Thanks for the suggestion! My videos are recorded on at 1440p which makes the text appear very small in the videos. I will consider increasing the text size for future videos!
Hello, Thanks for the great Tutorial. Is it possible to set different initial Temperatures for the Components? I Want to simulate the cooling of a weld
Hello, to look at the cooling of a weld you will have to preform a transient simulation. After initialization you can go to initialize>patch and patch different temperatures to cell zones or regions of the model. After starting the simulation the heat will be dissipated over time as the simulation progresses.
Hey nice video and nice explaination ever on youtube. I have been working on brake disc analysis and my objective is to analyse the effect of turbulence on heat transfer on brake disc due to various vane geometries. Could you please upload a video on this. This may help a lot. Thank you
Hello, Joule heating can be preformed in Ansys using the electric potential model. To my knowledge, there is no Seebeck model natively in Fluent. This could definitely be done with a UDF, if you are feeling ambitious. However, the joule heating, Seebeck, and thermal modeling could all be done natively in Ansys Mechanical with the thermal-electric solver. If you still want to track the fluid flow you will have to couple the results of this model model into Fluent. This is a fairly complex model. Let me know if you have any other questions that will help get your model going!
Hello, Workbench can be a little funky with exporting data to CFD-Post. In Fluent, you can try manually exporting the results by going to File>Export to CFD-Post and selecting the variables you want to export on the right side. You can then open a new CFD-Post window and load the results.
Hello Antoine, could you provide some additional information on what you want to do? Is the heat flux going into the chip or out of the chip? In the current tutorial the walls of our chip are inside the air domain. They form a coupled wall shown as "wall-chip-enclosure" and "wall-chip-enclosure-shadow" in the boundary conditions. Since this is a coupled connection some heat flux from the chip will be transferred into the air as it flows past, from the chip walls themselves. You can check exactly how much by going to results>flux>total heat flux and selecting the chip walls.
Yes, definitely! You can model the chip outside of the fluid domain and only have the heatsink inside the fluid flow. The walls of the chip outside the domain can be sent to adiabatic, or 0 heat flux.
Hello, Could you please describe in a short video how to do the same cooling study but in large scale for data Center hall containing many racks and servers
Hey Ahmed, if you're modeling a datacenter HVAC system you often build up a series of simulations, instead of modeling each individual heat sink in each server. It starts my modeling a single server and calculating the total heat output of each server. You can then estimate the total heat output of all the servers in a rack. Each rack will be modeled as a fan boundary condition with a specific pressure jump and temperature jump. These racks are modeled inside the larger data center and the HVAC system in the room is analyzed. If this is what you are interested in I can add it to my list of future videos!
Informative tutorial thanks for your effort 🙏🏻 if I may ask I have question what would you suggest for ansys fluent simulations working on desktop or just laptop with advanced features equal to those of this desktop like Asus laptops? Ps :I'm consider buying on of the above mentioned computers
Hello, I always prefer desktops because they are on average more powerful and can handle the heat of running at full power for a long time during a simulation. In either case, look for a CPU with a good amount of cores, good CPU clock speed, and as much RAM as you can get. Fluent has some GPU, compute capability, but it's still not worth investing in an expensive GPU for Fluent. Also, don't forget a lot of storage! Simulations and results produce some very large data sets. Make sure you have enough room!
If you use the enclosure tool, as shown in the video, there is no need. The tool will automatically apply Boolean operations and remove the volume of the solids from the enclosure.
Hello, this problem was something that I came up with. You can find some example problems online or from textbooks. A favorite book of mine is "Computational Fluid Dynamics, A Practical Approach" by Tu et. al.
This is very helpful thanks so much, but could you tell what is the thermal BC at the wall-chip-enclosure and wall-enclosure-heatsink, because as default it is zero heat flux so no heat transfer between the gas? Thanks
Hi, what will be the effects of scaling up or down all the components while still leaving the input values same (i.e., TDP and inlet airspeed). Also, is using a velocity inlet like in your analysis similar to using a fan with a fan curve? Thank you for this video 👌
Scaling the heatsink will change the amount of heat it can dissipate. Heat transfer is a function of both surface area and fluid speed. The velocity inlet condition will set the pressure and flow rate according to the size of the inlet and downstream geometry to maintain the given velocity. Fan curves usually provide a flow rate vs. pressure graph. In this case the flow rate will depend on the downstream pressure and the velocity will be set depending on the flow rate and size of the inlet. Depending on the specific problem will dictate what kind of inlet condition to use.
