[Dr Youssef Hafez]

This developed turbulent flow codes is a finite element computer model, written in FORTRAN which was developed to solve the Reynolds equations of motion and continuity for steady and fully developed mean turbulent flow. The Finite Element mathematical treatment of this matter is reported in the following Ph.D. reference at CSU, Fort Collins, Co, the USA (please the link to this dissertation is https://mountainscholar.org/handle/10217/235417). Later on this code was extended to three dimensional and unsteady flows. The turbulent stresses appearing in the Reynolds equations are modelled in terms of mean velocity nonlinear gradients and turbulent viscosity. The non-linear algebraic stress model used in closed channels is totally different from the one used in open channels with new algebraic open surface proximity functions. A two equation turbulence model consisting of the turbulent kinetic energy (K), and its rate of dissipation (ε) evaluates the turbulent viscosity that appears in the algebraic stress models.
For wall bounded flows especially at corners (such as at an airplane's body to wing), my code succeeded in an excellent manner in simulating the main velocity, secondary velocities and turbulence structure (turbulent viscosity, K, and ε). In addition, distributions of the non-gravitational pressure and turbulent stresses are well predicted too. You may contact me for samples of the results of simulating turbulent flow in a square duct. Along the corner bisector the maximum secondary velocity divided by the average shear velocity is calculated as 0.31 which is in agreement with experimental data of Brundrett & Baines (1964) of about 0.32. For the wall bisector a value of 0.21 is predicted versus a measured value of 0.20 by Gessner and Emery 1980. Bulging of the velocity contours toward the corner is well predicted which is important in reducing separation due to adverse longitudinal pressure gradients. No up-winding or (over/under) relaxation are used. Non-linear Newton–Raphson that has quadratic convergence is used for dealing with the system of nonlinear equations. The boundary shear stress is calculated and is shown to be affected by the secondary velocities.
- For open channel flows: a very distinct feature in my nonlinear k-ε model is the use of an-isotropic turbulent viscosity in which the turbulent viscosity in the vertical direction differs from that in the lateral direction, a feature not existing in any existing CFD code to my knowledge. The model succeeded in predicting the depression in the main velocity maximum to be at 0.6 from the bed in a channel with aspect ratio of 1:2. The secondary velocity structure is also well predicted . Another important feature is prediction of the cellular secondary cells due to periodic roughness changes along the bed or walls. This helps in investigation of flow over ribbed surfaces.
Now, how much does the source code for this turbulent flow CFD code worth in US $?
I appreciate honest answers and offers.
For more details I can be reached at my email:
youseefhafez995@gmail.com

[sbaffini]

I think it's worth mentioning the question originally was on researchgate, where a couple of answers were already given. Hopefully, this will get the traction I expect it to have.

[Dr Youssef Hafez]

Dear Readers,

Please note the question was already asked on RESEARCHGATE WITH VALUABLE REPLIES AS SEEN IN THE FOLLOWING LINK:

https://www.researchgate.net/post/Ho...976c04da0a7c83

[LuckyTran]

Do you have a recently developed product based on the know-how you had and probably accumulated since your PhD days, in which case the question is how much would a fresh CFD software be worth today?

Because the source code, if it was written while you were at CSU is IP of CSU and you really should be talking to an attorney before proceeding any further. I am assuming here the typical situation that CSU have all rights reserved because that is the de-facto standard. Supposing that you have already had these conversations and have acquired these rights (I have no clue why a body would give up IP if it was worth anything), let's say you are going to sell to someone a non-exclusive pdf of just the fortran code, it would have the value of an e-book, say $10, unless the source code contains some technical panacea like blueprints to the atomic bomb in it.

You have to think of it in terms of value-added, what anyone plans to do with this source code and how it will be consumed. If you are talking to a book publisher for example, who wants to add a chapter in their book containing your code, and they plan to sell one million copies of the book, then it's worth a bit more, say $100k.

It's different if you plan on commercializing your code and creating, maintaining, and servicing the resulting product. Source code by itself is just symbols on print, no different than a book you get at the bookstore.

