otc_1999-insulationtestmethod_tno

Copyright 1999, Offshore Technology Conference

This paper was prepared for presentation at the 1999 Offshore Technology Conference held inHouston, Texas, 3–6 May 1999.

This paper was selected for presentation by the OTC Program Committee following review ofinformation contained in an abstract submitted by the author(s). Contents of the paper, aspresented, have not been reviewed by the Offshore Technology Conference and are subject tocorrection by the author(s). The material, as presented, does not necessarily reflect anyposition of the Offshore Technology Conference or its officers. Electronic reproduction,distribution, or storage of any part of this paper for commercial purposes without the writtenconsent of the Offshore Technology Conference is prohibited. Permission to reproduce in printis restricted to an abstract of not more than 300 words; illustrations may not be copied. Theabstract must contain conspicuous acknowledgment of where and by whom the paper waspresented.

AbstractThis paper describes a long term programme of researchwhich has resulted in the development of a simulated servicetest facility incorporating a direct measurement system whichcan be used to confirm the heat transfer characteristics ofsubsea factory applied and field joint insulation coatingsystems produced under normal production conditions. Thesimulated service test was originally developed by ShellResearch at their Thornton Research Centre in the UK in thelate 1980’s. To date these test facilities and accompanyingprocedures have been successfully used by industry to studythe behaviour of thermal insulation coating systems on lengthsof steel pipe internally heated to temperatures up to 140 °Cand when subjected externally to hydrostatic pressures up tothe equivalent of water depths of 1450m under real timeconditions. It is already becoming apparent that the specialisedform of the heat flux measuring device which has beendeveloped during the course of this research programme canbe used in other applications.

IntroductionTraditionally externally applied coating systems for offshorepipelines, subsea flowlines, steel catenary risers, etc. havebeen used mainly to protect the steel pipe from attack by thesurrounding seawater environment. Normally the majority ofthe length of such pipelines/flowlines/risers are protected by acoating system applied under factory controlled conditionswhereas the field joint areas within the factory applied coatingsystem are protected by a coating system, applied under fieldconditions, which may comprise of entirely different types ofmaterial.

Subsea developments have placed a requirement on the

development of coating systems for flowlines which willprovide them with protection against corrosion attack, heatloss, instability under hydrodynamic loadings, mechanicaldamage from third party activities, etc., often in combination.These coating systems are required to operate in situations inwhich the produced fluid temperatures are increasing in excessof 140°C and/or in water depths in excess of 1500m.

Solutions to satisfy these often conflicting demands are notreadily available so new materials need to be either identifiedor developed and built into cost effective thermal insulationcoating systems. This requirement applies not only to factoryapplied coating systems but equally to field joint coatingsystems. The requirement to continually develop coatingsystems for pipelines/flowlines/risers has emphasised the needto have in place approaches, including test procedures, whichensure acceptable levels of quality control and can also beused to establish whether in fact they will be fit-for-purpose.

The significance of the difference between quality controltesting where the constituent parts of an insulation coatingsystem are tested under standardised conditions and fitness-for-purpose tests where either the constituent parts of a coatingsystem or the complete coating system are tested under in-service conditions must be emphasised. The acceptancecriteria to be adopted in these test programmes should reflectthe composite nature of these coating systems.

It is therefore important to be able to predict the likelyextent of the damage or disruption to flowline coating systemsresulting from the mechanical loadings, etc. experienced whenpipelines/flowlines/risers are being installed and alsothroughout their design life when they are in-service on theseabed.

However, not all coating systems are suitable for all fields,and a proper understanding of the requirements and thesystems is critical.

A comprehensive range of test arrangements andcorresponding test procedures have been developed, withsupport from mainly the international oil and gas industry, atHeriot-Watt University to allow the fitness-for-purpose ofcoating systems to be studied in detail at full scale. Many ofthese test procedures now form part of the qualificationrequirements for coating systems for subsea developmentsinternationally.

This paper describes a long term programme of research

OTC 11040

A Direct Measurement System to obtain the Thermal Conductivity of Pipeline InsulationCoating Systems under Simulated Service ConditionsD. Haldane, Heriot-Watt University, F. van der Graaf and A. M. Lankhorst, TNO Institute of Applied Physics

2 D. HALDANE, F. VAN DER GRAAF, A.M. LANKHORST OTC 11040

which has resulted in improvements in the simulated servicetest which is normally used to confirm the heat transfercharacteristics of factory applied and field joint insulationcoating systems produced under normal production conditions.

