# Temperature Pull Down Simulation using Fluent

 Register Blogs Members List Search Today's Posts Mark Forums Read

 October 27, 2017, 02:31 Temperature Pull Down Simulation using Fluent #1 Member   Vignesh Lakshmanan Join Date: Nov 2016 Posts: 79 Rep Power: 8 Hello All, I am working on estimating the time taken for temperature pull down inside a deep freezer using Fluent. The deep freezer, which is functional, is switched off. The Lid is opened and left open for sometime, as the ambient air enters inside. The lid is closed. The freezer is switched on, whose Te = -20 deg C My aim is to estimate the time taken to cool down the air at 43 deg C (ambient) to -18 deg C. Im assuming entire freezer is filled with ambient air. For starters, my initial condition is air at 43 deg C and the walls are maintained at a constant temperature of -20 deg C. On solving for natural convection using Boussinesq approximation, the time taken for the required pull down is approx. 115 s, say 2 mins. Is this right? I ask this because experimental analysis show it takes around 2 hours. If my result according to the way I have modelled the problem is right, what actually is fluent solving? By the way, I plot area - weighted average of temperature over an iso-surface inside the domain versus flow time, which directly gives the time taken for cooling. Is this interpretation right? My another concern is, once the freezer is switched on, the walls wont be at -20 deg C, as the evaporator needs some time to reach the specified evaporationg temperature. The air temp of 43 deg C reduces simultaneously along with the the wall temp, until wall temp reaches steady state of -20 deg C. The cooling of air continues further. How to model this temperature - time dependence as Boundary condition in fluent. Thanks in Advance. With Regards Vignesh

 October 31, 2017, 00:50 #2 Senior Member   Lucky Join Date: Apr 2011 Location: Orlando, FL USA Posts: 5,065 Rep Power: 60 Maybe it is related to how you evaluate when the entire refrigerator has reached -20°C. Can you check, using the htc obtained from the simulation perhaps, what the time constant is using a lumped capacitance approach? Probably this number, after you compute it, will be very close to 115s like you have numerically or it will be much closer to the experimental 2 hrs. My guess is it will be much closer to 2 hrs for the next reason... What is the size of your refrigerator? The thermal penetration dept, even if you assume a type 1 semi-finite solid with fixed surface temperature will be very small since the thermal diffusivity of air is 10^-6 m^2/s. That is, in 2s, it is unlikely the entire refrigerator has responded to the -18°C boundary condition. Regarding #7. I have a similar concern. I don't have a good feel for what a typical timescale is for the condenser to reach -18 °C. However, HVAC systems are super-efficient when the delta T is small which seems to counter the argument. That is, they are very quick at reducing the temperature a small amount initially but much slower to lower the temperature further (a sort of exponential decay or diminishing returns scenario).

October 31, 2017, 05:34
#3
Member

Vignesh Lakshmanan
Join Date: Nov 2016
Posts: 79
Rep Power: 8
Dear Lucky Tran,

How am I supposed to do this?
Quote:
 Originally Posted by LuckyTran Can you check, using the htc obtained from the simulation perhaps, what the time constant is using a lumped capacitance approach? Probably this number, after you compute it, will be very close to 115s like you have numerically or it will be much closer to the experimental 2 hrs. My guess is it will be much closer to 2 hrs for the next reason...
My freezer volume is 500 Litres hard top with PUF thickness of 60 mm

As you said, I am getting this exponential behavior of temperature vs time and Heat flux vs time . Physics wise, I think the model is correct. But I am not sure about the result

 November 1, 2017, 10:25 #4 New Member   Join Date: Oct 2017 Posts: 26 Rep Power: 7 I doubt that a freezer can go from 43C to -18C in 115 seconds... Air does not conduct heat fast at all, so two hours definitly makes more sense to me. As for the wall temperature, it will be so fast compared to the cooling of air that you could get away with neglecting it I think.

