Gasspredning over vann konsekvensog risiko-beregninger

Transcription

1 OIL & GAS Gasspredning over vann konsekvensog risiko-beregninger Seminar om effektene og farene ved gassutblåsning under vann Ptil, Stavanger Asmund Huser 4. September DNV GL September 2014 SAFER, SMARTER, GREENER

2 Gasspredning over vann konsekvens- og risiko-beregninger Content Objectives and Background Why consequence and risk assessment History of development project Overview of calculation methods Detailed CFD methods Comparisons between CFD methods Recommended modeling Look-up tables and PlumePro Comparisons between CFD and integral methods Field examples Dewatering of pipeline QRA example Summary and conclusions Wind- and turbulence profile Gas release area, bubblezone Subsea pipeline Subsea gas leak Sea surface Figure Error! No text of specified style in document.-1 Problem report considers only dispersion of gas above sea and the inter 2

3 Gasspredning over vann konsekvens- og risiko-beregninger Content Objectives and Background Why consequence and risk assessment History of development project Overview of calculation methods Detailed CFD methods Comparisons between CFD methods Recommended modeling Look-up tables and PlumePro Comparisons between CFD and integral methods Field examples Dewatering of pipeline QRA example Summary and conclusions Wind- and turbulence profile Gas release area, bubblezone Subsea pipeline Subsea gas leak Sea surface Figure Error! No text of specified style in document.-1 Problem report considers only dispersion of gas above sea and the inter 3

4 Objectives of analyses In planning of subsea equipment and pipeline operations involving potential gas releases Safe distances from releases Need estimates in QRA to find low risk drilling operations and design of subsea arrangement Need for subsea shutdown valves? QRA for Drilling rigs When accident is a fact, need quick good estimates Gas hazard for people on board Gas hazard for approaching vessels Stability hazards for anchored floating vessels 4

5 Possible hazards from subsea leaks Drilling and production platforms Gas from a subsea template or pipeline drifting to platform Gas on rig igniting Flashfire on rig Fire escalation to risers and structure on surface Explosion inside rig Loss of stability of anchored surface vessels Subsea installations and pipeline leaks Gas on surface vessels during operations Flashfire on vessels Explosion in gas on board surface vessels Fire burning back to the sea surface 5

6 Definition of hazards Gas Can cause asphyxiation When ignited burning back to sea surface Fire Flashfire when ignited - threat to personnel in the gas cloud Long lasting fire - Can threat structure and risers on the surface Explosion Explosion only when gas is in congested and/or confined area Bubble plume Can cause strong current of gas and water on the surface away from leak Can cause anchored vessels to take in water Bubbles are small and not a threat to buoyancy Balance between upward flow caused by gas bubbles and gravity 6

7 Snorre A November Water depth 300 m Estimated leak rate, kg/s Gas measured on platform but not ignited by flare During low wind, when wind is shifting direction, more gas was measured Safety zones established 2000 m radius and 3000 ft high (ref Ptil report) Estimates with PlumePro gives max 200 m long and 30 m high 7

8 Project history 2006 onwards 2006 Petroleum Safety Authorities Norway (Ptil) workshop 2007 new Ptil workshop SINTEF develop subsea plume models using CFD 2008 DNV finds reasons for model differences Code developers modify codes to simulate dispersion above sea 2009 SINTEF subsea Model development and testing 2009 DNV survey and Best Practice development 2011 DNV Find recommendations on turbulence model 2013 DNV PlumePro developed JIP by Sintef called SURE initiated 2014 New JIP with subsea CO2 release proposed by DNVGL Work sponsored by Statoil and Gassco 8

9 Gasspredning over vann konsekvens- og risiko-beregninger Content Objectives and Background Why consequence and risk assessment History of development project Overview of calculation methods Detailed CFD methods Comparisons between CFD methods Recommended modeling Look-up tables and PlumePro Comparisons between CFD and integral methods Field examples Dewatering of pipeline QRA example Summary and conclusions Wind- and turbulence profile Gas release area, bubblezone Subsea pipeline Subsea gas leak Sea surface Figure Error! No text of specified style in document.-1 Problem report considers only dispersion of gas above sea and the inter 9

10 Overview of calculation methods - dispersion above sea Integral methods Made for standard scenarios w jet release Phast, UDM Based on open fields, flat terrain, no geometry Quasi transient Fast simulations, less than 1 min. CFD Generalised model and boundary conditions KFX, FLACS, CFX, Fluent, StarCD 3D grid and geometry Solves local flow and thermodynamic equations Full transient development possible Approximate turbulence models Longer simulation times 1-10 hours 10

