CO2 rich mixture with N2 fails with segmentation fault

[Pablo-Salinas]

Description

Trying to obtain the properties of 0.97CO2+0.03N2 using a C++ codeto call Coolprop I get segmentation faults depending on the input P and T. I checked with gdb and debuging the library and it seems internal to Coolprop as it is allocating a vector:
std::vector<double, std::allocator >::operator[]
The code below reproduce the problem. If instead of 264.65 K one uses, say 263.15, things are fine. By they way, I coudln't find a way to provide a guessed density, is that possible or one can only help the solver by providing the expected phase? Thank you.

Steps to Reproduce

#include "CoolProp.h"
#include
#include <stdlib.h>
using namespace CoolProp;
int main() {

std::vector<std::string> fluids = {"CarbonDioxide", "Nitrogen"};
std::vector<double> z = {0.97,0.03};
double temp = 264.65; // 263.15 is fine
double pres = 6.51e6;
std::vector<std::string> outputs = {"DMASS", "H", "Z", "PHASE", "VISCOSITY"};

std::vector<double> T(1, temp), p(1, pres);
std::vector<std::vector<double>> coolsol =PropsSImulti(outputs, "T", T, "P", p, "HEOS", fluids, z);
for(int i=0;i<size(outputs);i++)
{
    std::cout << outputs[i] <<" "<< coolsol[0][i] << std::endl;
}
// All done return
return EXIT_SUCCESS;

}

Versions

v6.6.0 and also tested 6.4.2

[jowr]

Hi - thanks for reporting the issue.

The Halley solver used in solver_dpdrho0_Tp seems to overshoot and tries to calculate with a negative molar density which then fails ungracefully.

This may take some time to debug :/

[ibell]

Some comments here:

1. Definitely the failure mode should be an exception not a crash (this is not good)
2. Definitely the pure fluid EOS are not helping since nitrogen EOS doesn't have a spinodal sometimes
3. REFPROP is recommended for mixtures in general
   3a) Help improving the mixture routines is welcomed(!)
4. The mixture viscosity model in CoolProp should not be trusted, don't use it except for as a very rule-of-thumb model