Change FDS default value for SURFACE_DIAMETER?

[IFAB-AS]

Hi FDS-Team,

we simulated lots of water mist in recent years. During the latest project (hot aisle spray pattern simulation in a data center), I noticed something unexpected. If the water hits components (e.g. large pipes and other components in the spray path), it can drop down from these components. But if no value for SURFACE_DIAMETER is set, the droplets use the same diameter as they had before. Meaning a 100 micrometer droplet will stay a 100 micrometer droplet.

If the air is moving below (e.g. 5 m/s), you will get a completely different distribution at the floor level or in the air volume due to this as the real droplets would be much larger and less affected by the air. I noticed you can set the SURFACE_DIAMETER manually to adjust, but doesn't it make sense to use a standard value here to get more reasonable results with the standard setting? E.G. SURFACE_DIAMETER=2000?

I have not found any discussions in Github/google groups on this, and I expect that most people in the industry are using the unset SURFACE_DIAMETER value. What do you think?

[drjfloyd]

Converting this to a discussion.

[drjfloyd]

It is not surprising to me that there has been no discussion on this. Most simulations using water sprays are not water mist systems but standard sprinklers with larger droplet sizes. In those uses cases the quantities of interest are typically activation times and the impact of the spray on the smoke layer. Dripping is not important for either of these.

Keep in mind that FDS makes very simple assumptions on how droplets behave once they hit a solid obstruction. It is not forming rivulets, there is no dependence on movement speed on the type of surface, no penetration of water into porous surfaces, etc. You are likely to have errors in where FDS determines drips occur vs where they would actually occur. That aside, in your case I would agree that setting a larger SURFACE_DIAMETER is a more reasonable approach.

[IFAB-AS]

The issue is not that FDS should model the complete physics of droplet formation after surface impact. The current default assumption for the diameter (secondary droplets keep the original diameter) is unlikely to be representative for the water mist cases, and the user does not notice.

If the user does not specify SURFACE_DIAMETER, FDS still makes a physical assumption. However, the user is usually not aware of this and typically has no information to determine a better value.

Would it make sense to use a default value in FDS in the typical range of secondary droplets instead? One value that is representative for both (sprinkler and water mist) secondary droplets? For example, a value of 2000 µm or some distribution within 1-3 mm would probably be closer to reality for many practical cases than keeping the original droplet diameter.

[mcgratta]

I believe that originally there was a default droplet diameter, but we added in the current option later at some user's request. The diameter of a surface droplet matters initially because a heat transfer/evaporation computation is still performed. But when enough water collects on a particular surface cell, it is assumed to form a film and the droplet diameter no longer matters.

As an aside, there is only one case in the Validation Guide, I think, that addresses directly the issue of water evaporation from a hot surface. In that case, the surface is very hot and the water evaporates so rapidly that it is not possible to test these various strategies of re-assigning the droplet diameter. If you know of such experiments that we can use, let us know. It would be nice to tighten up this bit of code.

[IFAB-AS]

In https://www.researchgate.net/publication/231143777_Measurement_of_surface_tension_by_the_dripping_from_a_needle they used a needle + dripping to measure the surface tension. The average diameter for 50 droplets with 0.88 g would be approximately 3.2 mm (Demineralized water and spherical droplets assumed, see Table 2).
For trees, the median D_V50 for throughfall was measured in https://wires.onlinelibrary.wiley.com/doi/10.1002/wat2.1225. Values for various plants are given in table 2 (throughfall contains various droplets, which means rain, splashing, canopy drip etc., 1 mm - 6.3 mm were measured depending on the tree). D50 for the canopy drip (which is nearly what we are seeking) is >2-3 mm if we have heavy rain: "Throughfall categorized as canopy drip usually comprises more than half of the total volume of throughfall, and thus D50 is usually higher than 2–3 mm. For open rainfall, these larger raindrops are rare and the volume percentage is low and only observed at high rainfall intensities; for example, a 3 mm D50 value is observed at a rainfall intensity >100 mm h-1"

So maybe a good default value would be SURFACE_DIAMETER=3000, what do you think? If you need a specific type of measurement, I can also spray water against a surface or something else and collect the droplets. Counting the number of droplets and measuring the total weight would also give a good default value.

