Heat transfer

Potential Operating problems

Heat loss to the ambience Thermal insulation in oil and gas production systems are required for the following main reasons:

Avoid operation below wax appearance temperature (WAT) to avoid wax deposition totally. (When the flowing temperature drops below WAT wax deposition will occur). !inimise wax deposition rate and thereby minimise wax pigging frequency in case operation below WAT cannot be avoided. "nsulation helps because below WAT the rate of wax deposition is proportional to the heat loss to the ambience. Avoid operation below hydrate formation temperature Allow shutdowns without cooldown into hydrate formation region. (Wax formation during shutdown is a lesser problem as wax deposited during a now flow situation tends to be very soft and is easily removed during restart).

#ther (not so common) reasons are: Avoid liquid dropout (water and$or condensate) in wet gas pipelines. Avoid excessive pressure drop due to high viscosity at low temperatures Avoid separation problems due to more stable emulsions at low temperatures. Thermal insulation is expensive. %or deep water risers it can even be a feasibility problem with sufficient insulation to achieve cooldown times required for hydrate prevention during unplanned shutdowns. To avoid expensive &over'insulation( it is therefore important to ma)e correct temperature calculations. *ven worse is it if temperature calculations are wrong on the optimistic side as this may lead to serious operational problems e.g. cooldown to hydrate formation before all complete hydrate prevention can be achieved limited turndown ability too low heating duty in the processing facilities etc. Joule-Thompson and expansion The following thermal problems are caused by changes in pressure: With gas the +oule Thompson effect may give temperatures far below ,ero -.. due to high pressure drop over valves during startup (e.g. over production cho)e valves). /uring shutdown of a well the upper part of the well tubing will cool down rapidly because the ambient formation temperature is near to the seawater temperature the tubing is normally not thermally insulated and the tubing is gas filled. This can give the following problems: 0ow temperature material problems 1ydrate and$or ice formation

1igh required hydrate inhibitor in2ection rate 3equired hydrate inhibitor in2ection rate above in2ection capacity %or liquid and high pressure gas the +oule'Thompson effect may give increased temperature which may be problem if there are high temperature design temperature problems. /uring depressuri,ation severe cooling may be caused by isentropic expansion (lost expansion wor) for the fluid at the entrance of the pipe which 4pushes4 i.e. does wor) on the fluid downstream due to expansion ) in addition to the +oule' Thompson cooling. This may give temperatures below minimum design.

Design Problems
Transient problems The following thermal problems are caused by changes in pressure: 0ong time to heat up a well. The biggest problem in this respect is to calculate the transient temperature correctly. "t can also be a problem to start up because of low temperatures over extended times which requires special precaution with hydrate protection (normally with continuous methanol in2ection during startup) and with material problems. This problem is increased if there is a need for significant cho)ing during startup. 0ong time to heat up a flowline. /uring startup it may be required with in2ection of methanol. Alternatively can the line be heated up (e.g. with hot oil circulation or with heat tracing) before well production startup. Modeling problems and uncertainties The following general sources to error are considered the main problems: Wrong energy equation. %or example is the gravity term not included in many commercial pac)ages. 56T with blac)'oil model and simplification of the energy equation. The pressure effect on temperature is often not included as it is normal to assume that the enthalpy change is given by .p7dT and that the energy equation is .p7dT 8 d9 8 heat loss to the ambience over the section d:. %or rough calculations this is often good enough approximation but especially for wells and risers with substantial effect of gravity and pressure drop it is not good enough. ;lac)'oil modeling in general. To account for deviation from ideal fluids (+oule Thompson effect) flashing$evaporation and condensation is difficult with blac)' oil models. <ormally it is however agreed that precise temperature calculations for gas and gas'condensate systems is not possible with blac)oil models (should only be used for oil systems).

.ompositional 56T. "f not made properly compositional models may be worse than blac)oil especially near the critical point. The enthalpy for a fluid is a direct function of fluid density and it is well )nown that liquid densities are often very wrong with *#= models li)e =3> and 53. Wrong liquid densities results in wrong enthalpies for the liquid. (=ince it is easier to tune liquid densities with blac)oil models than with compositional models better enthalpies and hence temperatures may be calculated with blac)oil models. 6ery often the composition is not )nown and then there is nothing to gain by using compositional models). ?ncertainties in well models (discussed under models for wells). ?ncertainties for pipelines and bundles (discussed under pipeline and bundle models).

