Achieving sustainable performance with data
Reconciling all data to achieve sustainable performance
Data strategies often focus on technology such as Big Data, IoT and AI. This can cloud fundamentals for achieving sustainable performance. In this article we explore how using these technologies can contribute to actions and processes that improve performance over time.
Obstacles to sustainable performance
Several obstacles to performance related to data can be identified in the factory:
- Access to information as close as possible to needs and decision-making – the right information and the right person at the right time in the right place.
- Ability to process and analyze to capture relevant information and identify required actions.
- Effective rollout in the field.
Identify sustainable performance data drivers
Analyzing performance is essential for identifying potential obstacles and being able to take a holistic approach to the interaction between data and performance. For production, there are two possible types of action:
- Controlling production processes. In this case, best past production conditions are duplicated to enhance performance. Known performance levels are used to apply optimal operating conditions. The aim is to maximize productivity, minimize consumption of materials and energy, and reduce environmental impacts.
- Improving production processes. In this case, the aim is to understand existing processes to improve them incrementally or through developing significant changes. This action achieves new levels of performance that cannot be achieved with simple process control.
In both approaches, teams apply actions generated by their own ideas or ideas from other teams.
These ideas are generated by the combination of the necessary information with industry experts. The information is the result of the combination and transformation of available data.
This is summarized in the diagram above.
Focus on the data value chain
The road from data to a tangible impact on performance can be long. The obstacles mentioned stem from inefficiency in the different steps required for turning data into actions.
Data collection, structuring and processing:
Often addressed separately, these steps can be a major obstacle to smooth operations. Actions for problem solving or means for improvement are often neglected because the steps are too time-consuming or complex. Successful analysis is not possible without reliable and relevant information.
Analysis involves processing information to generate ideas. Expertise can be limited by the volume of information to be processed, but also by the depth of analysis required. What is required for effective analysis? To begin with, we need to be able to visualize information from different angles, including with statistical approaches. Data visualization is key to understanding your production processes better. It also identifies anomalies for the improvement of data quality.
More advanced analytical approaches using Machine Learning (AI); algorithms, and statistics identify areas of improvement without fail. This fosters progress and removes obstacles associated with volume and complexity. As a result, expert analytical skills are optimized. The more the results can be visualized and interpreted, the more effective the analysis. This interpretation will contribute to how experts understand the processes.
Supervising processes is the basis for process control. This can be achieved through control charts, using the 6 Sigma approach, or predictive models that generate instructions for operational staff. Be reactive in the application of new guidelines to optimize performance.
Improving processes can trigger significant changes in operating conditions, changes in facilities through investments, and the like. The ensuing necessary engineering phase requires accurate information to ensure scaling is appropriate.
➔ Automating data collection, structuring, and transformation to provide reliable, timely, and continuous information can have a major impact on the entire chain.
➔ Provide visualization and analysis tools for experts and give them autonomy by ensuring complete and interpretable data.
➔ Provide data sets for data analysts/scientists to develop and validate their models.
➔ Develop tools to manage guidelines and models for timely rollout in the field.
➔ Provide engineering teams with the information necessary for their designs.
A partial approach focusing on analysis only and which overlooks elements upstream the data transformation chain (collection, structuring, processing, etc.) may have limited results. Benefits of a comprehensive approach include significantly more impact and the possibility to effectively reach as many people in the company as possible.
Digitization has two advantages: improved performance from a technical point of view, and a more agile and responsive organization with smoother operations from a managerial point of view.
Author: Mathieu Cura