Hello Pr, it is a good job, I have question about the properties using to calculate inlet velocity when using nanofluid, I mean we use base fluid or nanofluid properties to calcuate inlet velocity at any Reynolds number?
Hello Влад Калчанов, there is a few ways to achieve this in Fluent. The first would be to use parameterization in Spaceclaim/Workbench. This will allow you to change the basic dimensions of the heat sink like length, width, and height of the fins. You can then use the parametric solver in workbench to run a series of configurations. However, this is more parametric modeling than true freeform topology optimization. To do this in Fluent you can use the Fluent Adjoint Solver. You can check out the Fluent manual on how to utilize the Adjoint solver for topology optimization. I also think this would make a great tutorial, so I will add it to my list for future videos!
@cfdkareem...I want to calculate the thermal and viscous entropy. Please guide how to find that data? I found the entropy in the volume-avg option by selecting Temp. But I don't know how to calculate thermal and viscous entropy. Thanks
Hello, in this case, with such a low Reynolds number, the viscous heating was assumed negligible. If you want to calculate viscous entropy you will have to turn that option on under viscous models > Laminar > Viscous heating, and recalculate.
@@cfdkareem Yes, I found the viscous heating option in viscous model. I found the following statement: Viscous Heating (if enabled) includes the viscous dissipation terms in the energy equation. This option is recommended when you are solving a compressible flow. Note that this option is always turned on when one of the density-based solvers is used; you will not be able to turn it off. In my case, my flow is laminar and incompressible. I am using pressure-based solver as well. In that case, how can I calculate "Viscous entropy" and Thermal Entropy"? Looking forward to your kind response. Thanking you.
hi great video, how would you recommed approaching the same problem with water instead of air, and making it a multiphase problem where some of the water boils off, i am currently trying to do this but i cant seem to achieve convergence of the fluxes [energy]
Hey Muhamd, Thanks for the suggestion. Spray cooling is a fairly complex topic. It can be done using Discrete Phase Modeling (DPM) or VOF-to-DPM. I will add it to my schedule for future videos!
Make sure you have 3 separate bodies in space claim. Sometimes when using the pull tool you can accidentally merge some bodies. You can use the split tool to make sure there are 3 separate bodies for the heat sink, chip, and fluid domain.
I have 40x40x15 mm waterblock with heater attached in the front face. I input heat source to 21mmx35mmx5mm heater alumunium block for 2,000,000 W/m3. And then I set operational temperature to 25degC. The water inlet is 20degC. Why would the heater temp only reach around 20degC. I already doubled the W/m3 and it still around 20degC. Can you help me?
Actually, I have tried to rework your steps in video and has the same result. In other hand, I tried to reuse the geometry and use Fluent with Fluent Meshing with the same parameters in the next steps. And the temperature result is only around 21.8degC. What happened here?
@@willywahyanto9774 Hey Willy, It will be difficult to give you an exact answer without looking at your setup, but I can give some suggestions. First, what do you mena by operational temperature? You should only set the inlet temperature and the heat source. Second, make sure you used shared topology in your geometry setup and that your walls are coupled. If the walls are not transfering heat it will not allow heat transfer to the air. Third, confirm your scale is correct. It is easy to transfer a geometry in meters that should be in mm. This will make your volumetric heat source much smaller than expected. Finally, confirm that the simulation is converging and you are giving enough time for iteration. If you interupt the simulation too quickly the temperatures can be lower than expected. Let me know if any of these help, or if you have any other questions! -Kareem
Hey Karen, if you want to add thermal resistance between the chip and heatsink you can use wall thickness. 1) Create a new material for your interface. 2) Under boundary conditions find the wall between the chip and heatsink. 3) Change the wall thickness to your interface thickness, say .001 m, and change the material (in the lower left corner of the wall dialogue box) to the interface material. Fluent will calculate the thermal resistance using the material properties and the wall thickness. If the interface is thick, say over 1 mm, I would consider adding shell conduction, which will allow in-plane conduction through the interface. Check out the Fluent manual for shell conduction for more details. Thanks!
