[FLASH-USERS] Flow Reversal for Windtunnel with Gravity

Dean Townsley Dean.M.Townsley at ua.edu
Fri May 4 00:35:09 EDT 2012 

Hi Nitesh,

It seems to me like you will need to initialize the non-jet portion of 
your domain in hydrostatic equilibrium so that it doesn't just start 
falling.  A ballistic trajectory with an initial velocity of 15cm/s will 
give you only 0.1 cm rise before it peaks and starts to fall back down.  
And nominally you have fluid pushing down in addition to just gravity.  
Though I'm unsure if your domain is initialized with fluid at rest or 
not, but regardless it will start to fall pretty quickly because there 
is no pressure gradient.

Flash does have a "hydrostatic" boundary condition that works like the 
ones discussed in the paper I referenced, though you would have to put 
in the "hole" for your jet but that might not be too hard a 
modification.  Putting things in hydrostatic equilibrium also should be 
pretty easy since you are using an ideal gas EOS, so the profile is 
analytic with dP/dz = -rho g.  The necessary pressure gradient isn't 
even really that big, but anything is big compared to zero.  With zero 
gradient the fluid on the domain will just immediately start to free-fall.

I think in your second boundary condition the material is probably 
trying to support itself, since the pressure at the bottom boundary is 
free to increase, but there is not enough damping for this to ever find 
an equilibrium.