11 Modelling concepts Status Integral models for jets and plumes in open areas Benefits: Quick to use Drawbacks: Slow area-release not possible, no geometry, not transient CFD models commonly used for gas dispersion above sea Benefits: can model area release, transient effects, geometry Drawbacks: long simulation times Standard k-epsilon gives different results is not fully standard Higher order turbulence models exists full Reynolds stress and LES Open experiments with slow area release on sea surface does not exist in literature Models are not validated with area releases and large scale experiments Physical effects are included in CFD model and therefore CFD models are regarded more accurate than integral models 11

12 Modeling concepts summary Integral models Main drawbacks: Cannot capture surface dynamics and geometry effects Plume with transient release rate. Main strengths: Very fast and simple to program Still workhorses for QRAs when the source is a jet, not for area release. CFD models Main drawbacks: Longer setup and run times Main strengths: More flexible to capture physical effects Can include geometry effects and transience 12

13 Gasspredning over vann konsekvens- og risiko-beregninger Content Objectives and Background Why consequence and risk assessment History of development project Overview of calculation methods Detailed CFD methods Comparisons between CFD methods Recommended modelling Look-up tables and PlumePro Comparisons between CFD and integral methods Field examples Dewatering of pipeline QRA example Summary and conclusions Wind- and turbulence profile Gas release area, bubblezone Subsea pipeline Subsea gas leak Sea surface Figure Error! No text of specified style in document.-1 Problem report considers only dispersion of gas above sea and the inter 13

14 What is a Computational Fluid Dynamics (CFD) model? Solving Navier Stokes (RANS) equations with K-e turbulence model Specify wind inlet boundaryconditions with atmospheric wind profile Specify surface roughness Specify stability class Specify area release Specify gas density Simulate transient or steady state 14

15 LEL max cloud length (m) Blind tests from kg/s 40 kg/s 3m/s 7m/s 3m/s 7m/s DNV FLACS Scandpower FLACS DNV CFX Safetec CFX Scandpower KFX Lilleaker Fluent 0 Case 1 Case 2 Case 3 Case 4 DNV FLACS Scandpower FLACS DNV CFX Safetec CFX Scandpower KFX Lilleaker Fluent

16 CFD models gives different results, reasons Turbulence model differs Inlet wind turbulence conditions are different Long domains are modelled differently Area releases are different 16

17 Turbulence model differs Transport equation for turbulent dissipation, epsilon: Buoyancy production definitions are different: Strength of buoyancy production is different: KFX and CFX: 1 (gives more dissipation and less turbulence: longer plumes) FLACS: Equations Code G k B C C k C k P C x x x u i t i i i FLACS a G k C C k f C k Pf C x x x u i t i i i KFX b G k C C k C k P C x x x u i t i i i CFX CFX KFX FLACS C C,, 3, 3 1

18 Buoyancy production of turbulent dissipation One parameter, C 3, is not specified in the standard k- model Different modelling formulations in literature and codes Dramatic impact on plume heights for low wind speeds (next slides) Recommendations: C3 Code Default Recommended KFX 1 1 FLACS CFX 1 1 C 3, FLACS 1 C3, KFX, CFX 18

19 Buoyancy prod. 1 m/s, D, 450 kg/s Default FLACS Default KFX KFX with C3 as FLACS 19

20 Buoyancy production. 3 m/s, D, 450 kg/s Default FLACS KFX with C3 as FLACS Default KFX 20

21 Buoyancy production. 5 m/s, D, 42 kg/s Default FLACS KFX with C3 as FLACS Default KFX 21

22 Observations Results are similar when buoyancy production term is the same CFD models gives the same results as long as the turbulence models are the same > other model parameters is also modelled the same way Which model should be used? 22

23 Seen from above Side view Comparison with Briggs formula for plumerise CFD model setup: Wind Outlet Release Chimney stack Wall Z X Wind Wind Outlet Release Symmetry Y X Valid for buoyant plumes Low velocity release of smoke from chimneys No near buildings 23