General - definition og problems

When defining the problems the following questions should be raised: Which problems are foreseen with respect to design fabrication and installation@ Which operating problems may occur@ .onsider both low or high temperatures Which mechanical (hardware) problems may occur during operation@

Are simplified methods acceptable or are more detailed analyses required@ .an the customer provide overall ?'values to be used for the steady state calculations@ "f so the calculations problem is significantly simplified. "s the well bore defined such that detailed calculations are possible@ /oes compositional data exist@ "f not accurate wellhead temperatures can hardly be calculated (unless data for enthalpy are available). Are transient calculations needed to establish %W1T versus time after start'up@ What are the problems with respect to calculation uncertainties@

General - abbreviations and definitions

Abbreviation %W1T .p =pecific heat capacity 8 .v =pecific heat capacity 8 !assflow m.p ? 5roduct of massflow and heat capacity )g$s W$> +$)g$> /escription %low wellhead temperature ?nit -. +$)g$>

#verall heat transfer coefficient (normally referred W$mA$>

to inner pipe diameter) 1 5r 3eynolds t t *nthalpy 5randtl number 8 .p7$) 3eynolds number 8 7?7/$ Temperature Time 6iscosity Thermal *xpansion coefficient 8 +oule Thomson coefficient 8 (dT$d5)1 /ensity Thermal diffusivity 8 "sothermal compressibility 8 -.$5a >g$mC mA$s B$5a +$)g ' ' -. = <s$mA B$>

The following temperature definitions (which may require different conservative assumptions) are suggested for flowing wellhead temperatures: !aximum operating %W1T. This will vary from well to well and with time during the life of the field due to changes in the production rate and composition. This definition has implications for si,ing of downstream facilities calculation of corrosion and scaling rate etc. The input parameters and calculations should be conservative on the high side. !aximum maximum (4maxmax4) operating %W1T. This is the maximum over time and for all the wells. !aximum design %W1T. This is the 4maxmax4 operating plus a safety margin to account for uncertainties. The same design temperature should be used for all wells and for the whole production life. "t has implications for well and flowline' riser design e.g. material selection and design of expansion loops. !inimum operating %W1T. As with maximum operating %W1T this will vary from well to well and with time during the life of the field due to changes in the production rate and composition. "t is used for si,ing of downstream facilities for example separators (definition of insulation requirements heating duty etc.). =ometimes the minimum operating temperature will be used for capacity calculations and linesi,ing (in cases with very high viscosities such that low flow conditions gives higher pressure drop than high flow cases). %or pipeline design possible +oule Thomson cooling over the production cho)es must be calculated. <ote that the lowest temperature in the flowline will normally not be given by the normal flowing condition. "nstead it may be given by the ambient temperature in which case the calculation of minimum %W1T is of no interest. #r it may be given by the flowing temperature shortly after a cold startup. %W1T is then equal to ambient temperature at the :mas tree and the fluid composition may often be

gas due to separation in the well. The question is what pressure drop that occurs over the cho)e during the cold start'up. !inimum minimum (4minmin4) operating %W1T. This is the minimum over time and for all the wells. !inimum design %W1T. This is the 4minmin4 operating or shutdown %W1T minus a safety margin to account for uncertainties. "t is used for well and pipeline design e.g. material selection and design of expansion loops. The same design temperature should be used for all wells and for the whole production life. "t has implications for well and flowline'riser design li)e material selection and design of expansion loops.

Stead state energ balance - basic theor

The following energy balance is generally valid for pipe flow (including frictional heat): (B) (D) Total fluid massflow ()g$s) d1 .hange in mixture fluid enthalpy (+$)g) over the axial length d: d: 5ipe length (m) g gravity constant 8 E.FB m$sA "nclination angle (from the hori,ontal) (sin EG- 8 B.G) d*) .hange in )inetic energy 8 (u is the velocity) over d: (+$)g)

1eat loss to the environment (W) from the pipe section of length d: !echanical wor) (W) performed by the fluid over section length d: ? #verall heat transfer coefficient referred to inner pipe diameter (W$mA$>) "/ "nner pipe diameter (m) T Average mixture fluid temperature in the section of length d: (-.) Tamb Average ambient temperature outside the section of length d: (-.) To calculate temperature drop from equation (B) the enthalpy can be substituted with the following equations:

(H) and evaporation.

assuming constant composition i.e. no condensation

(I)

8 .p 8 specific heat capacity (+$)g$>)

(J) +oule Thompson .oefficient (>$5a) Thermal *xpansion coefficient 8 6 =pecific volume 8 B$ mC$)g %luid density )g$mC (B$>)

The enthalpy change can also be expressed with the following equation: (K)

*i "nner fluid thermal energy !eal fluids %rom equations B'J assuming no mechanical wor) performed by the fluid and no change in composition:

(L) An alternative form using the +oule Thompson coefficient (8 expansion coefficient () is: (F) "deal gas With ideal gas 8 B$T or 8 G. %rom equation (H) and assuming no mechanical wor): (E) d1 8 .p x dT5 (BG) #ther useful relations: ) instead of the thermal

(BB)

8 ?niversal gas constant 8 FHBI +$mol$>

%rom equation BG the following interesting points are deduced: %or ideal gas the fluid temperature does not increase due to wall friction. %or ideal gas with no heat loss to the environment T 8 'B$.p x g x sin x : when neglecting the )inetic energy term.

