Page 122 - Fundamentals of Magnetic Thermonuclear Reactor Design
P. 122

106     Fundamentals of Magnetic Thermonuclear Reactor Design


                            ∫  P (, r ⋅dl = ∫∫  j ( t rd S =  I t(),
                                                   ⋅
                                                ,)
                                    )
                                  t
 ∫asPsrc(t,r)⋅dl=∬Sjsrc(t,r)⋅dS=  as  src  S  src
 I(t),
            where I(t) is a given time dependence of the total current flowing through a
            selected cross-section S within the boundary ∂S. Substituting P for P  in the
                                                                     tot
            system of equations and summing the right-side terms that describe the ex-
            ternal sources, the final system of equations is obtained to describe eddy and
            halo currents in tokamaks. Examples of generated 3D meshes are given in Figs.
            A.4.2.2–A.4.2.4, Appendix A.4.2.
            4.4.5  Computation of Potentials: Global and
            Local Model Integration
            Let us discuss the procedure for computing the time-dependent distribution of
            potentials, eddy currents, fields, EM forces and heat releases in detail. A set of
            independent parameters I  (t, r ), (i = 1, 2,…, NI) which, together with their de-
                                    i
                                i
            rivatives, are assumed to be smooth continuous functions and can be introduced
            to describe external currents. These parameters give the spatial and temporal
            distributions of external magnetic fields.
               At the beginning of the simulation, we assume that the design and structural
            materials for a magnet system are pre-determined. This allows material proper-
            ties to be taken into account and the calculated domain to be partitioned into
            a set of appropriate FEs. Every FE (MFE is the total number of finite element
            mesh) is characterised by eddy current density vector I (t). To find it, we can use
                                                        j
            computation algorithms and procedures that employ either the potential P de-
            scribed earlier, or EM field vectors B and D, or the magnetic vector and electric
            scalar potentials.
               In any case, interpolation on FE meshes leads to a system of linear (or lin-
            earised) ODEs with constant coefficients. For vector P it takes the form

                                                  {(
                                  P t)}
                                          P t)}
 L{P˙(t)}+R{P(t)}=F{Ii(t)},     L{(    + R{(   =  F It)},
                                                    i
 {P(t)}     where  P t{( )} is a column vector of the unknowns, the point above the vector
            denotes differentiation over time, and the L, R and F matrices are known. The
            system includes the initial and boundary conditions. We therefore have a linear
            Cauchy problem, a partial solution to which is obtained numerically.
               The methods for solving linear ODEs are well known. Time distributions are
            obtained by numerical integration of the system with the right side defined by
            a given current scenario, I (t). Sets of relevant modules are incorporated in the
                                 j
            TYPHOON and TORNADO codes.
               Let us now discuss another approach based on the use of a linear ODE with
            the right side having the form of a superposition of known functions. The total
            field, determined by the instantaneous values of currents I (t) and eddy currents
                                                          i
            I (t) can be written as
             j
   117   118   119   120   121   122   123   124   125   126   127