoing R&D project show that theautomation algorithms being developedallow accuratecontrolof the touchdownpositionand installed slackforlargediameterand small diametercables.

5. CONCLUSIONS

1. The Makai and the SEADYN cable models havebeen shown to predict the behavior of micro cabledeployments with in-line bodies under realisticdeploymentscenarios.

2. The Makai cable model runs 4 to 13 times fasterthan real time for the cases analyzed involving smalldiameter fiber optic cables with no body in the watercolumn. Theseexecutiontimes wereone to two order ofmagnitude smaller than those of SEADYN. Theseexecutiontimes were achieved on an HP ApolloModel720/25 Desktop Workstation using no codeoptimization during compilation (speed comparable tothat achievedon a DOS based PC 486/66). It is worthpointing out that these speeds were achieved with thesame program used for laying conventional largediameter submarine cables. A two to three foldexecution speed improvement can be expected byoptimizing the input variables for the cable model andby using a high level of optimization during thecompilation.

3. The Makai cable model program has adequate speedto simulate deployment of bare small diameter fiberoptic cables and to be used in conjunction with all theother programs of the Makai Integrated Control System(ICS) to control the deployment of micro cables in real-time.

4. When a heavy body attached to a small diameterfiber optic cable is in the water column, the currentMakai cable model used in these simulationswas only1.6 times faster than real time (using no optimizationduring compilation). Further improvements inexecution speeds and reliability for deployment of in-line bodies can be achieved by further optimizing thecode for specific cases of small diameter fiber opticcables. These improvementsare necessary in order toachieve real-time control capabilities when the cablemodel is incorporatedinto the Makai rcs.

5. The results computed by the Makai cable modelusing noisy data show that the model is able to predictthe behaviorof the cable under abrupt changes of cablepayout rates. No convergence problems occurredduring the simulationsbecauseof noisydata. Althoughnot used in these runs, the Makai rcs has the ability tofilter the inputdata in real-timebeforerunning the cablemodels in order to increase speed and accuracy of theresults.

CASES (ALL EXPERIMENT SEADYN MAKArFROM UDX2) (min) (min) (min)1. Body I 10.0 N/A 6.22. Pack 4 70.5 N/A 5.43. Body 4 11.5 N/A 7.24. Pack 5 32.5 N/A 6.8

ACKNOWLEDGMENTS

The authors wish to thank the Naval FacilitiesEngineering Center for providing the data from theUDX experiment and allowing their public release.Also, thanks to Western Instrument Corporation forcompleting the simulation runs with SEADYN.

REFERENCES

[1] Andres, J.M., Jefferies, S.R and Jensen, D.N.,HawaiiDeep WaterCableProgram: Results of theAt-Sea Test Cable Lay, Froc. Marine TechnologySocietyConference'90, Washington,D.e.

[2] Van Ryzin, I.e., Resnick, AM. and Jefferies,S.R,A Real-TimeIntegrated Control Systemfor LayingCables in the Ocean,FrOG.MfS'90, Wash., Doe.

[3] Andres, I.M., Jefferies,S.R and Gillenwaters,G.,Validation of a Real-Time Cable DeploymentControl System for Slack Cable Laying, FrOG.0ceans'93, Victoria,Canada.

[4] Jefferies, S.R, and Andres, J.M, A 3-D CableModel for Cable Lay Operations, FrOG.MarineTechnologySocietyConf. '90 , Washington,D.e.

[5] Western Instrument Corporation,Evaluation of theMakai Cable Deployment Model, for the NavalCivil Engineering Laboratory,Port Hueneme, CA;Contract No. N47408-92-D-7004,Delivery Order0021, CDRLAO014,May 1993.

[6] Western Instrument Corporation, DeepwaterDeployment, Phase III, Experimental Test Plan(Final Report), for the Naval Civil EngineeringLaboratory, Port Hueneme, CA; Contract No.NO0123-89-D~46, Delivery Order 0088, CDRLAOO4, March 1992.