Numerical Model of Simulated Service TestThe first simulated service test arrangment was developed byShell Research at their Thornton Research Centre in the UK inthe late 1980’s. In 1990 when Shell Research decided toeffectively withdraw from this area of research a nominallyidentical test arrangement was established at Heriot-WattUniversity with the support of the oil and gas industry. Theoperation of the test arrangement was mainly dependent on theuse of heat flux sensors (HFS’s) to measure the flow of heatthrough the insulation coating system on the test pipe whensubjected to simulated service conditions. These heat fluxsensors, which were originally designed and manufactured byTNO-TPD in the Netherlands, were supplied in the form of arigid epoxy resin flat disc approximately 30 mm in diameterand 3.5 mm thick with a centrally positioned 10mm diametersensitive area. It became apparent that the difficultiesexperienced in mounting the rigid flat form of such a heat fluxsensor on the external curved surface of the insulation coatingsystem on a test pipe were contributing to the errors whichwere being found in the results from tests conducted oninsulation coating materials with a known thermalconductivity.

In order to develop an in-depth understanding of thebehaviour of such a test arrangement when fully operationalnumerical simulations of the heat transfer and water flow inthe simulated service test arrangement were carried out usingthe Computational Fluid Dynamics (CFD) model TNO-WISH3D.The CFD-model enabled the surface temperature and heat fluxdistributions to be obtained along a length of insulated steelpipe immersed in pressurised water. Particular attention wasgiven to studying the following:1. The magnitude of the errors in the measured heat flux undertest conditions.2. The influence of the physical presence of the HFS on theresulting heat flux measurements.3. The influence the HFS may have on the thermally inducedwater circulation within the test arrangement and thesignificance of any resulting error in the measured heat flux.Numerical Model. The general purpose finite-volumePatankar1-based CFD-model, TNO-WISH3D, which was usedwas capable of simulating flow, heat transfer and transport ofmass species in enclosures. The simulation can includeradiative heat transfer and/or combustion. Flows with steadyor transient boundary conditions or sources can also behandled. The code uses a 2D or 3D Cartesian grid or a 2Dcylindrical grid for the geometrical description of the testarrangement. The fluid domain may contain obstacles and inthese obstacles heat conduction can be taken into account(conjugate heat transfer).Problem definition. In the present simulations the testarrangement was assumed to be axially symmetric. Thetemperature distribution in the steel pipe, the steel blanking

plate, the insulation coating system, the HFS and the waterinside the pressure vessel was determined. Moreover, thewater velocity field due to natural convection (flow inducedby density differences caused by a temperature gradient)inside the pressure vessel was calculated. The fluid velocitiesin the pressure vessel were found to be very low. Therefore,the fluid flow was not in the turbulent regime, thus allowinglaminar flow calculations. Buoyancy effects were accountedfor by prescribing a linear temperature dependent relation forthe density of water (see Table 1). For decreasing temperaturethe density of water increases leading to flow in the directionof the gravitational force and vice versa for increasingtemperature. The temperature of the water within the pressurevessel normally remains above the density maximum point (4oC).Cases analysed. Two cases were considered both included alength of steel pipe complete with an externally applied layerof insulation coating material:1. No HFS was present on the external surface of theinsulation coating system.2. A HFS was mounted on the external surface of theinsulation coating system.The first case was included as a basis of comparison for theresults obtained from the second case thus enabling theinfluence of the presence of the HFS to be established. Theresults from the first case were also compared with the resultsobtained using published heat transfer data.Boundary conditions and material properties. One of theboundary conditions used was the temperature of the oilcirculating around the internal surface of the steel pipe. Thiswas assumed to be constant at 104 oC (considered uniform onthe internal surface of the steel pipe). The temperature ofcooling water was also assumed constant at 8 oC (considereduniform on the external surface of the pressure vessel). Thetemperature of the blind flange of the pressure vessel wasfound from laboratory based tests and was defined in themodel as being 48 oC. The material and fluid properties usedin the simulation are given in Table 1.Geometrical description and numerical discretisation. Thegeometry of the computational domain and the numericalmesh is given in Fig. 1. The mesh was locally refined forenhanced resolution near the HFS, the interfaces and in thosepositions where large velocity or temperature gradients wereexpected. The same mesh was used for both cases examined inthe numerical simulations. The HFS was positioned on theexternal surface of the layer of insulation coating material midway along its length. Typical dimensions of the coated testpipe and the HFS are given in Table 2.Results. The simulated contour plot of the temperaturedistribution for the case with the surface-mounted HFS isgiven in Fig. 2 (temperature scale from 0 - 100 oC). Thermalstratification occurs close to the joint between the blind flangeof the pressure vessel and the coated test pipe.

