Green sand mold
Relying on human-interface based decisions to control high-volume, high-density, high-speed green-sand molding systems significantly lowers the complex understanding of the sources of errors, argues Deepak Chowdhary.

Bringing Big Data to Green-Sand Processing and Control

Cloud-based software promises to reduce casting rejects and optimize additive consumption by applying predictive and prescriptive data analytics, machine learning and IoT functions to green-sand molding

The Smart factory "vision" is gaining global acceptance. It involves integrating data analytics, machine learning and IoT by incorporating sensors, SCADA and line-speed or near-real-time data-based analytics for process optimization. Predictive and prescriptive data decision-support, predicting highly accurate outcome has shifted the paradigm in manufacturing and many other industrial and consumer sectors. For manufacturers, rapid proliferation and advancement of information technologies, such as cloud computing, allows them to generate and store huge volumes of data.

In addition, the digitalization of process automation has empowered industries to collect and transmit near-real-time data. Cloud computing ensures data availability without geographic or physical limitations. By leveraging the power of data-based analytics and machine learning, industries are generating significant tangible and intangible profits.

But, according to Deepak Chowdhary, “It is a matter of concern that modern foundries with all their modern machinery and technically demanding castings processes are unfamiliar with the power of analytics and are heavily dependent on human experience; especially in the control of molding processes — which is rapidly shrinking.”

Chowdhary is the inventor and developer of Sandman, a patented Cloud-based software service offering algorithmically derived mathematical modelling that foundries use to optimize molding sand control.

“A unique and complex model framework for prescriptive analytics predicts dose-by-need additive quantity for each batch of sand,” he noted, “helps to reduce over/under dosing of the system sand. This allows foundries to operate their sand systems in an optimal and dynamically balanced manner that is both sustainable and scalable.”

Optimizing additive and sand usage thus also helps reduce toxic waste. Chowdhary argues that the need for data-based analytics is critical in green-sand molding.

Sand is also the highest contributor to casting defects due to inherent process variables. Almost 80% of all castings worldwide are produced by green-sand casting processes, owing to the low operating cost and wide availability of low-cost raw materials. Therefore, Chowdhary continues, “relying on human-interface based decisions to control high-volume, high-density, high-speed green-sand molding systems significantly lowers the complex understanding of the sources of errors, and the potential to resolve them. If a foundry is checking 10 sand properties daily and there are even three significant types of defects occurring repetitively, it’s like taking decisions with a multivariable factor of 103!,” he emphasized.

Moreover, foundries lose precious time if they continue to ignore the transformational change happening with digitalization and analysis. Having sophisticated molding process lines is not enough if the execution of the process continues to rely on the experience or expertise of individuals. That experience will not last, Chowdhary argues.

Furthermore, modern university-level metallurgical courses are not focused on shop-floor management, or the ‘why’ of a process or the inputs. Today, learning (in classrooms, but also online and via social media) is increasingly weighted on AI-, ML-, and data-related subjects. What is called for then? “Resolute on-boarding of modern, data-driven analytics and decision support, machine learning, and deep learning technologies to limit the human interface to only intuitive and human-intelligence control,” he stated. “Leaving machine learning and AI to do the complex and heavy lifting of process consistency, repeatability and accuracy is the way forward for the metalcasting industry.”

SandmanSandman software, two graphs

Examples showing actual data-based process control via analytics and its results.

The Green Sand System Optimization software application leverages historical data and continuously "learns" from current data to predict influencing sand properties that need correction, in order to optimize the molding process outcomes in castings. With this input to the process, foundries are set to produce better castings with lower rejection rates, fewer variability in sand properties, and optimized “dose-by-need” sand additive consumption.

Also, the software integrates IoT functionality using sensor and SCADA-based technology to generate real-time data. The machine-learning algorithm establishes a cause-and-effect relationship between sand parameters, casting rejections, and additives consumption data. With this data, it offers optimized solutions for significantly accurate process control outcomes.

On deeper, more granular levels of data, the predictive modeling can identify optimal sand properties and corresponding optimal rejections for each type of casting and rejection, according to Chowdhary.

The Sandman software concept is supported by case studies to validate its usefulness for profit-making by process optimization. It keeps the molding-sand treatment process repeatable, adaptable, scalable, and dynamically stable for better quality castings, improved productivity, and on-time castings supply.
Contact Deepak Chowdhary at [email protected] or visit www.mpminfosoft.com, or  www.sandman.co.in

TAGS: Molds/Cores
Hide comments

Comments

  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Publish