Grid guidelines

When this applies

Recommendation

Notes

ALWAYS

Stretch factor⁺⁺ ≤ 1.2.

Stretch factors⁺⁺ beyond 1.2 have not been widely tested and are not generally required. However, it may be necessary to use a slightly higher stretch factor⁺⁺ for a dispersion scenario which otherwise requires an impractical number of grid cells, or where the grid has been adjusted to fit a wall. It is not anticipated that results will be adversely affected provided the factor remains below 1.4, but this has not been widely tested.

ALWAYS

Grid does not extend unnecessarily below an impermeable ground surface (and ground covers whole of the horizontal domain).

ALWAYS

Cell size at all ROI(s)⁺ ≤ 5 m.

This ensures that all values are meaningful (further recommendations below will result in a smaller value for most scenarios).

This may need to be increased to avoid an impractically high number of grid cells for dispersion and fire simulations that cover a very large domain. Cell sizes of up 20 m have been used successfully for simulations with ROI(s)+ that are 500 m from the source.

ALWAYS

At least 3 cells between the total domain boundary and any ROI⁺.

This makes it less likely that values are impacted by any unphysical boundary effects.

ALWAYS

Monitor points should not be in the same grid cell as a solid wall (unless the wall is thicker than a single grid cell or forms the simulation boundary, e.g., the ground).

The wall may snap to a cell wall that means the MP is on the other side of it from what was intended.
Note that MPs may be placed in a cell that is bounded by a wall (or panel) with thickness = 0 if the wall (or panel) is positioned on a grid plane.

ALWAYS

Monitor points should be in unblocked cells.

Ideally, monitor points should be in porosity=1 cells. If this is not possible, then we should try to maximise the porosity of the MP cell. If neighbouring cells have higher porosity, then we should move the MP so that it is just inside a neighbouring cell. If neighbouring cells also have low porosity, then a finer grid resolution may be needed to capture any space around the MP.

Geometry present

Large objects should be aligned with the grid and positioned so that edges fall on grid planes.

Applies to objects that occupy > 1.5 grid cells, e.g., walls and decks.

Multiple geometry objects

No unintended partial openings between objects (i.e., no leakage)

Pipes

Solid walls around ‘minus primitive holes’ must be at least 1 cell thick.

This ensures no leakage through the wall (2 cells may be better when a cylinder is ‘subtracted’ from another cylinder)

When this applies

Recommendation

Notes

Explosion, blast and fire

Cell size in all domains ≥ 1 cm.

Cell sizes smaller than this have not been widely tested are there is a possibility that the sub-grid models may fail, in particular the sub-grid model for premixed combustion (explosions) may severely overpredict burning velocities for very fine resolutions.
Where a scenario is particularly fine-scaled (or where grid refinement is needed around a very small feature, e.g., a small point leak), then cells may be smaller than this, however it is recommended that results are checked carefully and if any artifacts are present that indicate model breakdown, then a coarser resolution should be used. Finer resolutions have been tested for some fire scenarios and accuracy has not been adversely impacted.

Explosion

Aspect ratio^** for cells in core domain^* ≤ 2.

The longest cell side should be no more than twice the length of the shortest cell side.

Explosion

Core domain^* should cover:

• The expanded gas cloud(s)

• Any geometry that extends beyond the cloud(s) and is likely to impact on the explosion

• Any vents⁺⁺⁺

• All near-field ROI(s)⁺⁺

Allow for 5% expansion of the gas cloud in every direction in which expansion is possible.

In some cases with multiple gas clouds, it may be more practical to create separate core domains for the individual gas clouds.

Some judgement may be required to determine whether geometry beyond the expanded gas cloud is likely to impact on an explosion and a sensitivity study may be helpful.

See recommendation below for how to deal with far-field ROIs which would result in an impractically big core domain.

Explosion

Size for cells covering the expanded gas/dust cloud should be ≤ D/15 in each dimension, where D is the extent of the expanded gas/dust cloud in that dimension.

When multiple clouds are included in the same core domain, this recommendation should be met for every cloud.