The contour plot for the absolute velocity distribution, dueto natural convection, for the case where a surface-mountedHFS is present is shown in Fig. 3. The absolute velocity(defined as the square root of the sum of the quadratic values

OTC 11040A DIRECT MEASUREMENT SYSTEM TO OBTAIN THE THERMAL CONDUCTIVITY OF PIPELINE INSULATION COATING SYSTEMSUNDER SIMULATED SERVICE CONDITIONS 3

of both velocity components) is scaled from 0 - 5 mm/s. Notethe disturbance of the boundary layer due to the presence ofthe HFS.

Fig. 4 presents a detailed view near the HFS showingvelocities in the boundary layer and the disturbance in the flowdue to the presence of the HFS. The length of the referencevector is 10 mm/s. The size of each vector is proportional tothe fluid velocity. The direction of the vector corresponds withthe direction of the flow. The velocities are of the order of 10mm/s or lower. Along the vertical surfaces, natural convectionboundary layers occur. As the velocity vectors are plotted onthe grid points of the non-linear numerical mesh, clusters ofvectors occur in regions where the grid is refined.Measurement errors. The most significant results obtainedfrom the simulations are summarised in Table 3. Theundisturbed temperature and heat flux values are those shownfor the case where no HFS was present.

When a HFS is introduced the changes in both thetemperature at the HFS/insulation interface and the heat fluxare significant, compared to the base case where no HFS ispresent. However, when Equation (1) is used to determine thethermal conductivity (k-value) of the insulation coatingmaterial, the accuracy of the approach is very high (of theorder of 0.1%).Numerical simulation versus analytical approach. Theanalytical solution for a one-dimensional conduction problemin a composite cylindrical structure is given by Bird et al.2.The (modified) expression for the radial heat flux evaluated onthe insulation surface r = r2 (for explanation of variables seeTable 2) is given by Equation (1):

Q" = (Ts-Tc)/[r2{ ln(r1/r0)/kS + ln(r2/r1)/kINS + 1/r2h }]….(1)

In this equation (as well as in the simulations) the heat transfercoefficient at the internal surface of the steel pipe wasassumed to be very large. This is true as the recordeddifferences in the temperature of the mineral oil flowing intoand out of the test pipe was small (0.3 oC). This equation infact gives the heat transfer coefficient from the surface of theHFS exposed to the surrounding cooling media circulating at atemperature of 8 oC. The resulting value for this heat transfercoefficient obtained from the numerical simulation was foundto be approximately 28 W/m2.K. This results in a value for theheat flux Q" = 198.4 W/m2 which almost exactly agrees withthe value found in the numerical simulation (see Table 3).

In the subsequent work use has been made of the value of hobtained from the simulations. The value of Q" however isrelatively insensitive to the value of h, if h is large. In thepresent situation h is large compared to the heat resistance ofthe insulation coating material. Thus, if h was 100% largerthan the value obtained from the numerical model, Q" wouldbe only 3.8% higher, leading to an uncertainty in the value ofkINS of 3.8%.Conclusions from numerical simulations. The conclusionswhich resulted from the numerical simulations were asfollows:

1. The results obtained from the numerical model were ingood agreement with those obtained using the analyticalapproach and the predicted results were in excellent agreementwith the results obtained from the laboratory basedinvestigations.2. The k-value of the insulation material can be determinedwith very high degree of accuracy if the heat flux and thetemperature of the HFS/insulation interface are measuredaccurately.3. The HFS should be as thin as possible. These simulationshave been carried out assuming a 3.5mm thick HFS. In laterdevelopments the minimum thickness of the heat flux sensorhas been reduced to 1 mm.4. The distance between the external surface of the insulationcoating system and the internal wall of the pressure vessel isnot critical and has no effect on the surface heat transfercoefficient as the boundary layers are very thin. This is true aslong as the rising and descending boundary layers do notinteract.