[Dr Youssef Hafez]

I thank you for your valuable comments. I think since I was not supported financially by CSU or by any project of any of its faculty members and since the code was not published in the PhD Dissertation, therefore I am the sole holder of all copyrights to the code I developed by myself. The PhD dissertation included the Finite Element treatment of how to solve RANS equations and results of application of the code to duct and open channel flow cases to show the success of the code versus experimental data. Since then the PhD code has been substantially developed. In the PhD code, the fully developed (Y-Z plane, where all x-derivatives are neglected) turbulent flow was considered under steady conditions. The Y-Z plane is the cross sectional plane of the duct or channel while the X-coordinate is the longitudinal axis of the conduit. However, subsequent developments included adding unsteady and full three-dimensional conditions. Also, a scalar transport component was added to solve the advection-diffusion-source transport differential equation. Another-two dimensional codes are also developed; one in the X-Y plane (horizontal resolution) and the other in the X-Z plane (vertical resolution).


My intentions are in line with what you stated at the end which is “I am looking for someone to buy the code and/or the other different codes”. If any one is interested please let me know. Of course the code has the format of being a research code suitable to solve nonlinear turbulent flows (of course laminar flows could be easily handled if required) ; therefor it needs to be cast in commercial form. It could be used to solve non-Newtonian fluid problems. Of course I accumulated the know-how of being able to develop codes of nearly most problems or just about solving numerically any differential equations such as for example Three-dimensional stress-strain of solid bodies.

I am also thinking of writing a book “Finite Element-Project-based Analysis” in which I can present about 20 to 30 different applications of the Finite Element Method (FEM) starting for the formulation of the governing equations, to the finite element treatment to solve the problem at hand, to how the system of algebraic nonlinear equations is solved down to how the code is developed and written. Various and very diverse problems or case studies will be presented in elasticity, fluid mechanics (CFD), hydraulics, Heat transfer, computational Hydraulics (flow in rivers, open channels, lakes) , contaminant transport (air pollution, water quality, soil quality), water distribution networks, nonlinear pendulum motion among others. These problems or case represent over 30 years of doing research and consultation using the Finite Element Method to solve practical real-life problems. I noticed most FEM books present the mathematical theory with very little into the solution process and the structure of the code in spite that these are very essential parts and are the corner stone in the problem solving process The proposed book will represent the Finite Element Method from A to Z unlike typical FEM books which don not discuss case studies and/or programming details except for few books who merely present simple practical problems (most one and two-dimensional problems). However, I don’t know a suitable or publisher who might be interested in this as here at home not much interest in this type of scientific books is there while in the developed western countries lots of interest in such kind of books is there. I hope if someone could help me in that respect, that’s really would be appreciated.

[FMDenaro]

Consider that the performance of a CFD code is a further value in the market. That means the capability of the code to run in parallel.

[LuckyTran]

Specifically regarding your PhD code... In the U.S., if you were a graduate student at a public university, the IP belongs to the university if it is any way perceivable that the IP was due to your participation in the university. You are 100% not the owner unless the IP is regarding a recipe for baked brownies. Unpublished works are also protected and even more protected than published works. If you haven't already received an explicit statement in writing that the IP has been licensed to you, and it is clear from statements that you do not have this in writing, then CSU would retain all these rights. Again, highly recommend you seek legal assistance if you are seriously considering publishing your old university work.

If you want to circumvent this, then you need to start over from a clean sheet and reverse-engineer your own work. And then the question really is, how much would a new cfd code be worth and not about your PhD code.

When someone posts stuff on the internet without asking for permission, the reason they can get away with it is because they're not asking for $, meaning they value the content at exactly $0. Nobody will sue you for $0. Here, you are specifically asking for $ and so I advise you to not get into trouble. Just go get the written statement from CSU that it's okay.

Still it is incredibly rare that anyone actually publishes source code instead of building an application and selling that instead.

[Dr Youssef Hafez]

Just a quick note.
Sir, I am afraid you are talking about different situation or case than mine. There are several reasons and justifications to my argument the least of which is that I was an international student sent by my government to study at a US university. The rules you are talking apply to US citizens or foreigners funded by US college departments or industries or projects through the college faculty members. Again, the code is not part of the dissertation at all, and anyone can use codes like ANSYS in his/her dissertation without ownership of the university, in fact one person used my code in his PhD at CSU itself. Also I have no publishing restrictions by CSU at all because I was not funded by CSU. You are taking about a common business model of typical scholarship programs in US colleges but my case is different. CSU or any college has ownership of the dissertation itself but not to the tools or methods used. There are no copy write statements on my dissertation on CSU printed papers. At the end it is a new code that is totally different as it is unsteady and fully three-dimensional in addition to two dimensional models and scalar transport models. The PhD just gave me the know -how but after which and while I was outside USA that all these models are developed without any funding from any one and I am the sole owner of my models. So I hope this clarification closes this issue and attention is given to the real issue which is the value of the new CFD code.

