I.N. Glushkov, M.V. Zhestev, V.A. Belkin
The article is dedicated to the two approaches to the simulation modeling of complex engineering systems. These simulation modeling approaches are based on principles of sequential and distributed simulation. The authors of the article consider the typical schemes of the sequential and distributed simulation model construction and to the function of the modules which comprise these schemes.
The development of the sequential simulation modeling technologies was due to initial monoprocessor computers orientation. This became the basis of the typical computing schemes of simulating algorithms that have been elaborated by now. However as a result of a blistering technological development of the powerful microprocessors, building super computers on their basis and netting of local and global computer networks the second trend in simulation modeling began to evolve. This second trend is related to the methodology and technological aspects of the distributed modeling which is based on the methods of distributed data processing.
The transition from the sequential to the distributed simulation modeling was due to the necessity to solve the following practical task: the use of the computing resources of the several processors for the realization of a simulation experiment (the distribution of the simulation model components over processors (computers) and their joint participation in the simulation experiment); the use of the local memory of other processors (computers); the activation of the several operation concurrently on several computers (the time expenditures reduction on the simulation experiment); the integration of the already prepared simulation models and the their participation in the joint simulation experiment (the reuse of the program code); reliability augmentation while performing the simulation experiment, as the performance can be continued on other processor (computer) in case of failure of one of the processor or computer which realizes one of the components of the simulation model. The article shows that the main obstacle in using the distributed simulation model is the necessity to organize the interoperation of the processing units including the synchronization of their performance. The approaches presented in the article have found practical use while developing the integrated simulation stand of the special assignment complex managerial and engineering system. This simulation stand includes subsystem and asset models which function under control of the single supervisor.