Heat Flux SensorsBackground to HFS’s. In general a HFS consists of a thinsheet of filler material in which a thermopile is embedded. Thethermopile is formed by a number of differentialthermocouples connected in series. The thermocouples of theembedded thermopile are positioned in such a way that thecold junctions are located on or near one surface of the sheetof filler material and the hot junctions on or near the other.

A typical section through a HFS is showndiagrammatically in Fig 5. If a heat flux passes through fromone surface of the sheet of filler material to the other thendepending on the thermal resistance of the filler material, asmall temperature difference (gradient) is generated. Thissmall temperature-difference will generate a relatively highvoltage signal from the thermopile The resulting heat flux iscalculated using a calibration value which is determined foreach HFS by subjecting the HFS to a known heat flux andmeasuring the resulting output voltage signal.

In practice, HFS’s are either built into or mounted on thesurface of the body through which the heat flux is to bemeasured. The selection of a HFS for a particular applicationneeds to be considered carefully as errors can be present evenalthough they have been correctly installed.

Errors due to presence of HFS. The HFS has a certainthermal resistance which is mainly dependent on the thicknessand thermal conductivity of the filler material. The filler-material effectively consists of two areas; one comprising onlyof the filler material (low thermal conductivity) and the otherin which the filler material is either partly or totally filled withthermopile material (higher thermal conductivity).

In its final form the HFS comprises of a sheet of fillermaterial with an effective thermal conductivity which isusually higher than that of the filler material alone. Changes inthe thermal conductivity of the filler material used in themanufacture of a HFS will influence the extent to which theresulting heat flux value deviates from the correspondingundisturbed value3 i.e. no HFS present. Similarly, when a HFS


is mounted on the surface of a body and exposed to a fluid(air, water, etc.) where heat is transferred by natural or forcedconvection, the resulting heat flux value will also differ fromthe corresponding undisturbed value3. These effects arecommonly referred to as the “macro heat flux disturbance”and will give rise to measurement errors. To reduce sucherrors a HFS is commonly manufactured in sheet form butwith the thermopile located in only the centre of the sheet (e.g.in the form of a disc). The non-sensitive outer part of the HFSis commonly referred to as the “guard”.

In practice the outer part of the HFS consists only of thefiller material which results in a difference between thethermal conductivities of the centrally located sensitive areaand the guard. The magnitude of the resulting errors can bereduced by appropriate sizing of the sensitive area and theguard for a given HFS.

Errors due to internal construction of HFS. In practicethe thermopile is made by winding a constantan wire (50µm or30µm in diameter) around a rectangular strip of syntheticmaterial. Half of each winding is plated with a layer of purecopper of uniform thickness, using a special plating process(see Fig. 6 and Fig. 7). The junctions between the bare and thecopper plated constantan act as junctions of a thermocouple.Leads are welded to the ends of the thermopile which is thenwound into place around the strip of synthetic material. Eachhalf winding of constantan and coppered constantan acts as a“cold bridge” which results in a hot/cold spot on the respectivesurfaces of the sheet of filler material. The significance of thisdisturbance is dependent on the thermal conductivity of thefiller material in which the thermopile is embedded.

Also, this effect is influenced by the heat transfer coefficientwhen the HFS is exposed to a fluid on one of its surfaces. Thisis commonly referred to the “micro heat flux disturbance”. Inpractice it means that for different applications differentcalibration values may have to be applied. The significance ofthis disturbance is dependent on the design of the HFS (choiceof materials and dimensions). For a wide range of HFSapplications it is preferable to minimise the influence of themicro heat flux disturbance effect or in other words produceHFS’s in which the calibration value is independent of theapplication.

k-value measurement. Fig. 8 shows a length of the cross-section of an insulated length of steel pipe, complete with aHFS, exposed to oil and water on adjacent surfaces. Therelation between heat flux, temperatures, dimensions andthermal conductivities is expressed by Equation (2) which is asimplified form of Equation (1).

Q" = (Ts-Tc)/[r2{ ln(r1/r0)/kS + ln(r2/r1)/kINS}]……….…(2)

A value for the thermal conductivity of the layer ofinsulation material, KINS, can be calculated for a known valueof Ks provided the values for Q", (Ts-Tc), r0, r1 and r2 can bemeasured. Consequently a direct measurement system requiresthe measurement of Q" and (Ts-Tc). It was therefore decided todevelop a heat flux sensing device which could be used to

measure the values for Q" and Tc directly. Tc was measuredusing differential thermocouples embedded within the heatflux sensing device immediately adjacent to the appropriateinterface. A separate temperature sensor was also developed tomeasure the value of Ts and was mounted on the internalsurface of the steel pipe directly opposite the position wherethe heat flux sensing device was located.

