Upload
leque
View
232
Download
2
Embed Size (px)
Citation preview
Ventilation rate with two different CFD-codes
Jan Terje Birkeland
FLACS User Group meeting, Bergen, May 3, 2017
Ventilation rate with two different CFD-codes
08.05.2017 2
• Requirements to ventilation in areas where gas leakages may occur
• Recommendations from software providers
• Comparison of ventilation simulations with KFX and FLACS
Ventilation rate with two different CFD-codes
WHO WON?
08.05.2017 3
Not a competition…
Not compared to measured ventilation rates
Not a master’s thesis
Main deliveries from an engineering perspective
4
QRA Technical
safety
CFD:
Explosion Gas dispersion Ventilation Fire
Working environment
/ Environmental aspects
08.05.2017
Why compare?
• Illuminate an –initially- relatively simple and normal issue in offshore development projects where CFD tools are used
• There is often limited opportunity (time/resources) to evaluate all aspects of the modelling, however quality check and grid sensitivity is performed
• Guidelines from software provider are therefore contributing to assuring quality
08.05.2017 5
Requirements to ventilation in areas where gas leakages may occur, NCS
• Facility regulations §14: – Ventilation in indoor- and outdoor areas shall cover the need for air
exchange and provide acceptable air quality.
• Guideline to Facility regulations §14: – To fulfill the requirements for ventilation as mentioned in the first
subsection, the standards NS-EN ISO 15138, NORSOK H-003 og S-001, chapter 16.4 should be used.
08.05.2017 6
Requirements of natural ventilation in areas with danger of gas leakage, NCS
• NORSOK S-001: – Natural ventilation in hazardous areas shall be as good as possible and
shall as a minimum provide an average ventilation rate of 12 AC/h for 95 % of the time. The ventilation rate shall be provided throughout the area to avoid stagnant zones.
– Potential stagnant zones shall be evaluated and precautions taken where considered necessary.
• NORSOK H-001 and NS-EN ISO 15138: approximately same wording
𝐴𝐶𝐻 =𝑄
𝑉 × 3600
08.05.2017 7
Recommendations for ventilation simulations
• Simulation volume
• Grid
• Boundary conditions
This presentation does not include all of the recommendations found (e.g. avoiding long, narrow cells in area of interest)- only the most relevant for the comparison are presented.
08.05.2017 8
Recommendations- simulation volume
08.05.2017 9
Geometry area
Sufficient size - must be evaluated by CFD analyst, although some guidelines exists. • FLACS: Minimum 2-3 x the geometry length in all directions.
Run sensitivity in case of doubt
• KFX: Create enough room downstream obstructions to avoid recirculation on the outlet boundary
Simulation volume
Recommendations - grid
08.05.2017 10
Uniform grid ( ca. 1 m)
Grid stretched towards the boundaries (applies to all directions)
• FLACS: Uniform grid at the geometry. 1.5 m cells are OK – outside of the geometry, the grid is stretched towards the boundaries • KFX: Use dense grid at high velocity gradients
Recommended boundary conditions FLACS:
08.05.2017 11
Inflow +X: WIND
Outflow -X: NOZZLE
Parallell -Y: WIND
Parallell +Y: WIND
Ground/Sea -Z: NOZZLE or EULER with solid surface
Parallell +Z: WIND
Parallel: WIND: Forces the wind to flow in parallel, acts approximately as a symmetry condition.
WIND
NOZZLE
WIND
NOZZLE
Ensure that the chosen boundary conditions are the ones that fit best for the issue.
