On best practice for CFD in the oil and gas sector Technical safety applications

Transcription

1 On best practice for CFD in the oil and gas sector 1 On best practice for CFD in the oil and gas sector Technical safety applications Dr Prankul Middha 4 th May 2017

2 Abercus On best practice for CFD in the oil and gas sector 3 Abercus is an independent, privately-owned consultancy specialising in advanced engineering simulation within the energy sector computational fluid dynamics (CFD), finite element analysis (FEA), the development of bespoke software tools and teaching/training.

3 On best practice for CFD in the oil and gas sector 4 Agenda Introduction Development of best practice Interpretation of CFD predictions Application area helideck turbulence Summary.

4 On best practice for CFD in the oil and gas sector 5 CFD modelling benefits and challenges One of the major benefits of CFD is that it is a first principles approach, which enables a large degree of flexibility on the applications to which it can be applied However with this flexibility come great responsibility CFD is a tool that can be misused The abstraction and derivation of the mathematical model is entirely down to the analyst/engineer The issue of verification and validation is hugely important for gaining confidence in the CFD approach.

5 On best practice for CFD in the oil and gas sector 6 CFD modelling benefits and challenges With other industry-specific black-box simulation tools, the abstraction and derivation (and possibly validation) of the mathematical model has generally been undertaken by the software vendor This is the main benefit of the black-box approach it s consistent, repeatable and robust for the application for which it s been developed This is also sometimes a limitation of the black-box approach engineers may use the software inappropriately, beyond the envelope for which the tool has been validated (perhaps through ignorance and a lack of training, or because it s the only software tool they have to hand).

6 On best practice for CFD in the oil and gas sector 7 CFD modelling benefits and challenges Often, the issue is not whether CFD can model something it s the validation of the CFD approach for the application of interest Ultimately, combining the first-principles flexibility of CFD with the benefits of a black-box approach is where the industry will head It s important to recognise the envelope of applicability for the tools used and choose an appropriate fit-for-purpose tool for the application of interest do not blame CFD tools if they don t yield a useful prediction they are verified for solving equations, so its probable that the conceptual model has not been correctly defined.

7 On best practice for CFD in the oil and gas sector 8 Agenda Introduction Development of best practice Interpretation of CFD predictions Application area helideck turbulence Summary.

8 On best practice for CFD in the oil and gas sector 9 Need for development of best practice CFD and FEA are becoming increasingly used within the oil and gas sector (technical safety and other things such as flow assurance) Overall guidance relating to use of such tools is at fairly high level, at best (in general) or may not be appropriately transmitted No coordinated system of documentation on how to undertake practical CFD analysis in many of the disciplines identified. As a consequence: Experienced CFD engineers will have their own methods and workflows which will differ, so CFD predictions will be inconsistent across our industry For CFD engineers entering the industry there is no coordinated literature that can assist when establishing new capability. Leads to variation in results and in some cases poor analysis

9 On best practice for CFD in the oil and gas sector 10 Example of poor practice Some examples shown in FLUG presentations from yesterday HPC complacency can also play a role: Attempt to capture all of the topsides congestion explicitly within the CFD mesh, resulting in an excessive mesh comprising around 60 million cells (not FLACS) Not possible to refine the CFD mesh adequately in the regions that were of interest, even with the HPC resource

10 On best practice for CFD in the oil and gas sector 11 Example of poor practice Severe expansion in mesh causes unrealistic distortion of plume envelope

11 On best practice for CFD in the oil and gas sector 12 Example of poor practice Inappropriate meshing can have other consequences: rapid mesh expansion away from the discharge location predicted plume envelopes more spherical than one would expect More worryingly: at lower concentrations the envelope is predicted to stretch back to the vessel, presumably along a line of refined cells near the geometry of the burner boom

12 Example of poor practice On best practice for CFD in the oil and gas sector 13 Inappropriate meshing can have other consequences:

13 On best practice for CFD in the oil and gas sector 14 Steps forward NAFEMS NAFEMS ( is the International Association for the Engineering Modelling, Analysis and Simulation Community NAFEMS focusses on the practical application of numerical engineering simulation techniques such as finite element analysis, computational fluid dynamics, and multibody simulation Establishment of an O&G focus group within NAFEMS to address these concerns, with the focus: to review existing guidance in the oil and gas sector relating to the use of simulation, and identify where there are gaps within the existing guidance develop a coordinated series of How To guides to provide practical guidance for simulation users in the oil and gas sector.