November 6, 2017, 05:16
#5
Member

Vignesh Lakshmanan
Join Date: Nov 2016
Posts: 79
Rep Power: 8
Quote:
 Originally Posted by LuckyTran Maybe it is related to how you evaluate when the entire refrigerator has reached -20°C. Can you check, using the htc obtained from the simulation perhaps, what the time constant is using a lumped capacitance approach? Probably this number, after you compute it, will be very close to 115s like you have numerically or it will be much closer to the experimental 2 hrs. My guess is it will be much closer to 2 hrs for the next reason... What is the size of your refrigerator? The thermal penetration dept, even if you assume a type 1 semi-finite solid with fixed surface temperature will be very small since the thermal diffusivity of air is 10^-6 m^2/s. That is, in 2s, it is unlikely the entire refrigerator has responded to the -18°C boundary condition. Regarding #7. I have a similar concern. I don't have a good feel for what a typical timescale is for the condenser to reach -18 °C. However, HVAC systems are super-efficient when the delta T is small which seems to counter the argument. That is, they are very quick at reducing the temperature a small amount initially but much slower to lower the temperature further (a sort of exponential decay or diminishing returns scenario).
Dear Lucky Tran,

What I meant was, what HTC to use? That of the steady state or transient run? In both the cases, the HTC at the end of simulation (ie, when the transient run has reached steady state) is of the order of 10^-11. Which value should I use

November 6, 2017, 05:18
#6
Member

Vignesh Lakshmanan
Join Date: Nov 2016
Posts: 79
Rep Power: 8
Quote:
 Originally Posted by Foxhunter I doubt that a freezer can go from 43C to -18C in 115 seconds... Air does not conduct heat fast at all, so two hours definitly makes more sense to me. As for the wall temperature, it will be so fast compared to the cooling of air that you could get away with neglecting it I think.
Dear Foxhunter,

In case I neglect it, already the wall will be having non uniform temperature distribution, wouldnt it be too much approximation? This might increase the error in my results right

November 6, 2017, 05:23
#7
New Member

Join Date: Oct 2017
Posts: 26
Rep Power: 7
Quote:
 Originally Posted by ViLaks Dear Foxhunter, In case I neglect it, already the wall will be having non uniform temperature distribution, wouldnt it be too much approximation? This might increase the error in my results right
Yes it will be more of an approximation. It just depends on how accurate you results have to be.

November 6, 2017, 05:30
#8
Member

Vignesh Lakshmanan
Join Date: Nov 2016
Posts: 79
Rep Power: 8
Quote:
 Originally Posted by Foxhunter Yes it will be more of an approximation. It just depends on how accurate you results have to be.

November 23, 2017, 01:10
#9
Member

Vignesh Lakshmanan
Join Date: Nov 2016
Posts: 79
Rep Power: 8
Hi All,

I have managed to finally simulate the problem . Next part is how to solve the problem. Using natural convection, the time taken for pull down is too long (1.5 hrs). In order to enhance the HT , it is decided to include a fan inside the domain (Refer attachment), so that it will help in improving the rate of convection . But the problem is how to simulate this one? I tried steady state MRF + Natural convection together, no use.If i need to use moving mesh for fan, the time step is too small to run the natural convection. Kindly give me some inputs regarding this.

Regards
Vignesh
Attached Images
 Capture-cfd.PNG (32.4 KB, 10 views)

November 23, 2017, 11:20
#10
New Member

Join Date: Oct 2017
Posts: 26
Rep Power: 7
Quote:
 Originally Posted by ViLaks Hi All, I have managed to finally simulate the problem . Next part is how to solve the problem. Using natural convection, the time taken for pull down is too long (1.5 hrs). In order to enhance the HT , it is decided to include a fan inside the domain (Refer attachment), so that it will help in improving the rate of convection . But the problem is how to simulate this one? I tried steady state MRF + Natural convection together, no use.If i need to use moving mesh for fan, the time step is too small to run the natural convection. Kindly give me some inputs regarding this. Regards Vignesh
I don't know how to simulate the fan, but perhaps calculate the theoretical increase in airspeed, and adjust the Cp value of air? So the air speed will stay the same in the program, but it absorbs heat quicker, giving an indication of the HT?

November 24, 2017, 00:39
#11
Member

Vignesh Lakshmanan
Join Date: Nov 2016
Posts: 79
Rep Power: 8
Quote:
 Originally Posted by Foxhunter I don't know how to simulate the fan, but perhaps calculate the theoretical increase in airspeed, and adjust the Cp value of air? So the air speed will stay the same in the program, but it absorbs heat quicker, giving an indication of the HT?
Dear Foxhunter,

I have planned to run the fan seperately and compute the air flow rate, which then can be given as velocity input, such that the fan sucks from and delivers within the deep freezer.

