using a ships propeller for hull condition monitoring mkt1

8/10/2019 Using a Ships Propeller for Hull Condition Monitoring Mkt1

1/20

Presented at ASNE Intelligent Ships Symposium IX - May 25, 2011, Philadelphia, PA, USA

MACSEA Ltd. 1

Kevin P. Logan

Using a Ships Propeller for Hull Condition Monitoring

ABSTRACT

As a ships hull condition degrades due tomarine fouling, more power and fuel are neededto maintain service speeds. A by-product of theincreased fuel consumption is increased GreenHouse Gas emissions. Rising fuel costs, hullmaintenance expenses, and mountingenvironmental regulations make hull conditionmonitoring a crucial tool for prudent shipoperators to eliminate energy waste due to hull

fouling, reduce carbon emissions, and eliminatethe carriage of invasive species between ports.

Methods for using a ships propeller as a powerabsorption dynamometer employ the propeller asa measuring instrument to estimate either speedor power. The calibration is typically performedfor clean hull conditions, allowing the resultingpropeller model to be used to track shipperformance degradation due to hull foulingagainst a standard clean-hull baseline.

The propeller power absorption technique ispresented, along with the salient results of usingit to monitor two Navy sisterships over a year-long time period. This information may beuseful for Navy decision makers responsible forhull/propeller maintenance and hull paintselection.

INTRODUCTION

Most ship owners and operators realize that hullfouling causes drag-related speed loss and

increased fuel consumption when more power isrequired to maintain ship service speeds. Hullfouling is also a topic of growing environmentalconcern and international regulation as it relatesto green house gas emissions and the carriage ofaquatic invasive species on fouled hulls.No ship operator can afford to waste energy.Ship performance losses due to hull and

propeller fouling can be substantial but havehistorically been difficult to quantify, sincechanging ship and environmental conditionsintroduce a large degree of variability inperformance data that makes separation of hulland propeller effects a difficult task. Forexample, beyond the hull and propellercondition, ship performance measurements willchange with draft, trim, rudder activity, wind,waves, currents, water depth, etc. This paperdescribes a method of separating out theseeffects and using a ships propeller as a tool forearly detection of hull fouling.

Economic Penalties of Hull Fouling

Most recently, (Schultz et al. 2011) performed acomprehensive estimation of the economicimpact of hull fouling on the US Navys ArleighBurke-class destroyers (DDG 51). Estimatedhull fouling-related costs were based on a rangeof factors, including extra fuel consumption, hullcoatings, coating application and removal, and

hull cleaning expenses. As input to this work,predictions of full-scale ship resistance andpowering were made for antifouling coatingsystems across a range of roughness and foulingconditions (Schultz 2007). These estimates werebased on scale-model tank testing resistancemeasurements and similarity law analysis. Basedon this, the overall costs associated with hullfouling for the Navys present coating, cleaning,and historical fouling levels were estimated to beapproximately $56M per year and $1B over 15years for the entire DDG 51 class (56 ships). It is

clear that hull fouling is a major expense for theentire Navy fleet. Depending on foulingseverity, fleet-wide hull fouling-related costscould fall in the $180M - $540M per year range(Schultz et al. 2011).

The Navy has recognized the importance of hullcondition monitoring in their energyconservation guidelines to the fleet (NAVSEA


2/20


MACSEA Ltd. 2

2010). It is stated therein that fuel consumptionincreases caused by hull and/or propeller foulingare often the largest single cause of excess fuelconsumption. It is also recommended that hulland propeller cleaning be condition-based, butrather than directly measuring performance

losses, as described in this paper, the Navyregularly schedules hull inspections by divers.These visual inspections provide usefulinformation on hull condition, but do not offerquantified performance loss data that can befurther employed for optimal hull and propellermaintenance strategies.

The Naval Ships Technical Manual CHAPTER081 (NAVSEA 2006) provides that latestofficial guidance on waterborne underwater hullcleaning criteria to maintain a vessels optimum

performance.

Modern Hull Coating Technologies

Marine bio-fouling increases the frictional dragof a ships hull, thereby requiring more powerand fuel to maintain operational speeds. Marinegrowth can be retarded through the use of anti-fouling paint and until about a decade ago, themajority of ship hull coatings were tributyl-tin(TBT) self-polishing co-polymers. Through asteady release of the TBT toxin, ships could bekept free of fouling for up to 5 years.Unfortunately, due to TBTs negativeenvironmental side effects, the InternationalMaritime Organization (IMO) imposed a ban onTBT paint, leading paint manufacturers todevelop non-toxic alternative paint systems.Amongst these paint alternatives are low-copperand copper-free ablative antifouling (AF), foulrelease (FR) coatings, and surface treatedcoatings (STC).

ABLATIVE ANTIFOULING

Ablative/self-polishing AF paints have typicallycontained biocides mixed into co-polymer paint.The surface layer of paint is gradually dissolved(ablated)by the seawater, revealing freshbiocides that were buried beneath. For thisreason, several coats of paint can be built up toprovide effective protection for many years.

Since they contain biocides, normal ablativetypes offer excellent anti-fouling performance.Surface roughness after the self-polishingprocess is designed to be less than whenoriginally applied.Most manufacturers ofablative AFs have several types, designed for

different ablation rates, depending on fast,medium, or slow speed service.

(Yebra and Catala 2011) discuss the advantagesof new formulations of silylated acrylate (SA)coatings being offered by most major paintcompanies. This self-polishing, biocide paint isreportedly effective for up to 90 months, withdynamic exposure test results to date showingthat it is half as rough as a comparable copperacrylate. But as these authors point out, there is alack of reliable studies linking AF performance

to fuel consumption and a scarcity of accurateperformance monitoring systems that can beused to quantify the economic benefits ofcompeting coating systems.

