Open jowr opened 10 years ago
I'm not afraid of higher order polynomials, but are there better models for the viscosity of liquids? Have you heard of Eureqa(http://www.nutonian.com/)? You might want to take a viscosity dataset and chuck it in there and see what comes out. I had the same idea before, but they obviously have a much better idea what they are doing than I do!
On Wed, Oct 30, 2013 at 10:41 AM, jowr notifications@github.com wrote:
Viscosity fits are of a rather poor quality. The simple equation exp(A/(T-B)+C) can describe the general trends. At high temperatures, many fluids exhibit a relative error of up to 100%, that is not really acceptable... There are two ways out: use a polynomial with degree > 7 inside the exp function or use a Chebyshev Rational function, e.g. rank 4. The latter gives the best fit results but is a little messy to implement.
— Reply to this email directly or view it on GitHubhttps://github.com/ibell/coolprop/issues/65 .
Interesting software, but I used another one. That is where the Chebyshev Rational suggestion comes from. For now, I updated the plots to include the relative error. I also added them to the source control again so everyone can see the plot quality for the fluids they are using...
I was a little too pessimistic. The worst fit is the one for T72 with an error of "only" 25%...
Viscosity fits are of a rather poor quality. The simple equation exp(A/(T-B)+C) can describe the general trends. At high temperatures, many fluids exhibit a relative error of up to 100%, that is not really acceptable... There are two ways out: use a polynomial with degree > 7 inside the exp function or use a Chebyshev Rational function, e.g. rank 4. The latter gives the best fit results but is a little messy to implement.