As you said, I can calculate the theoretical speed, but why should I adjust cp accordingly? I am modelling natural convection, so that cp is calculated according to density changes right?

November 27, 2017, 07:30
#12
New Member

Join Date: Oct 2017
Posts: 26
Rep Power: 7
Quote:
 Originally Posted by ViLaks Dear Foxhunter, I have planned to run the fan seperately and compute the air flow rate, which then can be given as velocity input, such that the fan sucks from and delivers within the deep freezer. As you said, I can calculate the theoretical speed, but why should I adjust cp accordingly? I am modelling natural convection, so that cp is calculated according to density changes right?
Heat transfer = Massflow * Cp * Delta T
In your case, you cannot alter the Massflow since you can't use/simulate a fan. To still change the Heat transfer, you could change the Cp value by looking at the mass flow increase. So calculate the theoretical massflow, let's say it increases by a factor of two. To simulate the new heat transfer, multiply the Cp of air by the massflow factor, two. See where I'm going?

November 28, 2017, 00:34
#13
Member

Vignesh Lakshmanan
Join Date: Nov 2016
Posts: 79
Rep Power: 8
Quote:
 Originally Posted by Foxhunter Heat transfer = Massflow * Cp * Delta T In your case, you cannot alter the Massflow since you can't use/simulate a fan. To still change the Heat transfer, you could change the Cp value by looking at the mass flow increase. So calculate the theoretical massflow, let's say it increases by a factor of two. To simulate the new heat transfer, multiply the Cp of air by the massflow factor, two. See where I'm going?
Yes!!! Thanks !!!
But I was able to simulate fan and natural convection together !!! There was considerable increase in heat transfer !!!

Regards
Vignesh

November 28, 2017, 04:39
#14
New Member

Join Date: Oct 2017
Posts: 26
Rep Power: 7
Quote:
 Originally Posted by ViLaks Yes!!! Thanks !!! But I was able to simulate fan and natural convection together !!! There was considerable increase in heat transfer !!! Regards Vignesh
Happy to have helped!

February 23, 2020, 16:02
#15
New Member

Join Date: Jul 2019
Posts: 27
Rep Power: 5
Hello ViLaks,

I was reading over your thread and I have a problem where I need to achieve a desired temperature inside of a room and was wondering if you were able to achieve the desired internal temperature of -18 C and how much of the air inside was changed to -18 C.

I have a few questions:
1. How did you specify your initial conditions to be 43 C?
2. How did you incorporate the fan condition?
3. From later in the thread it was discussed to alter the cp of air, is this what you did as well?
4. Would you be able to lay out what boundary conditions you used (inlet, outlet, fan, turbulence, etc.)

Thanks so much.

Quote:
 Originally Posted by ViLaks Hello All, I am working on estimating the time taken for temperature pull down inside a deep freezer using Fluent. The deep freezer, which is functional, is switched off. The Lid is opened and left open for sometime, as the ambient air enters inside. The lid is closed. The freezer is switched on, whose Te = -20 deg C My aim is to estimate the time taken to cool down the air at 43 deg C (ambient) to -18 deg C. Im assuming entire freezer is filled with ambient air. For starters, my initial condition is air at 43 deg C and the walls are maintained at a constant temperature of -20 deg C. On solving for natural convection using Boussinesq approximation, the time taken for the required pull down is approx. 115 s, say 2 mins. Is this right? I ask this because experimental analysis show it takes around 2 hours. If my result according to the way I have modelled the problem is right, what actually is fluent solving?

 February 24, 2020, 00:53 #16 Member   Vignesh Lakshmanan Join Date: Nov 2016 Posts: 79 Rep Power: 8 Hi Bran, I used ANSYS Fluent for simulations. To specify initial conditions, I just initialised the problem at 43 °C. I did not proceed with implementing fan for various reasons. I just used a higher capacity compressor and a higher evaporator length. But to answer your question, I used "fan" boundary condition in fluent to simulate fan in steady state. Upon convergence, I switched off fan and switched on Energy equation and transient formulation. Since my problem is inside a closed domain, I did not have any inlet and outlet boundary conditions. Regards VL

 Tags fluent, natural convection, pull-down