FOUL RELEASE

Foul release or low surface energy coatingsact to prevent hull fouling by providing a low-friction surface onto which marine organismshave difficulty attaching. If vessels arestationary for short periods of time, fouling may

occur, however; there will be a weak bondingbetween the fouling organisms and the coatingsurface. The organisms may be subsequentlyremoved by either the hydrodynamic forces onthe hull when the vessel transits at a sufficientlyhigh speed or by underwater cleaning. Thelifetime of foul release coatings may be limitedas they are mechanically soft and easilydamaged. Underwater cleaning must be donewith soft brushes which may not remove allfouling organisms, for example after longer in-port periods. In addition, mechanical damage to

the hull may result in local unprotected areasthat will eventually require touch-ups in dry-dock.

SURFACE TREATED COATINGS

Surface treated coatings consist of large glass-platelets suspended in a reinforced vinyl esterresin. The hull coating is conditioned by divers


3/20


MACSEA Ltd. 3

following ship launch using mechanical brushes.Additional cleanings are performed on an asneeded basis and, due to its non-toxic properties,without adversely affecting the marineenvironment. Since the hull surface is hard andcan withstand many cleanings, long-term service

life is possible (e.g. 20 years), and the coatingbecomes smoother after each cleaning.

A comprehensive summary of antifoulingcoating technology as of 2004 is provided in(Yebra, Kiil, and Dam-Johansen 2004).Although there are several non-toxic paintsystems currently available, the apparent marineindustry consensus is that there is no painttechnology yet available that is comparable inperformance to TBT-based underwater hullcoatings. This has resulted in a significant

amount of R & D to find better solutions.Definitive results can only come fromoperational ship performance monitoring, asopposed to laboratory testing. The economicchoicebetween alternative paint systems is not astraightforward decision in todays climate, aseach paint manufacturer claims similar fuelsavings can be obtained by using their products.Clearly, an accurate hull monitoring capabilitythat measures actual performance differences toprove or disprove marketing claims, rather thanrelying on paint company-sponsored studies,

could be put to good use in support of shipoperators purchasing decisions.

Naval vs. Commercial Ship Operations

As noted in (Schultz et al. 2011), naval vesselsrepresent a unique challenge to paintmanufacturers because of the extent of their timespent pierside, compared to commercial ships.This is a particularly important operationalcharacteristic when considering paint containingno antifouling properties, such as FR or STC. Asdiscussed in this paper, viable solutions for fast-moving commercial vessels may not work wellfor naval vessels. The U.S. Navy continues toinvestigate antifouling coatings offeringeffective fouling control for ships operating inall geographic areas, with the goal of reducinghull cleaning and dry-docking frequency.

Environmental Regulatory Pressure

Ship hull fouling can impact the environment inseveral ways:

1)

extra power and fuel consumption areneeded to maintain ship service speeds,

thereby increasing green house gasemissions,

2)

periodic hull cleaning for fuel economycan pollute marine environments withtoxic paint residuals, and

3)

aquatic invasive species resident on aships fouled hull may be transportedfrom port to port, damagingecosystems and creating hazards forlivelihoods, human health, and localeconomies.

The IMO, among others, will continue to imposeboth direct and indirect regulatory pressure onship operators to keep their hulls clean. Ongoingglobal regulatory issues are covered in somedetail in (Hydrex 2010, 2011).

Purpose of Paper

Ship performance monitoring traditionally hasbeen a complex subject, requiring a deep anddiverse background in naval architecture, marine

engineering, mathematics, statistics, and morerecently, computer science. The subject has along and rich history, spanning the better part ofa century and leaving a wide trail of researchliterature. Surprisingly enough, shipperformance analysis still seems today to besomewhat of a black art, requiring highlyspecialized knowledge and skills.

The author spent nearly seven years researchingthe subject during the oil crisis of the 1970s(Logan 1980), and after pursuing other areaswas surprised to learn that practical techniquesfor assessing the performance of a ship at seahave changed relatively little during the past 35years. For example, using logbook data,recorded periodically by a ships crew, not onlycreates an added work burden, but moreimportantly creates a dependency on crewdiligence for data quality that, more often thannot, prevents high accuracy results from being


4/20


MACSEA Ltd. 4

obtained. This can mean the difference of savinga few percent in fuel or not. Another example isthe reliance on various resistance models (e.g.wind and wave effects), developed decades agousing model testing, to predict performancelosses for an altogether different full-scale ship!

The constraints on ship performance analysistechniques that were practical and useful 35years ago were generally related to data quality.With modern sensor and computer technologies,as well as the extensive automation systemsinstalled on todays vessels, these constraints nolonger apply. High speed data acquisition, low-cost computers, and advanced databasetechnology are now at the practitioners disposalto eliminate past barriers to achieving the highaccuracy ship performance models needed to

measure even small effects to save fuel.

This paper highlights an innovative technique ofusing a ships propeller, coupled with moderndata acquisition and database methods, toaccurately track power increase due to hullfouling.It is amusing to note that thisinnovation was developed in 1926! Contrastis made to more recent and complex resistance-based modeling. The propeller power absorptiontechnique is demonstrated through the analysisof two naval sisterships being used to evaluate

alternative hull coatings in an ongoing study forthe U.S. Navy Military Sealift Command.

SHIP PERFORMANCE

MONITORING

Factors Effecting Ship Performance

There are many ship and environmental factorswhich can influence the speed/power

performance of a ship. These include:

Ship factors:Draft, trim, steering/rudderactivity, operating transients (e.g. speeding up orslowing down), hull condition, propellercondition

Environmental factors:Currents, wind, waves,water density and temperature, water depth, airhumidity and temperature, barometric pressure

The effects of these factors are complex andhave been studied for decades by the

international maritime research community.Historically, mathematical models have beenderived to estimate the effects of each variable.Once developed, the models are then used tocorrect speed and/or power performance to somestandard baseline set of conditions (e.g. calmweather, specific draft/trim condition, etc.).Model development is an expensive, time-consuming effort, usually performed throughship model testing in a towing tank facility.Model development and validation through full-scale ship trials requires a large data set covering

a multitude of ship and environmentalconditions, which may take many years toaccumulate.

Traditional Approach

The technical literature abounds with referencesdescribing ship resistance modeling approachesto ship performance analysis. Detailedtheoretical foundations can be found in (Todd1967) and (Carlton 2008), as well as numeroustechnical papers from various conferences andsymposia in the interim years. Depending on thefocus of various authors, there appears to besignificant variation in the components of shipresistance included in the total ship resistanceequation.