Hello Amey1723, you can do this by changing the rectangular pattern of the fins in spaceclaim. At 4:00 I discuss how to create the pattern for the fins. In your case, create one fin at .5 mm, select the fin for the pattern, and make 8 copies. If you have already created the geometry you will want to make sure you recreate the external fluid domain to capture the new heatsink geometry. Also, note that these smaller fins and flow channels will require a larger mesh to capture the geometry correctly! Let me know if you have any other questions! -Kareem
If 1 W increases the temperature by that amount, how is it possible to cool a CPU with a TDP of 50W? As I believe most CPUs have a much larger TDP than 1 W.
It mostly boils down to powder density, surface area of heat sink, and air speed. In our problem we have a small chip, with a small heat sink, and a very low air velocity. For a computer CPU the heat is spread out over a larger surface area by the internal heat spreader (IHS). This reduces the power density and allows us to have a large contact area for out heat sink. CPU heat sinks are also much bigger with much more surface area. Finally, CPU fans attached to the heatsink will move the air much faster than 1 m/s over the heatsink. All these factors add up to us being able to safely dissipate power from 50-100W.
you have no idea how much youve helped me . thanks a lot
very great tutorial with more explaination than other vids, I salute you.
Thanks!
thank you for your clear and engaging explanation Sir !! i truly appreciate you teaching style
Wonderful tutorial. What a life saver mate.
Glad it helped!
Best video I found on this by far! It would be helpful if you made your PC's text size larger.
Thanks for the suggestion! My videos are recorded on at 1440p which makes the text appear very small in the videos. I will consider increasing the text size for future videos!
This and Mehdi Khatamifar's video are the best combo for Fin heatsink analysis on entire youtube
very very good tutorial with best way of explanation. Thanks.
Glad it was helpful!
Loved the video and explanation, thanks!
finally thank you so much for this video
It was very helpful. Thank you very much.
Very useful 👍. Thank you 😊
Very informative....👌
thankyou
Hello, Thanks for the great Tutorial. Is it possible to set different initial Temperatures for the Components? I Want to simulate the cooling of a weld
Hello, to look at the cooling of a weld you will have to preform a transient simulation. After initialization you can go to initialize>patch and patch different temperatures to cell zones or regions of the model. After starting the simulation the heat will be dissipated over time as the simulation progresses.
Hey nice video and nice explaination ever on youtube. I have been working on brake disc analysis and my objective is to analyse the effect of turbulence on heat transfer on brake disc due to various vane geometries. Could you please upload a video on this. This may help a lot. Thank you
Hello, first thanks a lot for the turorial it helped me a lot. I had a question, how can I add Seebeck effect and Joule heating in the solution model?
Hello, Joule heating can be preformed in Ansys using the electric potential model. To my knowledge, there is no Seebeck model natively in Fluent. This could definitely be done with a UDF, if you are feeling ambitious. However, the joule heating, Seebeck, and thermal modeling could all be done natively in Ansys Mechanical with the thermal-electric solver. If you still want to track the fluid flow you will have to couple the results of this model model into Fluent. This is a fairly complex model. Let me know if you have any other questions that will help get your model going!
thamks a lot
How to plot Nusselt no. I recorded it during simulation calculation but it doesn't show in cfd post options.
Hello, Workbench can be a little funky with exporting data to CFD-Post. In Fluent, you can try manually exporting the results by going to File>Export to CFD-Post and selecting the variables you want to export on the right side. You can then open a new CFD-Post window and load the results.
@cfdkareem thanks, will try that. I had manually noted it.
Your videos are really a great help for us. ♥️
Hi, nice video! Would be possible to add a heat flux on the chop wall? is correct select heat flux both for chip and chip-wall to be is same?
Hello Antoine, could you provide some additional information on what you want to do? Is the heat flux going into the chip or out of the chip?
In the current tutorial the walls of our chip are inside the air domain. They form a coupled wall shown as "wall-chip-enclosure" and "wall-chip-enclosure-shadow" in the boundary conditions. Since this is a coupled connection some heat flux from the chip will be transferred into the air as it flows past, from the chip walls themselves. You can check exactly how much by going to results>flux>total heat flux and selecting the chip walls.
is it possible to do this analysis with the chip not being included in the fluid enclosure? with only the heat sink part being exposed to the liquid?
Yes, definitely! You can model the chip outside of the fluid domain and only have the heatsink inside the fluid flow. The walls of the chip outside the domain can be sent to adiabatic, or 0 heat flux.
best
Hello,
Could you please describe in a short video how to do the same cooling study but in large scale for data Center hall containing many racks and servers
Hey Ahmed, if you're modeling a datacenter HVAC system you often build up a series of simulations, instead of modeling each individual heat sink in each server. It starts my modeling a single server and calculating the total heat output of each server. You can then estimate the total heat output of all the servers in a rack. Each rack will be modeled as a fan boundary condition with a specific pressure jump and temperature jump. These racks are modeled inside the larger data center and the HVAC system in the room is analyzed. If this is what you are interested in I can add it to my list of future videos!