[bigfootedrockmidget]

Were you registered as a student at the university? If so, then everything you write for the PhD, during the PhD, even during your spare time, is owned by the university, unless you've made an explicit agreement with the university that this is not the case. It does not matter where the funding comes from.


The REAL issue here is if you own the code or not. It does not matter what you think or say, what matters is what a lawyer says. You should absolutely, definitely, talk to a legal advisor.



When you write a book on CFD and make this code part of that book, like for instance the fortran codes that came with Ferziger and Peric' book, the publisher also wants to make sure that you have the IP to the code.

[Dr Youssef Hafez]

Sir, what u r saying applies to faculty staff because they get paid by the university and that exists in university's contract or generally when you work at some company or organization they pay you salary so they are entitled to what you do or invent and that is stated clearly in the work contract. But this is not my case at all.
I don't know that when you register at a university they made you sign a form that contain copyright of any work you do?.
Again, my case is different, I did not registered at CSU, there was a project between my country and CSU and all registration and administrative things were done by some people so I did not sign any form or no mention of this issue was made to me at all.
I wonder how can I had 5 faculty staff in my Ph.D. committee and no one of them even mention that I have to put the code in the dissertation, nor the university mention anything. It was 1995 so I don't know rules have been changed since then to what u r saying or not. If the code is not reported in the dissertation how can the university prove that this code they own. Plus any one can make an application without the source code being published. So I don't see any legal stands or grounds for the university in my case. They did not pay me, and it was the other way I was paying the university so you can not apply the same rules pertaining to students with scholarships from the university. It seems the gentlemen who spoke had scholarships so they talk based on scholarship rules.
At the end the code which I speak about is totally different than the Ph.D. code, and there are several brand new codes as I explained before. I mentioned the Ph.D. just as a reference for the mathematical treatment of the Finite Element method which is published everywhere and to show the results of applying the Finite Element methodology which is a common science that not owned by any university !.
P.S. I don't live in the USA since the last 20 years at least.

[LuckyTran]

It is perceivable that a code can change after 20 years. Hence, why I asked in my original question whether you are talking about a new code or the actual source code from your PhD days that you have saved as a backup for 20 years. No one ever needs to tell you that you don't own their stuff, that's the meaning of the phrase all rights reserved. The code from your PhD days is not legally your IP, that discussion is indeed closed. If you really want me to explain it out in detail I can, but a lawyer will do it much faster.

Back to the original question of how much any general code would be worth when published in a book... If you plan work with a publisher like Wiley / Elsevier / etc., you're looking at cents per copy of the book in royalties. Making up numbers, the codes will take up probably 10% of the book, since you'll probably write 90% new paragraph of text and pictures and all the other content one expects to find in a book. And since it sounds like there are multiple codes involved, the valuation of one specific code is even less. When you do a printing run, you'll decide to print # number of copies of the book based on an estimate of how many copies you think will be sold and then you can get an idea of how much it's worth in totality in that context.

[Dr Youssef Hafez]

Thanks for the information about CFD book.