Development of Beltform Heat Flux Sensing Device.Introduction. It was realised early in 1995 after a lengthyperiod of development and optimisation of HFS’s in theiroriginal form i.e. a rigid epoxy resin flat disc, were toocomplex to be universally accepted for everyday use in suchtests. Heriot-Watt University in conjunction with TNO-TPDinitiated a further programme of research aimed at developinga heat flux sensing device which would form the basis of adirect measurement system. This sensing device was intendedto be used mainly for the determination of the thermalconductivity of insulation coating systems on lengths of steelpipe subjected to simulated service conditions. Thebackground and the subsequent use of this sensing device aredescribed in the following sections.Specified requirements. The initial design requirements werespecified as follows:1. To be flexible enough to conform to the external curvedsurface of insulation coating systems with a range of diameters(100mm - 410mm). It had to be provided with a means bywhich it could readily be held in position around the curvedsurface of the insulation coating systems.2. To include a means by which interface temperatures couldbe measured.3. The sensor and cable connections were to be of watertightconstruction when subjected to operational pressures of 100bar (water).4. The accuracy of the heat flux measurement was to be betterthan 2% but in any case it was accepted that this would bedependent on the operational limits of currently availablecalibration devices.5. To be capable of measuring heat fluxes in the range 100 -500 W/m2.6. To be suitable for use with 20mm - 75mm thick layers ofinsulation coating materials with k-values in the range 0.1 -0.5 W/m.K.7. To be suitable for use in environments in which the surfaceheat transfer coefficients were in the range 20 - 40 W/m2.K.Manufacturing aspects. A number of developments whichhad taken place at TNO-TPD were made available to theproject:1. Thermopile ribbons in widths of 2mm - 0.5mm andthicknesses of 0.12mm - 0.05mm were now available. Itshould be emphasised that the effective thermal conductivity,sensitivity, thickness and dimensions of the sensitive areawithin the HFS are all influenced within certain limits by thetype and length of thermopile ribbon.2. Improvements in the moulding technique required to embedthe thermopile ribbon in materials such as polyurethane hadbeen introduced. This presented the possibility of producing


HFS’s with the required flexibility, toughness, water-tightnessand hydrostatic pressure-resistance. It therefore had becomepractical to manufacture HFS’s in a flexible thin belt formcapable of being used in the above test environments.Numerical modelling. In order to determine the optimumconfiguration for the beltform heat flux sensor the numericalmodelling tool described earlier was again employed. Fig. 9shows an enlarged part of the cross-section of a HFS mountedon the external surface of a section of insulated steel pipesubjected to hydrostatic pressure on one surface. The HFS wasassumed to be flat in the model because of the high ratiobetween the outer diameter of the test pipe and the thicknessof the HFS. It was assumed that heat transfer between theouter surface of the sensor and the surrounding water wouldresult in heat transfer coefficients in the range 10 to 50W/m2.K. The embedded thermopile was represented within thecross-section of the HFS as a series of parallel strips.

In the model, the calibration value was defined as follows:

C = Q"/∆T………...……………………………………(3)

By varying the values for d and L the significance of themicro heat flux disturbance can be studied and expressed interms of h and C. Also the effective thermal conductivity ofthe sensitive area (the area of filler material in which thethermopile ribbon is embedded) was calculated to study thesignificance of the macro heat flux disturbance.Thermopile layout. The results obtained from the numericalsimulations resulted in the following configuration for thethermopile (see Fig. 10):1. Thermopile-wire diameter: 0.03mm2. Height of the strip/width of the thermopile ribbon: 0,5mm(positioned symmetrically between both the surfaces of theHFS)3. Pitch of the thermopile windings: 0.14mm4. Thermal conductivity of the filler material : 0.2 W/m.K5. d= 1mm6. L= 5mmCharacteristics. The resulting characteristics of the beltformHFS (see Fig. 11) were as follows:1. Overall dimensions: 380mm x 100mm x 1mm.2. Size of sensitive area: 40mm x 60mm.3. Typical calibration value: approximately 30 W/m2/mV4. Internal electrical resistance: 2 - 2.5 kOhm.5. Resistant to hydrostatic pressure up to 100 bar, possibly upto 200 bar.6. The beltform HFS was to be held in position around theexternal curved surface of the insulation coating system bymeans of a plastic membrane placed over the HFS and held inposition with the assistance of two stainless steel springs.The HFS and the plastic membrane are shown in Fig. 12 andFig. 13.Performance. The predicted performance of the HFS obtainedfrom the numerical simulations is summarised as follows:1. The effective thermal conductivity of the sensitive area of