RESISTANCE MODELING

The estimation of ship resistance is crucial indetermining required engine power and theselection of the correct propeller to move a ship

at its design speed. Ship powering prediction is awell-developed field and resistance modelingplays a central role.

The power required to overcome total resistanceis defined as:

(1)


5/20


MACSEA Ltd. 5

Where:

,

Total resistance is comprised of a number ofcomponents related to various sources ofresistance that have complex interactions witheach other.

The following components are commonlyconsidered to comprise total calm-waterresistance, RT(Todd 1967):

1)

frictional resistance2)

wave-making resistance3)

eddy resistance

4)

air resistance

Additional resistance will come into play when aship experiences heavy weather andconsiderable research has been performed toquantify these effects. (Carlton 2008) referencessome of these techniques, but in general, mostpractical methods for estimating weather-relatedresistance rely on model testing, either forderiving regression equations or empiricalcorrection factors.

RECENT WORK

One of the most recently published worksdescribing the practice of ship performancemonitoring can be found in (Aas-Hansen 2011).Although newly published, the book describesthe decades-old, traditional approach ofattempting to quantify the effects of hull andpropeller fouling by first removing or correctingfor all other factors that affect ship resistance.These factors include those previouslymentioned (i.e. wind, waves, etc.).

The book describes detailed mathematicalmodels founded on known principals of navalarchitecture and previous experimental work byseveral researchers, but lacks informationregarding the expected accuracy of shipperformance predictions based on thesetechniques.

In a recent study of the economic impact of hullfouling for the US Navy (Schultz et al. 2011),resistance models were used for predictions offull-scale ship resistance based on scale-modeltank test measurements and similarity lawanalysis (Schultz 2007).

PROBLEMS WITH TRADITIONAL

APPROACH

Because not all researchers include allcomponents of resistance in their work, thepower prediction results are inconsistent.Typically, a separate model is developed foreach resistance component and componentinteraction effects are largely ignored.

Resistance models developed by many

researchers remain unvalidated against actualships. The correlation allowance is essentially acorrection factor applied to align powerestimates from model tests with results fromship trials (Bose and Molloy 2009). Correlationallowances are sometimes considered asproprietary information by model basins; assuch, technical details of their derivation remainobscure. Correlation allowances vary with theextrapolation method used and are somewhatdependent on the tank facility at which testing isperformed.

Model-ship extrapolation relies heavily on thework of Froude from more than 100 years ago.FroudesLaw of Comparison was establishedexperimentally and is not a rigorous physicallaw.

(Bose and Molloy 2009) provide an examinationof the accuracy of frictional resistancecoefficient values. They stated that historicallythese values have been found from experimentalresults and that there is no absolutely accurate

way to separate the frictional resistance from thetotal resistance measured during a ship modeltest. It is therefore important to know the levelof uncertainty in the powering predictionprocess from these sources. Uncertainty in theextrapolation process stems not only from theresults of the model tests, but also in theassumptions made in formulating the method,the analysis methods employed to estimate


6/20


MACSEA Ltd. 6

parameters in the analysis, and in selection ofvariables.

(Pedersen and Larsen 2009) note the difficultiesof traditional ship performance analysistechniques requiring the estimation of a number

of unknown friction-related coefficients, oftenfrom limited model test or empirical full-scaletrials data. As an alternative to resistancemodeling, their research outlines a method ofusing artificial neural networks to predictpropulsive power from the same theoreticalvariables influencing ship resistance, such asship speed, relative wind speed and direction, airtemperature, and sea water temperature. Usingactual ship datasets segmented by draft and trimconditions, they report propulsion powerprediction accuracy of greater than 2.7%,

compared to range of 18-28% using traditionalresistance techniques (Pedersen and Larsen2009).

PROPELLER MODELS

Calibration of Propeller as Speed/Power

Measuring Device

Methods for using a ships propeller as a speed

or power measuring device were established atleast eighty years ago. A propeller calibrationprocedure using actual ship data results in aquantitative model specifically chosen torepresent clean hull and calm weatherconditions. Once this calibration has beenperformed, the resulting propeller model can beused to track speed and/or power changes overtime by continually comparing the current shipsperformance to the baseline clean hullperformance. Significant differences from thebaseline will be indicative of hull and/or

propeller degradation. The analytical methodsare primarily based on the power diagram workof Telfer (Telfer 1926, 1964).

TELFERS METHOD

(Telfer 1926, 1964) developed a simple yeteffective method for calibrating a shipspropeller as a power absorption dynamometer

over a range of slip conditions. He defined atorque constant, Qc, and developed a slipfunction to allow it to be used to monitor for hullfouling. The torque constant was defined as:

(2)

where:

He then defined a simple linear relationshipbetween Qcand apparent slip, Sa, as follows:

(3)

The calibration procedure essentially involvesdetermining the slope and intercept of equation(3) from ship trials, presumably duringacceptance trials or following hull and propellerconditioning in drydock.

As with any similar type of linear curve fit, thewider the variation of the independent variablein the test data set, the more accurately themodel can be fitted. Because slip may not varysubstantially during a sea trial, Telfer developedthe following expression for the torque constantat the 100% slip condition (denoted as theconstant C):

(4)

where:

The average values of the torque constant and

slip, , , were determined from sea trials,providing a second data point with which tocalculate the unknown coefficients in equation(3).


7/20


MACSEA Ltd. 7

Using the known data points, (1, C) and

( , ), and simple algebraic equations forstraight lines, the slope, b, can be determined asfollows:

(5)

The intercept can then be determined:

(6)

This completes the calibration for the clean-hull condition. Equation (3) can now be used tomonitor hull/propeller condition. There arevarious ways in which this can be accomplished.(Telfer 1964) chose to monitor the changes inthe slope, b, of the model, noting that b willincrease with hull fouling. A different methodwill be presented in the following.

Telfers work has not been highly publicizedover the years, although it has been the basis ofship performance monitoring by severalresearchers, including (Townsin, Moss et al.1975), (Bustard 1978), (Townsin, Byrne et al.1980), (Logan et al. 1980), and (Townsin andSvensen 1980).

