Process optimization has become a powerful tool for the design and improvement of flowsheets. In the past decade, two different approaches have emerged for flowsheet optimization. With modular simulators, unit models are treated as black boxes and are fully converged with every flowsheet pass. With an equation-oriented approach, on the other hand, models are converged simultaneously with the optimization problem but here the model structure is not exploited in the procedure and specialized unit algorithms cannot be used. Here we present a simple extension to simultaneous modular optimization that allows for simultaneous convergence of the unit models with the optimization problem while exploiting the unit structure of existing Newton-based process models. At the heart of this approach is a reduced space Successive Quadratic Programming (SQP) strategy that requires only weak interactions with the flowsheet. In addition to the development of a flowsheet decomposition strategy, we present a case study of the Hydrodealkylation (HDA) process that demonstrates the potential of this approach.