Recently, I was contacted by a friend in industry who performs FEA stress analysis (I am still in grad school). Due to a new client of his, he will be analyzing parts that experience fluid stresses. He was wondering if I would do some freelance CFD in order to provide pressure data for his stress analysis. The problems are generally 3-D, but not especially complex. I have experience with both 2-D and 3-D gridding and solutions, both steady and unsteady, and have cleared the use of the school's computing resources. I was also asked by the same friend how much the going rate for such CFD analysis, as his company is intending to pay for my services. I did not know what to tell him, and was hoping some of you might give me an idea of what is a fair rate. Work on my end would include importation of models, gridding, problem setup, running the solver, and providing some post-processing/variable data.

Hmm this is a tough one to answer because your costs bear no relation to the "competition" and you are proposing something that is not morally defensible or, possibly, legal.

The facilities you are proposing to use have almost certainly been provided by the taxpayers for the purposes of teaching/research. It is probable that as a legal entity the "grad school" is not allowed to engage in this sort of activity (assuming the "grad school" has a charitable non-profit status but this is speculation). It is also very likely that the CFD software you are proposing to use was provided very cheaply or at no cost for the purposes of teaching and expressly forbids commercial activity. However, I know some CFD suppliers would allow the codes to be used for commercial activity if they are paid the full commercial price for the software.

As an individual, so long as you are not employed (an assumption), I think the legal risk lies with the grad school.

Are you signing a contract with this company? Who is responsible if the predictions are wrong, something is built and lots of people die? That is, who is pays the insurance premiums. A not insignificant amount of this type of work fails to be viable because the insurance costs are too large for small one-off pieces of work.

If you are keeping everything informal then, again, the company is likely to be bearing the risk (I am not a lawyer though!).

Under the circumstances I would suggest the only fair rate is to cover costs (i.e. you take home experience) with the grad school possibly receiving a donation of something. You did use the word "fair"!

Alternatively, if you want to make money ("fairly") I would suggest the company obtains the CFD software and you use it charging a typical short-term labour only rate. Look at the rate for computer programmers locally for some idea of what is reasonable. Or failing that, work out an hourly rate from, say, double the typical annual salary of an equivalent worker (your friend perhaps?). I would be wary of accepting a very low rate if there is any chance you might want to do this sort of thing again - you will find it very hard to put it back upto a normal rate.

(1). There is no standard here, it is between you and your friend, as long as you both are satisfied. (2). If you are allowed to use the facilities and the codes, then it is all right. (otherwise?)

The normal consulting rate for analysis, is 1200-1500 $/day. This is the ongoing rate that all the software companys including Fluent Inc. and Ansys charge.

As for responsibility to the client, the software companies have a disclaimer which they use on software as well as services, meaning that they only provide a solution to a set of PDE's and they are not consulatants in the true sense of the word.

There is a problem with a univesity license, in that it can not be used for consulting.

I would try to get a legal copy of the CFD code that you would like to do the analysis with. Call them up and see if they are willing to lease you the code on a short term basis (say 3 months). I am sure that one of them will. then charge the client the ongoing rate (1200-1500/day) that I mentioned to you and pay for the lease yourself.

(1). In old days, the cost to use a super-computer was about one US dollar per second. Based on that, it will cost 28800 dollars for eight hours of CPU time. (2). Anyway, the commercial rate is determined by competition.

You can obtain CPU time at under 1 dollar per PC CPU hour if you buy a reasonable chunk. However, this does not mean your desired commercial software is going to be available.

(1). Unless the use of a particular commercial CFD code is required, then I think, there are many ways to do numerical analysis using various assumptions and codes (which can be provided by the client, or free software developed by government agencies, or author's own code, research codes developed by the professor and his students, etc.). (2). As a consultant, it is important to provide useful solutions, rather than solutions obtained from a particular commercial code with no warranty on its solutions. (3). Well, if the client wants your resume and report written only using MS Word, then that's a different story. (4). So, if one is thinking about doing consulting work, then it is a good idea to collect all available codes for free, and also develop his own codes. In this way, he can easily avoid the expensive licensing issue.

This is slightly off topic.

"As for responsibility to the client, the software companies have a disclaimer which they use on software as well as services, meaning that they only provide a solution to a set of PDE's and they are not consulatants in the true sense of the word."

Do you know if this type of clause has been successfully tested in court? I am not trying to implying it will fail but if it is solid it would be useful to know. The consequences of getting it wrong for industries such as aerospace and nuclear can be terminal in more ways than one.