The recommended maximum aspect ratio for core domain cells means that clouds with an aspect ratio of greater than 2 should be resolved with more than 15 cells in their maximum dimension.

In cases with small geometry enclosed within the cloud, a sensitivity study is recommended to ensure that the geometry is sufficiently resolved.

Explosion and blast

Distance between the core domain^* boundary and total domain boundary ≥ 3 × the core domain^* extent in every direction.

This reduces the likelihood of boundary effects impacting on calculations in the core domain^* and is particularly important for scenarios with PLANE_WAVE boundary conditions.

Explosion and blast with far-field ROI(s)⁺

Core domain^* cell size should be maintained from the core domain^* to the ROI(s)⁺.

Pressures are calculated at lower resolution over stretched parts of the grid so this recommendation is to avoid peaks being dampened.
If this results in an impractically high number of grid cells, then the grid should be stretched beyond the core domain* until cells have increased in size by a factor of two. For example, a grid with 1 m cells in the core domain* may be stretched outwith the core domain until the cell size reaches 2 m. Cells between this point and and any far-field ROI(s)+ should all have a size of 2 m.

Gas explosion when an external explosion is likely

Minimum distance from vents+++ to core domain* boundary:

• 3 x vessel length in vent-opening direction

• 7 cells in all other directions

For a gas explosion, an initial coarse-resolution simulation may help to determine whether an external explosion is likely. The possibility of external explosion should always be considered when the fuel mix includes hydrogen.

Gas explosion with vents⁺⁺⁺

Minimum distance from vents⁺⁺⁺ to core domain^* boundary:

• 7 cells in vent-opening direction

• 5 cells in all other directions.

Explosion with vents⁺⁺⁺ (including vents covered by panels)

A minimum of 3 cells is recommended across every vent (in every direction).

When this is impractical, then it is important that vents are covered with at least one cell in every direction that has porosity=1.

Explosion with vents⁺⁺⁺ (including vents covered by panels)

The edges of all vents should be on grid planes.

This is particularly important if only one cell is used to resolve a vent ( we recommend 3 cells whenever possible).
When some of a surface constitutes a panel surface that protrudes (or is recessed) slightly from the surrounding surface, then this may lead to grid planes being impractically close together. For these cases, it is recommended that the panel surface is adjusted in the geometry to sit in the same plane as the surface that surrounds it.

Any scenario with an ignition/explosive location

Ignition should not be on a grid line and should be in an unblocked cell.

See note above about recommendation for MPs in unblocked cells.
For blast scenarios, the explosive location may be in a partially blocked cell but an initial simulation should be used to check that the high temperature and pressure region do not partially penetrate the wall.

When this applies

Recommendation

Notes

Explosion, blast and fire

Cell size in all domains ≥ 1 cm.

Cell sizes smaller than this have not been widely tested are there is a possibility that the sub-grid models may fail, in particular the sub-grid model for premixed combustion (explosions) may severely overpredict burning velocities for very fine resolutions.

Where a scenario is particularly fine-scaled (or where grid refinement is needed around a very small feature, e.g., a small point leak), then cells may be smaller than this, however it is recommended that results are checked carefully and if any artifacts are present that indicate model breakdown, then a coarser resolution should be used. Finer resolutions have been tested for some fire scenarios and accuracy has not been adversely impacted.

Dispersion and fire (including pool fire)

The leak direction should be aligned with a grid axis.

The OPEN_SIDES setting ensures that this is the case, however it may not be possible for scenarios with multiple leaks. In these cases, direction cosines can be used to specify the direction for leaks that cannot be aligned with the grid.

Dispersion and fire

Leak should not be positioned on a grid line in the leak direction.

For example, a leak that flows in Z direction should not be positioned on a Z grid line. In the simulation, the leak is assigned to a cell and if it is positioned exactly on a boundary then it may be assigned to the incorrect cell and the leak may flow out of the wrong cell.

Dispersion and fire with area leak(s)

Leak edges should be on grid planes.