MODIFIED APPROACH

The approach taken in this work more closelyfollows that of (Bustard 1978), although thesame principals as described by Telfer apply.Instead of developing the Qc-Sarelationship asTelfer does, Bustard develops an equivalentlinear relationship between SHP/n3and Va/n,where Vais the inflow velocity to the propeller.In order to do so, a suitable value for Taylorwake fraction, w, must be estimated fromperformance data, since:

(7)

The relationship between SHP/n3and Va/nderives from the well-known equations for thepropeller torque and advance ratio coefficients:

(8)

(9)

where:

SHP= shaft horsepowerV= ship speed through water

Kq= torque coefficient= density of seawatern= rotational speed

D= propeller diameterJ= advance ratiow= Taylor wake fraction

The propeller open water characteristics define alinear relationship betweenKqandJwithin thenormal working range of the propeller. Similar

to Telfer, this is essentially a torque-slipfunction and once known, can be used toestimate shaft power from rpm and speed or,alternately, to estimate speed from rpm and shaftpower (Logan et al.1980). Figure 1 illustratestypical propeller open water characteristiccurves. The typical operating range of {J, Kq}values is relatively small and their relationship islinear within this range.

Figure 1 - Typical Propeller Open Water

Characteristics RelatingKqandJ

SinceKqandJare linearly related within thenormal working range of the propeller, bycombining the constant terms relating toseawater density, propeller diameter, and wake


8/20


MACSEA Ltd. 8

fraction, the preceding equations (8) and (9)indicate that the power ratio defined by SHP/n3will be linearly related to the speed ratio definedas V/n.

The intent is to define a propeller power

absorption model for a standard set of baselineconditions, which will include clean-hull, calmweather, fixed draft and trim, and constantconditions for all other slip factors. In this case,wake fraction in equation (9) may be assumedconstant.

The essence of Telfers original work is thatshaft torque and rpm are related to propeller slip,which embodies speed through the water.

Factors that affect propeller slip include:

Draft Trim

Propeller pitch (for ships withcontrollable pitch propellers (CPP)

Current

Weather-induced ship motions

Rudder-induced ship motions

Operating transients (rapid power andspeed changes)

Hull and propeller fouling

As noted in (Bustard 1978), the propeller doesnot sense wind, waves, fouling, or any otherresistance. It does sense a reduction in the rate ofinflow. All factors above that influence slipalso impact the inflow water velocity to thepropeller. The ship performance problem is oneof separating out these individual contributors toslip (inflow) variation.

Regarding draft and trim, it is easy enough tocalibrate the propeller based on the mostcommon loading conditions and then use thetechnique only for those (baseline) conditions toassess hull condition. For a tanker, two separatecalibrations can be performed (e.g. loaded andballast conditions). For other vessels, it is easyenough to determine the most frequent loadingconditions from past voyage data and thencalibrate to those conditions. Note that thetechnique easily lends itself to determining therelative power absorption at different draft and

trim conditions as historical data is archivedcovering the full range of variation of eachparameter. This lends itself readily to trimoptimization applications. A similar approachcan also be taken to quantify performance lossesdue to heavy weather.

For the purposes of hull condition monitoring,the effects of current, weather-induced shipmotions, rudder-induced ship motions, andtransients can be eliminated through advanceddata filtering to exclude data containing theseeffects from the analysis. Baseline conditions areestablished to allow filtering of all data notmeeting the baseline criteria. For example, acomparison of GPS-derived speed over groundto speed log-derived speed through water willindicate the magnitude of current and a threshold

tolerance can be established. Heavy seas areusually accompanied by strong winds, so windspeed provides a commonly availablemeasurement by which to filter out a calm-weather data set to allow more accuratecalibration. For ships with fixed-pitchedpropellers, slip variation due to heavy weathermay actually help with the calibration, as it willprovide data with a wider range of slip variationfrom which to estimate the coefficients inequation (3) (Telfer 1964).

Hull and propeller fouling are the long-termeffects of interest. Presumably, the data used forcalibration represents clean-hull conditionsimmediately following drydock. So hull foulingeffects are not present at the time the calibrationis performed (although roughness due to thequality of paint application will have aninfluence). After such data filtering, all thatremains left to consider from the slipcontributors listed above is propeller pitch, aspertaining to CPP ships.

After filtering performance data to meet theestablished baselines discussed above, the mainsource of slip variation remaining in data forCPP ships will be propeller pitch. In this case,the torque-slip relationship of eq. (3) may berevised without losing meaning as follows:

(10)


9/20


MACSEA Ltd. 9

In this work, the term will be referred to as

thePower Ratio.

Furthermore, since the Power Ratio is equivalent

to the propeller torque coefficient, Kq, from eq.(8), and it is linearly related to the advance ratio,J, a Speed Ratiosimilar to (Bustard 1978) canalso be defined:

(11)

For the CPP type ship, equations (10) and (11)relate propeller shaft power, revolutions, andship speed through the water to the onlyremaining source of slip after data filtering tobaseline conditions, that being propeller pitch. Inthis work, it was found that the reciprocals of thespeed and power ratios yielded slightly moreaccurate estimates than the forms of equations(10) and (11) and were used without changingthe validity of the foregoing description.

High-Accuracy Performance Analysis

As simple as it may appear, the technique wasused to consistently predict ship power with lessthan 0.5% error, even from manually collected

ship logbook data (Telfer 1926). Telfer furtherindicated that similar levels of accuracy wereobtained for several ships under his study.

(Bustard 1978) also reported good accuracyusing his slightly modified technique in theanalysis of the tankerEsso Edinburghdata,showing an average absolute error of 1.0%across the first six voyages of a seven-monthdrydocking cycle. The technique readily showedthe onset of fouling during the sixth and seventhvoyages, after which drydocking was shown to

restore performance. Accuracy was determinedby comparing model-based SHP to thatmeasured via torquemeter.

