KANEKA Corporation - Automated Drying Process Temperature Control with ML to Optimize Resources

Takashi

Team members:

• Shingo Kinaka, Manager, DX & Carbon Neutral Team
• Mr. Hiroaki Ichiriki, DX & Carbon Neutral Group, Global Production & Engineering Integrity Center
• Mr. Yuya Hitotsuda, Corporate Manufacturing Integrity Center

Country: Japan

Organization: KANEKA Corporation

Line of Business (Solutions Vehicles) Vinyls and Chlor-Alkali, Performance Polymers (MOD), Performance Polymers (MS), Foam & Residential Techs, E&I Technology, PV & Energy management, Performance Fibers, Medical, Pharma & Supplemental Nutrition, Foods & Agris.

Awards Category:

• Best MLOps Use Case

Business Challenge:

Resin represents an important part of the business and requires to dry before being sold to manufacturers.

Within drying installations, the operator adjusts the operating conditions under several constraints (temperature, exhaust humidity, feed rate, etc.)

This process requires frequent manual adjustments to take into account the fluctuations in supply, ambient temperature, and other factors, as well as exhaust humidity restrictions. Newcomers have difficulty because it requires know-how, and it is difficult to accurately predict the behavior of the system and stick to it, resulting in wasteful operation.

The mission of the Production Department is to reduce waste and standardize operations across Kaneka's production facilities, therefore took on the challenge to optimize the resin drying process.

Business Solution:

Our objective is to predict the future state of the process with gradient boosting according to the temperature setpoint,

A new software has been installed on the equipment to transmit real-time information to the database. The data workflow takes in the temperature setting in various parts of the dryer. It induces an intermediate feed rate and the dryer exhaust humidity.

The following goals are defined to operate in optimum conditions:

1. Increase intermediate supply
2. Ensure exhaust humidity is below the standard
3. Lower heater steam usage

Therefore the algorithm will compute the optimum parameters in line with the targets through inverse analysis.

With AI prediction and the calculation of optimum conditions, feedback is provided to the equipment which will update the temperature settings so as to improve production and steam usage, without any manual intervention.

A guidance screen also enables employees in the Production department to monitor the current and past conditions to check for any inconsistencies.

Day-to-day Change:

The introduction of the Dataiku system has enabled us to consistently adjust the temperature in our daily work and has eliminated the need to assign people to tasks; the Dashboard can be displayed and monitored, so temperature control can be seen at a glance, allowing us to focus on other tasks. The Dashboard display and monitoring has also enabled us to quickly see the current temperature control and free up time for other tasks.

Business Area Enhanced: Manufacturing

Use Case Stage: In Production

Value Generated:

The three main improvements made in dryer operations in this use case are:

• Improved steam consumption (cost reduction).
• Increased production capacity through constant operation under optimum conditions.
• Standardization and automation have moved away from depending on personalized operations.

Value Brought by Dataiku:

Using Dataiku, we were able to quickly develop data flows and facilitate communication with engineers in the field using a graphical explanation screen.

We have been able to cut back on development time by half, compared to a custom solution, by using standard package functions. We also expect to continue improving this solution with the latest technologies which Dataiku will integrate.

As an end-to-end platform, it also features a dashboarding capability, so that we didn't have to introduce any other GUI tools. Our in-house personnel is empowered to analyze data, develop models, and communicate insights - which accelerates momentum across the organization to delve into other day-to-day improvements for further efficiency.

Value Type:

• Reduce cost
• Reduce risk
• Save time
• Increase trust

Value Range: Hundreds of thousands of $

JamesO

very cool use case!