Coupled 3D-1D problems arise in many practical applications, in an attempt to reduce the computational burden in simulations where cylindrical inclusions with a small section are embedded in a much larger domain. Nonetheless the resolution of such problems can be non trivial, both from a mathematical and a geometrical standpoint. Indeed 3D-1D coupling requires to operate in non standard function spaces, and, also, simulation geometries can be complex for the presence of multiple intersecting domains. Recently, a PDE-constrained optimization based formulation has been proposed for such problems, proving a well posed mathematical formulation and allowing for the use of non conforming meshes for the discrete problem. Here an unconstrained optimization formulation of the problem is derived and an efficient gradient based solver is proposed for such formulation. Some numerical tests on quite complex configurations are discussed to show the viability of the method.
翻译:在许多实际应用中,出现了3D-1D问题,目的是减少模拟过程中的计算负担,在模拟中,一个小部分的圆柱体融入被嵌入一个大得多的领域,然而,从数学和几何角度来说,这些问题的解决可能并非微不足道。事实上,3D-1D的结合需要在非标准功能空间运作,而且模拟地理比例对于多重交叉域的存在可能很复杂。最近,为这些问题提出了一个基于PDE的、受限制的优化配方,证明它是一种完善的数学配方,并允许使用不兼容的模具处理离散问题。在这里,为这种配方提出了一种不受限制的优化配方,并提出了一种基于高效梯度的溶液。讨论对相当复杂的配置进行一些数字测试,以显示方法的可行性。