Similar accuracy levels were found in this workand will be discussed further. As a preview, theaverage speed and shaft horsepower errors foundhere for one test ship were -.10% and -.13%,respectively, over a baseline data set of 3325

records. The average absolute errors (i.e.absolute value function removed negativevalues) were 1.8% for speed and 0.9% for shafthorsepower over the same 3325 record dataset.

Ship performance monitoring methods typically

have difficulty eliminating data scatter due tovarying ship and environmental conditions,hampering the ability to detect smallperformance losses early in the hull foulingevolution. Higher accuracy allows detection andquantification of hull-related losses at theearliest possible time, allowing improvedmaintenance planning and more fuel savings tobe achieved. As pointed out in (Schultz et al.2011), even small percentage losses can result inlarge fuel penalties.

Separation of Hull and Propeller Effects

The separation of hull fouling from propellerfouling is generally not possible without bothshaft torquemeter and shaft thrustmetermeasurements. Torquemeters are in commonuse, but thrustmeters are rarely installed.

In an attempt to develop a method by which hulland propeller effects could be separated withoutrelying on a thrustmeter, (Wan, Nishikawa, andUchida 2002) developed an experimentalmethod by which the relationship between thetorque coefficient, KQ, and the thrust coefficient,KT, was determined experimentally throughmodel testing with varying degrees of propellerroughness.

By knowing how the KQ-KTrelationshipchanges with fouling (i.e. surface roughness),KQwas calculated from onboard torquemeasurements (Uchida and Nishikawa 2005). KTwas then estimated from the KQ-KTrelationshipwithout relying on thrust measurements. Uchidafound that increased fouling will generally causeKQto increase and KT(and propeller efficiency)to decrease.

For a given propeller installed on a ship,practical considerations will preventexperimental determination of surface roughnesseffects as outlined in (Wan, Nishikawa, and


10/20


MACSEA Ltd. 10

Uchida 2002). Hence, for the work described inthis paper, as well as in most other reported shipperformance analysis techniques, only thecombined effects of hull and propeller foulingcan be determined. Fortunately, it is thecombined effect on fuel economy that most ship

operators are concerned with, hence the rare useof thrustmeters on ships. Most prudent shipoperators have adopted a time-based propellermaintenance approach relying on regular diverinspections and cleanings.

The remainder of this paper describes theapplication of the techniques just described in anongoing study to evaluate the relativeeffectiveness of alternative hull coatings on twoNavy sisterships operated by the Military SealiftCommand (MSC).

HULL PAINT STUDY

In 2009, MSC initiated a comparative study ofthe cost benefits of foul release paint over thebiocide ablative paint traditionally used on MSCvessels.

Two identically designed, 667 foot fleet oilers,the USNS Kanawha (T-AO 196) and the USNSBig Horn (T-AO 198), were selected as the test

ships based on their drydocking schedules.

The Kanawha was coated with a biocide-free,fluoropolymer foul-release coating during herSummer 2009 drydocking. The Big Horn waspainted with a tin-free, self-polishing, biocideanti-fouling coating during Fall 2009. Both shipsare outfitted with dual CPPs, but of slightlydifferent designs, which influences theirrespective baseline speed/power performances.

Automated data acquisition systems were

installed on both ships to monitor hullperformance throughout the study period. Datawas regularly transmitted ashore for analysis.The Kanawhas propeller models werecalibrated from a 3326 record data set collectedduring the September-November 2009 timeperiod immediately following drydock.

Big Horns propeller models were calibratedfrom a much smaller dataset (392 records)collected during December 2009 throughFebruary 2010 time period immediatelyfollowing her drydock. The smaller dataset isattributed to the limited number of days Big

Horn operated at sea during this time period.

Data Management

ONBOARD DATA ACQUISITION

Automated data acquisition systems wereinstalled on both test ships during their lastdrydocking periods in the fall/winter of 2009.No additional effort on the part of the ship crewswas required for collecting the data needed forthis study. This was accomplished by leveragingtwo existing shipboard systems, the DEXTERShip Health Monitoring System (MACSEA2011) and MSCs Shipboard AutomatedMaintenance Management (SAMM) system,both of which were already installed on the testships. The DEXTER system acquires real-timemachinery and navigational performance data,while SAMM records periodic manual entries oftypical logbook data, such as drafts, sea state,etc. SAMM also manages periodic datareplication to a shore side data center.

DATA ELEMENTS

Table 1 lists the salient data items for this study.All data are timestamped and archived once perminute.

Table 1 Salient Ship Performance Data

Description Source Units

Ship Speed Over Ground GPS KTS

Ship Heading - True GPS Degrees

Ship Heading - Magnetic GPS Degrees

Ship Speed Through Water Speed Log KTS

Wind Angle Anemometer Degrees

Wind Speed Anemometer (K/M/N)

SHP (Port/Stbd) Torque meter SHP

Shaft Torque (P/S) Torque meter FT-LBS

Shaft RPM (P/S) MCS RPM

Propeller Pitch (P/S) MCS Degrees

Engine BHP (P/S) MCS BHP

Engine RPM (P/S) MCS RPM

Throttle (P/S) MCS %

Engine Fuel Consumed (P/S) MCS Gal/HR

Forward Draft SAMM Feet

Aft Draft SAMM Feet

Sea State SAMM Beaufort #

Sea Direction SAMM Degrees


11/20


MACSEA Ltd. 11

STANDARD BASELINE CONDITIONS

As previously discussed, there are many shipand environmental factors which influence thespeed/power performance of a ship. Theseinclude:

Ship factors:Draft, trim, steering/rudderactivity, operating transients (e.g. speeding up orslowing down), hull condition, propellercondition

Environmental factors:Currents, wind, waves,water density and temperature, water depth, airhumidity and temperature, barometric pressure

Within the context of this study, it was desirableto eliminate the effects of all major factors

except the combined hull/propeller effects. Withperformance data being recorded once perminute, the abundance of data allows the use ofdatabase filtering to remove records that do notmeet standard baseline conditions defined intable 2.

Table 2 Baseline Conditions for Ship

Performance Analysis

Draft and Trim

Baseline draft and trim conditions wereestablished on the basis of values at which theship normally operates the majority of the time.

