4 Replies Latest reply on Jun 30, 2011 12:47 PM by James Allen

    Simulating fire / smoke in FloSim

    James Allen

      Hi,

       

      To support a recent enhancement request Im looking to gauge user interest in simulating fire / smoke in Flosim for development of HVAC system designs. The more people we get onboard the greater likelihood the development team at Dassault Systemes will include this feature in a future release.

       

      We currently use Ansys CFX to model all of our fire simulations in enclosed car parks and tunnels (large well ventilated enclosures). The methodology we use is an inert heat source. Combustion models are also used but that is beyond the scope of this discussion.

       

      I have recently developed a way of simulating an inert fire in Solidworks Flosimulation with very good results. We can produce results in a much shorter timescale than it would take in CFX (within 1 day compared to many days in CFX) so it is of real interest to us to
      invesigate this further.

       

      To use Flosimulation for all fire simulation work we would need to be able to simulate smoke. One method of doing this in CFX is to simulate the smoke as an additional variable defined as a scalar quantity (with units kg/kg) i.e. smoke is expressed as a
      mass fraction and assumed to have the same properties as the primary fluid.

       

      The scalar quantity which is user defined needs to be linked to the transport equation so
      it is carried through the fluid domain with the convective and forced flow terms. The
      additional variable is defined from a user expression using a yield factor and the heat of
      combustion of a given fuel.

       

      In post processing we create a user defined variable to calculate visibility which can be
      done simply by emperical formulae from the mass fraction of the scalar.

       

      So in summary there are three features we would like to see developed:

       

      1. ability to create an additional variable that can be linked to the transport equation.
      2. ability to create user expressions to define 'smoke massflow' and 'visibility'
      parameters.
      3. ability to create a user variable as a post processing feature which is equal to the
      expression 'visibility' (with units in metres).

       

      Cheers taking time to read this . I look forward to reading your comments and ideas on this subject.

       

      James

        • Re: Simulating fire / smoke in FloSim
          Joe Galliera

          I find this interesting, especially in extending the functionality of the software by implementing unique solutions to problems.  You write about recently developing a method of doing this in Flow Simulation, but you do not share this method.  I like the idea of the visibility parameter; possibly good for the HVAC module.

            • Re: Simulating fire / smoke in FloSim
              James Allen

              Hi,

               

              Thank you for your reply. The method of implementing a heat source to represent the fire is quite simple. Determining whether or not the simulation is producing sensible results is a bit more tricky and requires some knowledge of fire dynamics.

               

              Ive outlined the basic steps and some important considerations:

               

              The principle is to define a volume region in which the heat source is imposed.This can be done by creating an unmerged body and deactivating the body within Flosim. A finer computational mesh is then appied to this region, using manual localised meshing, along with the heat source. It is important to ensure that buoyancy is activated in the general settings. Radiation is also an important consideration since it accounts for about 30 to 40% of the heat from the fire so an appropriate sub model for this should also be chosen.

               

              For realistic temperatures in the mid to far field from the fire it is important that the size of fire source region is such that it yields a realistic maximum temperature. This can be achieved by running a series of numerical experiments with the temperature in the fire region defined as one of your goals.

               

              To produce realistic results a maximum temperature within the flaming region should not exceed typically 1200 to 1300°C although it should generally be in this order of magnitude. Emperical correlations for centerline velocity and temperature should also be used to check that the simulation is producing sensible results.

               

              The simplest way to represent the fire is as a circlular pool of fuel i.e the heat source region would be cylindrical or cone shaped.

               

              Definition the flame height can be determined by empericial correlation. A good reference source for this and other fire dynamics related info is Drysdale, Dougal, (1998). "An Introduction to Fire Dynamics, Second Edition"

               

              The above strategy is for a steady state fire i.e.one that doesn't change spatially with respect to time. For a growing fire the volume of the heat source would need to grow relative to the heat production rate up to a maximum volume. This would require a user defined equation which is linked between the fire growth rate and a dimension (diameter of the source) withn Solidworks. In this case the finer mesh region within the fire source would be defined by the maximum volume of the fire i.e the mesh is fixed and does not change dynamically with respect to time. This functionality would need to be included as part of the development if transient fires were to be considered.

               

              James.

                • Re: Simulating fire / smoke in FloSim
                  Bill McEachern

                  On your transient growing fire you can probably do that with existing functionality where you could set a temperature goal at some location where once the temperature at the location (say average temp of the volume of the next ignition region) reaches a certian vaule you could ignite another body representing a step increase  in fire volume and energy output. Obviously the fineness of the steps would be somethng to explore to get a reasonable estimate.

                   

                  On your steady state fire you could also add another species of air to track the smoke - same properties- different name. Obviously it is an approximation as you would need to  control the smoke volume. One could suck in the appropriate amount of air to say the bottom of the fire and then  reinjecting the "smoke" species sucked in air plus gas genreation fromteh fuel. Say put a small cylinder as s sink near the periphery of hte heat generation volume and another "cone" to eject the smoke species also contained within the heat generation volume. Obviously you would have to tune this up a bit to ensure correct behavior (you don't want to be sucking in the smoke you just ejected) but it might be away to get what you are after as far as a steady state fire is concerned. You could then plot the smoke as a function of mass or volme fraction provided youhave some sense of how smoke fraction relates to visibility which you probably do. It is not all that elegant but it might be a workable approximation.

                   

                  Obviously I don't do fires so I might be missing some key subtlety in the proposed approach that would make it unworkable. Thoughts?

                    • Re: Simulating fire / smoke in FloSim
                      James Allen

                      Hi Bill,

                       

                      Sorry its taken me so long to reply, ive been a bit caught up with projects so haven't been able to find time to read your post.

                       

                      I like your ideas and I think the steadystate fire is a good workaround solution. The translation from mass / volume fraction to visibility would have to be tablulated in post processing rather than something you could show visually in the software which is where the software development bit would come in. Its a good idea though and one Im going to explore a bit more.

                       

                      The transient fire is a bit more tricky. The heat source from the fire would need to grow spatially in three dimensions but the temperature of the fluid would not necessary reach the ignition point exactly where you want it to. For example at some small distance adjacent to the fire base, say on the upstream side, the temperature in the simulation may not reach the ignition temperature where as in reality the fuel would ignite due to radiative heat exchange.

                       

                      To ensure that the heat was imposed within the correct region of the mesh and the fire grows at a certain rate regardless of local flow conditions (i.e. an inert fire model), the maximum fire size and growth rate would need to be from user defined input. The growth of the fire is usually specified using idealised growth rates which are based on fire experiments where the mass loss rate of fuel is calculated as a function of time. The user would need to be able to change the diameter of the heat source as a function of the heat release rate up to a maximum value. This maximum diameter would be linked to the volume of the source which is set at a value which yields a realistic maximum temperature.

                       

                      So for these reasons I don't think you could control the location where the heat was imposed by using a temperature goal. For an inert (non reacting) fire model the fire volume has to be described in same way by the user i.e. either as a cylinder or cone shape so that the heat is imposed in the correct location of the mesh.