Sand Mould Identification Pattern for Casting Traceability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Modern foundry production lines face challenges in achieving traceability of castings for retrieval of production and quality data without affecting production rates, as existing solutions are limited in detecting identification patterns after industry-standard finishing treatments and require additional process steps, leading to increased costs due to rejection of defective castings.
Innovation Solution
The implementation of individually adjustable indicator elements with rounded edges and a machine learning-based image detection system that provides digital images of cleaned castings, allowing for extensive identification pattern combinations and accurate detection post-finishing, integrated with a computer-controlled database for data storage and analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If identification elements are positioned in sand mould parts before assembling, then casting identification is achieved, but production rate is reduced
Solution Approach 1:
The identification pattern is impressed in the sand mould part during the compaction process itself, before the mould is assembled and before casting. This preliminary action integrates the identification marking into the existing moulding operation without requiring separate positioning steps, thereby maintaining production rate while achieving reliable casting identification.
2Reliability
If identification elements are used, then casting identification is achieved, but additional process steps are required
Solution Approach 1:
The identification pattern impression is merged with the sand mould compaction process. The same compaction force that forms the sand mould also impresses the identification pattern from the pattern plate into the sand mould part. This combining of functions eliminates the need for separate identification marking process steps.
3Extent of automation
If standard image detection is used, then identification patterns are detected, but detection accuracy decreases after finishing treatment
Solution Approach 1:
The identification pattern is designed with specific parameters optimized for machine learning detection: high contrast between the pattern and background, geometric shapes with distinctive features, and placement on casting surfaces that remain accessible after finishing treatment. These parameter changes enable accurate automatic detection even after shot blasting or other finishing processes remove surface material.
4Adaptability or versatility
If more indicator elements are provided, then more identification combinations are achieved, but device size increases
Solution Approach 1:
Instead of increasing the number of indicator elements in a linear arrangement, the system uses a two-dimensional grid arrangement of indicator elements on the pattern plate. This dimensional change allows for exponentially more identification combinations (N×M grid provides N×M unique positions) without proportionally increasing device size, as the elements are distributed across a compact rectangular area rather than requiring linear space.
Data Source
AI summary
The foundry production line includes a sand moulding machine, a melt pouring device, a shakeout machine, a finishing apparatus, an inspection station and a computer controlled database system. A pattern plate is provided with a sand mould identification device including a plurality of individually adjustable indicator elements adapted to impress an identification pattern in a sand mould part. Each indicator element has rounded edges and indicates a direction. An automatic image detection system includes an imaging device arranged at the inspection station and being adapted to provide a digital image of an individual identification pattern formed in a cleaned casting. The automatic image detection system includes a computer system running a computer program developed by means of machine learning to analyse the digital image and thereby detect the individual identification pattern.