Simulation volume:
Recommended boundary conditions KFX:
08.05.2017 12
Inflow +X: WIND
Outflow -X: WIND
Parallell: WIND: Lets the wind flow out if the direction is out (Free boundary)
Ground/Sea: SOLID
WIND
WIND
WIND
WIND
Simulation volume:
Parallell -Y: WIND
Parallell +Y: WIND
Parallell +Z: WIND
Basis for comparison of FLACS and KFX
• Same 3D model:
FPSO modelled in FLACS,
imported to KFX
• Same grid,
same
simulation volume
08.05.2017 13
Starting point for comparison of FLACS and KFX
• Same scenarios: 7 m/s wind from two directions
– Some more prevention of flow in module with wind direction 3 than wind direction 1
– Some more wind pushed up above the blockage, towards +Z boundary, with wind direction 3 than wind direction 1
08.05.2017 14
Starting point for comparison of FLACS and KFX
• Same wind profile at the inlet boundary
• Sea surface
• Incompressible flow in KFX, compressible in FLACS- will normally not affect ventilation simulations
• Both use k-ε turbulence model
• There are some differences we cannot affect: – Calculation of porosity
– Sub-grid modelling
– Modelling of boundary conditions
08.05.2017 15
Comparison FLACS and KFX
Initial result: needed a larger simulation volume in FLACS than in KFX in order to avoid recirculation on the outlet boundary:
08.05.2017 16
Y boundary : 11 x geometry lengths away
X boundary : 4 x geometry lengths away
Uniform grid (1 m) around
FPSO
Stretched grid towards the boundaries
- Z boundary: sea surface +Z boundary: 10 x geometry lengths away
KFX basecase
FLACS basecase
As KFX, but downstream X: 7x and Y: 17x geometry lengths away
Comparison FLACS and KFX
Wind direction
Volume flow [m3/s] FLACS
Volume flow [m3/s] KFX
% difference KFX / FLACS
1 842 892 +6%
3 269 217 -19%
08.05.2017 17
Basecase: WIND boundary conditions
• Varied horizontal extent of simulation volume
• Varied height of simulation volume
• Varied between boundary conditions WIND and NOZZLE for +Z boundary in FLACS
– WIND acts approximately as a symmetry boundary
– NOZZLE lets the airflow out
• Only boundary condition WIND for +Z in KFX
– Is set as «free boundaries» if the flow is outwards
• Tested boundary conditions SYMMETRY & EULER in FLACS
Additional sensitivities FLACS and KFX
08.05.2017 19
Used simulation volumes, horizontally
08.05.2017 20
Medium: 4x / 11x geometry
Small: 2x / 5x geometry
Large: 7x / 17x geometry in downstream direction (FLACS only) XL: 10x / 23x geometry in downstream direction (FLACS only)
Used simulation volumes, vertically
08.05.2017 21
+130 → 200 (approx 3 x above geometry height)
+680 → 750 (approx 10 x above geometry height)
Geometry, h=70 m
Approx. height above geometry for +Z boundary:
Ground/Sea
Initial scenarios compared
Cases Simul. volume +Z boundary FLACS +Z boundary KFX + Z boundary
1 Medium 750 - WIND/Free
3 Medium 200 - WIND/Free
11 Large 750 WIND -
13 Large 200 WIND -
08.05.2017 22
• Basecase FLACS is case 11
• Basecase KFX is case 1
Basecase according to recommendations
KFX sensitivities simulation volume
08.05.2017 23
Basecase according to recommendations
Variation is within ± 5 %
Lower +Z boundary→ some decrease in volume flow
Less simulation volume → some increase in volume flow
FLACS sensitivities Sim. Vol. and BC
08.05.2017 24
Basecase according to recommendations
Variation is larger than for KFX, but simulation volume is too small (recirculation on outflow boundaries)
Further scenarios FLACS
Cases Simul. volume +Z boundary FLACS +Z boundary
21 XL 750 WIND
22/12/2 XL/Large/Medium 750 NOZZLE
13 Large 200 WIND
14/4 Large/Medium 200 NOZZLE
15 Large 1500 WIND
16 Large 1500 NOZZLE
17 Large 750 SYMMETRY
18 Large 750 EULER
08.05.2017 25
Result of further sensitivities
08.05.2017 26
Varies within about ± 10%, and the trends are corresponding :
• NOZZLE lets more airflow out from the simulation volume the closer the border and the less the simulation volume is
• WIND and too low +Z boundary pushes more airflow in
• SYMMETRY on +Z boundary gives similar results as with WIND, EULER a bit lower
• XL sim vol and 1500 +Z boundary had little impact
Basecase according to guidelines
Noted about results & time and resources vs. accuracy
• FLACS requires longer distance to downstream borders than KFX in order to keep the results from being affected - at least for this FPSO example.
• The results become somewhat different, probably not because of following the guidelines for FLACS and KFX, but inherent code differences.
• For modules with inadequate ventilation, these differences can mean acceptable or unacceptable ventilation when 12 ACH is considered directly.
• BUT - More important - focus on how design may improve instead of running a lot of simulations to get within the requirements (there will always be some uncertainty with the simulations).
• Choice of simulation volume and grid resolution affects the accuracy of the results and time/resource need for the job
• Choose that which is adequate / good enough for the purpose
• We, the users, should keep in mind that guidelines are not strict rules, and think for ourselves
08.05.2017 27