Informative tutorial thanks for your effort 🙏🏻 if I may ask I have question what would you suggest for ansys fluent simulations working on desktop or just laptop with advanced features equal to those of this desktop like Asus laptops?
Ps :I'm consider buying on of the above mentioned computers
Hello, I always prefer desktops because they are on average more powerful and can handle the heat of running at full power for a long time during a simulation. In either case, look for a CPU with a good amount of cores, good CPU clock speed, and as much RAM as you can get. Fluent has some GPU, compute capability, but it's still not worth investing in an expensive GPU for Fluent. Also, don't forget a lot of storage! Simulations and results produce some very large data sets. Make sure you have enough room!
Hi, do we need to subtract the air domain by solid domain using boolean operation modelling?
If you use the enclosure tool, as shown in the video, there is no need. The tool will automatically apply Boolean operations and remove the volume of the solids from the enclosure.
From where ,I get this types of problems ?
Hello, this problem was something that I came up with. You can find some example problems online or from textbooks. A favorite book of mine is "Computational Fluid Dynamics, A Practical Approach" by Tu et. al.
@@cfdkareem Thank you
This is very helpful thanks so much, but could you tell what is the thermal BC at the wall-chip-enclosure and wall-enclosure-heatsink, because as default it is zero heat flux so no heat transfer between the gas? Thanks
oh I found it, shared topology :|
thanks a lot, really helpful
Hi, what will be the effects of scaling up or down all the components while still leaving the input values same (i.e., TDP and inlet airspeed).
Also, is using a velocity inlet like in your analysis similar to using a fan with a fan curve?
Thank you for this video 👌
Scaling the heatsink will change the amount of heat it can dissipate. Heat transfer is a function of both surface area and fluid speed. The velocity inlet condition will set the pressure and flow rate according to the size of the inlet and downstream geometry to maintain the given velocity. Fan curves usually provide a flow rate vs. pressure graph. In this case the flow rate will depend on the downstream pressure and the velocity will be set depending on the flow rate and size of the inlet. Depending on the specific problem will dictate what kind of inlet condition to use.
Hello Pr, it is a good job, I have question about the properties using to calculate inlet velocity when using nanofluid, I mean we use base fluid or nanofluid properties to calcuate inlet velocity at any Reynolds number?
Hello, you would use the properties of the nano fluid to calculate the inlet velocity. Thanks!
Thank you so much Pr.@@cfdkareem
for three day, with bugs and errors, but in the end it worked
Great to hear! Could you describe your bugs/solutions you faced? I'll make sure to cover them in future videos.
Do you know if we can combine this forced convection with topology optimization for better heatsink design?
Hello Влад Калчанов, there is a few ways to achieve this in Fluent. The first would be to use parameterization in Spaceclaim/Workbench. This will allow you to change the basic dimensions of the heat sink like length, width, and height of the fins. You can then use the parametric solver in workbench to run a series of configurations. However, this is more parametric modeling than true freeform topology optimization. To do this in Fluent you can use the Fluent Adjoint Solver. You can check out the Fluent manual on how to utilize the Adjoint solver for topology optimization. I also think this would make a great tutorial, so I will add it to my list for future videos!
@cfdkareem...I want to calculate the thermal and viscous entropy. Please guide how to find that data? I found the entropy in the volume-avg option by selecting Temp. But I don't know how to calculate thermal and viscous entropy.
Thanks
Hello, in this case, with such a low Reynolds number, the viscous heating was assumed negligible. If you want to calculate viscous entropy you will have to turn that option on under viscous models > Laminar > Viscous heating, and recalculate.
@@cfdkareem Yes, I found the viscous heating option in viscous model. I found the following statement:
Viscous Heating
(if enabled) includes the viscous dissipation terms in the energy equation. This option is recommended when you are solving a compressible flow. Note that this option is always turned on when one of the density-based solvers is used; you will not be able to turn it off.
In my case, my flow is laminar and incompressible. I am using pressure-based solver as well.
In that case, how can I calculate "Viscous entropy" and Thermal Entropy"?