When this is not possible, e.g., for an elliptical leak, the leak area should be contained within a whole number of grid cells without necessarily filling every cell (FLACS-CFD will correct for this).

It is important that there is at least one fully unblocked (i.e., porosity =1) cell over the leak.

Dispersion and fire with area leak(s)

The centre of the core domain^* for each leak should be located at the centre of the leak (in the plane of the leak).

Multiple core domains may be required for scenarios with multiple leaks.

Dispersion and fire with area leak(s)

In the plane of the leak, the core domain^* should extend beyond each leak edge by at least one cell.

In the leak direction, the extent of the core domain* should equal the extent of the core domain* in the plane of the leak.

All cells in the core domain should be the same size and shape, but this shape need not be cubic since cells may be elongated in the leak direction by up to an aspect ratio of 5.

Dispersion and fire with area leak(s)

Every leak should be resolved by at least 3 cells in the plane of the leak.

Dispersion and fire with area leak(s)

Cells in a core domain^* that is centered on a leak should have sides <= 4 m in the plane of the leak.

The recommended aspect ratio^** means that this also implies a maximum cell size in the leak direction in the core domain^*.

Dispersion and fire with point leak(s)

The cell containing the leak, plus one cell either side of this (in the leak plane), plus two cells in front of it (in the leak direction) should all be refined, i.e., cells in this area should all be the same size and shape (so not stretched).

All refined cells should be the same size and shape, but this shape need not be cubic since cells may be elongated in the leak direction by up to an aspect ratio of 5.

Dispersion and fire with point leak(s)

Refined cells around the leak should have sides of length (1.25 × expanded leak area)^0.5 in the plane of the leak.

The recommended aspect ratio^** means that this also implies a maximum cell size in the leak direction in the refinement region.

This refinement is not necessary when only results in the far-field are of interest.

Dispersion and fire with point leak(s)

Point leaks should not be in the same grid cell as a solid wall.

See note above for the same recommendation for MPs.

Fire and dispersion with leak(s)

Aspect ratio^** for cells in core domain(s) and refinement region(s) around area or point leak(s) ≤ 5.

Cells may be elongated in the direction of the leak/flame propagation.

In some cases, for example a large flame from a small jet leak, such a small aspect ratio may be impractical. In such cases, the aspect ratio should be kept as low as possible.

Dispersion and fire with time-varying leak(s)

Recommendations here should be met throughout the evolution of the leak.

The time-area profile for time-varying leaks should be checked.

Dispersion and fire with time-varying leak(s)

Leak area(s) should be covered by an odd number of cells in each direction in the plane of the leak.

This prevents leaks them from shrinking to nothing.

Fire with near-field ROI(s) (including pool fire)⁺ and impingement

Any surface on which a flame from a jet or pool fire impinges should be aligned with the grid.

Any scenario with an ignition/explosive location

Ignition should not be on a grid line and should be in an unblocked cell.

See note above about recommendation for MPs in unblocked cells.

For blast scenarios, the explosive location may be in a partially blocked cell but an initial simulation should be used to check that the high temperature and pressure region do not partially penetrate the wall.

Any scenario with an ignition

Ignition should not be in the same grid cell as a solid wall.

See note above about about the same recommendation for MPs.

When this applies

Recommendation

Notes

Pool

Vertical pool region^*** should extend from pool-substrate interface to the height of the bund plus 6 cells.

The height of the pool surface should be used if there is no bund.

Pool

Size of cells in the vertical pool region^*** ≤ pool diameter /25.

For a spreading pool, the pool diameter is the maximum pool diameter.

Pool

Horizontal core domain^* should cover the pool surface area.

For a spreading pool, use the maximum pool surface area.

Pool

Number of cells to resolve pool diameter ≥ 15.

Pool

Cell size in the horizontal core domain ≤ 2 m.

Pool

Total vertical domain extends from the core domain boundary to a distance of 10 × the core domain extent for fire scenarios. For evaporation-dispersion scenarios, substitute ‘vertical refinement region’ for ‘core domain’

Note, no vertical core domain is generally required for pool scenarios with no fire, provided that the recommendation for the vertical pool region is met.