"deal li#uid %or an ideal liquid the thermal expansion coefficient is ,ero. %rom equation J or from 8 G equation H then becomes (no mechanical wor) included:) (BD) d1 8 .p x dT M d5$ 8 .p x dT M 6xd5 ( )

The )inetic energy term with liquid can be neglected hence equation B becomes: (BH) The gravity term (BI) The following points are deduced for an ideal liquid: <o effect of elevation change on temperature %rictional pressure drop increases fluid temperature by: dT 8 ' d5friction $ $ .p )

cancels out with

i.e:

%rom equation K and BD and the fact that 8 G it is seen that This again implies that .p 8 .v (for solids and real liquids .p N .v too)

General transient modeling

(15)

*) >inetic energy 8 *p 5otential energy 8

(u is the velocity) in cell of length d: (+$)g) (h is height in m) in cell of length d: (+$)g)

%or a situation with flow through pipes the transient term on the right side of eq. BI is

small and can be neglected if there is no accumulation or reduction of mass in a section. 1owever for gas and multiphase flow mass is often accumulated or reduced with time and then the right term should be included. To calculate change in internal energy d*i can be done by the following equations: (BK) assuming constant composition

(BL) (BF) 8 "sothermal compressibility (B$5a)

6ery often e.g. for wellflow the left side of equation BI is ,ero and then the only transient phenomenon with respect to energy and thermal calculations that needs to be included is the transient heat'up of the formation around the well bore. This is further discussed in the chapter on conduction in the earth$formation.

Pipelines
Stead state heat transfer coefficients The heat loss d9 OW$(meter of pipe)P is calculated from: (BE)

(DG) )n 8 1eat conductivity for layer n OW$m$>P ?i 8 #verall heat transfer coefficient referred to the inner tubing pipe diameter with all layers included out to the cement OW$mA$>P. 7 "/ 8 "nner pipe diameter OmP "/n 8 "nner diameter for layer n OmP #/ 8 #uter diameter of most external layer OmP hi 8 .onvective film coefficient on the pipe inside OW$mA$>P

ho 8 .onvective film coefficient outside external layer OW$mA$>P $nburied pipelines %or unburied pipelines the external film coefficient ho is normally very large compared to the inner film coefficient and conduction through the different layers including scaling and fouling. The last term (B$ ho) in equation BK can then be neglected. This applies even for a bare steel pipe for which convection on the inner wall (hi) and scaling$fouling will determine the overall ?'value. "f nevertheless the external film coefficient is of interest it can be calculated from the following equation:(from /avenport T... and .onti 6.+. &1eat Transfer 5roblems *ncountered in the 1andling of Waxy .rude #ils in 0arge 5ipelines( +ournal of the "nstitute of 5etroleum( 6ol. JL <umber JJJ'!ay BELB. (DB)

%or transient situations heat'up and cooldown of the pipe walls must be included. The time constant for thermal transients is in the same order as the retention time for the fluid through the pipe. 1ence transient modeling is normally done with a dynamic pipeline simulator li)e #0QA. %uried pipelines %or buried flowlines the heat transfer from the pipe to the soil here called ho can be calculated as follows (from /avenport T... and .onti 6.+):

(DD) "f 1 RR / (deeply buried pipelines) then:

1 8 1eight from seabed to center of pipe %or buried flowlines it ta)es long time to heat up the soil to reach a steady state ?'value. To calculate the ?'value over time ho can be calculated as follows (from /avenport T... and .onti 6.+):

(DH)

To find the time to reach steady state equation HG and HB are set equal and the equilibrium %ourier number is calculated. <ote that these transient models does not include heat'up of the pipe walls. This ta)es much shorter time and should be calculated with numerical codes li)e #0QA.