[drjfloyd]

Size of droplets form a drip is function of drip rate and the surface. I have seen numbers ranging from 1 to 4 mm. The primary purpose of SURFACE_DIAMETER is not the drip size, it is the size of the droplet that is formed on the surface for doing heat transfer and deciding when enough water is present to consider it as a film. I would expect that the size on the surface is going to be dependent on the surface and the size, impact velocity, and rate of the droplets hitting the surface. With surface heat transfer being the primary purpose, the default of being the drop size may still be reasonable.

[IFAB-AS]

Ok I see what you mean, the SURFACE_DIAMETER is set when the particle is hitting the surface, not when it is leaving the surface. What about using two values? SURFACE_DIAMETER can stay at it was before, and an additional DRIP_DIAMETER can be used for the dripping? This would match with the description from the user guide, p. 229 "However, when the droplet leaves the solid surface in the form of a “drip,” it reforms into a droplet, and its new diameter can be specified via SURFACE_DIAMETER on the PART line in units of μm."

E.g. in part.f90 approx. l.2540 something such as:

IF (LP%CFACE_INDEX==0 .AND. BC%IOR/=0) THEN

         LP%WALL_INDEX = CELL(IC_NEW)%WALL_INDEX(-BC%IOR)

         IF (WALL(LP%WALL_INDEX)%BOUNDARY_TYPE/=SOLID_BOUNDARY) THEN

            IF (LPC%ADHERE_TO_SOLID) THEN

               SELECT CASE(BC%IOR)

                  CASE( 1)
                     BC%X = BC%X - 0.2_EB*DX(BC%IIG)
                     LP%W = SQRT(2._EB*GRAV*DZ(BC%KKG))

                  CASE(-1)
                     BC%X = BC%X + 0.2_EB*DX(BC%IIG)
                     LP%W = SQRT(2._EB*GRAV*DZ(BC%KKG))

                  CASE( 2)
                     BC%Y = BC%Y - 0.2_EB*DY(BC%JJG)
                     LP%W = SQRT(2._EB*GRAV*DZ(BC%KKG))

                  CASE(-2)
                     BC%Y = BC%Y + 0.2_EB*DY(BC%JJG)
                     LP%W = SQRT(2._EB*GRAV*DZ(BC%KKG))

                  CASE(-3,3)
                     LP%U = -2._EB*LP%U
                     LP%V = -2._EB*LP%V
                     BC%Z = BC%Z - 0.2_EB*DZ(BC%KKG)

               END SELECT

            ENDIF


!**********************************************************************

            IF (LPC%LIQUID_DROPLET .AND. LPC%DRIP_DIAMETER>0._EB) THEN
               R_D_OLD = LP%RADIUS                          ! radius before dripping
               R_D     = 0.5_EB*LPC%DRIP_DIAMETER           ! new radius
               LP%PWT  = LP%PWT*(R_D_OLD/R_D)**3            ! ** mass conservation**
               LP%RADIUS = R_D
               LP%MASS   = FOTHPI*LPC%DENSITY*R_D**3
            ENDIF


!**********************************************************************


            BC%IOR = 0
            LP%WALL_INDEX = 0

         ENDIF


      ELSEIF (LP%CFACE_INDEX==0) THEN

         LP%WALL_INDEX = 0  ! The droplet is not stuck to a WALL cell

      ENDIF

[mcgratta]

I will add this bit of code to be consistent with the idea behind SURFACE_DIAMETER. Its default value will be -1, meaning that it should have no effect unless the user chooses a value.

[mcgratta]

Pull Request