14 On best practice for CFD in the oil and gas sector 15 What will be included in terms of guidance? High-level guidance for simulation engineers and the wider engineering team, Discussion of the wider physics that should be considered Flowcharts to help to guide on whether simulation is fit for purpose Checklists to ensure that if simulation is fit for purpose, all of the important physics are captured Detailed guidance specifically for simulation engineers Worked examples with real numbers relating to the underlying physics that can usefully be used for the purpose of CFD verification. Blind tests to identify inconsistencies Identification of shortcomings of existing CFD software.

15 On best practice for CFD in the oil and gas sector 16 Blind tests and user variability Blind tests are an extremely useful vehicle to identify the variability in the CFD approach due to differences in interpretation of high-level guidance and user variability Blind tests were undertaken by a French working group and this showed significant variations even among the members of the working group, who are all considered to be experts in the field of atmospheric dispersion modelling using CFD. There were several differences identified by the French group, including how to model buoyancy effects, the roughness of the terrain, the release source, and the degree of mesh refinement (Harmo, 2013).

16 On best practice for CFD in the oil and gas sector 17 Example from technical safety space

17 On best practice for CFD in the oil and gas sector 18 Final words Currently several consultancies, large O&G contractors, regulatory bodies and operators are a part of this effort Some CFD providers are also participating Would be good if Gexcon considers to join the activity as FLACS is an important tool for O&G work! Potential to influence work done and also have an independent means of propagating best practice within the industry.

18 On best practice for CFD in the oil and gas sector 19 Agenda Introduction Development of best practice Interpretation of CFD predictions Application area helideck turbulence Summary.

19 On best practice for CFD in the oil and gas sector 20 Interpretation of CFD predictions Once a converged CFD prediction has been achieved, the interpretation of the prediction is also important As an example, consider a buoyant plume dispersing within the atmospheric boundary layer.

20 On best practice for CFD in the oil and gas sector 21 Interpretation of CFD predictions It is well known that the flow structure within a jet or plume in cross flow comprises a pair of counter rotating vortices As a consequence, the highest concentration within the plume, at the centre of each of the vortices, may not lie on the centreline of the dispersing plume.

21 On best practice for CFD in the oil and gas sector 22 Interpretation of CFD predictions In Abercus experience, one common method for conveying CFD predictions for atmospheric dispersion is to present contours of plume concentration/temperature on a plane through the plume source and aligned with the wind direction. View angle for next slide Wind direction Source Projection plot plan view

22 On best practice for CFD in the oil and gas sector 23 Interpretation of CFD predictions For this example, the contour plot along the centreline of the plume provides a good indication of the plume behaviour, in terms of the concentration within the plume and the distance over which the plume extends. Contour plot centreline of plume

23 On best practice for CFD in the oil and gas sector 24 Interpretation of CFD predictions For lower wind speeds, however, the counter rotating vortices may be more apparent, as in the example opposite In this case, a contour plot on the centreline will not pick up the true extent of the plume A projection plot, however, does properly capture this behaviour. View angle Wind direction Projection plot plan view Source

24 On best practice for CFD in the oil and gas sector 25 Interpretation of CFD predictions For lower wind speeds, however, the counter rotating vortices may be more apparent, as in the example opposite In this case, a contour plot on the centreline will not pick up the true extent of the plume A projection plot, however, does properly capture this behaviour. Contour plot centreline of plume

25 On best practice for CFD in the oil and gas sector 26 Interpretation of CFD predictions The envelope plot for the orange concentration, for example, extends further and lower than is apparent from the contour plot on the plume centreline shown opposite. Envelope plot orange contour Contour plot centreline of plume

26 On best practice for CFD in the oil and gas sector 27 Interpretation of CFD predictions Similarly for the yellow concentration Envelope plot orange contour Contour plot centreline of plume

27 On best practice for CFD in the oil and gas sector 28 Interpretation of CFD predictions and the green concentration Envelope plot orange contour Contour plot centreline of plume

28 On best practice for CFD in the oil and gas sector 29 Interpretation of CFD predictions and the cyan concentration. Envelope plot orange contour Contour plot centreline of plume

29 On best practice for CFD in the oil and gas sector 30 Interpretation of CFD predictions Essentially, what we call a projection plot is the superposition of the sequence of envelope plots, and the benefit of this kind of plot is that it always shows the true extent of the plume (projected in the direction of the view) start off with a blue background, which is representative of the background concentration then superimpose each envelope plot, starting with the lower concentrations.

