The success of large projects depends on how people associated in teams and organisations perform their tasks and cooperate to achieve a shared goal. Success depends on an effective socio-organisational design that defines organisational structures; human resources in the form of people's abilities, roles and responsibilities; and establishing processes and contracts that will drive measures of success. Predicting the behaviour of large projects is difficult and the performance of the socio-organisational design is likely to be validated only when it is put in to practice. Simulation offers a way to explore the likely behaviour of social systems without the drawback of negative effects on real situations. Agent-based models, in particular, can be applied to the simulation of organisations. This paper proposes an agent-based framework to simulate large projects, in particular those that involve the development of complex technical systems. The proposed framework is then applied to a simulation using Repast, an agent-based toolkit, to demonstrate how the framework can be applied to investigate feedback loops in the social system, and how these feedbacks affect the schedule, cost and quality of final products. The paper concludes discussing how agentbased models and simulation using this framework can help managers and decision-makers to acquire a better understanding of social systems, to explore socio-organisational designs of large projects with better chances of success and avoiding designs that would have higher risk of failure.