Search

An algorithm is developed for solving the multi-vehicle subscriber dial-a-ride problem. This algorithm is tested on a set of data from Project Mobility in Baltimore, Maryland. Results of these tests are reported along with recommendations for future work in this area., Approved for public release; distribution is unlimited.

An algorithm is developed for solving the multi-vehicle subscriber dial-a-ride problem. This algorithm is tested on a set of data from Project Mobility in Baltimore, Maryland. Results of these tests are reported along with recommendations for future work in this area.

The optimization of system reliability of a series parallel system containing t types of components is found where the cost of purchasing the components is disregarded, a component can be assigned to more than one component position, and a limited supply of components is available for assignment. The optimal solution is found by ranking the reliabilities of the components of each type and searching over these ranks in the orders specified in this paper.

Esary and Proschan show that a lower bound to system reliability can be found by enumerating all min cut sets in the coherent structure, connecting the components in each min cut set in parallel and joining each of these parallel subsystems in series where replicated components are replaced by identical yet independently operating components. A module of a coherent structure is a subset of the basic components of the system which can be treated as a component of the system due to their substructure topology. In this paper, it is shown that a lower bound estimate of system reliability can be derived by decomposing the coherent structure about its modules and applying the Esary- Proschan lower bound procedure to each module and then to the resultant coherent structure where each module has been replaced by a single component whose reliability is the Esary-Proschan lower bound to that module.

A basic problem in studying Markov Chains is the determination of the class properties in each communicating class. In this paper, two algorithms for determining class properties are presented - the first algorithm finds all the class properties while the second algorithm merely finds the set of communicating classes and their classification. Extensions of these algorithms to other problems are noted and the computational efficiency of these algorithms is described. (Author)