This article presents a dataset generated from a cylindrical plunge grinding process conducted in an industrial piston ring manufacturing environment. The data aim to support studies focused on process capability, variability, robustness, parameter optimization, and modelling approaches in grinding research. The experimental conditions were defined following a Central Composite Design for k=4 factors and axial distance ρ=1.5, which established the levels of Wheel infeed rate, Dressing speed, Grinding wheel peripheral speed, and Dressing depth. These parameter combinations were carried out on the shop floor under real production constraints, ensuring that the collected measurements represent industrial operating conditions rather than laboratory simulations. Two external noise factors were considered during data collection, consisting of the different mandrels and the position of the piston ring within the production package. These noise sources reflect variations commonly encountered in manufacturing and allow researchers to investigate the robustness and sensitivity of dimensional responses. A CCD with thirty runs, being sixteen factorial points, eight axial points, and six center points, organized in two blocks of experiments, was performed, associated with four noise combinations, defined according to an experimental plan. For every condition, ten repeated measurements, represented by ten piston rings sampled from the package and assembled in different mandrels, were acquired using the dimensional control fixture routinely employed in the industry, forming a dataset of 1200 runs. This procedure ensured consistency with existing quality inspection practices and provided a rich structure suitable for repeatability and uncertainty analyses. The dataset includes raw measurements, process parameters, and experimental identifiers that enable multilevel exploration of machining performance. Its structure supports diverse analytical applications, including modelling of process capability, variability, evaluation of noise effects, optimization of input parameters, statistical analysis of repeated measures, and the development or validation of data-driven and machine learning methods. Owing to its industrial origin, the dataset offers realistic variability patterns and is relevant for comparative studies, benchmarking activities, and the development of predictive or robust design frameworks in manufacturing research.
Cesário et al. (Sun,) studied this question.