[GregNordin]

I am using "codeFixedValue" in "0/U" to implement a time-varying boundary condition (parabolic velocity profile at an inlet with time-varying max velocity). Writing out the code as a series of C++ statements works fine. However, if I consolidate some of the code into a function that I define and use in the code block, I get an error that includes the following: "error: a function-definition is not allowed here before '{' token". The '{' token is the opening curly brace for the function code.

Is it possible to define and use a function in the "codeFixedValue" code block? Or is there another way to achieve the same purpose by defining the function in a separate file and including it somehow?

[GregNordin]

I found a way to resolve my issue based on Ben Voight's answer at Simulating nested functions in C++. A full solution is shown below for 2D inlet parabolic fluid velocity with pulsed fluid flow in time. The key point is defining a "struct" (see "struct local" below) with a member function, "u_now", that is called further down in the code (see line with "u = local::u_now(...)").

    inlet
    {
        type codedFixedValue;
        value uniform (0 0 0);

        name inletParabolicVelocityTimeVarying;
        code
        #{
            const fvPatch &boundaryPatch = patch(); // generic
            const vectorField &Cf = boundaryPatch.Cf(); // generic
            vectorField &field = *this;                 // generic
            const scalar t = this->db().time().value();

            const scalar chan_width = 100e-6; // definition of the channel width
            const scalar Umax = 0.001;        // definition of the maximum velocity

            struct local
            {
                static scalar u_now(scalar t, scalar t0, scalar delta_t_start, scalar delta_t_on, scalar delta_t_stop, scalar Umax)
                {
                    const scalar t1 = t0 + delta_t_start;
                    const scalar t2 = t1 + delta_t_on;
                    const scalar t3 = t2 + delta_t_stop;
                    scalar u;

                    if (t < t0)
                        u = 0;
                    else if (t < t1)
                        u = Umax * (t - t0) / (t1 - t0);
                    else if (t < t2)
                        u = Umax;
                    else if (t < t3)
                        u = Umax * (1.0 - (t - t2) / (t3 - t2));
                    else
                        u = 0;

                    return u;
                }
            };

            // Ramp velocity up from 0, hold, then ramp back down to 0
            const scalar t0 = 0.01;
            const scalar delta_t_start = 0.01;
            const scalar delta_t_on = 0.01;
            const scalar delta_t_stop = 0.01;
            const scalar delta_t_between_pumps = 0.020;
            const scalar delta_t = delta_t_start + delta_t_on + delta_t_stop + delta_t_between_pumps;
            scalar u;

            scalar t0_now;
            int N = floor((t - t0) / delta_t);
            if ((t - t0) < 0.0)
                t0_now = t0;
            else
            {
                t0_now = t0 + N * delta_t;
            }
            u = local::u_now(t, t0_now, delta_t_start, delta_t_on, delta_t_stop, Umax);

            forAll(Cf, faceI) // loop over all the patch faces
            {
                const scalar y = Cf[faceI].y();             // y coordinate of the faces i
                const scalar normalized_y = y / chan_width; // compute radius from center patch

                field[faceI] = vector(4 * u * normalized_y * (1 - normalized_y), 0, 0); // define velocity value on the face i
            }

        #};
    }

[mill]

Hi,
Did you solve this problem, I also want to learn how to use codeFixedValue

[GregNordin]

Yes, my solution is posted above (dated March 15, 2022). I had to use a struct instead of a function, and then OpenFoam handled it fine.

[mill]

Originally Posted by GregNordin

Yes, my solution is posted above (dated March 15, 2022). I had to use a struct instead of a function, and then OpenFoam handled it fine.

Thank you!

[gbope7]

Hi Greg, did you ever have issues where an error arose that your function was an undeclared identifier? Like this error:

/Users/promisea/OpenFOAM/ibukunOluwa/testCases/blockTest/testNSTX/0/T.boundaryField.top:83:45: error: use of undeclared identifier 'readXPositionArray'
            std::vector<double> xPosArray = readXPositionArray(filePath);
                                            ^
1 error generated.
make: *** [Make/darwin64ClangDPInt32Opt/fixedValueFvPatchFieldTemplate.o] Error 1

[GregNordin]

@gbope7, sorry, I don't recall seeing an error like that.