Speed Through The Water

Only data records for which speed through thewater was 10 knots or greater were consideredfor analysis. This speed also happens to be thelower limit at which fouling release is designedto occur for this type of hull coating.

Eliminating Ocean Current Effects

The effects of ocean currents, weather, and tosome extent, rudder activity, can be eliminatedby analyzing only data in which speed over theground (SOG) and speed through the water

(STW) closely match. The presumption is thatwithout significant influence of theseperturbations, SOG and STW will be roughlythe same, as the ship traverses a relativelystraight course. The following speed differencemetric was used as a data filter:

(12)

Any data exceeding 3% agreement betweenSTW and SOG was filtered out from theanalysis.

Eliminating Wind and Wave Effects

Wind resistance and heavy weather/wave effectshave a significant impact on ship performance.Within the data items listed in table 1, there areseveral parameters available to filter out heavyweather data:

Sea state This is a manual entry in SAMMlogbook, entered as a Beaufort Number, assubjectively determined by the crew. Only datarecords for which sea state was Beaufort 3 orless were considered for analysis.

Wind speed Only data records for which windspeed was 15 knots or less were considered foranalysis.

Transient Operating Conditions

The propeller models discussed previously aredesigned for use in steady state operations.Hence, the models should not be used duringsudden changes in ship speed or RPM during

course changes or maneuvering situations whenslip variation occurs. With data being recordedonce per minute, consecutive records wereassessed to determine the Rate of Change (ROC)for both speed through the water (STW) andmean propeller RPM. Only data records withSTW ROC of .25 knots or less were consideredfor analysis. Only data records with mean shaft

Parameter Baseline Condition

Mean Draft [32, 35] feet

Trim [1, 3] feet

Speed through water 10 knots

Speeddiff= Water - Ground Speed 3%Sea State 3 (Beaufort)

Wind Speed 15 knots

STW Rate of Change .25 knots/minute

Mean Shaft RPM*Rate of Change .5 rpm

Total SHP (both shafts) [5000, 11,000] SHP

Port/Stbd Pitch Difference .3 feet

Port/Stdb RPM Difference .5 rpm


12/20


MACSEA Ltd. 12

RPM ROC of .5 revolutions or less wereconsidered for analysis.

Shaft Horsepower

The torque meters installed on each propeller

shaft provide an indication of shaft horsepower(SHP). Total SHP was calculated as the sum ofthe port and starboard SHP readings. In order toeliminate extreme power conditions whichinfluence accuracy, only data records for whichtotal power was in the [5000, 11,000] rangewere considered for analysis.

Synchronized Propeller Operation

The propeller models assume equal contributionof the port and starboard propellers and basicallytreat the combined effect of the two as a single

propeller. The ship generally operates with equalpitch and revolutions for the two propellers,however; this is not always the case. Since theeffects of unequal settings are unknown, a datafilter was established to ensure equal pitch andRPM readings on the two shafts. Only datarecords for which the difference in pitch settingsbetween the port and starboard shafts was .3 feetor less were considered for analysis. In addition,the difference in RPM readings between the portand starboard shafts was required to be within .5revolutions.

Propeller Calibration Models

The Kanawhas propeller models werecalibrated from a 3326 record data set collectedduring the September-November 2009 timeperiod immediately following drydock. Due tolimited ship operations, this time period wasnecessary to establish an adequate range of pitchvariation in the filtered data set used for thecalibration.

As previously noted, the reciprocals of the speedand power ratios (i.e. n/Vand n3/P) providedslightly higher accuracy, so they were used asthe basis for model development. Figure 2shows a plot of the power ratio versus pitch forthe Kanawha baseline dataset, filtered to thestandard conditions listed in table 2, along withits linear regression model. The power ratio is

highly correlated with pitch, with the linearmodel accounting for over 97% of the variancein the data. Figure 3 shows a plot of the speedratio versus pitch for the Kanawha baselinedataset, filtered to the standard conditions listedin table 2. The linear regression model is a good

fit, accounting for approximately 95% of thevariance in the data

Similar high correlation was also found for thepropeller models developed for the Big Horn.

Key Performance Indicators (KPIs)

The calibrated propeller relationships are used toestimate clean-hull speed through the water andshaft horsepower over time. By comparing these

estimates to actual measured ship values, theKPIs of speed loss and power increase can bemonitored, providing a clear signal of hullfouling. Power increases attributable to foulingcan be readily converted into fuel and emissionincreases and tracked as well.

INCREASED POWER ABSORPTION

The power ratio, Pratio, is linearly dependent onpropeller pitch under the established baselineconditions. The derived model, representative ofclean hull, baseline conditions, can be used toestimate SHP from measured RPM and pitch asfollows:

(12)

where:

Pratio= c1+ c2*H (model)n= mean shaft RPMc1, c2= model coefficients

A deviation metric is tracked over time to

monitor power increase:

Power(t) = SHP(t) P(t) (13)

As the hull/propeller condition degrades,

Power(t)will increase.


13/20


MACSEA Ltd. 13

Figure 2 Pratio Propeller Calibration Model (Kanawha)

Figure 3 Vratio Propeller Calibration Model (Kanawha)

R = 0.974

0

20

40

60

80

100

120

140

16 17 18 19 20 21 22 23

Pratio(N^3/SHP)

Pitch (ft)

R = 0.949

0

2

4

6

8

10

12

16 17 18 19 20 21 22 23

Vratio(N/V)

Pitch (ft)


14/20


15/20


MACSEA Ltd. 15

performance data. An underwater hull cleaningwas performed in late January 2001 that restoredKanawha performance back to baselineconditions as shown in figures 6 and 7.

Figure 6 - Power versus Time

Figure 7 - Speed Loss versus Time

Figure 8 compares Kanawhas Speed-Powerperformance before and after hull fouling tookplace. From December 2009 through June 2010,her performance remains constant, following awell-defined speed-power curve. Subsequent to

the fouling that occurred during the summer of2010, a marked upward curve shift occurred, asevident by the August-September 2010 data.

Figure 6 showed that Kanawha incurred a powerpenalty due to hull fouling during the September2010 through early January 2011 time period.

