I've been surfing and prodding a lot, and I'd like to know if anyone has created some logical comparison matrix to help sort out CFD purchase decisions. Most of the top dollar packages can handle my HVAC applications, but perhaps there is a more cost-effective package. This kind of question seems, to me, to be one that likely has been asked before.

How important is productivity to you? If you're going to use a package for purely academic purposes I'd advise going for something cheap and cheerful such as Phoenix. If you want to use the package to increase commercial efficiency then FLOVENT from Flomerics seems to be the best tailored HVAC CFD code around.

At then end of the day metrics to help you choose a code may include:

Man hours of experience of CFD in the company that is selling the code

Man hours of HVAC industrial experience in the company that is selling the code

Ratio of support staff to number of seats of the code

It is almost impossible to quantify code quality, useability, accuracy etc. Choose at least 2 codes and do a trial. Good luck!

(1). Very good answer. (2). First of all, we will have to assume that the user of a commercial code knows what he is doing. Otherwise, a blind person and a black box is not going to lead him anywhere. I think, this part is essential. (3). Based on this, if the user has the time to write his code, it would be ideal because he is the one who is going to use it. He should know his problem, and what he is looking for. This is all right when there is no time constraint. You can work on the problem when there is plenty of time available. (4). From this point on, the road will be different. If the problem you are interested in happens to be the same problem the code vendor has been working on for a while, then I would say the code should be ideal for the user. It is like having an additional group of people working on the same problem for you. (5). Beyond this point, it is really hard. Trying to find a particular feature from a particular code to match the users' need at a particular time requires some research. And the results may not be universal. (6). The bad thing about the code is: it does not run by itself. It does not correct itself, or even update itself. The code is always lagging behind the real world and waiting to be upgraded by the engineer. So, you see, if your problem can be solved by existing codes, then the problem must be an old one. If one is still interested in solving an old problem, then he must have a lot of time to spend. So, he might as well sit down and write his own code. (7). For any new problems, you will have to depend on human experience. And there is no replacement for a human brain. (the code is a dead thing, the reason why it can also play chess is because there are human brains behind the code.) (8). There is a practical way to lower the price of a code, that is to reduce the number of the features in a code and increase the volume of the sale. That's not a good idea either because he is going to work harder with the code and compete against other users of the code. It is a tough world.

I like (and appreciate) both responses thus far. My application is steady-state stratification analysis in large atriums, which we've done with relative success before, sans CFD. As I told the Flovent rep, commercial HVAC engineers have a tough question on this, because we want to have CFD available, but subcontracting it out to experienced CFD practitioners is likely still the way to go, barring the case where very intense, unique scenarios are your specialty.

(1). You can look into the possibility of packaging the CFD code for your specific applications so that engineers in your field can easily handle it, if that is your concern.(2). So, you could try this approach with the CFD code vendor to come up with a more user-friendly package for your applications.