Looking forward to your kind response. Thanking you.
hi great video, how would you recommed approaching the same problem with water instead of air, and making it a multiphase problem where some of the water boils off, i am currently trying to do this but i cant seem to achieve convergence of the fluxes [energy]
Boiling is a fairly complex simulation which will require multiphase and the boiling/condensation model. Is this what you are trying?
Hi, can u teach me on how to pulse injection cooling/spray cooling
Hey Muhamd, Thanks for the suggestion. Spray cooling is a fairly complex topic. It can be done using Discrete Phase Modeling (DPM) or VOF-to-DPM. I will add it to my schedule for future videos!
It s very excellent.
Hi! Thank you for video. I've only 2 regions in Fluent Meshing. Why? Is it because I'm doing share topology in spaceclaim?
Make sure you have 3 separate bodies in space claim. Sometimes when using the pull tool you can accidentally merge some bodies. You can use the split tool to make sure there are 3 separate bodies for the heat sink, chip, and fluid domain.
Hii,
could you please design a prismatic Li-ion battery pack of 4P23S in SolidWorks/Ansys with mechanical support?
I have 40x40x15 mm waterblock with heater attached in the front face. I input heat source to 21mmx35mmx5mm heater alumunium block for 2,000,000 W/m3. And then I set operational temperature to 25degC. The water inlet is 20degC. Why would the heater temp only reach around 20degC. I already doubled the W/m3 and it still around 20degC. Can you help me?
Actually, I have tried to rework your steps in video and has the same result. In other hand, I tried to reuse the geometry and use Fluent with Fluent Meshing with the same parameters in the next steps. And the temperature result is only around 21.8degC. What happened here?
@@willywahyanto9774 Hey Willy, It will be difficult to give you an exact answer without looking at your setup, but I can give some suggestions. First, what do you mena by operational temperature? You should only set the inlet temperature and the heat source. Second, make sure you used shared topology in your geometry setup and that your walls are coupled. If the walls are not transfering heat it will not allow heat transfer to the air. Third, confirm your scale is correct. It is easy to transfer a geometry in meters that should be in mm. This will make your volumetric heat source much smaller than expected. Finally, confirm that the simulation is converging and you are giving enough time for iteration. If you interupt the simulation too quickly the temperatures can be lower than expected. Let me know if any of these help, or if you have any other questions! -Kareem
Hey can tell me if I add in between core and heatsink some interface which condition I have to change
Hey Karen, if you want to add thermal resistance between the chip and heatsink you can use wall thickness.
1) Create a new material for your interface.
2) Under boundary conditions find the wall between the chip and heatsink.
3) Change the wall thickness to your interface thickness, say .001 m, and change the material (in the lower left corner of the wall dialogue box) to the interface material.
Fluent will calculate the thermal resistance using the material properties and the wall thickness. If the interface is thick, say over 1 mm, I would consider adding shell conduction, which will allow in-plane conduction through the interface. Check out the Fluent manual for shell conduction for more details.
Thanks!
@@cfdkareem thanks
Can you do the analysis like that type using heat sink using interface material
I want to increase the number of fins from 4 to 8 and to decrease the width from 1mm to 0.5mm how to edit. please help
Hello Amey1723, you can do this by changing the rectangular pattern of the fins in spaceclaim. At 4:00 I discuss how to create the pattern for the fins. In your case, create one fin at .5 mm, select the fin for the pattern, and make 8 copies. If you have already created the geometry you will want to make sure you recreate the external fluid domain to capture the new heatsink geometry. Also, note that these smaller fins and flow channels will require a larger mesh to capture the geometry correctly! Let me know if you have any other questions! -Kareem
If 1 W increases the temperature by that amount, how is it possible to cool a CPU with a TDP of 50W? As I believe most CPUs have a much larger TDP than 1 W.
It mostly boils down to powder density, surface area of heat sink, and air speed. In our problem we have a small chip, with a small heat sink, and a very low air velocity. For a computer CPU the heat is spread out over a larger surface area by the internal heat spreader (IHS). This reduces the power density and allows us to have a large contact area for out heat sink. CPU heat sinks are also much bigger with much more surface area. Finally, CPU fans attached to the heatsink will move the air much faster than 1 m/s over the heatsink. All these factors add up to us being able to safely dissipate power from 50-100W.
bro can you give me your mail or any contact socila media because i have some Doubts which only you can help me🥺
Hello, you can contact me at kareemcfd@gmail.com. Please give me a few days to respond! Thanks