the HFS was found to be keff = 0.212 W/m.K.2. The variation in the value of C was of the order of 1% whenh varies from 10 to 50 W/m2.KThe narrowest part of the guard was selected to be 30mm tominimise errors resulting from the macro heat flux disturbanceeffect described earlier. The resulting significance of themacro heat flux disturbance was obtained by studying thevariation in the heat flux Q" passing through the HFS.Equation (2) was used to find the theoretical thermalconductivity of the pipe insulation material which was thencompared to the actual value. The results obtained were asfollows:1. The influence of the heat transfer coefficient h in the range10 - 50 W/m2.K was found to be negligible.2. As an example, assuming an insulation layer thickness (r2-r1) of 15mm the discrepancy in the results was of the order of0.2%. This discrepancy was found to increase for higherinsulation layer thicknesses (up to 60mm) to approximately1%.Calibration method. Each HFS is individually calibrated bysubjecting it to a known heat flux electrically generated in ahigh performance twin-plate test arrangement developed atTNO-TPD. A description of the calibration technique has beenpublished previously3.

Simulated Service TestA typical diagrammatic view of such a test facility is shown inFig. 14. It should be emphasised that this test cannot be usedto predict the long-term creep characteristics of insulationcoating systems as the test is conducted under real timeconditions.Test arrangment and test procedure. In this short-term test,lasting typically 28 days, the flow of heat through theinsulation coating system and the radial contraction/expansionof the insulation coating system is monitored under conditionswhich simulate those to which the insulation coating systemwould be subjected to when in-service on the seabed.Normally in the test the insulation coating system on the testpipe is subjected to a temperature gradient across its fullthickness and a hydrostatic pressure which are identical tothose which it will experience when in-service on the seabed.Preparation of coated test pipe. Ideally the test pipe shouldbe prepared from a length of steel pipe coated under normalproduction conditions and not from a length of steel pipecoated under laboratory conditions. The first stage in thepreparation of the length of coated pipe selected for the test isto weld a steel blanking plate to one of its ends and preferablythen encapsulated it in a suitable insulation coating material. Atransition collar, complete with ring flange, is butt weldedonto the other end of the coated test pipe. The transition collarand the ring flange are used to modify the outside diameter ofthe steel pipe to enable it to be bolted to the underside of theblind flange of the pressure vessel. At this stage thetemperature sensor(s) is attached to the internal surface of thesteel test pipe.


Mobilisation of pressure vessel. The end of the coated testpipe fitted with the ring flange is then bolted to the internalface of the blind flange of the pressure vessel. At this stage theheat flux sensor device(s) is attached to the external surface ofthe coated test pipe (see Fig. 15). The pressure vessel blindflange, complete with the instrumented test pipe, is then boltedin position on the open end of the pressure vessel. The annulusbetween the coated test pipe and the internal surface of theshell of the pressure vessel is filled with simulated sea waterafter the electronic displacement transducers have been boltedonto the ports on the shell of the pressure vessel.

The pressure vessel is then placed in the cooling jacket andthe anti-freeze mixture already in the cooling jacket is toppedup to the required level. The anti-freeze mixture in the coolingjacket is continuously circulated through chiller units duringthe test to bring the temperature of the simulated sea watersurrounding the coated test pipe in the pressure vessel down tothe test temperature.