Figure 8 - Kanawha Speed-Power Performance

As a fleet oiler, the Kanawha is required tomaintain fixed speeds during underway fuelreplenishment operations. The data shows that in

order to make up for the approximate one knotspeed loss due to hull fouling during September2010 operations, the ship had to generate 2700additional shaft horsepower (see figures 6 and7). As a percentage of normal operating powerlevels (i.e. clean-hull), 2700 SHP was about35% more power than what would be needed ifthe ship had a clean hull, as the September 2010data in figure 9 indicates.

Figure 9 - Kanawha Power Penalty Sept 10

Figure 10 expresses the average monthlyincreased power as a percentage of clean-hullpower from the onset of hull fouling inSeptember 2010 until just before the underwatercleaning was performed in January 2011.

-2000

-1000

0

1000

2000

3000

4000

5000

PowerChange(SHP)

KANAW HA BIG H ORN

Kanawha Hull Cleaning

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

16,000

SHP

September 2010 Data Sample

Measured SHP Clean Hull SHP

35% Increased Energy Due to

Hull Fouling-1

-0.5

0

0.5

1

1.5

2

SpeedChange(kts)

KANAWHA BIG HORN

Kanawha Hull Cleaning

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

16,000

0 2 4 6 8 10 12 14 16 18

SHP

Speed Through Water (kts)

Dec-09 Jun-10 Aug-Sep 2010

Dec '09 - June '10

Aug-Sept'10


16/20


MACSEA Ltd. 16

Figure 10 Historical Kanawha Power Penalty

Figure 10 indicates that a significant amount ofadditional energy was consumed as a result of

hull fouling. When converted into equivalentfuel cost, the savings available from a proactivehull and propeller maintenance strategy aresubstantial, particularly when considering alarge fleet. In addition, since burning a gallon ofdiesel fuel generates about 22 pounds of CO2emissions (EPA 2011), potential fuel savingsdirectly translate into quantifiable emissionsreductions.

By using high accuracy propeller powerabsorption models, hull and propeller fouling

can be detected early in the fouling evolution.Added energy costs can be quantified such thatprudent maintenance decisions can limit excessfuel consumption within the constraints ofdeployment requirements.

SUMMARY & CONCLUSIONS

The Cost of Hull and Propeller Fouling

Ship operators realize that hull fouling causesspeed loss and increased fuel consumption.There is also growing environmental concernand international regulation relating to greenhouse gas emissions and the carriage of aquaticinvasive species on fouled ship hulls.

Ship performance losses due to hull andpropeller fouling are difficult to quantify

because changing ship and environmentalconditions generate performance variations thatmakes separation of hull and propeller effects adifficult task.

The Navy has recognized the importance of hull

condition monitoring in their energyconservation efforts and has stated hull and/orpropeller fouling are the largest cause of excessfuel consumption. Depending on foulingseverity, fleet-wide hull fouling-related costshave been estimated to be somewhere between$180M - $540M per year. Spending a fraction ofthis for improved monitoring solutions willcertainly offer a rapid return on investment.

Hull Coating Solutions

Marine growth can be retarded through the useof effective anti-fouling paint, such as tributyl-tin (TBT) self-polishing co-polymers, however;due to TBTs negative environmental sideeffects, the International Maritime Organization(IMO) imposed a ban on TBT paint, leadingpaint manufacturers to develop non-toxicalternative paint systems. Amongst these paintalternatives are low-copper and copper-freeablative antifouling (AF), foul release (FR)coatings, and surface treated coatings (STC).Unfortunately, marine industry consensus is thatthere is no paint technology yet available that iscomparable TBTs performance and asignificant amount of industry R & D isunderway to find better solutions. One thingappears certain; the IMO and other national andstate organizations will continue to increaseregulatory pressure on ship operators to keeptheir hulls clean.

Market Need for Independent and

Transparent Hull Performance

Monitoring

The choicebetween alternative hull coatings isnot an easy one, considering that shipperformance is difficult to measure and that themajor paint companies all claim similar fuelsavings by using their products. Clearly, anaccurate hull monitoring capability thatmeasures actual ship performance, rather than

35

31

25 25

0

5

10

15

20

25

30

35

40

45

50

Sep '10 Oct '10 Nov '10 Dec '10 Jan '11

Power(

%)

Increased Power Absorption Due to Hull Fouling

InPort


17/20


MACSEA Ltd. 17

relying on paint company-sponsored studies,could be put to good use in proving ordisproving those marketing claims.

Naval vessels represent a unique challenge topaint manufacturers because of the extent of

their time spent in port, compared to commercialships. Viable coating solutions for fast-movingcommercial vessels may not work well for navalvessels.

Modern Ship Performance Monitoring

Alternatives

Ship performance monitoring traditionally hasbeen a complex subject, requiring knowledge ofnaval architecture, marine engineering,

mathematics, statistics, and more recently,computer science. The practice still seems to besomewhat of a black art, requiring highlyspecialized knowledge and skills, as it has fordecades. A need exists within the maritime andnaval communities for a transparent andindependent methodology for measuring actualship performance over time, indexed to a clean-hull baseline.

In the past, the major constraints on shipperformance analysis techniques were generally

related to data quality. With the modernautomation systems installed on todays vessels,these constraints no longer apply. High speeddata acquisition, low-cost computers, andadvanced database technology are now at thepractitioners disposal to eliminate past barriersto accurately measuring hull and propellerperformance.

RESISTANCE MODELING

Ship powering prediction is a well-developedfield and resistance modeling has played acentral role. The estimation of ship resistance iscrucial in determining required engine powerand the selection of the correct propeller tomove a ship at its design speed. Total resistanceis comprised of a number of components relatedto various sources of resistance that havecomplex interactions with each other, includingfrictional resistance, wave-making resistance,

eddy resistance, air resistance, and heavyweather effects. Depending on the focus ofvarious researchers, there has been significantvariation in the components of ship resistanceincluded in the total ship resistance equation,making power prediction results inconsistent. In

addition, a separate model is typically developedfor each resistance component and componentinteraction effects are largely ignored.Resistance models developed by manyresearchers remain unvalidated against actualships. The correlation allowance is essentially acorrection factor applied to align model-basedpower estimates with results from ship trials.Beyond this, model development is anexpensive, time-consuming effort, usuallyperformed through ship model testing in atowing tank facility.

