Eng-Tips is the largest engineering community on the Internet
Intelligent Work Forums for Engineering Professionals
-
Congratulations waross on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!
How to stimulate the cooling rate of the centre of a large componment? 1
- Thread starter leizhou
- Start date
- Thread starter
- #3
Try this approach: Thermal modulus= Volume/Area
then for time to cool for example is worked by times the modulus by the approprate co-efficant and square it i.e
minutes=[modulus x 1.64]²
this coeffecient here is for time for solidication of steel in a sand mould from say 1600°C to 1500°C
I think the value in air- cooling by memory is about double i.e. = 3.28
There is a shape issue involved here. that makes a small correction if desired for accuracy
I would calculate a 150x150x150 as having a modulus as Mc=2.5cm then correct it 'thermally' to Mth=3.1cm to allow for this 'corner effect'
but it may not be need for what you are doing
it useually only a small error.
For intrest a plate 700x700x71 has the same cooling rate or modulus
the key is the co-efficents
look up a book [R.Wlodawer] directional solidification of castings or simular if you need to go further
Cheers..
then for time to cool for example is worked by times the modulus by the approprate co-efficant and square it i.e
minutes=[modulus x 1.64]²
this coeffecient here is for time for solidication of steel in a sand mould from say 1600°C to 1500°C
I think the value in air- cooling by memory is about double i.e. = 3.28
There is a shape issue involved here. that makes a small correction if desired for accuracy
I would calculate a 150x150x150 as having a modulus as Mc=2.5cm then correct it 'thermally' to Mth=3.1cm to allow for this 'corner effect'
but it may not be need for what you are doing
it useually only a small error.
For intrest a plate 700x700x71 has the same cooling rate or modulus
the key is the co-efficents
look up a book [R.Wlodawer] directional solidification of castings or simular if you need to go further
Cheers..
There are various technical sources that can be used to calculate cooling rates of objects (slab or cylindrical) based on formulas that are based around surface area/volume and thermal diffusivity and conductivity. Several sources are listed below;
(1) Steel casting handbook
(2) Heat Transfer Handbook
(3) Transport Phenomena Book related to Metallurgical Processes
Unless the boundary conditions are defined,mode of heat transfer factored in, thermal calculations are difficult. However, I know people who do this for a fee. Your case is simple,as it involves only solid state cooling and there is no liquid to solid transition.
You can assume near ideal conditions and perform your work.
My suggestion was most simple,you could source a block of 6' cube,drill a hole and give it to any heat treater to record the cooling curve. This is not expensive at all.
Chocolates,men,coffee: are somethings liked better rich!!
(noticed in a coffee shop)
You can assume near ideal conditions and perform your work.
My suggestion was most simple,you could source a block of 6' cube,drill a hole and give it to any heat treater to record the cooling curve. This is not expensive at all.
Chocolates,men,coffee: are somethings liked better rich!!
(noticed in a coffee shop)
- Thread starter
- #7
Thanks for all your advise,
The problem is that I will have several different size and shape. That is why I can not try my direct waste a large block. And also it will cost me 250pound unless I wait unitil the next time when I heat treat all the parts. And that is too late for me.
For the simulation, I was told that it can be done in (pro-engineer, anysis this kind of softerware).
Do you have any experience on that please? Meanwhile, I am checking some book, they are base on 1D or 2D simple calculations.
Many thanks.
Lei
The problem is that I will have several different size and shape. That is why I can not try my direct waste a large block. And also it will cost me 250pound unless I wait unitil the next time when I heat treat all the parts. And that is too late for me.
For the simulation, I was told that it can be done in (pro-engineer, anysis this kind of softerware).
Do you have any experience on that please? Meanwhile, I am checking some book, they are base on 1D or 2D simple calculations.
Many thanks.
Lei
If 250 pound sterlings is high,what kind of a project is it.Any student assignment.ou are looking for some very specific data,not commonly available.Hence you were advised to conduct the tests.Computational analysis need to be verified by experimental results,else there can be gross misintepretations.
I am not being pessimistic,but I am trying to show the pitfalls.
Chocolates,men,coffee: are somethings liked better rich!!
