Experimental Identification and Multiobjective Optimization of the Vibrocompacting Process of Composite Substances

Vibration is the most popular means of compacting material mixtures such as fresh concrete, powder and granulated materials. To achieve a high compaction rate and high strength of formed product - paving blocks, bricks etc., the vibration must be combined with high pressure. There is an extensive literature on the effects of such features as frequency, amplitude and acceleration, and a lot of literature about rheological models of raw concrete. The results depend on material characteristics such as the difference between its initial density after preparation and its maximum packing fraction, the coefficient of friction between the grains, the angularity of particles etc. [1]. So the theoretical results are not capable of giving full and adequate information for the optimization of the vibrocompacting process according to quality of products, costs and other criteria of manufacturing effectiveness. Optimal vibrocompacting parameters usually must be determined experimentally [2]. In the present work a more general methodology of experimental identification of mathematical models of compacting process and consequent multiobjective optimization is introduced. This methodology consists of the following steps: 1. The choice of input variable parameters X for the model building. Some of them are controllable (vibration frequency, pressure, water/cement ratio...), other parameters have an effect on process, but are only monitored, and must be considered as given (dimensions of product to be formed, granulometric parameters of used materials...). 2. Carrying out the natural experiments according to sequential, Mean Square Error-optimal experimental design [3]. We used material testing machines Instron and Zwick with specially constructed mould-plunger devices for raw concrete forming. The measured results are load and displacement (compaction) graphs versus the time. 3. Approximation of load and displacement graphs with minimal number of parameters Y. Exponential functions with 2-3 parameters give sufficiently accurate approximation. 4. Building the mathematical models for the approximation of the dependence of the parameters Y on input factors X. Nonparametric approximation methods like kriging should be used to reduce the number of experimental tries [3]. 5. Analytical or experimental formulation of optimization objectives (compacting rate, manufacturing cost, product cost etc.). Providing of multiobjective Pareto-optimization. 6. Validation of optimization results using additional experiments. This methodology is demonstrated for the choice of optimal vibration frequency, pressure and process duration. The objectives are compacting rate, consumed energy, pressing force. The same methodology can be used for powder and granulated material compacting