However, Sir you have not presented any proof or evidence of what you say about code ownership or written university's rules that apply to my case. You keep just stating and re-iterating your opinion without providing any support or legal evidence. I have not seen any discussing or refuting to my arguments, for example no body presented screenshot of university rules in this matter. All I saw are just speculations and opinions based on some personal experiences which I repeatedly said my case is different than yours. In terms of CFD, the boundary conditions in my case are different than the boundary conditions of students with university's scholarship, so the solutions or the situation must be different. I started to feel that there might be a conflict of interest here and some people do not want to see a new CFD code. I have not heard that universities are selling codes ?. Just a question, if some one wants to purchase this code which even does not have a name (no official name or registered in any patent) how can CSU sell it, as it does not have and what is the exact content of the code since it has never been published. Even patents have limited time say 5 or seven years, and this issue happened 27 years ago. Since the code is never published in any form or anywhere how it is defined and based on what??. Apparently if the code were published in the Ph.D. dissertation then your arguments may hold as there can be a reference and clear definition of the code. I understand a code should be registered in a patent for a strong proof of ownership. My Ph.D. does not have any statement of copy right or no statement such as the "All rights are reserved". So I see every thing is different in my case and few gentlemen are trying so hard so hard to discourage me and give threatening statements. And who to decide connection between the old and new codes, these are technical matters that Lawyers can not determine. The new code is totally different and has been done many years after finishing Ph.D. So why all of these discouraging arguments.

P.S. this is my thread and I have the right to respond to any comment.

[LuckyTran]

International students are U.S. persons and U.S. laws do apply to them.

One does not have to write down all rights reserved, the rights are reserved automatically even when it is not written. For example, even if I do not tell you that you are forbidden from stealing my work, that does not imply that I have given you permission to use my work. That is what is happening here. CSU has not explicitly allowed you to publish this work.

What is very special about your case is that you have confessed in writing multiple times that this code is from your PhD days and is not something new and we are explaining to you, this means you don't own that particular IP.

And I am fairly certain that I do not share a conflict of interest with your code. I have been very encouraging of you publishing new code in a way that won't get you into trouble. I am not threatening you with legal action. I'm simply telling you, that you need to go to CSU, and have them write you a letter or email that says it is okay for to publish your old work. When you go to a book publisher, the editor will ask you to do the same for every piece of content you plan to publish. You can publish it, but there are more steps to secure the rights to the IP.

I understand that when you went to school nobody bothered to explain these rules about IP ownership to you. It is in fact a huge issue and that's why I had to go through the training during my (more recent compared to you) student days. It is now required in the U.S. that entities provide training on IP because of situations similar to yours. And if I wasn't forced to take that training, I probably wouldn't have any clue either. The thing about laws and legal issues is, we implicitly agree to abide by them or face the consequences. No one ever has to explain the law to you before they can punish you for it. As the saying goes, you have the right to a legal attorney...

[sbaffini]

I think the legal matter here is not so obvious or relevant.

In fact, what may seem obvious from a common law perspective, would instead spectacularly fall apart on civil law grounds (and viceversa, obviously). If anything, this is a matter of international law and/or specific agreements between two institutions that existed at a very specific point in time. Assuming that a common law model was arbitrarily in place doesn't seem correct to me.

Also, in such cases one typically assumes a guilty frame that, honestly, is always far from proven. Can the university prove that the code was actually made in the Ph.D. context? Can the university actually define what the Ph.D. context is? Is there a daily time window or place where someone's work is accountable for the Ph.D. and some other where it isn't? Is all the IP produced by an individual during the Ph.D. years owned by the institution granting the title? Can the university prove that such code wasn't instead already there before the Ph.D.? Has the university actually checked that the said work wasn't, indeed, already IP of someone else? What if student and university have conflicting versions and none of them can prove their own?

But even assuming a 100% guilty use of the code, let's be honest here, is the university reasonably willing to chase a case, again far from clear and possibly linked to international law, without at least few tens or hundreds thoudands being involved? I don't think so.

Finally, if it ever becomes a trouble, such codes can be rewritten in few weeks in such a manner that any claim becomes unreasonable.

What I think the matter is, here, is how to determine if a code has actual value and how to determine it. As I stated on RG, I think this is only a matter that you can resolve.

Any code for a book is certainly added value, but I would not think of it like actual value, at least no more than a dedicated chapter on a specific subject that few other treat.

So, besides the book example, the code value is what someone else actually pays for it. At the moment it is 0. It is the combined effort of making the code and selling it that, like the Schrodinger's famous experiment, makes the value collapse on its actual state.

And you are the only one who can actually come up with a rationale to buy your code, in whatever form. We can only point you at what the alternatives out there are and how they work.

Note that the effort in selling the code involves finding possible use cases as well as possible clients and convincing arguments for them, besides all the bells and whistles that make the code actual selling material.