&ells
The following main modeling problems and uncertainties exist: Transient effects in the earth$formation for a single well 1eat transfer between different wells 3epeated shutdown and restart *ffect of cooling from the ocean. (6ertical heat conduction in the earth$formation around the well bore). 1eat transfer in vertical and deviated annuli where there a mixture of conduction natural convection and radiation contributes to the heat transfer. Qaslift with cold gas in the annulus. (cross current heat exchanging) /ownhole temperatures varies with depth /ownhole temperatures varies with location especially where there is a large difference in sea depth between individual wells Qeothermal gradient and reservoir. This gradient may vary considerably over the field especially if the sea depth also varies considerably over the field. =tatic bottom hole temperature. The reservoir temperature at the same depth may vary over the field due to difference in geothermal gradient and$or sea depth. %lowing bottom hole temperature (%;1T). "f there is a large inflow pressure drop %;1T may be much higher or lower than the reservoir temperature. 'onvection and radiation in the annulus To model natural convection in annuli is not straightforward because: "t is a combination of conduction convection and radiation. <o literature is available on models for natural convection in concentric very long annuli except one by 1asan A.3. and >abir ..=.: =5* paper DDEIF (BEEB) where a model for the convection between two parallel plates is suggested used. "t is for vertical plates and it is un)nown how it wor)s for concentric cylinders. Wells are often deviated wells which may alter the convection flow significantly. The annulus is often partially occupied with gas and the volume of gas is difficult to assess. This may the most important source of error.

%ouling. With seawater scaling starts at temperatures above HG'IG -.. 5recipitation of halites (normally sodium chloride) due to in2ection of methanol. This may reduce convection considerably. 3adiation over the annulus. This may be an effect for high temperature wells. 1asan A.3. and >abir ..=.: =5* paper DDEIF (BEEB) presents a model but source of errors are numerous li)e uncertain emissivity which depend on surface finish and view factors among other variables.

The convection film coefficient in the annulus from 5etroleum *xperts &5rosper(:

(DI) rci 8 "nner diameter of inner casing rto 8 #uter diameter of tubing The radiation term is given by (from 5*T*:):

(DJ) 8 =tefan';olt,mann constant e 8 *missivity The effect of radiation is only significant in case the annulus is filled with gas (which it may become over time also for a brine filled annulus sue to gas migration and liberation into the annulus). The conduction term is given by:

(DK)

The total heat transfer coefficient in the annulus (DL)

is given by:

'onduction in the earth(formation 1eat loss in the wellbore never reaches steady state due to the large thermal mass in the formation around the well. "t attains a quasi'steady state in which the rate of heat loss is a monotonically decreasing function of time. ;ecause of this transient effects will always be present and they must be considered also for steady state calculations. =ome problems with this are: What is the formation conductivity@ /ata for thermal conductivity for the earth (roc) or soil) around the well is normally not available and they can vary a lot depending on the porosity and moisture content. As an example heat conductivity for soil can vary from G.H to H W$m$> depending on the moisture content. With very wet soil a value close to that of water of G.K W$m$> would be expected but a common value used for thermal calculations is D W$m$> for both soil and roc). "f #0QA is to be used how far into the formation is it required to calculate the conduction. "f very thic) layer is required the calculation time will drop. What is the influence of wells near to each other (heat transfer between the wells). 1ow to model repeated shutdowns and restarts (with many wells near each other). The recommended equation for calculation of the convection in the annulus is given in the chapter &recommended model for steady state applications. )orced convection bet*een fluid and inner tubing *alls The forced convection heat transfer coefficient from the fluid to the inner pipe wall is given the following equation:

(DF) Overall $-value in stead state *ith transient effect in the formation 1ence all the resistance to heat transfer is in the formation outside the *ffect of earth$roc) around the well bore %rom .hiu >. and Tha)ur =... =5* paper DDFLG (BEEB) the heat loss from the well bore is:

(DE) (HG)

(HB) The last term in this equation includes all layers with conduction heat transfer only. The conduction in the annulus is included in . 8 1eat conductivity of the cement$earth$formation (W$m$>) 7 )n 8 1eat conductivity for layer n OW$m$>P 8 %luid temperature (-.) 8 6irgin (undisturbed) earth temperature (-.) 8 "nner tubing radius 8 Thermal diffusivity (mA$s) 8 Time after initial well startup (seconds) ?i 8 #verall heat transfer coefficient referred to the inner tubing pipe diameter with all layers included out to the cement OW$mA$>P. 7 "/ 8 "nner tubing pipe diameter 8 D x OmP "/n 8 "nner diameter for layer n OmP #/ 8 #uter diameter on outer casing OmP hi 8 .onvective heat transfer coefficient on the pipe inside OW$mA$>P +ote , "t is assumed that the cement has the same heat conductivity and thermal diffusivity as the formation. "f this is not the case the ?value out to the outer casing (referred to "/ of the tubing) must be replaced with the ?'value when the cement is included. "n the original paper by .hiu >. and Tha)ur this was actually done. 1owever since the cement is quite thic) it will contribute significantly to the transient effect and it is therefore considered better to include it in the f(t) term. *ffect of heat conduction through the earth in vertical direction "f a well is restarted after a process shutdown of limited duration equation DH cannot be used. To account for this it is referred to .hiu >. and Tha)ur =... =5* paper DDFLG (BEEB).