I hope that helps,

Dean



On 05/03/2012 07:14 PM, Nitesh Attal wrote:
> Hello Dean,
> I would have a look at the paper you mentioned. It looks like my 
> previous discussion with Dangwook was not displayed on the user forum 
> and it is as follows.
> _____________________________________________________________________
> Hello Dongwook,
> Let me describe the problem again just to make sure we are on the same 
> page,
> The details are as follows------------------------
> Coordinate System 2D cylindrical (r and z)
> Left BC-Reflecting
> Right BC-Outflow
> Top BC-Outflow or Diode
> Bottom BC-User
> _________________________________
> User BC details -
> If r<=0.355cm then
>     Vz = 800cm/s
>     Vr = 0 cm/s
>     P = 1atm
>     T = 350K
>     Species - N2 and O2 (AIR) (I plan to change this Later)
> else
>     Vz = 15cm/s
>     Vr = 0 cm/s
>     P = 1atm
>     T = 300K
>     Species - N2 and O2 (AIR)
> _________________________________
> Domain Initialized with Air at 300K and 1atm pressure.
> Direction of Gravity = Z
> Acceleration = (-981.0)
> This means that the gravity is acting against the inlet flow.
> Domain Size - r = 0 to 15 and z = 0 to 21
> _________________________________
> Solvers - Split PPM
> EOS - Multigamma
> _________________________________
>
> I have tried the following,
> 1) Gravity OFF = No flow reversal
> 2) Gravity ON = Flow Reversal
> 3) Gravity ON but only for z>0, this means in the guard cells used to 
> apply the lower BC experience no gravity (acceleration is zero) = Flow 
> reversal
> 4) Gravity ON (Increase Co-flow) = Flow reversal
>
> From the above I concluded that there is pressure rise in the domain 
> when the gravity is turned on because of which flow reversal is 
> observed. Thus I changed the the Inlet BC as follows,
> _________________________________
> If r<=0.355cm then
>     Vz = 800cm/s
>     Vr = 0 cm/s
>     P = *Pressure at (guard+1) cell*
>     T = 350K
>     Species - N2 and O2 (AIR) (I plan to change this Later)
> else
>     Vz = 15cm/s
>     Vr = 0 cm/s
>     P = *Pressure at (guard+1) cell*
>     T = 350K
>     Species - N2 and O2 (AIR)
> _________________________________
>
> This results in NO flow reversal but the pressure in the domain keeps 
> on increasing with time which I dont want to simulate. I want to 
> simulate an atmospheric jet.
> Thank you for helping me out.
> Nitesh
> On 5/3/2012 6:56 PM, Dean Townsley wrote:
>> Nitesh,
>>
>> From your description it sounds like the simulation is doing just 
>> what it is supposed to, just the gravity you are applying is much 
>> stronger than what you might have expected.
>>
>> In order for the fluid to remain static when a simulation is started, 
>> it must be initialized in hydrostatic equilibrium -- i.e. it must 
>> have a pressure/density/temperature gradient of the appropriate size 
>> and direction.  Without this the fluid will just start to "fall" 
>> toward such an equilibrium as soon as the simulation starts.
>>
>> There is some discussion of initializing fluid in hydrostatic 
>> equilibrium in Zingale et al. (2002ApJS..143..539Z 
>> <http://adsabs.harvard.edu/abs/2002ApJS..143..539Z>), though it is in 
>> the astrophysical context.  This also discusses relevant boundary 
>> conditions to support the fluid.
>>
>> If the gravity is too strong, it is possible for an inlet boundary 
>> condition to be inconsistent, i.e. the imposed pressure may not be 
>> high enough to actually support the material in the domain against 
>> gravity.  In this situation the pressure is not consistent with 
>> sustained inflow and things will just get weird.
>>
>> I have implemented a top-to-bottom flow-through hydrostatic domain, 
>> but it is pretty touchy and it doesn't sound like this is what you 
>> are looking for.  For a buoyant jet it seems like you want the 
>> hydrostatic background to be mostly static, not steadily flowing.  
>> Though honestly I'm unsure what you are trying to reproduce.  Maybe 
>> if you have a reference for an example?
>>
>> Dean
>>
>>
>> On 05/02/2012 12:43 PM, Nitesh Attal wrote:
>>> Hello,
>>> 1) I use a constant acceleration of gravity (-981.0) along x 
>>> direction. The x-velocity decreases throughout the domain at early 
>>> time and as time progress the rate at which it decreases increases. 
>>> Also, pressure inside the domain increases where as we are fixing 
>>> the pressure at the inlet boundary (xl). This causes the reverse 
>>> flow, and I am wondering if it has anything to do with the gravity 
>>> implementation or its compatibility with the inlet boundary. What 
>>> should I use for the inflow boundary?
>>> 2) (i)When I changed constant of gravitational acceleration to 
>>> +981.0 along x direction. The velocity inside the domain rises 
>>> rapidly and there is no flow reversal.
>>>      (ii) I would get back on that
>>>      (iii) I would get back on that
>>> 3) I am using FLASH4-beta but observed this in the previous releases 
>>> also(FLASH3.3 and FLASH4-alpha)
>>> On 5/2/2012 1:06 PM, dongwook at flash.uchicago.edu wrote:
>>>> Hi Nitesh,
>>>>
>>>> I am not sure if anyone has replied to your email, but if not, I am 
>>>> sorry
>>>> that your email has not been anwered yet.
>>>>
>>>> Just few questions to understand your issue:
>>>>
>>>> (1) To what direction do you apply your constant gravity? Is this 
>>>> to the
>>>> negetive x-direction, and as a result, do you see flow reversal at the
>>>> xl-boundary?
>>>>
>>>> The Windtunnel uses inflow boundary condition at the xl-boundary, 
>>>> and this
>>>> should not allow any matter to leave across the boundary. But if your
>>>> gravity is applied to the negative x-direction, it may as well be 
>>>> the case
>>>> that flow reversal would happen (depending on how strong your 
>>>> gravity is
>>>> relative to the inflow velocity), but the inflow boundary condition 
>>>> at the
>>>> xl-boundary doesn't seem to be a sensible thing to have.
>>>>
>>>> (2) In case that your gravity is NOT to the negative x-direction 
>>>> but still
>>>> experiencing the flow reversal, it could be useful for you to 
>>>> understand
>>>> your problem by trying several things:
>>>>
>>>>    (i) change gravity directions and identify which one has the 
>>>> flow reversal,
>>>>    (ii) change the magnetitude of gravity and identify the problems 
>>>> as a
>>>> function of its magnitude,
>>>>    (iii) use an alternative solver, for example, if you use the 
>>>> split PPM
>>>> then use the unsplit hydro solver, or vice versa.
>>>>
>>>> (3) By the way, what FLASH do you use? Do you use the most recent 
>>>> release?
>>>>
>>>> Hope this helps.
>>>>
>>>> Best,
>>>> Dongwook
>>>> =========================================
>>>> Dongwook Lee, Ph.D., Research Scientist
>>>> The Flash Center for Computational Science
>>>> The University of Chicago
>>>> 5747 S. Ellis Ave., Room 319
>>>> Chicago, IL 60637
>>>> (773) 834-6830
>>>>
>>>>> Hello All,
>>>>> I intend to perform a Multi-species laminar buoyant Jet simulation.
>>>>> I began with the turning on Gravity in the Supplied Windtunnel test
>>>>> problem.
>>>>> The Windtunnel test problem when ran with constant Gravity results 
>>>>> into
>>>>> complete flow reversal (towards xl-boundary). Even when the 
>>>>> xl-boundary
>>>>> is made outflow flow reversal is observed.
>>>>> Is this a known issue? Could some one suggest how to overcome this?
>>>>> Thanks,
>>>>> Nitesh
>>>>>
>>>>
>>>
>>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://flash.rochester.edu/pipermail/flash-users/attachments/20120503/41d6804e/attachment.htm>

More information about the flash-users mailing list