30 On best practice for CFD in the oil and gas sector 31 Interpretation of CFD predictions Projection plot Envelope plot

31 On best practice for CFD in the oil and gas sector 32 Interpretation of CFD predictions Now let us compare the projection plot with the contour plot on the plane along the plume centreline

32 On best practice for CFD in the oil and gas sector 33 Interpretation of CFD predictions Projection plot Contour plot centreline of plume

33 On best practice for CFD in the oil and gas sector 34 Interpretation of CFD predictions The projection plot shows the plume extending further and lower than is apparent from the contour plot along the plume centreline This is the same CFD prediction, but it just demonstrates that due consideration ought to be given to the interpretation of the CFD data The projection plot properly captures the plume extent and behaviour, whereas the contour plot may be misleading.

34 On best practice for CFD in the oil and gas sector 35 Interpretation of CFD predictions > LFL 40% LFL passes close above platform LFL 50 % LFL 40 % LFL 35 % LFL 30 % LFL 25 % LFL 20 % LFL 15 % LFL 10 % LFL 5 % LFL 0 30% LFL extends as far as platform Projection plot Contour plot centreline of plume

35 On best practice for CFD in the oil and gas sector 36 Interpretation of CFD predictions This different interpretation of the CFD data may lead to different inferences from being drawn This is an example of the kind of discussion ongoing within the NAFEMS group.

36 On best practice for CFD in the oil and gas sector 37 Agenda Introduction Development of best practice Interpretation of CFD predictions Application area helideck turbulence Summary.

37 On best practice for CFD in the oil and gas sector 38 Application Area helideck turbulence Based on CAP437 / NORSOK C-004 One of the criteria relates specifically to the vertical component of turbulence It is known that the ABL is highly anisotropic - is it OK to simulate this using an isotropic two-equation turbulence model such as the popular k-epsilon model? Some discussion also in ISO requires that an anisotropic, a Reynolds stress model (RSM) (also known as a differential stress model or DSM) is used for CFD calculation of turbulence over helideck.

38 On best practice for CFD in the oil and gas sector 39 Application Area helideck turbulence There are benefits of using the simpler two-equation models: RSM / DSM based turbulence models are computationally more expensive isotropic turbulence models (k-epsilon) are encouraged due to their ease of convergence and the fact that flow stabilizes quickly Several popular offshore technical safety CFD codes (including FLACS) do not currently include this capability Question: does it make a difference?

39 On best practice for CFD in the oil and gas sector 40 Application Area Helideck Turbulence Recent work presented by two groups comparing helideck turbulence modelling comparing two equation turbulence models (SST and RKE) with RSM/DSM * They have shown that the CFD predictions for turbulence environment can be significantly different for certain conditions * Robinson, et.al. (Hazards 25 conference, 2015) * Venkataraman & Nyridy (OMAE conference, 2015)

40 On best practice for CFD in the oil and gas sector 41 Application Area Helideck Turbulence Use of different approaches can potentially have a large impact E.g. CFX modelling comparing the two approaches (2 equation and RSM)(Robinson et al, 2015) Two-equation turbulence model (isotropic) RSM model (anisotropic)

41 On best practice for CFD in the oil and gas sector 42 Application Area Helideck Turbulence CFX modelling comparing the two approaches (Robinson et al, 2015) 2 equation approach: large areas over the helideck would be deemed to fail the CAP437 standard with the v value in excess of the 1.75 m/s limit However, the DSM approach shows that the helideck would in general pass the CAP437 standard.

42 On best practice for CFD in the oil and gas sector 43 Application Area Helideck Turbulence In recent months, Abercus has repeated two helideck turbulence studies using RSM turbulence modelling, specifically because the original study was undertaken in FLACS and the client has deemed this to be inadequate Our feedback to Gexcon is that some clients do not want to use FLACS for such studies These are clients who have a good understanding of the issue Can Gexcon put inclusion of RSM model in the development plan to enable FLACS to be used? Recognise this as an opportunity to improve FLACS! It might even be easier to implement the RSM than to expend effort justifying the use of a simpler two-equation modelling approach.

43 On best practice for CFD in the oil and gas sector 44 Agenda Introduction Development of best practice Interpretation of CFD predictions Application area helideck turbulence Summary.

44 On best practice for CFD in the oil and gas sector 45 Summary The presentation has aimed to further stress a need for efforts to improve the reliability of CFD predictions O&G focus group at NAFEMS can be a good way forward Proper interpretation of CFD predictions is important There is a need to properly evaluate how FLACS can be used for predictions of helideck turbulence, and recognise opportunities to improve the FLACS code.