*ffect of heat conduction in the formation in vertical direction "f the effect of the heat loss from the well bore in vertical direction through the formation is to be included it is referred to Aarvi) A. and =trand ;. &5rediction and !easurement of wellhead temperatures( ;13 Qroup BEEF !ultiphase Technology. This effect is however believed not relevant before after very long times since the heat diffusion into the formation then must have reached very far S to a distance in the same order as the well depth. After such long times repeated startup'shutdowns and heat transfer to$from other wells will be more significant than vertical heat loss.

Thermal modeling for bundles and riser to*ers

The following heat transfer mechanisms are relevant: %orced convection between the fluid and the inner wall (wax or steel pipe) .onduction through wax and the steel pipe .onduction through the steel pipes to the fluid or solid in the tower or sleeve pipe (or carry pipe if there is no sleeve pipe) or other layers and external convection to the ambience. <atural convection in eventual the fluid between the different production pipes .onduction in eventual solid between the different production pipes *xternal convection to the ambience .ompared to gas or a solid insulation material in the space it will be best to have water in the space between the production pipes due to the following: Water has much higher heat capacity which gives increased cooldown time. Water (or another liquid) gives good heat transfer between the production pipes such that one line can be )ept warm during shutdown or production at a low rate from the other(s) lines that are producing. With water in the space between the production lines and the thermal insulation outside (in figure B for example the space between and around pipe B D and H) it is possible to simplify the thermal modeling significantly by assuming that the water inside the carry' pipe in the space between and around the production pipes has a homogenous temperature. The thermal modeling is then reduced to a simple symmetrical problem. A further simplification which often is close to correct is to assume that all everything the main insulation layer has the same temperature and that the heat stored in the insulation layer can be neglected. Then steady state and transient simulation can be done with an analytical model. .alculation of cooldown is described under simplified cooldown modeling). 1owever to )eep the weight down such that it can float during tow'out it may be

required to have low'density insulation material in all available empty spaces including the space between the production lines. This complicates thermal modeling significantly and rigorous finite element or finite difference models are required. Simplified stead state calculations for bundles and to*ers ?sing figure B as an example with bare steel pipes B D and H and the space between and around these pipes filled with water. Then the heat transfer between these pipes are so much higher than the heat loss to the ambience that it can be assumed that the temperature is the same for all pipes and the water around them. =implified calculation of temperature drop in axial direction can then be based on the sum of &!.p( (massflow times .p) with the following equation (which is often used for simple individual pipeline calculations also):

(HD) Qiven the assumptions above plus the following: <o mechanical wor) involved (through for production and export pipelines without pumps or compressors as part of the system under consideration) <o effect of composition changes due to evaporation and condensation. <egligible )inetic energy (normally the case for production and export pipelines). *ither a near ideal liquid or little effect of gravitational energy "f it is a near ideal liquid little effect of frictional heating Then this will give good results for the temperature calculations. The ?'value for the total bundle can be defined in a number of ways as long as it gives the correct total heat loss per meter pipe and per -. (8 ? x x /). To avoid errors it would be best to define ?' values per meter pipe but since the normal convention within heat transfer is to define ? as heat transfer per mA (area) and per -. one should stic) to that (although some are defining ? as heat loss per meter). "f the ?'value is to be defined for one single pipe inside the bundle only this should be done such that the calculated temperature drop will be correct (e.g. when using equation HD or with more rigorous models). This is a bit tric)y and the following points may help to do this correctly: "f the requirement for steady state temperature loss for a single pipeline is ?B then if this pipeline is inside a bundle the required ?'value for the same line will be G.J 7 ?B if there is one more producing line inside the bundle and B$H 7 ?B if there is two more producing lines (total of three) inside the bundle assuming that the mass flow times .p are the same in the different lines. "f only one line is producing inside the bundle the ?'value for the single producing line is proportional to the reference diameter. *.g. with carry pipe

diameter D times the inner diameter of a production line the ?'value referred to the carry'pipe diameter is G.J times the ?'value referred to the inner diameter of the single producing line. "f equation HD is integrated it gives the following simple equation for axial temperature drop:

(HH) TD 8 ;ul) fluid temperature out of a pipe with length 0 O-.P Tamb 8 Ambient temperature O-.P / 8 5ipe diameter OmP ? 8 .orresponding overall heat transfer coefficient OW$mA$>P 0 8 5ipe section length OmP mx.p 8 5roduct of fluid massflow and specific heat OW$>P <ote that it is assumed constant ambient temperature within the section. This is not correct but the error is normally negligible if the temperature gradient is not very steep. !igorous transient and stead state calculations O-G. #0QA with the bundle module can be used for rigorous steady state calculations as wells as for transient calculations. The heat transfer between different lines in a pipeline bundle can be calculated with the bundle module. #ne such bundle consists of one carrier line containing the other lines one or more multiphase pipelines (#0QA branches) and a number of additional single phase lines. The carrier line and the additional lines can contain different fluids flowing at different flow rates. 6ery often the carrier line will transport a warm fluid that heats up all the other lines including the multiphase line. The heat transfer between the different lines is calculated based on the flow rates in the different lines their inlet temperatures the ambient conditions (exterior to the carrier line or one of the additional lines) and the thermal characteristics of the various fluids and pipe materials. The flow and temperature in the multiphase lines are calculated using the standard #0QA model and the presence of the carrier fluid only modifies the boundary conditions for the outer wall layer. The ambient temperature is replaced as boundary condition by the heat transfer from the carrier fluid to the outer #0QA wall layer. The bundle model which calculates the heat exchange between the carrier fluid and the fluids in the other lines uses the temperature of the outer wall layer as boundary conditions related to the #0QA lines. The heat transfer from the carrier fluid to the outer wall layer of the #0QA line (9=#"0) can be plotted and if the T;?< (bundle temperature) for the olga'line is

plotted the plotted temperature will be the temperature of the outer wall layer and not the fluid temperature as for the other lines. The WA00 temperature calculation option must be specified when the bundle module is used due to the coupling explained above. The steady state pre'processor in #0QA does not ta)e into account the presence of a bundle and the steady state results are valid for the multiphase pipeline with ambient conditions as specified with the 1*ATT3A<=%*3 )eyword. /uring the transient calculations with the bundle module the ambient conditions are considered to be exterior to the line which has the largest diameter. 1eat transfer from each flow line i into the fluid in the carrier line (?i is the overall 1.T..oeff.) is governed by: 9i2 8 ?i7D7 T 73i7@U27(Ti2 ' TB2) Where U2 is the section length 3i is the inner radius and T is temperature. The indices i and 2 designate line and section respectively with i8B for the carrier line. The ?'value is the combined value of the heat transfer coefficients from the fluid of the individual pipe to its wall (either given or calculated) through the wall and from the wall to the carrier fluid. A minimum heat transfer coefficient for fluid'wall heat transfer can be given using the 1*ATT3A<=%*3 )eyword. The coefficient given applies to both sides of the specified wall..rossovers of fluid from the outlet of one line to the inlet of another can be specified. The fluid can be reheated at the crossover. The #0QA line cannot be part of a crossover as the other fluids are single phase. ;oth the flow rate of and the heat input at the crossovers can be regulated by controllers. =everal bundle groups can be specified in either one branch or in different branches. .rossovers between bundle lines in different branches are possible. The mass flow as well as the inlet temperature in each of the single phase lines can be given as a time series. ;undles filled with conductive material (not flowing) between the different lines can be simulated in an approximate way by using the soil module with multiple #0QA'branches. Other tools The steady state 4.onoco bundle model4 is programmed on basis of the =5* paper JKLBE: 4An analytical !odel for %lowing ;undle =ystem4 Thomas +. /anielson and 0loyd /. ;rown BEEE. .ode and information ref. >ristin %al)$<#.3.$A;;. Another programme developed in the T!% (transient multiphase flow) +"5 which is both steady state and transient will also be available. +atural convection in fluids bet*een pipes %or a hori,ontal cylinder with free convection on outer surface the <usselt number is given as

*quation B ' %ree convection / 8 #uter diameter OmP h 8 <atural convection heat transfer coefficient OW$mA$>P ) 8 1eat conductivity for layer n OW$m$>P Qr 8 Qrashofs number 5r 8 5randtl number r 8 %luid (here water) density O)g$mCP m 8 %luid (here water) viscosity O<s$mAP b 8 Thermal expansion coefficient OB$>P T 8 Temperature difference between wall and bul) fluid (here water) To calculate h iteration is required because h varies with 9 8 total heat transfer. (h is a function T which is given by the 9).

