Introduction: Systems for fully closed loop insulin delivery (AP) lack bolus administration at mealtimes, resulting in higher postprandial glucose excursions than hybrid APs. In order to mitigate this disadvantage an algorithm can be used for the automatic delivery of bolus insulin (BI) in case a meal is recognized from the CGM recordings. The use of a full AP with automatic BI delivery is tested in simulations against a hybrid AP with manual bolusing and a full AP without any meal boluses. Methods: The simulation approach used for this purpose combines data from real clinical trials with mathematical models. A cohort of five insulin-treated T2D patients is used to test the feasibility of the different AP concepts. Results: Using a full AP with automatized bolusing more patients were able to achieve their therapeutic goals (defined here as HbA1c < 7%) than for the case without meal boluses (4/5 vs. 1/5). Time in hypoglycemia is at a comparably low level for both considered full AP versions (0.9±1.5% vs. 0.4±0.8%), whereas time in hyperglycemia (21.9±5.4% vs. 30.9±7.9%) and HbA1c (6.8±0.3% vs. 7.4±0.5%) is significantly reduced for the case with automatized bolusing. However, as can be expected, a hybrid AP approach with manual bolusing does result in even better glycemic outcomes. Conclusion: Results show that automatized bolusing does indeed lead to enhanced glycemic outcomes in full AP systems.