Field Joint Coatings for Deep Sea PipelinesField Joint Coatings for Deep Sea PipelinesRobrecht Verhelle1, Luk Van Lokeren1, Samir Loulidi1, Helen Boyd2, Guy Van Assche1Robrecht Verhelle1, Luk Van Lokeren1, Samir Loulidi1, Helen Boyd2, Guy Van Assche1

1. Vrije Universiteit Brussel, Physical Chemistry & Polymer Science, Pleinlaan 2, 1050 Brussels, Belgium; 1. Vrije Universiteit Brussel, Physical Chemistry & Polymer Science, Pleinlaan 2, 1050 Brussels, Belgium;

2. Heerema Marine Contractors, Vondellaan 55, 2300 PH Leiden, The Netherlands.

Heerema Marine ContractorsHeerema Marine ContractorsHeerema Marine Contractors (HMC) is contracted to install Although the individual pipe sections (12 m) are coated with aHeerema Marine Contractors (HMC) is contracted to installpipelines in the sea. The metallic pipes, generally of carbonsteel, need not only to be protected against corrosion, but alsoto be insulated to maintain the temperature of the pipe

Although the individual pipe sections (12 m) are coated with afactory-applied coating along their full length, the coating is cut backat the ends before welding them together during a J-lay or reel-layinstallation. After welding, a field joint coating is applied over theto be insulated to maintain the temperature of the pipe

contents and assure the flow. Therefore a multilayer polymercoating is applied.

installation. After welding, a field joint coating is applied over thewelded area. Ensuring optimal application conditions for the coatingduring an offshore installation is far from straightforward.coating is applied. during an offshore installation is far from straightforward.

Pipe sections Surface cleaning FBE application Injection moulding Field jointPipe sections

Welded together

Surface cleaning

Grit blasting

FBE application

Corrosive protection

Injection moulding

Thermal insulation

Field joint

Needs cooling

Objectives Cross section model with dimensions (mm) and boundary conditions:

symmetric, outflow, convective cooling h , convective cooling hObjectivesIn order to optimise the application process ofthe field joint coating, deep insight into the cure

symmetric, outflow, convective cooling h1, convective cooling h

2

In the first part of this research project,the cooling process of a field joint

and crystallisation kinetics, together with a goodcomprehension of the heat transfer in the fieldjoint is required. Experimental data on the raw

the cooling process of a field jointcoating is simulated, computing thetemperature and crystallinity profiles,throughout the coating, as a functionjoint is required. Experimental data on the raw

materials, acquired by thermal analysis, will beused to determine the crystallisation1 and cure2

kinetics model, which will consequently be

throughout the coating, as a functionof time using the cure andcrystallisation kinetics model obtainedfrom experimental data.

L 1200.00 TS 15.70

Lcutback 313.00 TIMPP 52.59kinetics model, which will consequently beimplemented in the computational finite elementmodel.

from experimental data.Lcutback 313.00 TIMPP 52.59

Lchamfer 91.10 T3LPP 9.19

Loverlap 50.00 T4L 39.40

1. J.D. Hoffman, R.L. Miller, Polymer 1997, 38,

3151-3212

2. G. Van Assche, A. Van Hemelrijck, H. Rahier, B.model. Loverlap 50.00 T4L 39.40

T5L 4.00

2. G. Van Assche, A. Van Hemelrijck, H. Rahier, B.

Van Mele, Thermochim. Acta 1995, 268, 121-142

Computational Methods

Dependent variable u Source term f

Computational MethodsAll computations are performed in COMSOL

ODE parameters for the crystallisation kinetics modelIn order to obtain stable and low time-consuming computations, preferably thePARDISO solver was used, together withDependent variable u Source term f

N

All computations are performed in COMSOLMultiphysics. The crystallisation kinetics modelwas incorporated as a set of ODEs,3 all of form ( )

dt

dT

dT

TdN

dt

dTqN

)(1

1

1)( 0α

α

α−+

−+−

PARDISO solver was used, together withthe BDF timestepping method.

Nat

Where u is the dependent variable, d the

ft

ud

t

ue aa =

∂

∂+

∂

∂2

2 dtdTdt1 α

−

α−1

)( NTqFurthermore, to avoid mathematicallycorrect but physically unrealistic data for

the relative crystallinity α (i.e. α ∈ [0,1]),α

F

Where u is the dependent variable, da thedamping coefficient, ea the mass coefficient andf the source term. Since our model only has first

α−1

( ) ( )QFPNFG at +−− 214 2απ

)(TG

the relative crystallinity α (i.e. α ∈ [0,1]),and the amount of nuclei N (i.e. N > 0),both parameters were limited usingtransformation functions:F

P

f the source term. Since our model only has firstorder time derivatives, all mass coefficients ea

are always zero. Furthermore, all equations arewritten so that the damping coefficient d equals

)(TG

α−1

)(TFNqtransformation functions:

2

1

2

)(+=

berfα NeN log=

Q

written so that the damping coefficient da equals1.

α−1

)(2 TNqF22

+=α eN =

3. J.M. Haudin, J.M. Chenot, Intern. Polym. Process

2004, 19, 267-274 & 275-286

ResultsA1-2-3-4-5: Centre of the FJC

Temperature and relative crystallinity profileswere computed for different geometries (e.g.with and without a mould, representing anwith and without a mould, representing animmediate removal/opening of the mould afterthe injection and a complete cooling in themould), different pretreatments (preheating of

B1-2-3-4-5: Centre of the cutback

mould), different pretreatments (preheating ofsteel pipe and factory applied coating) anddifferent start and boundary conditions (e.g.

Temperature (left) and relative crystallinity (right) profile different start and boundary conditions (e.g.temperature of melt, mould and air).

Points of interest were selected in the modelC1-2-3-4-5: Cutback

Temperature (left) and relative crystallinity (right) profile

of the FJC after 120 min.

PerspectivesPoints of interest were selected in the modelwith the perspective to be compared withindustrial test results. This validation step, aconfrontation of the computational results with

Perspectives

In the last quarter of 2014, the computedtemperature and crystallinity profiles will beconfrontation of the computational results with

the experimental results on the industrial scale,is planned in the last quarter of 2014. D1-2-3-4-5: Parallel to the chamfer

temperature and crystallinity profiles will becompared to industrial test results.

Shrinkage during cooling and crystallisation will beis planned in the last quarter of 2014. D1-2-3-4-5: Parallel to the chamfer

Shrinkage during cooling and crystallisation will beimplemented in the model, using experimental(lab-scale) results in order to predict and to

Locations of computed temperatures and crystallinities

in the field joint coating. Relative crystallinity Temperature

(lab-scale) results in order to predict and toevaluate internal and interfacial stresses.

Filling of the mould will eventually be implemented.

This research was carried out under project number M31

in the field joint coating. Relative crystallinity Temperature Filling of the mould will eventually be implemented.

This research was carried out under project number M31

6 12469 in the framework of the Research Program of

the Materials innovation institute M2i (www.m2i.nl).

Excerpt from the Proceedings of the 2014 COMSOL Conference in Cambridge