Simplified cooldo*n modeling

Transient temperature calculations An analytical solution is easily obtained for transient temperature calculations if the following conditions are satisfied:

%or steel pipes pipe'in'pipe bundles and towers with all thermal insulation outside the outer steel pipe or carry pipe 1omogenous and equal temperature in the fluid and pipe walls surrounded by the thermal insulation. 1eat stored in the thermal insulation can be neglected. This is often the case as mass of the thermal insulation is low compared to mass of the steel walls and the fluid inside. .onvective heat transfer in the fluid surrounded by the thermal insulation is high compared to the total ?'value. "f there is no forced convection this assumption may not be true but it will give the most conservative cooldown time (the fastest cooldown). "f only natural convection dominates (no forced convection) the overall ?'value may be significantly reduced compared to only conductive resistance to heat loss in the thermal insulation. 1owever calculation of natural convection is by nature very uncertain. "n addition axial fluid motion will normally occur and give an uncertain contribution to forced convection.

The analytical solution for the cooldown time () is: (seconds) 8 mcp$(?7 T 7/)7ln((tin'tamb)$(tend'tamb))

mcp 8 =um of mass times specific heat capacity for all pipe walls and fluid surrounded by the thermal insulation "/ 8 /iameter with which the ?'value is referred to. ? 8 #verall heat transfer coefficient referred to /. "f the assumptions above are not valid it is recommended to apply #0QA.

%undles and riser to*ers

Description of concepts An example of a bundle (=tatoilVs Qullfa)s =outh bundle) is shown below:

%igure B. .ross section of typical pipe bundle An example of the riser tower is shown below:

5/3 8 5roduction flowline Q03 8 Qas lift riser 103 8 1eating line return %igure D. .ross section of a typical riser tower %or the bundle a sleeve steel pipe which is thermally insulated on the inside is wrapped around the flowlines. The sleeve is filled with seawater. This ma)es sense since air inside would have given much less thermal mass and more rapid cooldown. Also it would give very low heat transfer between the heat'up pipe and the flowlines. Around the sleeve'pipe is a carrypipe also filled with seawater. The heat loss from the carry'pipe to the ambience is much larger than from the sleeve'pipe to the carry'pipe. 1ence water temperature between the carrypipe and the sleeve pipe is approximately J -.. A system is provided for heat'up of the flowlines in the bundle by circulation of hot water in the dedicated circulation pipes inside the bundle. This heat'up facility can be used to melt any wax and hydrates. (Wax melting temperature is significantly higher than wax appearance temperature (N IG -.) which is again significantly higher than the hydrate formation temperature (DG ' DJ -.).

T pical data
Heat capacit / conductivit / viscosit and densit 5arameter 5roduced water #il and condensate .p .onductivity 6iscosity +$)g$-. W$m$> .5 IBFL BFGG G.K G.BJ G.J'B G.J'BGGG (H is common) /ensity )g$mC BGHJ LGG'BGGG (FJG common for oil)

Wax Qas .arbon steel BH W .r steel Titan Wet concrete =oil Wet formation roc) "nner carcass .oflon Tetawire$armering 3ilsan %abrictape 55 adhesive *5/! 55 55foam 56.foam 5* 5olycloropren 1/ syntactic foam 5olyurethane foam 6i)oterm <eopren Asphalt

BFGG DJGG JGG JGG JDG DGGG DGGG DGGG IKG DGGG IKG DHGG BGGG DGGG DGGG DGGG DGGG BGGG DGGG BGLG BBHG BJEG BGGG BGGG BFGG

G.BJ G.GJ IJ DD BG G.J 7 B7 H7 IJ G.D G.E G.H B.D G.H G.H G.D G.D G.D G.D G.BH G.BD G.GDF G.DL G.DK G.BJ

<.A. G.GD <.A. <.A. <.A. <.A. <.A. <.A. <.A. <.A. <.A. <.A. <.A. <.A. <.A. <.A. <.A. <.A. <.A. <.A. <.A. <.A. <.A. <.A. <.A.

FJG'BGGG 8 G.LJ75(bar) LFGG LFGG IJIG DJGG BJGG DJGG IDIL BLBL LGJF BGIG KHG BDGG BDGG EGG LIG BGGG EGG BJFG JGG FJ BGGG BGGG EJG

<ote 7 A normal heat conductivity for the soil and earth$formation around the well bore is D W$m$>. %or transient calculations of heat flow into the formation a thermal diffusivity of B.e'K mA$s is recommended. T pical $-values 5art of production system Wells (uninsulated) ?ninsulated trenched and bac)filled flowlines Typical ?'value (W$mA$>) BG DG