PROPELLER POWER ABSORPTION

MODELS

Methods for using a ships propeller as a speedor power measuring device were established atleast eighty years ago, but were not highlypublicized even though several ship performanceresearchers used them in the 1980 timeframe.The unique characteristics of these methods aretheir simplicity, high accuracy, and their use ofactual ship performance data, instead of scale

models. A propeller calibration procedure resultsin a quantitative model representing clean hulland calm weather conditions. The propellermodel can be used to track speed and/or powerchanges over time by continually comparing thecurrent ships performance to the baseline cleanhull performance using KPIs that are easilyunderstood by both ship crews and theirmanagement counterparts.

From a theoretical standpoint, the propeller openwater characteristics define a linear relationship

between the torque coefficient,Kq, and theadvance ratio,J, within the normal workingrange of the propeller. This is essentially atorque-slip function and once known, can beused to estimate shaft power from rpm andspeed or, alternately, to estimate speed from rpmand shaft power. The ship performance problemis one of separating out the individualcontributors to slip or water inflow velocity. In


18/20


MACSEA Ltd. 18

this regards, a propeller does not sense shipresistance caused by wind, waves, fouling, etc. Itonly senses a reduction in the rate of waterinflow. All factors that influence slip also impactthe inflow water velocity to the propeller and themajority of these can be filtered out of analysis

datasets using modern database technology. Thecombined effects of hull and propeller foulingwill be the only ones remaining after thefiltering process is performed.

Propeller models as described herein can also beused to determine the relative power absorptionat different draft and trim conditions such thatoptimal trim for any ship loading condition canbe determined. A similar approach can also betaken to quantify performance losses due toheavy weather which has direct use in weather

routing applications. The predictive accuracyachievable with this technique has beenconsistently reported by several researchers tobe in the 0.5% to 1.0% range, significantlyhigher than that reported for resistance models.Higher accuracy allows detection andquantification of hull-related losses at theearliest possible time, allowing improvedmaintenance planning and more fuel savings tobe achieved.

It should be noted that the separation of hull

fouling from propeller fouling is generally notpossible without both shaft torquemeter andshaft thrustmeter measurements. However, ifthrust measurements are available, the propellermodel has direct applicability for propellerfouling, as well.

Demonstration Through Navy Hull Paint

Study

The propeller power absorption technique hasrecently been demonstrated during an MSC-initiated comparative study of foul releaseversus ablative antifouling paint performance.Two identically designed fleet oilers had theirhulls painted during drydock and were outfittedwith automatic data acquisition systems. Theirperformance has been monitored for the past 18months. Power and speed predictions based onthe calibrated propeller relationships of each

ship were used to track speed loss and powerincrease based on actual ship performancemeasurements.

The onset of Kanawha fouling on her foul-release coated hull was readily apparent in

September 2010 after the ship sat at birththroughout most of that summer. At that time, anadditional 2700 SHP was needed to maintainservice speeds, as compared to clean-hullpowering conditions. Kanawhas powerperformance for ensuing months was alsosignificantly higher than clean-hull conditions.An underwater hull inspection performed inearly November 2010 verified the presence ofhull fouling evident in Kanawhas performancedata and MSC immediately planned anunderwater hull cleaning, which was

subsequently performed in late January 2011.Subsequent analysis clearly showed thatKanawhas performance was restored back tobaseline conditions.

Big Horns performance data indicated that herhull remained clean after over one year with theablative antifouling coating.

Final Conclusions

Rising fuel costs, hull maintenance expenses,and mounting environmental regulations makehull condition monitoring a crucial tool forprudent ship operators to eliminate energy wastedue to hull fouling, reduce carbon emissions,and eliminate the carriage of invasive speciesbetween ports.

Propeller models offer many advantages to Navydecision makers responsible for hull/propellermaintenance and hull paint selection:

The techniques are well-established (albeitnot well-publicized)

They provide direct measurements ofperformance loss on actual ships in terms ofeasily understood KPIs

They use sensor data that is available onmost ships

Propeller power absorption calibration,while based on propeller/propulsion theory,


19/20


20/20


MACSEA Ltd. 20

Symposium on Shipboard Energy Conservation80, SNAME, New York, September 1980.

Uchida, M. and E. Nishikawa, AdvancedEducation & Research on Marine Propulsion New Method for Analyzing Propulsion

Performance in Service, Proceedings ofInternational Association of MaritimeUniversities, WIT Press, Sixth Annual GeneralAssembly and Conference, Malm, Sweden,October 2005.

Wan, B., E. Nishikawa, and M. Uchida, TheExperimental and Numerical Calculation ofPropeller Performance with Surface RoughnessEffects, Journal of the Kansai Society of NavalArchitects, Japan, No. 238, pp 49-54, 2002.

Yebra, D.M., and P. Catala, RedefiningAntifouling Coating Technology Part 1,Materials Performance, NACE International,Vol. 50, No. 4, April 2011.

Yebra, D.M., S. Kiil, and K. Dam-Johansen,Antifouling Technology Past, Present, andFuture Steps Towards Efficient andEnvironmentally Friendly AntifoulingCoatings, Progress in Organic Coatings 50,Elsevier B.V., 2004.

ACKNOWLEDGEMENTS

The author is grateful to MSC for their supportof the hull paint study and their consideration ofthe techniques presented in this paper.

Kevin Loganis President of MACSEA Ltd., a29-year old company specializing in ship healthmonitoring solutions, including hull andmachinery Condition-Based Maintenance

systems.Mr. Logan holds a BA in Mathematics,with graduate studies in operations research andsystems modeling. He has performed decades ofapplied research in ship performancemonitoring, machinery diagnostics, and artificialintelligence technologies. He has publishednumerous technical papers and is a member ofASNE, SNAME, IEEE, the Society for

Machinery Failure Prevention Technology, andthe International Neural Network Society.

using a ships propeller for hull condition monitoring mkt1

Documents