Predictive Maintenance: IAC Partners and L2F, accompany you to predict (rather than heal)

In the era of big data, the Internet of Things and real-time data analysis, maintenance becomes predictive and naturally fits in as one of the building blocks of industry 4.0.

By predicting failures, using algorithms and “Machine Learning”, this maintenance 4.0 makes it possible to optimize operational costs and schedules, increase the availability of the production tool and ultimately increase margins

What are the challenges of the future maintenance?

Jean-Baptiste Guillaume: First curative (an element is repaired as soon as a problem occurs), then preventive (following an established process, actions are planned), the future of maintenance will be predictive: instead of waiting for breakdowns or performing costly actions without a problem being noticed, the data will be used to predict the actions to be performed. This revolution will require many changes within organizations and in the short term, can only be implemented by coupling it with preventive maintenance.

What are the benefits of predictive maintenance?

JBG: Predictive maintenance allows the deployment of a maintenance strategy at the right time, limiting the movements of specialized operators but also the need for spare parts, thanks to the anticipation of breakdowns, made possible by the installation of sensors on the products/machines and the analysis of data in real time. Thus, predictive maintenance will eventually make it possible to eliminate production stoppages and/or unplanned operations.

The analysis of the product’s state of health in real time and the guarantee of a minimum level of operation/service also makes it possible to develop new business models and thus new sources of income, through the development of a service offer associated with the product.

What maintenance services or solutions of the future do you offer?

JBG: IAC Partners (strategy consulting firm) and L2F (Data Science specialist, winner of the Google Kaggle Cup), have developed a global methodology that allows manufacturers to implement predictive maintenance in their organization.

This methodology covers all fields of maintenance and is based on 5 pillars:

-Validation of the interest of predictive maintenance and its economic stakes for the company: what can be saved and for how long?

-Analysis of the organization, its maturity for the implementation of a predictive maintenance solution, the necessary resources

-The collection and analysis of the available data, failing which the identification of the necessary data and the means to capture it

-The construction and operation of a high-performance algorithm

-The implementation and monitoring of processes by people and the appropriate tools to validate concrete gains

• Predictive maintenance has many advantages but requires a certain number of evolutions within an organization

What are the obstacles and weaknesses in using the maintenance solutions of the future?

JBG: Predictive maintenance has many advantages but requires a certain number of evolutions within an organization before it can fully benefit from it. To do this, three main changes are needed:

• The integration of predictive maintenance from the product design phase allows the right data to be collected and software developments to be optimized by integrating the appropriate sensors.
• As the definition and marketing of new business models oriented towards “maintenance-in-as-a-service”, the manufacturer will then play the crucial role of a “turnkey” solution provider.
• The implementation of a data analysis team allows the continuous development and improvement of prediction algorithms

What are the criteria for acquiring the maintenance solutions of the future?

JBG: To deploy predictive maintenance solutions, it is necessary to first establish a representative sample of the fleet in service. The first step is therefore to target a deployment scope by quantifying the associated economic gains. Once validated, the next step is to select the type of data and associated sensors within a digital ecosystem that is often new to the organization.

Can you give an example of an application in the industry? What was the problem? What were the results?

JBG: Braskem, one of South America’s leading manufacturers of thermoplastic resins, has deployed a predictive maintenance system throughout its production site. Data from all machines on the site are centralized and analyzed on a common platform. Vibration analysis on a pump system made it possible to anticipate a failure that would have required 5 days of repair and therefore a loss of revenue of approximately $6 million.

Pouvez-vous indiquer un exemple d’application dans l’industrie ? Quelle était la problématique ? Quels étaient les résultats ?

JBG : Braskem, un des leaders sud-américains dans la fabrication de résine thermoplastique a déployé sur l’ensemble de son site de production un système de maintenance prédictive. Les données provenant de l’ensemble des machines du site sont centralisées et analysées sur une plateforme commune. L’analyse des vibrations sur un système de pompe a notamment permis d’anticiper une panne qui aurait nécessité 5 jours de réparation et donc des pertes de revenus d’environ 6 millions de dollars.