I have an impression of how, under certain conditions, your 2D code, properly rearranged for unstructured grids, could be of some value as part of a general 3D code. But this is me, and you may disagree, and there would be no point in discussing it.

It is not really different from the sell of a house (or any other good). House owners tend to consider their house to have an intrinsic value, while it only has it for them.

[Dr Youssef Hafez]

Sir, Well said every word by You Dr. Paolo Lampitella, thank you so much. I raise my hat to the wisdom, rational, the reasoning and sanity that comes in every word and every sentence.


I hope now that the discussion turn to how to value a CFD code and to provide your valuable opinion from every body in this respected CFD forum on the best ways to provide this code to the CFD community and to the CFD market or to whoever it may concern.

[FMDenaro]

My suggestion is to focus on the methodology inside the code, you developed it in the FEM framework and you should address the advantages of your structure compared to other FEM codes.
It is well known that some issues are:
- large memory required
- well assessed theory for linear problems but less for non linear ones.
- problems with highly convective flows.
- time integration method outside the FEM framework.


Thus,
prepare some standard test-cases, well accepted in the CFD community, show the resulting accuracy order and computational performances. The case of the square channel is not relevant.

Is the code able to read grid generated by other softwares?



Then,
could be your code relevant for didactic aims ? Why?

[andy_]

Originally Posted by Dr Youssef Hafez

Now, how much does the source code for this turbulent flow CFD code worth in US $?

I cannot think of a potential use/customer and without one it will have zero effective value. It might be used as a base to develop something of value but that would usually involve the author.

Many decades ago in the 70s and 80s Fortran CFD codes written by established university professors did have value to industry and were purchased from universities or in one case from an individual after they had retired. I am not aware of any CFD codes from PhD projects that were ever purchased without being sponsored. Many of these codes were shared within groups and sometimes between groups just not sold. Commercial CFD companies were careful to write their own code. Writing code doesn't take long if the algorithms are known and the author experienced.

Today there is lots of CFD source code around, industry no longer develops much internally and lots of commercial CFD companies have sprung up. CFD source code in itself is just not particularly valuable. It becomes valuable as part of a solution to a problem clients will pay for.

If you do find a buyer please post back because I have got loads of CFD and CFD-related research code going back 40 years which I had assumed has no value except to myself. Would be good to be wrong.

[Dr Youssef Hafez]

To Dr. Denaro, F.M.:
I thank you for pointing out these important issues and I will try to reply one by one as each issue raised by you requires detailed reply.
1. ٌRegarding Memory and storage Requirement:
All numerical methods for solving differential equations start by discretizing the solution domain into a number of cells or elements whose numbers depends on the correct geometrical representation of the physical domain and on the required level of accuracy in order to obtain a reliable solution to the phenomenon under consideration. These cells or elements are formed by joining together grid points or nodes. The governing equations are written in algebraic form on the cells or elements level as element matrices and then these element matrices are assembled into global matrices [A] {X} = {Q}, where [A] is the coefficient matrix which represents the properties of the system or its stiffness, {X} is the unknown solution vector, and {Q} is the forcing function vector.
Significant number of the elements of the coefficient matrix, [A}, has zero values and in this case the matrix is called sparse matrix. Indeed this calls for a way to reduce the size of the sparse coefficient matrix by storing as much as possible the elements with non-zero values. This enables P.C. computers to run the equation solver in a more efficient manner and in a suitable running time. Consequently, any equation solver will be based on the way in which the non-zero value elements of the coefficient matrix are stored.

