Early design decisions have a significant influence in the final success of the project. One of the most important decisions is to determine the system architecture, as it highly impacts the performance of the system. System architecture optimization can be used to determine the best possible architectures through the formulation of an optimization problem, allowing to explore the design space without traditional bias and conservatism. MDO can be used to evaluate the performance of each architecture, allowing to consider the interactions between the multiple and coupled disciplines involved in the design process. To do this, MDO platforms have to satisfy multiple requirements, including the automatic readjustment of the MDO problem for each system architecture. They also have to be adapted to collaborative MDO, so that they can be used in real industrial cases. Before this research, there was not MDO platform that satisfied all these requirements, impeding the integration of system architecture optimization in the industry. The MDO platform consisting on MDAx and RCE was adapted to collaborative MDO and satisfied all requirements to be used as architecture evaluator, except the automatic readjustment of the MDO problem. To fill the previous gap, the main objective of this research has been to extend MDAx backend code to allow the formulation of these dynamic MDO problems, allowing to use it in the system architecture optimization process. To achieve this, first the possible modifications that the system architectures can cause in the MDO problem, called architectural influences, are determined. Then, some possible implementation strategies MDO platforms can use to deal with these influences are presented. After that, the actual implementation process used to extend MDAx backend code is widely discussed. Afterwards, a benchmark problem based on Fourier series is used to verify the implementation. A real engineering problem, based on the design of a space multistage rocket, is also used as validation to show the potential tool, and more generally, of the methodology. Finally, some conclusions and possible future steps are drawn. In conclusion, this research allows to reduce the existing gap between system architecture optimization and MDO by obtaining an MDO platform that can be used as an architecture evaluator. Also, the different requirements identified for the inclusion of architectural influences, as well as the benchmark problems discussed, are aimed to help developers to extend their MDO platforms to be adapted to system architecture optimization, reducing the barriers for its implementation in the industry.