(noticed in a coffee shop)
I am not being pessimistic,but I am trying to show the pitfalls.
Chocolates,men,coffee: are somethings liked better rich!!
(noticed in a coffee shop)
the modern method of simulation the rate of heating or cooling uses finite elements- the conductive heat transfer diff eqn is harmonic and well suited to finite elements. The cubic shape is well suited to standard basis functions. The convective boundary condition imposes certain requirements on the time step used , ralated to the size of the grid element and the fourrier and peclet number. Too large a time step will cause numerical instability, so if you use a canned commercial finite element program, be sure to read the section on selecting the time step- or else you will need to run the program several times for increasingly smaller time steps until the answers converge on a duplicate solution.
Older methods used curves gleaned from the exact solution of the conductive differential equation for a discrete number of geometries and boundary conditions, then you would interpolate between curves for the solution to the particular case at hand. Refer to heat transfer textbooks by Kreith; they usually reprint the solution curves that were originally drawn in the 1930's
Older methods used curves gleaned from the exact solution of the conductive differential equation for a discrete number of geometries and boundary conditions, then you would interpolate between curves for the solution to the particular case at hand. Refer to heat transfer textbooks by Kreith; they usually reprint the solution curves that were originally drawn in the 1930's
- Thread starter
- #10
-
1
- #11
I misunderstood your post the first time, I apologize for that. I will have it deleted to prevent my stupidity being shown here. 
I have done enough finite element simulation in the past. Two best ones you shouldn't go wrong with are Ansys and Abaqus. But they both need months practice to be able to work on any real problems. But any other FEA software available to you could also be used as long as it is capable of doing heat transfer modeling. Not all FEAs support heat transfer.
As arunmrao said, you still need to do experiment to verify your simulation even if you get a expert Ansys user to do the simulation for you. Simulation is very tricky stuff - needing a lot of tailoring work back and forth before you get any reasonable results. Reasonable or not, testing is the only way to know.
If there is no phase transition involved, FEA can abosolutely serve your purpose with their available functions, from a technical viewpoint. I am wondering the same thing. If any FEA already got this built in?
You also can built your own subroutine and couple it together any commercial FEA softward. But it may cost years of hard work to do that.
I have done enough finite element simulation in the past. Two best ones you shouldn't go wrong with are Ansys and Abaqus. But they both need months practice to be able to work on any real problems. But any other FEA software available to you could also be used as long as it is capable of doing heat transfer modeling. Not all FEAs support heat transfer.
As arunmrao said, you still need to do experiment to verify your simulation even if you get a expert Ansys user to do the simulation for you. Simulation is very tricky stuff - needing a lot of tailoring work back and forth before you get any reasonable results. Reasonable or not, testing is the only way to know.
If there is no phase transition involved, FEA can abosolutely serve your purpose with their available functions, from a technical viewpoint. I am wondering the same thing. If any FEA already got this built in?
You also can built your own subroutine and couple it together any commercial FEA softward. But it may cost years of hard work to do that.
I'll check the local eng library this PM for a few heat transfer books and post the solution for a cube. My books are at my other house , 3300 miles away.
The exact solution for a cube is published and is "straighforward" using separation of variables. For your particular case, the steel block's side length, thermal diffusivity, and convective HT coefficient need to be known to solve for teh particular eigenvalues. Are you using air cooling or oil bath quench? What is the initila temp prior to quench - 1800 F? Rather than buy and train to use a FE program, one can more easily program an excel spreasheet to solve for the eigenvalues of the exact solution for a simple geometry such as a cube. The FE method is best suited for unusual geometries.
The exact solution for a cube is published and is "straighforward" using separation of variables. For your particular case, the steel block's side length, thermal diffusivity, and convective HT coefficient need to be known to solve for teh particular eigenvalues. Are you using air cooling or oil bath quench? What is the initila temp prior to quench - 1800 F? Rather than buy and train to use a FE program, one can more easily program an excel spreasheet to solve for the eigenvalues of the exact solution for a simple geometry such as a cube. The FE method is best suited for unusual geometries.
