The main objective of this research work is the smart optimization of the production process of corrugated pipes. A variety of modeling methods are used to achieve the project goals. We implemented a hybrid modeling approach combining analytical, numerical and data-based modeling for corrugated pipes. Multi-dimensional mathematical models were developed and implemented using symbolic regression analysis based on genetic programming for predicting the wall thickness distribution as function of the mold geometry and initial parison thickness. These models are later optimized by taking the thermal energy flow parameters into account. Thus, the production process can also be optimized for efficient cooling. Furthermore, to ensure the reliability and performance of the created pipe geometry, a mechanical performance analysis was also performed based on the models to determine the strength of a geometry. These results will help to develop guidelines for energy-efficient manufacturing a lightweight corrugated pipe that has the required mechanical performance with minimum material requirements. In addition, the findings can be used for the effective design and operation of corrugator and for the optimization of corrugated pipes.

In the progress of this project, a hybrid modeling approach that combines numerical and data-based modeling is implemented. The workflow is structured as follows: first, the mold block geometry with different mold shapes is analyzed. Then, the similarity theory is applied to determine the dimensionless influencing geometry and processing parameters, and then a comprehensive parameter design study is conducted to identify the most critical influencing parameters. The results of numerical simulation are later used as input for data-based modeling so that mathematical models can be developed. Furthermore, a mechanical performance analysis is performed based on the wall thickness models to determine the achievable strength of geometry. In order to gain insight into the physical conditions during the mold process, a selected mold block, equipped with a specifically devised sensor inset, will autonomously collect data for the analysis process.

In general, the expected results of this research work relate to the knowledge gained to optimize the manufacturing process and the final product itself, as well as to optimize the corrugator design. Using engineering simulation and data-based modeling to develop mathematical models for the prediction of the wall thickness distribution and the heat flux in the corrugator. Furthermore, guidelines and strategies for manufacturing a lightweight corrugated pipe that has the required mechanical performance with minimum material requirements will be provided.

So far, the initial stage of modeling the entire process of corrugated pipe extrusion has been achieved. The influences of major geometry parameters on the parison inflation process were identified and investigated by applying the theory of similarity, dimensional analysis, and a parametric design study. Correlations between independent and target parameters were established and utilized to estimate the wall thickness and its distribution in corrugated pipes. Multi-dimensional regression models of the wall thickness distribution as a function of mold geometries in extrusion blow molding of corrugated pipes were developed using heuristic approaches and implemented for mold design in the early design phase. The comparison of numerical simulation results and model predictions also confirmed the validity and feasibility of the regression models developed in this work. First comparisons with experimental trials delivered promising results. These results showed that the wall thickness predictions capture the reality as long as the velocity of the extruded parison approximately equals the line speed of the corrugator. For new processes, the proposed method may prove to be a valuable tool for minimizing the number of expensive and time-consuming experiments when evaluating (new) pipe designs and may add value well before the final product is produced. The developed models allow a target variable (of the corrugated pipe geometry) to be predicted without manufacturing and prototyping of a product. In addition, the regression models can cover a wide range of geometry variations as they are dimensionless, and as long as the new geometry is within the chosen dimensionless geometry parameter range. For very small and very large corrugated pipes, there is some risk that the dimensionless parameters fall within the extrapolation range for various reasons.