Among the well-known methods of reducing the size of the coefficient matrix is the method of skyline storage and its associated equation solver is called skyline equation solver, Thompson (1991). Several good equation solvers of this type exist but most of them seem to be owned by commercial companies and therefore they are not published. This makes it difficult for their inclusion in a newly developed code. The skyline storage converts the original square coefficient matrix, [A], into a column vector or row array.
To appreciate the significant storage saved by the skyline storage method, consider the following example. A mesh with 18000 elements and 21021 nodes and with 3 unknown velocities at each node, the total number of unknowns becomes 63063. Note that the pressure is treated differently and the equation of continuity is solved separately on the element level to obtain the pressure via the iterative penalty method. This requires a square matrix of 63063 * 63063 (Number of the rows x Number of the columns) for which the number of elements of [A]to be stored is = 63063 * 63063= 3,976,941,969 which is very large (getting close to 4 billion!). Now skyline storage will reduce this number to a vector matrix of dimensions 90,941,949 x 1 (i.e. one column matrix with only about 91 million rows). In other words, about 97.7 % of the original square matrix is not needed and discarded because it contains zeros. The storage percent is (90,941,949 /3,976,941,969)* 100 = 2.3 %. The numbering of nodes within an element determines the size of the Skyline storage through the maximum difference between node numbering within an element.
Another example is that for a mesh with 9000 elements and 9471 nodes and with 3 unknown velocities at each node, the total number of unknowns becomes 28,413 (= 3 x 9471). This requires a square matrix of 28,413 x 28,413 (No. rows x No. columns) for which the number of elements to be stored is = 28,413 * 28,413 = 807,298,569 which is still large (getting close to one billion!). Now skyline storage will reduce this matrix to a vector matrix of dimensions 40,446,099 x 1 (i.e. one column only matrix with about 40.5 million rows). The storage percent is (40,446,099/807,298,569)* 100 = 5%. For 2000 elements in 3D, the saving is about 10%. Thus, as the number of nodes or grid points increases significantly, the storage saving percentage increases also significantly. This trend favours the use of skyline storage especially when using very large number of grid cells or nodes. It should be noted that the domain or mesh was for a regular block where all elements are connected together but for multi-blocks the situation will be different. However, the essence of great saving by Skyline storage methods will remain the same.


References:

Thompson, E. G., 1991. CE 563 computational methods, class notes. Colorado State University, Fort Collins, CO, USA.

[andy_]

Originally Posted by LuckyTran

Because the source code, if it was written while you were at CSU is IP of CSU and you really should be talking to an attorney before proceeding any further.

The OP may have some strange views and I certainly agree with the advice to consult a lawyer (which I have done in the past concerning CFD code IP and it proved useful) but the ownership of IP is almost always determined by the details in employment contracts.

From what I think the OP has said he was working as part of an agreement between an institution in his country and the university. If his country paid a reasonable sum to the university it would be usual for the institution in his country to own 100% of the IP or if a more modest sum some form of joint ownership of IP. This assumes the OP was working for the institution in his country in some form.

If he was wholly self funded and he paid his PhD tuition fees himself then it is likely he owns the IP to his CFD code. If others contributed to his living costs and/or paid his fees then the paperwork/contracts will need to be checked for IP ownership.

In the UK (or more strictly the UK 25 years ago when I was reasonably on top of the subject) a PhD student employed by the university as a research assistant or by an external company would not own the IP to CFD code they wrote. Their employer would own it. There doesn't need to be a contract because it is UK law and other countries have similar although not necessarily identical laws. Contracts can give back rights. For example, universities would own an academics publications if they weren't explicitly given to the academic in their contract. Mine explicitly stated this did not apply to software! Contracts can also try to take away rights (e.g. prevent employment by competitors) but some of these can fail if challenged in court. Nor does it matter if all the work was performed at home in the evenings and weekends using ones own computers. If the type of work falls within the employees normal duties it is owned by the employer regardless.

However, if a PhD student was self funded or had a state grant then they would own the IP to the CFD code they wrote. The latter might vary a bit with country and, indeed, it may have changed in the last 25 years in the UK. So it is quite possible the OP owns the IP to his CFD code.

Of course, if he does find a buyer for the CFD code then that buyer is almost certainly going to insist on seeing proof that the OP actually owns the code. Just asserting that he does in the way he has done here is very unlikely to be sufficient. His employment status not only during his PhD work but also when he revised the code. If he has been employed to use his CFD knowledge then his employer may well own his revisions.

[andy_]

PS I see from the researchgate thread that the OP works for the Nile Research Institute. Assuming he has worked there for some time I think there is little chance he owns "his" code by default. Nonetheless, as mentioned by someone earlier, he may be able to negotiate an agreement with his employer. Doesn't always work though and I was told by the lawyer I consulted about a UK academic that deleted years of code when he couldn't reach an agreement with his university. In my case I and a colleague entered discussions with the university before writing any commercial CFD code (our existing research CFD code was not considered a suitable starting point) and concluded the university was not going to be a viable partner.