Kieth,

I'll start by answering your questions, and then chime in on other issues.

Consulting companies charge rates that range from about $80/hour to about $150/hour. The consulting company that I used to work for (~200 employees) had a sliding scale based on job title. An engineer with a few years experience ("Associate Engineer") was at the low end of the scale, while a "Senior Project Engineer" was billed at about $110/hour. Companies that provide more specialized analysis services tend to charge more, and this will vary by region. A company I know some people at in Connecticut charges ~$150/hour, and I would not be surprised to find that rates in California and New York are higher.

Independent consultants generally charge less (there is less overhead), and have more flexibility. Services like drafting would command rates of about ~$25-$30/hour. Analysis and design would typically be in the $40-$50/hour range.

Some people have brought up the issue of software. If the school owns an academic license for the code you will be using, then it really should not be used for a consulting project. The odds that you will be "caught", and that there will be negative consequences are low, but you have your own conscience to deal with. An option would be to get a short-term license. If you go that route, you should bid the project "fixed price". That way if it takes you very little time to complete you are assured that you'll get paid enough to cover the license.

Just my 2 cents as someone who has bid many projects.

Regards, Alton

I'd be pretty scared if the aerospace and nuclear industries used CFD alone. CFD should be used near the beginning of the design concept to focus ideas and to give a general location of the optimum product. Then experiments and testing need to be done to get a real product produced - using the CFD optimum as a starting point. If the aerospace industry built a plane solely from CFD, and never flight tested it, then I sure would never fly again.

My point is: CFD should never be the sole design step. Thus, there should be little or no liability that the CFD user has from a failed product. The company has the responsibility to test and improve the design BEFORE it is sold and AFTER the CFD step.

I am not sure your point is relevant. If a large corporation can shift the legal responsibility for a major cock-up onto a supplier and get it to stick in the courts it will do so. The question of moral responsibility is not relevant except, possibly, indirectly via the media. A supplier would be most unwise not to assess/cover the risk.

(1). It seems to me that you are not working in industries. (2). Based on my experience, many "world leading companies" simply do not follow your recommendattions. (3). Testing is rarely done because it is time consuming and very expensive. Their goal is to meet the stock holders expection. This is a very serious problem, but it is life or death for the company. (to keep the stock holder happy)

Please tell me which airplanes have been designed and built without testing, then loaded with hundreds of passengers, and then flight tested with the passengers on-board. I would like to know, so I will avoid ever flying on those (I'd really dislike being on that planes first flight ever). John Chien, I believe you worked in this field, so you should be able to answer. My aerodynamics professor really fooled us with those videos of planes being flight tested before passengers boarded...

I know this is off-topic, but a good CEO will attempt to make the company as profitable as possible. A poor CEO will attempt to make the stock holders happy while sacrificing profits. Too bad the world is filled with the latter.

Getting back on topic now. Suppose a company wants to design a new product. People brainstorm 5 possible designs. I as a CFD user show that one design will be the cheapest of those 5 designs. My company then sells the product without ever testing whether it will work. Why would I be the one with legal responsibility? Why not the company? Why not the CEO? Why not the other designers who came up with the design in the first place? Why not the people who chose not to make even a single test? I'm not a lawyer, but I'd think the CFD user would have less responsibility than most of the rest. That is like blaming the person who decided to buy the company car with vinyl instead of leather (cheaper) for the accident when the driver ran a red light.

In the future, there should be a CFD engineering license, so that when you stamp your CFDE license on your analysis report, you will assume the responsibility.

(1). That's should be the minimum requirement.

(1). Have you heard that FBI was investigating the case that wires were cut in many newly built commercial airplanes? It was in the news about a month ago. (2). That's only the tip of the iceberg. And that is only in one particular industries.

Interesting idea, but a PE stamp is rarely required in any situation. I've inked mine up only twice in ~4 years.

From what I've seen, design and analysis of Class 1 nuclear components are often stamped, as is work done for Government agencies. I think structural engineers use their stamps more than mechanical engineers.

Alton

(1). Good thinking. (2). So, structure engineer's work depends on the accurate input of the pressure and the temperature fields obtained from the CFD engineer. (3). His stamp is useful, only when the solution from the CFD engineer is useful. So, obsiously, there is a big hole there.

Actually, I mean structural engineers who design buildings, malls, parking garages, power plant buildings, etc.

Generally speaking, I don't think their work involves interfacing with CFD folks at all. For wind loads they use the uniform building code that has tables for determining wind pressure based on building height and location within the country. If you can multiply, you can get the wind pressure to use in designing a building.

Alton