For the analysis of access control networks, where capability lists, access control matrices, or RBAC permissions are available, it can be very useful to be able to determine which subjects can be able to know, or which objects can be able to store, data originating from objects in the network. This information can be used in order to answer questions of secrecy, integrity and privacy, related to the data flow analysis problem. On the basis of a logical method, we present a graphical formalism capable to represent such networks and for which the data flow problems can be defined. We present algorithms to calculate answers to data flow questions. Complexity analysis and simulations show that these questions can be practically answered for networks of sizes up to several tens of thousands of subjects and objects, which is the size of many real-life organizations. We also show that the results obtained can be used in the process of role engineering in Role based access control, for determining secrecy levels, as well as for eliminating or combining roles or objects. Finally, a method is demonstrated to go from capability lists to Label-based access control systems.