High uniformity large solenoid for magnetic isotope separation

The authors outline how the sequential quadratic programming algorithm, a superlinear, convergent, quasi-Newton method for solving highly constrained nonlinear programs, is used to optimize the design of large magnets requiring a high-uniformity field over a large volume. This method was used for determining the winding geometry and current distribution in designing a superconducting magnet system for a production isotope separation facility. The independent variables used for optimization are the inner and outer radii of the center coil and the two end coils, the length of the coil, the location and thicknesses of the flange, and the current density in the main and trim coils. The stored energy is not used as an optimization criterion since the conductor is self-supported and the required structure is not directly proportional. The optimization approach allows the selection of objective functions such as least volume, least energy, or least cost subject to the uniformity requirements constraints.