FEA technique is a commonly used method in the casting industry to know the hot spots,and provide adequate risering. However, I have seen many a program go wrong and resulting castings rejected.
You need a lot of experience before you start relying on the results,hence those well accepted and marketed FEA programs are pricey.
However,this particular case is a simple one and can be easily done knowing a few thermal properties. There are a number of well published papers in technical journals from which such data values can be taken.
Chocolates,men,coffee: are somethings liked better rich!!
(noticed in a coffee shop)
You need a lot of experience before you start relying on the results,hence those well accepted and marketed FEA programs are pricey.
However,this particular case is a simple one and can be easily done knowing a few thermal properties. There are a number of well published papers in technical journals from which such data values can be taken.
Chocolates,men,coffee: are somethings liked better rich!!
(noticed in a coffee shop)
Someone who has never seen Ansys or Abaqus before but who has finite element and programming experience should only take a couple of days to get proficient enough to solve that simple problem; I vaguely recall similar homework assignments in school. However, those programs cost $$$$ or even $$$$$. Not an option for someone who won't spend $$$ on a physical test.
Hg
Eng-Tips policies: faq731-376
Hg
Eng-Tips policies: faq731-376
Some FEA program are easy to get on and rock if they are integrated together well, I mean if it already got geometry building and meshing capabilities built-in. But those program are also usually pretty limited.
To get the best results, you usually have to bounce between several problem to do one simulation. For example, I had to use Pro/E for geometry, then hypermesh for meshing generating, finally use Abaqus as solver and Abaqus post-processor to view results. The interface is not perfectly seamless - ususually in those innoticeable sections of manual requiring very careful reading. Not mentioning that I had to use parallel computing in Unix system to reduce solution time, still hours/run. If with PC with enough memory sticks, one run can take several days.
Anyway, not try to scare people away from computer simulation. It is very time- and money- consuming at the beginning. Once it is upon running and confirmed by experiments, yeah, it is easy to run it again with different parameters.
To get the best results, you usually have to bounce between several problem to do one simulation. For example, I had to use Pro/E for geometry, then hypermesh for meshing generating, finally use Abaqus as solver and Abaqus post-processor to view results. The interface is not perfectly seamless - ususually in those innoticeable sections of manual requiring very careful reading. Not mentioning that I had to use parallel computing in Unix system to reduce solution time, still hours/run. If with PC with enough memory sticks, one run can take several days.
Anyway, not try to scare people away from computer simulation. It is very time- and money- consuming at the beginning. Once it is upon running and confirmed by experiments, yeah, it is easy to run it again with different parameters.
found this link by accident might be what your after:
The problem looks like a thermal problem to me and in my university I have seen thermal students using Gamit or Icepack software for this type of simulations, as said Ansys and Abacus are excellent softwares but you need to choose the boundary conditions and other parameters carefully to mimic the real time scenario.
ginsoakedboy
Mechanical
Heat Transfer text books will have solved examples. Your problem is a typical case of lumped mass, transient analysis. Some time in the library searching through heat transfer texts will provide you with some rough estimates.
Of course, as stated earlier, the determination of heat transfer coefficients and boundary conditions will prove to be the major challenge.
SolidWorks is sometimes bundled with CosmoWorks (an FEA software plugin) which may allow you to perform quick and dirty analyses. Always perform sanity checks with FEA results.
Also, study some electronics cooling literature. Those folks routinely solve these kind of problems.
Remember that you can greatly increase the rate of cooling by agitating the quench bath. Ultimately, I would perform a real test if time and cost allow.
Good luck.
Of course, as stated earlier, the determination of heat transfer coefficients and boundary conditions will prove to be the major challenge.
SolidWorks is sometimes bundled with CosmoWorks (an FEA software plugin) which may allow you to perform quick and dirty analyses. Always perform sanity checks with FEA results.
Also, study some electronics cooling literature. Those folks routinely solve these kind of problems.
Remember that you can greatly increase the rate of cooling by agitating the quench bath. Ultimately, I would perform a real test if time and cost allow.
Good luck.
- Status
Similar threads
- Locked
- Question
- Locked
- Question
- Locked
- Question