Pool fire

Distance from pool centre to total domain boundary in horizontal plane ≥ 5 × pool diameter.

Pool fire

Vertical core domain^* extends from top of vertical pool region^*** to ≥ 1.5 × flame height.

Note, no vertical core domain is generally required for pool scenarios with no fire, provided that the recommendation for the vertical pool region is met.

When this applies

Recommendation

Notes

Dispersion and fire (including pool fire)

The leak direction should be aligned with a grid axis.

The OPEN_SIDES setting ensures that this is the case, however it may not be possible for scenarios with multiple leaks. In these cases, direction cosines can be used to specify the direction for leaks that cannot be aligned with the grid.

Dispersion and fire

Leak should not be positioned on a grid line in the leak direction.

For example, a leak that flows in Z direction should not be positioned on a Z grid line. In the simulation, the leak is assigned to a cell and if it is positioned exactly on a boundary then it may be assigned to the incorrect cell and the leak may flow out of the wrong cell.

Dispersion and fire with area leak(s)

Leak edges should be on grid planes.

When this is not possible, e.g., for an elliptical leak, the leak area should be contained within a whole number of grid cells without necessarily filling every cell (FLACS-CFD will correct for this).

It is important that there is at least one fully unblocked (i.e., porosity =1) cell over the leak.

Dispersion and fire with area leak(s)

The centre of the core domain^* for each leak should be located at the centre of the leak (in the plane of the leak).

Multiple core domains may be required for scenarios with multiple leaks.

Dispersion and fire with area leak(s)

In the plane of the leak, the core domain^* should extend beyond each leak edge by at least one cell.

In the leak direction, the extent of the core domain* should equal the extent of the core domain* in the plane of the leak.

All cells in the core domain should be the same size and shape, but this shape need not be cubic since cells may be elongated in the leak direction by up to an aspect ratio of 5.

Dispersion and fire with area leak(s)

Every leak should be resolved by at least 3 cells in the plane of the leak.

Dispersion and fire with area leak(s)

Cells in a core domain^* that is centered on a leak should have sides <= 4 m in the plane of the leak.

The recommended aspect ratio^** means that this also implies a maximum cell size in the leak direction in the core domain^*.

Dispersion and fire with point leak(s)

The cell containing the leak, plus one cell either side of this (in the leak plane), plus two cells in front of it (in the leak direction) should all be refined, i.e., cells in this area should all be the same size and shape (so not stretched).

All refined cells should be the same size and shape, but this shape need not be cubic since cells may be elongated in the leak direction by up to an aspect ratio of 5.

Dispersion and fire with point leak(s)

Refined cells around the leak should have sides of length (1.25 × expanded leak area)^0.5 in the plane of the leak.

The recommended aspect ratio^** means that this also implies a maximum cell size in the leak direction in the refinement region.

This refinement is not necessary when only results in the far-field are of interest.

Dispersion and fire with point leak(s)

Point leaks should not be in the same grid cell as a solid wall.

See note above for the same recommendation for MPs.

Fire and dispersion with leak(s)

Aspect ratio^** for cells in core domain(s) and refinement region(s) around area or point leak(s) ≤ 5.

Cells may be elongated in the direction of the leak/flame propagation.

In some cases, for example a large flame from a small jet leak, such a small aspect ratio may be impractical. In such cases, the aspect ratio should be kept as low as possible.

Dispersion and fire with time-varying leak(s)

Recommendations here should be met throughout the evolution of the leak.

The time-area profile for time-varying leaks should be checked.

Dispersion and fire with time-varying leak(s)

Leak area(s) should be covered by an odd number of cells in each direction in the plane of the leak.

This prevents leaks them from shrinking to nothing.

When this applies

Recommendation

Notes

Pool

Vertical pool region^*** should extend from pool-substrate interface to the height of the bund plus 6 cells.

The height of the pool surface should be used if there is no bund.

Pool

Size of cells in the vertical pool region^*** ≤ pool diameter /25.

