Industry
Client Overview
"Amcor," a Fortune 500 manufacturing giant with over 200 facilities globally, was plagued by costly, unpredictable failures in its CNC machining centers. A single unexpected machine failure could halt a production line for hours, causing a ripple effect of delays and financial losses. Their existing maintenance schedule was based on fixed time intervals, which was inefficient-either servicing machines that didn't need it or failing to catch imminent breakdowns.
Client Testimonial
"CIS brought a level of rigor and expertise to this project that was truly impressive. They navigated our complex OT and IT environments, integrated data from a dozen different sensor types, and delivered a solution that our plant managers actually trust and use. The 22% reduction in downtime is a number that gets attention at the board level. This wasn't just an IT project; it was a fundamental improvement to our global operations." - John Miller, Global Head of Operations, Amcor
Problem
Amcor needed to shift from a reactive/preventive maintenance culture to a predictive one. They were data-rich, with years of sensor logs, but lacked the capability to turn that data into actionable predictions of machine failure.
Key Challenges
-
01
Data Silos & Variety : Sensor data (vibration, temperature, acoustics) was stored in disparate, on-premise historians with inconsistent formats.
-
02
Legacy Systems : The solution needed to integrate with a decades-old SAP ERP system for work order generation.
-
03
Scale & Complexity : The model had to be accurate across thousands of machines from different manufacturers and ages.
-
04
User Adoption : Plant floor engineers were skeptical of "black box" AI and needed a solution that was transparent and trustworthy.
Our Solution
CIS assembled a specialized Embedded-Systems / IoT Edge Pod combined with our Data Engineering experts to build an end-to-end predictive maintenance platform.
Centralized Data Lake : We architected and built a cloud-based data lake on AWS to consolidate all sensor and maintenance log data from Amcor's global plants.
IoT Edge Integration : We developed lightweight software agents deployed on edge devices at each plant to pre-process data locally, reducing latency and data transmission costs.
AI Model for Anomaly Detection : Our team developed a custom LSTM (Long Short-Term Memory) neural network, ideal for time-series data, to detect subtle anomalies in machine operation that were precursors to failure.
Predictive Work Order Generation : When the model predicts a high probability of failure within a specific timeframe (e.g., 7-10 days), it automatically generates a high-priority maintenance work order in Amcor's SAP system, including the likely point of failure and required parts.
Implementation & Execution
-
Phase 1 (Month 1)
On-site visits to two pilot plants to understand the physical environment and data infrastructure.
-
Phase 2 (Months 2-4)
Built the cloud data lake and the ETL pipelines. Successfully ingested and cleaned 5 years of historical data.
-
Phase 3 (Months 5-7)
Developed and trained the core LSTM model. Back-tested it against historical failures to prove its accuracy.
-
Phase 4 (Months 8-9)
Built the SAP integration and the user interface for plant managers, which visualized the machine health scores and alerts.
-
Phase 5 (Months 10-12)
Deployed the pilot solution in the two initial plants, running it in parallel with the old system to build trust and gather feedback.
-
Phase 6 (Ongoing)
Began a phased global rollout to 20 plants per quarter, with a dedicated support team managing the deployment and providing training.
Positive Outcome
1. 22% Reduction in Unplanned Downtime
Across the pilot plants, the system successfully predicted over 80% of major failures at least one week in advance.
2. 15% Reduction in Maintenance Costs
By avoiding unnecessary scheduled maintenance and reducing overtime for emergency repairs, costs were significantly lowered.
3. Improved Spare Parts Inventory
Predictive ordering of parts reduced both excess inventory and stock-outs of critical components.
4. Creation of a Strategic Data Asset
The centralized data lake is now being used for other optimization projects, such as energy consumption analysis and production quality improvement.
Why Choose Us
-
Verifiable Process Maturity
Managing a global rollout of this complexity is impossible without a CMMI L5 process.
-
Zero-Contractor Model
A stable, knowledgeable team was critical for the long-term, multi-phase nature of the project.
-
Enterprise-Grade Security
We handled sensitive operational data with the utmost care, adhering to our SOC 2 and ISO 27001 controls.
-
Business-Outcome Focus
The project's success was measured in "downtime reduction," a core business metric.
-
Flexible Engagement Pods
The specialized IoT and Data Engineering PODs brought the exact niche skills required.
-
Radical Transparency
Amcor's operations team had a dedicated dashboard showing model performance and project milestones.
-
De-Risk Your Investment
The pilot-first approach proved the solution's value and built internal buy-in before a full global investment.
-
Full IP & Data Ownership
Amcor owns the entire platform, from the edge agents to the AI model.
-
Future-Proof Technology
The solution was built on a scalable cloud architecture, ready to incorporate new plants and data sources.
Conclusion
This project demonstrates CIS's ability to bridge the gap between complex industrial machinery and cutting-edge AI, delivering a solution that creates massive operational and financial value for a global enterprise.