24 Comparing chimney plume simulations Best fit is obtained with KFX and CFX default C3e setting at low wind At high wind speeds the differences are not significant Recommendation: Use the C3e settings in KFX and CFX. FLACS needs to be modified. Note, check impact on other results in FLACS if C3e is changed Warning: This recommendation is based on only one comparison and full scale area release is recommended for validation 450 kg/s, 1 m/s 42 kg/s, 1 m/s 24

25 Best Practice: Air inlet profiles Use log law velocity profile Profile of eddy viscosity important for plume length Eddy viscosity is a function of k and Specify profiles of k and Eddy viscosity is single parameter which governs turbulence effect on dispersion Eddy viscosity profile described by two parameters: surface roughness and stability class Higher roughness gives more turbulence and more spread of plume and shorter plumes Lower stability class gives more turbulence and shorter plumes 25

26 Best Practice values for surface roughness and stability class Low surface roughness compared to over land Increase due to waves at higher wind Wind speed 10 m above sea (m/s) Roughness above Sea, z 0 (m)* Pasquill class Unstable conditions Pasquill class Stable conditions < A, B G 1.5 to A-B, C F, G 3.5 to B, C E, F 5.5 to C, D D, E 6.5 to to C, D D D D * Typical values, lower surface roughness can occur. 26

27 Maintaining turbulence downwind. 15 m/s, 42 kg/s (A) Default FLACS (B) Default KFX (C) KFX tweeked 27

28 Best practice: Maintain downwind turbulence in CFD code Relevant at high wind speeds or long domains Check that eddy viscosity profile is maintained downwind FLACS OK in tests KFX looses turbulence If eddy viscosity profile is not maintained: Try to increase surface roughness on sea surface (not on inlet) Carefully adjust C 28

29 Best Practice: Gas inlet profiles steady state release Tophat radius = 0.64 times the Bubblezone radius, or Tophat radius = 1.6 times the standard deviation of the Gaussian profile then the plume distances similar, typically within 5%. With these relations, the visible bubble zone has typically a Bubble zone radius = 2.5 times the standard deviation of the Gaussian profile. The top-hat flux is 0.8 times the Gaussian centre flux 29

30 Best Practice: Area release boundary condition 2D Gaussian profile can be specified in FLACS and KFX Release is defined by Total release rate (kg/s) Bubblezone radius (m) Need to be obtained from subsea bubble plume simulations 30

31 Gasspredning over vann konsekvens- og risiko-beregninger Content Objectives and Background Why consequence and risk assessment History of development project Overview of calculation methods Detailed CFD methods Comparisons between CFD methods Recommended modeling Look-up tables and PlumePro Comparisons between CFD and integral methods Field examples Dewatering of pipeline QRA example Summary and conclusions Wind- and turbulence profile Gas release area, bubblezone Subsea pipeline Subsea gas leak Sea surface Figure Error! No text of specified style in document.-1 Problem report considers only dispersion of gas above sea and the inter 31

32 PlumePro look-up tables Subsea release matrix of steady state results from SINTEF 2012: Plume radius Rate [kg/s] Depths Surface flux Rate [kg/s] Depths

33 Subsea data plotted Subsea leak rate (kg/s) as parameter Surface gas flowrate vs depth: Bublezone radius vs. depth: 33

34 PlumePro look-up tables Above sea matrix of steady state results from DNVGL 2012: 1/ 2 LFL plume length Bubblezone radius Wind speed Surface flux [m] [m/s] [kg/s]

35 1/2 LFL plume length [m] 1/2 LFL plume length [m] Above sea data plotted Release rate = 100 kg/s r=25 r=44 r=75 r=130 Above sea plume lengths vs. wind speed Bubblezone radius as parameter kg/s: 1000 kg/s: Wind speed [m/s] Release rate = 1000 kg/s r=25 r=44 r=75 r=130 r= Wind speed [m/s] 35

36 PlumePro details Based on CFD results from steady state simulations Subsea plume simulated by SINTEF Above sea gas dispersion simulated by DNVGL Matrixes of results are used Interpolations first in subsea matrix to find bubblezone radius and surface flux Further interpolations in above sea matrix to find cloud lengths and heights 36

37 PlumePro results 37

38 Gasspredning over vann konsekvens- og risiko-beregninger Content Objectives and Background Why consequence and risk assessment History of development project Overview of calculation methods Detailed CFD methods Comparisons between CFD methods Recommended modeling Look-up tables and PlumePro Comparisons between CFD and integral methods Field examples Dewatering of pipeline QRA example Summary and conclusions Wind- and turbulence profile Gas release area, bubblezone Subsea pipeline Subsea gas leak Sea surface Figure Error! No text of specified style in document.-1 Problem report considers only dispersion of gas above sea and the inter 38