The flexible hoses which are used to continuously circulatehot mineral oil through the bore of the coated test pipe are thenconnected to the two openings in the blind flange of thepressure vessel. The flexible hoses for the system used topressurise the simulated sea water surrounding the coated testpipe in the pressure vessel are then connected to the openingsin the blind flange of the pressure vessel. The remainingthermocouples and the pressure transducer are then alsoconnected to the blind flange of the pressure vessel.Instrumentation. The following instrumentation is normallyused in such a test:Pressure. The pressure inside the pressure vessel is monitoredusing a pressure transducer mounted in the blanking flange ofthe pressure vessel.Radial compressive deformation of coating. Electronicdisplacement transducers are mounted in the walls of thepressure vessel to monitor radial movements of the coatingsystem resulting from the presence of the hydrostatic pressureand the temperature gradient.Temperatures. The temperature of the steel test pipe ismonitored using specially designed temperature sensorsmounted directly on the internal surface of the steel pipe. Thetemperature of the mineral oil passing through the bore of thecoated test pipe is monitored using thermocouples containedwithin thermowells mounted in the flow and return pipework.Heat fluxes. The heat transfer characteristics of the insulationcoating system on the test pipe are obtained using the beltformheat flux sensing device whose development has beendescribed above.Data acquisition. The readings obtained from all the sensorsare recorded by a data acquisition unit which is programmedto permit sets of recordings to be taken at appropriate timeintervals throughout the duration of the test. The data filesfrom each test are available in spreadsheet format.

Establishment of test conditions. When the mobilisation ofthe test arrangement has been completed the anti-freezemixture in the cooling jacket is circulated through the chillerunits to bring the temperature of the simulated sea water

surrounding the coated test pipe in the pressure vessel down tothe test temperature.

When the temperature of the simulated sea watersurrounding the coated test pipe in the pressure vessel hasreached the test temperature the mineral heating oil is thencirculated through openings in the pressure vessel flange intothe bore of the coated test pipe. The temperature of the heatingoil is gradually brought up to and kept at the test temperatureusing electrical control circuits which are connected toremotely positioned immersion heaters and to sensorsmounted in the heating flow and return pipework.

When the required temperature gradient through the coatedtest pipe has been established and the response of the coatedtest pipe has stabilised a preliminary set of measurements areused to determine the thermal conductivity of the insulationcoating system. The simulated sea water surrounding the testpipe inside the pressure vessel is then gradually pressurised upto the test pressure provided the result from the preliminary setof calculations is acceptable. A control circuit is used to holdthe pressure constant at this value throughout the duration ofthe test.Termination of test. The test conditions are kept constant,within practical limits, for the duration of the test which istypically 28 days. On completion of the test the heatingcircuits are switched off thus allowing the internal temperatureof the coated test pipe to fall to the temperature of thesimulated sea water surrounding the test pipe in the pressurevessel. When the response of the test pipe to the removal ofthe temperature gradient has stabilised the hydrostatic pressurein the pressure vessel is reduced to atmospheric. Theperformance of the coated test pipe is normally monitoredthroughout the duration of the test up until it is finallyremoved from the pressure vessel.Calculation procedure. A typical set of results obtained froman actual simulated service test on a polyurethane based fieldjoint coating system are summarised below.Coated test pipe details:Outside diameter of steel pipe = 27.3.01 mmWall thickness of steel pipe = 22.2 mmThickness of insulation layer = 38.88 mmro = 114.305 mmr1 = 136.505 mmr2 = 175.385 mmResults from the test averaged over 400 hours:Temperature of internal surface of steel pipe, Ts = 89.73 °CTemperature under heat flux sensing device, Tc = 12.71 °CHeat flux = 339.08 W/m2

Calculations:Heat flow from insulation coating system, Q" = 373.66 W/mThe thermal conductivity of polyurethane, kINS using Equation(2) and assuming the thermal conductivity of the steel ks to be45 W/m.K, was found to be 0.194 W/m.K.Material assessment. It is strongly recommended thatconsideration is given to conducting a carefully plannedprogramme of standard physical and mechanical tests to beconducted on samples removed from areas of the insulation


coating system on completion of the test in order to confirmthat no significant degradation of the materials has taken placeand in particular to confirm the density of the insulant presentwithin the coating system.

Future developmentsThe design of the heat flux sensing device continues to beoptimised to meet the ever increasing demands of the industry.Tests conditions have become more demanding and there havebeen significant changes in the types of insulation materialsbeing considered as shown below:1. Internal pipe temperatures up to 140°C.2. Hydrostatic pressures up to the equivalent of a water depthof 1450m.3. Insulation coating materials containing micro and macrospheres.It should be emphasised the heat flux sensing device describedin this paper will not necessarily be suitable for use in testswhere its performance has not already been examined.

