有限元法和组元法在组合节点刚度研究中的比较

以下英文来自EC4的讲座《Structural Steelwork Eurocodes Development of a Trans-National Approach Course: Eurocode 4》lecture 9 第14页。中文是我的翻译。
问题：
1．有限元法和组元法的优劣，我还是不很了解。
2．现在研究组合节点刚度，是否仍能用有限元法

当然，在讲座材料中还有很多内容将到此两种方法，但我粗看了一些，仍不甚了了。

[中文译文]
对节点转动性能的描述需要考虑节点区所有的变形。而且，在分析模型中还需要考虑所有的局部塑性变形和不稳定性。
第一种研究方法是直接根据足尺试验结果发展的组元法（component method），由于影响因素众多，失败了。第二种研究方法是有限元模型法，由于“局部细节问题”，也未获成功。因而为了分析测定非线性节点反应，通过将“复合有限节点”细分为部件的宏观检查法证明是成功的。相比于有限元法，这些部件，作为节点的逻辑子系统，承受内力和弯矩的作用，而不是应力的作用。
下面章节还将详细的阐述，这些部件将被视为具有非线性的力―变形行为的“平移弹簧”。所有需考虑的部件能用相对便宜的“部件试验”来研究，并在此基础上发展理论模型。最后，将所有影响部件用”部件模型”（component model）组装，可以得到整个节点的反应。这种方法有很多优点，包括最少的试验成本，清晰的部件物理计算模型，设计者最大的自由度。设计者能够将可供利用的部件进行组合，以获得最经济的节点配置。很多足尺试验也验证了组元法的正确。
[英文原文]
A description of the rotational behaviour of a joint has to take into account all sources of deformations within the joint area. Furthermore all possibilities of local plastic deformations and instabilities have to be covered by such an analytical model.
The multitude of influencing parameters was the reason, why the first attempts to develop a component method directly based on full-scale joint tests have been doomed to failure. Also a second research tendency with the aim of finite element modelling did not really succeed because of local detailing problems. So for the analytical determination of the non-linear joint response a macroscopic inspection by subdividing the complex finite joint into so-called simple “components” proved to be successful. In contrast to the finite element method (FEM) the components - as logical subsystems of the joint

接上
在lecture 9第9页有以下一段话
By analysing full-scale joints it could be seen very quickly that the number of influencing parameters is too large. So world-wide the so-called component method is accepted universally as the best method to describe the joint behaviour analytically. In contrast to the common finite element method (FEM), which often fails to consider local load introduction problems, the joint here is divided into logical parts exposed to internal forces. So while the FEM works on the level of strains and stresses, the component method concentrates on internal forces and deformations of the component springs.

有限元法fails to consider local load introduction problems――如何理解？