39 380 kg/s, 1.5/F Phast plume sizes: H= 84 m, L = 141 m PlumePro plume sizes: H=136 m, L = 128 m Reason, Phast can not specify low release speed Release area larger with PlumePro 39

40 380 kg/s, 5/D Phast plume sizes: H= 50 m, L = 125 m PlumePro plume sizes: H= 37 m, L = 242 m Reason, Phast can not specify low release speed Release area larger with PlumePro 40

41 Gasspredning over vann konsekvens- og risiko-beregninger Content Objectives and Background Why consequence and risk assessment History of development project Overview of calculation methods Detailed CFD methods Comparisons between CFD methods Recommended modeling Look-up tables and PlumePro Comparisons between CFD and integral methods Field examples Dewatering of pipeline QRA example Summary and conclusions Wind- and turbulence profile Gas release area, bubblezone Subsea pipeline Subsea gas leak Sea surface Figure Error! No text of specified style in document.-1 Problem report considers only dispersion of gas above sea and the inter 41

42 Example: Dewatering of pipeline Objective, find safe distance on surface during dewatering When water is out, some gas will also come out Sea depth > 1000 m Bubblezone radius, approx. 300 m Initial gas flowrate, 800 kg/s Steady gas flowrate, 400 kg/s Simulated 3 wind speeds, 1.5, 3 and 10 m/s 42

43 Results at 3 m/s wind Initial flow rate, 800 kg/s Steady flow rate, 400 kg/s 43

44 Findings Max distance to ½ LFL is between 500 and 600 m Large surface bubblezone with 300 m radius 44

45 Example QRA Objective, find risk to platform from subsea and riser leaks Can investigate effect of SSIV Sea depth 300 m Approach Find probability of gas on platform given a subsea leak Run PlumePro with varying leak rates and find bubblezone radius Run KFX with varying wind speed and leak rate and find when gas hits platform 45

46 Example KFX simulations used in QRA Large subsea leak, bubblezone under platform, At which wind will gas reach platform? 46

47 Typical findings Wind speeds less than 10 m/s gas can reach bottom parts of platform Leak rates must be large or full bore rupture to reach platform Risk is small when number of risers are small. For large platforms with many risers, risk can be significant SSIV can not reduce initial leak rate subsea SSIV can reduce duration and ignition probability and hence fire and explosion risk Snorre accident show that for long lasting leaks will at some point in time wind speed will be reduced and gas can reach platform 47

48 Gasspredning over vann konsekvens- og risiko-beregninger Content Objectives and Background Why consequence and risk assessment History of development project Overview of calculation methods Detailed CFD methods Comparisons between CFD methods Recommended modeling Look-up tables and PlumePro Comparisons between CFD and integral methods Field examples Dewatering of pipeline QRA example Summary and conclusions Wind- and turbulence profile Gas release area, bubblezone Subsea pipeline Subsea gas leak Sea surface Figure Error! No text of specified style in document.-1 Problem report considers only dispersion of gas above sea and the inter 48

49 Summary and conclusions Above sea; Best practice for CFD modeling developed: Recommendations for buoyancy turbulence model given Bubble zone definitions given Air inlet boundary conditions specified Importance of maintaining turbulence downwind Can reproduce results using different codes Large scale area release validation experiments needed PlumePro tool developed to be used in quick calculations In comparison with integral methods longer gas plumes are observed with CFD Transient development important to consider for large leaks 49

50 Further work Above sea: Find best buoyancy turbulence model by comparing with relevant experiment Establish large scale experiment with area release and gas measurement above sea Update SINTEF matrix of cases Publish Best practice for CFD modeling of dispersion above sea CO2 subsea release JIP 50

51 DNV GL to address risks from offshore CO 2 pipelines New participants welcome (Petronas, National Grid, ENI and Petrobras are in JIP) Application to Climit Experiments in 3 and 10 m pond CO2 gas Measurements of gas concentr ation above sea High speed video below water Study subsea jet for ice and jet behaviour Compare with CFD Subsea and above sea Oxygen Water Tank Sensors Wind Direction 3m 3m 5m Separation Between Sensors 3m Separation Between Sensors 10m 10m 20m 45m 51

52 Thank You! SAFER, SMARTER, GREENER 52