Aesik
06-16-02, 12:22 AM
Here is the first official release of my little simulator. For those who want a little history on this program and its evolution, check out http://forum.oc-forums.com/vb/showthread.php?s=&threadid=75541&perpage=30&pagenumber=1
A few quick notes on the program: It's written in Fortran and is compiled as a simple DOS program (so sue me, I haven't bothered to learn how to write Windows interfaces for Fortran programs yet). If you run the .exe, it should just open up in a DOS box and run.
It's a finite difference (double precision) code that simulates the effects of water flow through a block. Once the flow characteristics are calculated, the code then predicts the theoretical temperatures along thousands of individual nodes in the block material. I've optimized node size and convergence criteria to be as accurate as possible while requiring the least amount of computation time.
The model is still fairly simplistic and simulates a block with a straight channel with one end heated to represent the CPU. The heated end is only at the inlet end in this version. The heat is dispersed along the length of the channel in somewhat exponential form; ie lots at the beginning and tapering off quite quickly. So far the use cannot input variable wall characteristics.
The object of this program is to be able to take in most of the variables waterblock builders are interested in and see how they effect the temperature of the CPU. It is NOT to simulate specific waterblocks. While it is definitely not the end-all-be-all for waterblock design, it is a useful (and fun!) tool too play with and see the effects of many different variables.
One major thing to note is that the program does NOT take into account head loss or gain due to geometry changes. For example, simply reducing the channel size while keeping the same flowrate will increase the velocity and thus the Reynolds number, and therefore give cooler CPU numbers. However, in an actual design, reducing the channel size will increase the head loss and lower the overall flow rate and raise the CPU temp. It is up to the user's best judgement to come up with flowrates.
I've spent a ton of time trying to get all this right. I've gone nearly bonkers while working on this little project and hope that it is of use to some of you.
I'll probably be leaving this program in its current state for awhile now because if I stare at it any longer I'll lose my mind. While I await feedback and discussion, I'll be working on my radiator/fan simulator. Hopefully in the future I'll be able to combine both to simulate an entire closed loop system.
I realize the program is still fairly rough. I'm hoping for some decent feedback on what you all would like to see added/changed in future versions.
Please give feedback here or directly to me at aesik@attbi.com.
A few quick notes on the program: It's written in Fortran and is compiled as a simple DOS program (so sue me, I haven't bothered to learn how to write Windows interfaces for Fortran programs yet). If you run the .exe, it should just open up in a DOS box and run.
It's a finite difference (double precision) code that simulates the effects of water flow through a block. Once the flow characteristics are calculated, the code then predicts the theoretical temperatures along thousands of individual nodes in the block material. I've optimized node size and convergence criteria to be as accurate as possible while requiring the least amount of computation time.
The model is still fairly simplistic and simulates a block with a straight channel with one end heated to represent the CPU. The heated end is only at the inlet end in this version. The heat is dispersed along the length of the channel in somewhat exponential form; ie lots at the beginning and tapering off quite quickly. So far the use cannot input variable wall characteristics.
The object of this program is to be able to take in most of the variables waterblock builders are interested in and see how they effect the temperature of the CPU. It is NOT to simulate specific waterblocks. While it is definitely not the end-all-be-all for waterblock design, it is a useful (and fun!) tool too play with and see the effects of many different variables.
One major thing to note is that the program does NOT take into account head loss or gain due to geometry changes. For example, simply reducing the channel size while keeping the same flowrate will increase the velocity and thus the Reynolds number, and therefore give cooler CPU numbers. However, in an actual design, reducing the channel size will increase the head loss and lower the overall flow rate and raise the CPU temp. It is up to the user's best judgement to come up with flowrates.
I've spent a ton of time trying to get all this right. I've gone nearly bonkers while working on this little project and hope that it is of use to some of you.
I'll probably be leaving this program in its current state for awhile now because if I stare at it any longer I'll lose my mind. While I await feedback and discussion, I'll be working on my radiator/fan simulator. Hopefully in the future I'll be able to combine both to simulate an entire closed loop system.
I realize the program is still fairly rough. I'm hoping for some decent feedback on what you all would like to see added/changed in future versions.
Please give feedback here or directly to me at aesik@attbi.com.