For a spreading pool, the pool diameter is the maximum pool diameter.

Pool

Horizontal core domain^* should cover the pool surface area.

For a spreading pool, use the maximum pool surface area.

Pool

Number of cells to resolve pool diameter ≥ 15.

Pool

Cell size in the horizontal core domain ≤ 2 m.

Pool

Total vertical domain extends from the core domain boundary to a distance of 10 × the core domain extent for fire scenarios. For evaporation-dispersion scenarios, substitute ‘vertical refinement region’ for ‘core domain’

Note, no vertical core domain is generally required for pool scenarios with no fire, provided that the recommendation for the vertical pool region is met.

When this applies

Recommendation

Notes

Explosion, blast and fire

Cell size in all domains ≥ 1 cm.

Cell sizes smaller than this have not been widely tested are there is a possibility that the sub-grid models may fail, in particular the sub-grid model for premixed combustion (explosions) may severely overpredict burning velocities for very fine resolutions.

Where a scenario is particularly fine-scaled (or where grid refinement is needed around a very small feature, e.g., a small point leak), then cells may be smaller than this, however it is recommended that results are checked carefully and if any artifacts are present that indicate model breakdown, then a coarser resolution should be used. Finer resolutions have been tested for some fire scenarios and accuracy has not been adversely impacted.

Explosion and blast

Distance between the core domain^* boundary and total domain boundary ≥ 3 × the core domain^* extent in every direction.

This reduces the likelihood of boundary effects impacting on calculations in the core domain^* and is particularly important for scenarios with PLANE_WAVE boundary conditions.

Explosion and blast with far-field ROI(s)⁺

Core domain^* cell size should be maintained from the core domain^* to the ROI(s)⁺.

Pressures are calculated at lower resolution over stretched parts of the grid so this recommendation is to avoid peaks being dampened.
If this results in an impractically high number of grid cells, then the grid should be stretched beyond the core domain* until cells have increased in size by a factor of two. For example, a grid with 1 m cells in the core domain* may be stretched outwith the core domain until the cell size reaches 2 m. Cells between this point and and any far-field ROI(s)+ should all have a size of 2 m.

Blast

Aspect ratio^** for cell sides in the horizontal plane in the core domain^* = 1.

If Z is vertical, then the lengths of the X and Y sides of cells in the core domain should be equal. Cells may be elongated in the Z-direction by up to a factor of 2 if the wave propagation in this direction is not of interest (otherwise cells should have an equal length in X, Y and Z).

Blast

Monitor points should not be cells with porosity = 0.5

FLACS-Blast treats porosities as binary, so porosity=0.5 cells are ambiguous.

Blast

The core domain^* must include:

All ROI(s)+

Everywhere where shock wave peak pressure > 1 barg

If some ROI(s)⁺ are very far from the blast and this therefore results in am impractical number of grid cells, then the recommendation below for ‘Explosion and blast with far-field ROI(s)⁺’ can be followed instead.

Blast

The initial high-pressure region should be resolved by at least 5 cells in every direction.

Any scenario with an ignition/explosive location

Ignition should not be on a grid line and should be in an unblocked cell.

See note above about recommendation for MPs in unblocked cells.

For blast scenarios, the explosive location may be in a partially blocked cell but an initial simulation should be used to check that the high temperature and pressure region do not partially penetrate the wall.

When this applies

Recommendation

Notes

Dust explosion with vents⁺⁺⁺

Minimum distance from vents⁺⁺⁺ to core domain^* boundary:

3 × vessel length in vent-opening direction

7 cells in all other directions

Recommended for all dust explosions to ensure that any external explosion is sufficiently resolved.

Any scenario with an ignition/explosive location

Ignition should not be on a grid line and should be in an unblocked cell.

See note above about recommendation for MPs in unblocked cells.

For blast scenarios, the explosive location may be in a partially blocked cell but an initial simulation should be used to check that the high temperature and pressure region do not partially penetrate the wall.

Any scenario with an ignition

Ignition should not be in the same grid cell as a solid wall.

See note above about about the same recommendation for MPs.