In dense orchard environments, apple-picking robots face significant challenges due to highly variable fruit postures, clustered fruit distribution, and severe occlusions from branches and leaves, all of which reduce the stability of suction-based acquisition and increase the risk of fruit damage. To enhance adaptability and operational reliability under such complex conditions, a pneumatic suction-twisting endeffector was developed. The device employed a compliant corrugated suction cup to achieve flexible negative-pressure adhesion and integrated a coaxial twisting-pulling mechanism that emulated the mechanical process of manual harvesting, thereby facilitating the fracture of the abscission zone and enabling low-damage fruit detachment. A quantitative model describing the relationship between blower flow rate and suction force was established based on the principles of negative-pressure adhesion. Furthermore, computational fluid dynamics was employed to analyze the effects of fruit posture, suction distance, and airflow rate on the suction flow field and surface pressure distribution. The simulation results indicated that variations in fruit posture influence suction performance primarily by altering the effective windward area and redistributing the pressure gradient over the fruit surface within the annular low-pressure field. As the suction distance increased from 1 mm to 5 mm, the surface pressure gradient decreased markedly, resulting in a substantial reduction in suction force. Increasing airflow rate strengthened the annular negative-pressure region and yielded a monotonic increase in suction force. Based on these analyses, the recommended blower operating range was determined to be approximately 15 ~20 kPa vacuum pressure with an airflow rate of around 240 m3/h. Field experiments conducted in dwarf high-density orchards demonstrated that the proposed end-effector achieved a maximum reliable suction distance of 4.41 mm and was capable of reliably grasping fruit under positioning errors of up to 5 mm. The optimal suction angle was determined to be 0°. The average suction-based acquisition time per apple was 1.97 s, with an overall capture success rate of 91% . Among different fruit sizes, medium sized apples (diameter 70 ~ 90 mm) exhibited the highest capture success rate, reaching 94.4% . No visible mechanical injury was observed on harvested fruit, confirming the feasibility and reliability of this design for soft and damage-free harvesting in complex orchard environments. The research result can provide theoretical support and technical reference for the structural optimization of pneumatic harvesting end-effectors and the development of intelligent robotic apple harvesting systems.