%lexible (.oflexip or Wellstream) un'insulated risers 5ipe'in'pipe (5"5) with maximum insulation Typical with maximum insulation for hard flexibles pipes without use of 5"5 concepts 1eat traced pipes e.g. &Thermotite(

BJ G.LJ 'B H I'F 7

Material and hard*are problems

High temperature The following material and hardware problems may occur due to high temperatures: Thermal expansion of wells flowlines and risers. %or the wells thermal expansion results in elongation of the tubing which must be designed for with the :mas tree. To avoid buc)ling of flowlines and export lines due to thermal expansion it is required to install expansion loops and sometimes use roc)'dumping. The very highest operating temperature must be conservatively calculated to assess the needs for expansion loops and roc) dumping. The cost is high and cooling should be considered. !elting of plastic and rubber$asphalt coatings. .orrosion. .orrosion rate increases exponentially with temperature. "n some cases there may be an absolute temperature limit which must not be exceeded. *.g. if the temperature goes above IG -. in seawater then pitting corrosion is a threat for stainless steel. "f the temperature becomes above the highest allowable design temperature shutdown must be made to avoid damage. 1owever note that it ta)es some time to heat up the walls and the flowline such that the high temperature will only last a short period (shorter than what will give the damage) and then shutdown can be avoided. -o* temperatures The material problem that may occur due to low temperatures is that temperatures below minimum design (e.g. for the steel plastic coatings and seals) may occur. 1ydrogen embrittlement in metals during sour service must also be considered.(sour service is when 1D= concentration is above a certain limit). Thermal insulation materials 1eat loss may be significantly increased due to destruction and deterioration of the thermal insulation materials as a result of the following:

.ompression of foam insulation materials if it is set under pressure. This can solved by having using a 5"5 concept such that the material does not need to be pressurised or having a very strong material that can ta)e pressure. .onvection inside gas filled foam insulation materials. "ncreased gas pressure (normally nitrogen or another inert gas) in foam insulation materials has been found to give significant increased convective heat transfer. This was the main reason for why the installed bundle on ;5Vs Troi)a field showed up to BG times the calculated heat loss during the design phase. !echanical damage during installation or creep during operation (due to thermally or chemical induced material changes) may cause water circulation through the insulation material. =mall holes in a plastic coatings may be enough to give significant cooling due to cold water circulation.

%uried pipelines The following problems with respect to heat loss may occur with buried pipelines: "ncreased heat'loss due to bare parts which may occur as a result of buc)ling (due to thermal expansion) and$or removal of soil due to e.g. seawater current. 0arge uncertainty in conductivity of the soil (very depending on moisture and porosity). ?ncertainty in effect of heat loss due to convection in the soil.

Summar og recommended methods

%or rigorous transient and steady state calculations #0QA is recommended for all cases. With bundles and towers #lga is presently the only method but very often simple hand calculations will be sufficient simplifications can often be done as explained above. Qas' lifted wells can be treated as bundles and can thus be modeled with #0QA and the bundle module. "n cases with bundles with conduction between the pipes the soil module in #0QA must be used in addition to the bundle and soil module. %or rigorous simulations with buried flowlines #lga with the soil module should be used. %or simpler pipeline calculations the heat transfer from the pipe to the soil here called ho can be calculated from /avenport T... and .onti 6.+ (ref $K$).

!eferences

.hiu >. and Tha)ur =...: X!odeling of Wellbore 1eat 0osses in /irectional Wells ?nder .hanging "n2ection .onditions V paper =5* DDFLG presented at the BEEB =5* Annual %all !eeting /allas #ct. E'E. pp JBL ' JDF. 1asan A.3. and >abir ..=.: X1eat Transfer /uring Two'5hase %low in Wellbores: 5art " S%ormation Temperature V paper =5* DDFKK presented at the BEEB =5* Annual %all !eeting /allas #ct. E'E. pp IKE ' ILF. 1asan A.3. and >abir ..=.: X1eat Transfer /uring Two'5hase %low in Wellbores: 5art ""' Wellbore %luid Temperature V paper =5* DDEIF presented at the BEEB =5* Annual %all !eeting /allas #ct. E'E. pp KEJ ' LGF. .arslaw 1.=. and +aeger +...: X.onduction of 1eat in =olids V #xford =cience 5ublications #xford ?.>. BEJE. ;arua =. : .omputation of 1eat Transfer in Wellbores in single and dual completions 4 =5* DDFKF. /avenport T... and .onti 6.+. 41eat Transfer 5roblems *ncountered in the 1andling of Waxy .rude #ils in 0arge 5ipelines( +ournal of the "nstitute of 5etroleum( 6ol. JL <umber JJJ'!ay BELB.