Nomenclatured= thickness of the HFS [mm]h= surface heat transfer coefficient [W/m2.K]Ks= thermal conductivity of steel pipe-material [W/m.K]Keff= thermal conductivity of the sensitive part of the HFS[W/m.K]KINS=thermal conductivity of the layer of insulationmaterial [W/m.K]L= distance between the rows of thermopile-ribbon in thesensitive part of the HFS [mm]Ts= temperature of the internal surface of the steel pipe[°C]Tc= temperature of the external surface of the layer ofinsulation coating material [°C]∆T=temperature difference as “seen” by the junctions ofthe thermopileQ"= heat flux [W/m2]

AcknowledgementsThe authors wish to thank Bart Paarhuis for carrying out thenumerical simulations relating to the “Macro and Micro heatflux disturbance” during his period of training at TNO-TPDwhich formed part of his studies in Physics at the Universityof Twente.

References1. Patankar, S.V., "Numerical Heat Transfer and Fluid Flow" ,

Hemisphere Publishing Corporation, Washington, (980).2. Bird, R.B., Stewart, W.E. and Lightfoot, E.N. “Transport

Phenomena”, J. Wiley & Sons, New York (1960).3. Van der Graaf, F., “Heat Flux Sensors”, chapter 8 of

“Sensor, a comprehensive Survey”, Volume 4 “ThermalSensors”, 1990 VCH.

4. Van der Graaf, F., “Building Applications of Heat FluxTransducers”, ASTM STP 885, Bales, E., Bomberg, M.,and Courville, G.E., (eds); Philidelphia: American Societyfor Testing and Materials, 1985, pp. 9-24.

Table 1−Material and fluid properties (* T in oC)Material Thermal

conductivity[W/mK]

Dynamicviscosity[kg/ms]

Heatcapacity[J/kgK]

Density

[kg/m 3]Water (H2O) 0.596 1.002 10-3 4185.0 998.23 - 0.15 T*Steel (S) 45.0 - - -Heat Flux Sensor (HFS) 10.0 - - -Insulation (INS) 0.18 - - -

Table 2−Dimensions of set-upPosition symbol radius

[mm]inner radius steel pipe r0 65outer radius steel pipe r1 84.1outer radius insulation r2 145.9

HFS thickness d 3.5

Table 3−Errors in the measured insulation valueCase THFS

[oC]∆T[oC]

Q" HFS

[W/m 2]∆Q"[%]

kINS

[W/mK]∆kINS

[%]Without HFS 15.10 - 198.0 - 0.18 -With surfacemounted HFS

14.28 -0.82 200.5 +1.3 0.1797 -0.1

Fig. 1−Geometrical description of the test arrangement and the mesh.

Fig. 2−Temperature distribution within the pressure vessel.

Fig. 3−Velocity distribution within the pressure vessel.

Fig. 4−Velocity vectors close to HFS.

Fig. 5−Schematic cross-section through a HFS.

Fig. 6−A schematic view of a typical length of thermopile-ribbon.

Fig. 7−Microscopic view of thermopile ribbon.

r0 r1 r2

Insulation

steel pipe-wall

water

oil

Ts

Tc

Fig. 8−Section through wall of insulated steel pipe.

surface exposed to waterL 0.1mm

0.5

mm

pipe insulation

{∆ T d

Fig. 9−Enlarged section through beltform HFS.

Fig. 10−Length of thermopile ribbon used in the beltform HFS.

380 mm

10

0 m

m

40

mm

30

mm60mm

Thickness : 1 mm

= thermocouple

Fig. 11−Beltform HFS, showing sensitive area, temperature sensors and overall dimensions (guard).

Fig. 12−Final form of beltform HFS.

Fig. 13−Plastic membrane used to hold the beltform HFS in position around test pipe.

Fig. 14−Typical simulated service test facility.

Compressor

Air DrivenLiquid Pump

Pressure Relief Valve

Thermocouples

Hot CircuitControllers

Multi-ElementElectrical Heaters

Chillers

Header Tan k

Pressure Vessel

Cooling Jacket

Isolating Valve

Coated Test Pipe

PressureGauge

Hot Circuit Pumps

Sight Glass

IsolatingValve

Isolating Valve

MEASUREMENTS RECORDED THROUGHOUT DURATION OF A TEST:Radial contraction/expansion of coating system.Heat flux through coating system.Hydrostatic pressure within pressure vessel.Temperature of internal surface of steel test pipe.Temperature of external surface of coating system.

Pressure Transducer

Fig. 15−Beltform HFS positioned on external surface of coated test pipe.

otc_1999-insulationtestmethod_tno

Documents