Toby Ingleton |
This article first appeared in the Summer 2017 issue of The Record.
Founded in 1988, Deschutes Brewery is the seventh-largest craft brewery in the US, with a distribution network spanning 30 states.
Historically, the brewery team tracked production in notebooks or in spreadsheets. To support growth, the company needed to enhance its production and distribution.
Deschutes Brewery adopted the PI System, a data management and operational intelligence solution from OSIsoft. Based on OSIsoft’s proprietary time-series database, PI System ingests data from virtually any source, including sensors installed on brewing vessels, and delivers this information in real-time to any application and user requiring the data.
The Deschutes team wondered whether the accumulated information could help optimise the brewing process by using advanced analytics tools available with the Microsoft Cortana Intelligence suite of cloud products. Deschutes, OSIsoft and Microsoft data scientists set about answering that question.
Within the brewing procedure, the nine-stage fermentation process lasts approximately 12 days, and presents multiple opportunities to optimise production. Deschutes decided to focus on the second stage – the transition between fermentation and free rise.
To gauge whether beer is ready for free rise, brewers manually measure liquid density every eight to ten hours. A miscalculation during this process can result in up to 72 hours of lost production time. However, an accurate transition prediction could save time without affecting quality.
Another challenge involved the ever-changing production line. To effectively apply data science practices to these processes, Deschutes required a flexible solution that would automate data preparation, and to which they could integrate predictive analytics with daily production processes.
Deschutes decided to join the Red Carpet Incubation Program hosted by OSIsoft and Microsoft, and extend the brewery floor to the Microsoft cloud to take advantage of scalable, automated machine learning services.
Onsite, the PI System now continuously collects information such as temperature, pressure, flow rate, and device state from vessels, and records fermentation readings performed by brewers.
Using the PI Integrator for Microsoft Azure, Deschutes is able to deliver analytics ready, contextualised and event data from the PI System to Microsoft Cortana Intelligence Suite for predictive analysis.
This data is presented in web-based Power BI dashboards and is also fed back to the PI System. This means production operators can work with data while using familiar tools such as PI Vision from OSIsoft. Brewers can now accurately predict when to transition virtually any brand of beer to the free rise stage.
Deschutes expects a considerable boost in productivity thanks to the solution, while the 20% increase in production capacity from existing equipment is expected to considerably increase annual profits.