SHOP4CF-Cheepsh: processes digitalization in the specialty food industry with Queizuar

Written by
inverbis analytics
15 de March de 2023 Max 3 min read

CHEEPSH is the acronym of a project that, under the umbrella of SHOP4CF, we have just started in collaboration with Queizuar, the leading producer of cheese with Protected Designation of Origin (PDO) Arzúa-Ulloa. Queizuar and Inverbis maintain a permanent process mining laboratory.

SHOP4CF (Smart Human Oriented Platform for Connected Factories) is a program funded by the European Union that, in several calls, supports the creation of people-centered software solutions in industrial companies. Queizuar and Inverbis’ project will transform the execution of processes in environments such as Arzúa-Ulloa cheese where the maintenance of traditional production conditions must be combined with a scale of production capable of reaching large audiences within commercial distribution.

CHEEPSH will solve some of the main problems of specialty food manufacturing related to process variability due to non-standardization of raw material (e.g. milk) and subjective decision making by workers, who lack up-to-date and accurate information, leading to the risk of a low quality product and production shrinkage with high variability.

For example, a Protected Designation of Origin cheese producer such as QUEIZUAR suffers daily from problems such as these in the multiple steps of the manufacturing and distribution processes:

1) PDO cheese producers use raw milk from local natural resources, and cannot standardize the properties of the milk. As a result, every day each tank of milk has a different acidity, fat – protein ratio, etc.

2) Lacking information about possible prior process variations, Cheese Masters have to make different decisions in each circumstance regarding key variables before moving to the next stage of the process, such as at what point the right temperature, texture, acidity or granularity is reached.

The combination of both facts leads to cheese products of different quality in an uncontrolled way, which can lead to loss of revenue and productivity (too many products not conforming to the desired quality standard and production losses).

These problems will be addressed by:

1) DATA COLLECTION WITH PLUG&PLAY SYSTEMS.

  • Collecting data in a simple way for low-skilled workers digitally through easy-to-use forms of simple choice (“Current step: cheese curing. Finish? Yes/No”).
  • Combining this worker response data with low-cost Internet of Things (IoT) sensor collection to track machinery. To this end, INVERBIS will use the FLINT component of SHOP4CF to facilitate interoperability of IoT devices and easily capture data from any sensor from multiple communication methods.
  • Combining this data with any other management system (e.g. ERP). The INVERBIS solution has a powerful data virtualization tool that allows this solution to easily access any type of database without having to develop specific data connectors for each system.

2) PROCESS MINING AND STATE-OF-THE-ART ARTIFICIAL INTELLIGENCE

Process Mining supported by Artificial Intelligence (AI) in order to analyze and optimize the process and provide decision support suggestions to workers adapted to the worker’s context and to the tasks previously performed by all workers involved in the process.

3) WORKER DECISION-SUPPORT INTERFACE

  • The provision of easy-to-understand automated contextual task guidance (e.g., “Let the milk rennet mixture reach a temperature of 60°”).
  • The provision of easy-to-understand alerts and calls to action (e.g., “The acidity level is too high. Please reduce mixing frequency by 10%”).
  • Workers will interact with the solution through an interactive point in each zone of the shop floor, equipped with an industrial touch screen and audible alerts.

The project started in February 2023 and will run until the end of autumn this year.

Read more…

How can you improve what you can’t measure?

How can you improve what you can’t measure?

In logistics, data is the foundation for optimizing processes and making informed decisions. But what happens when your data is incomplete? This real-world example highlights a common issue. This means, for example, you can’t determine how much time passes between a...