In the present digital era, artificial intelligence (AI) backed decision-process can transform the way asset-failures could be managed. A Proof-of-concept of scalable AI algorithm has been developed and field-verified, to demonstrate potential to implement "Just-in-time" (JIT) maintenance. The model has ability to absorb configuration variations; it is run on periodic basis, to reassess the findings, and identify changes in the operating behavior of the asset. It is a robust tool, for the field engineers. The objective of this research paper is to establish fundamentally different thought process for maintenance decision-makers for dynamic-diagnosis of faults using normal operating data. Each category of equipment has unique behavior, and hence must have customized solution. When linked through IIoT, the automated business decision for maintenance cost reduction can be applied on mass scale. The experimental results & field validations show that AI based diagnostics methodology outperforms traditional maintenance cost-management practices. It is a system which is self-learning, and therefore, as the model matures, dynamic strategy will evolve; and maintenance cost saving can extend beyond targeted 4% of net operating cost. The concept can be equally applied for any data-intensive-complex machineries and static-assets (Heat exchanger, Pipeline, Rotating equipment, etc.) across process industries.
Paper
Full text
Artificial Intelligence Application for Just in Time Maintenance
Semantic Scholar · Engineering · 2020
Abstract
In the present digital era, artificial intelligence (AI) backed decision-process can transform the way asset-failures could be managed. A Proof-of-concept of scalable AI algorithm has been developed and field-verified, to demonstrate potential to implement "Just-in-time" (JIT) maintenance. The model has ability to absorb configuration variations; it is run on periodic basis, to reassess the findings, and identify changes in the operating behavior of the asset. It is a robust tool, for the field engineers. The objective of this research paper is to establish fundamentally different thought process for maintenance decision-makers for dynamic-diagnosis of faults using normal operating data. Each category of equipment has unique behavior, and hence must have customized solution. When linked through IIoT, the automated business decision for maintenance cost reduction can be applied on mass scale. The experimental results & field validations show that AI based diagnostics methodology outperforms traditional maintenance cost-management practices. It is a system which is self-learning, and therefore, as the model matures, dynamic strategy will evolve; and maintenance cost saving can extend beyond targeted 4% of net operating cost. The concept can be equally applied for any data-intensive-complex machineries and static-assets (Heat exchanger, Pipeline, Rotating equipment, etc.) across process industries.