With the fast development of Internet and rapid acceptance of Web Service technology, more and more service resources have emerged. Service composition which integrates the functionalities of different services is a promising technique for developing applications across multiple organizations. However, in a distributed, dynamic and autonomous environment, such as a service composition-based system, the availability and reliability are big concerns in terms of nonfunctional properties. In this paper, we propose a novel service composition method, ANGEL, with the target of the improvement of system availability. We adopt redundant mechanism in ANGEL and propose a model to improve the property of the service availability. We model the multiple services selection problem based on redundant mechanism as a nonlinear mixed integer programming problem and therefore, we propose two heuristic algorithms to select multiple feasible services that have the same functions, but with better availability. In order to maintain the availability of composite services, we further introduce monitor and detection mechanisms. Through the comprehensive experiments, we find that our proposed techniques can indeed achieve better availability as expected.