Sustainability efforts lead to better efficiency, quality and profitability
Sight Machine, Microsoft, and Accenture on the MPI Vodcast
Jon Sobel, Co-Founder and CEO of Sight Machine, discusses manufacturing sustainability and energy optimization for manufacturers in this new podcast from Microsoft.
There are a lot of buzzwords floating around improving manufacturing sustainability – green manufacturing, IIoT, the circular economy, the Fourth Industrial Revolution. All of them have real meaning. But in the plant the reality is more basic. Having a continuous “Earth Day mentality” leads not only to sustainability but to better efficiency, quality and profitability.
Most manufacturers already have plenty of production data: terabytes of input streaming in continuously from hundreds to thousands of shop-floor sensors. Blend this insight-filled data together with information on quality, downtime, and energy, and voila, you have complete real-time visibility into production – the foundation for optimization.
Toros Tarim is the leading manufacturer of fertilizer for Turkey and neighboring regions. This large scale process requires a huge industrial chemical complex with each step in the production pathway (sulfur acid to phosphoric acid to ammonium diphospate for example) being dependent upon the former. Downtimes are extremely costly, with potential to impact all areas, including the on-site co-generated electricity whose surplus also bolsters the Turkish electrical grid. Toros approached Sight Machine as part of their digital transformation and lean journey with the clear expectation to start at building a contextualized data platform of all their data, facilitating an holistic view of their process network, before moving to data visibility and automated dashboarding of reports and key performance indicators (KPIs).
The next part of the journey is now underway as Sight machine and Toros partner to use their data platform to explore advanced analytics such as predictive and AI, process optimization (including for sustainability and emissions management), and maximizing plant reliability / minimizing asset downtime. This project was implemented entirely remotely during the COVID-19 pandemic and was possible because of the unique way Sight Machine is able to rapidly ingest and use automated tool sets to transform and contextualize tens of thousands of data fields in parallel.
Corrugated paper plant had trouble knowing which specific process parameters drove up scrap and customer claims. In addition, plant wanted to know when and which SKUs resulted in driving peak energy consumption.
Sight Machine integrated process data, energy data, quality data and customer claims data to provide integrated models that were then utilized to run advanced analytics including AI/ML algorithms.
AI/ML yielded insight within 2 weeks about causes of quality issues that had previously eluded company data scientists. Similarly, connections between process variation and energy use and defects and process variation represented new insight into mission-critical questions.
Steam usage had been high on a paper machine coming up from an outage with no identified root cause.
Variance analysis and correlation analysis identified a single valve that was found to not be 100% open. Root cause was found and corrective action was taken for a valve cut back.
This single insight and the corresponding best practice developed from it – one of dozens generated annually – yields millions per year at the enterprise level in steam savings and increases in production rate.
The lag between the causes of quality issues in a furnace and the emergence and detection of faults in glass 3 days later, makes it difficult for glass producers to manage yield and results in glass being returned to the furnace at every glass company. This results in increased energy use when glass has to be re-melted to remove defects.
By continuously analyzing 45 furnace parameters, Sight Machine predicted increases in fault density up to three days in advance with an extremely high degree of accuracy.
The furnace analytic predicted 78% of all defects from the entire process.