Conference Presentation, Keynote, Product Demonstration
Keynote by Caspar Herzberg, AVEVA Systems CEO | RAISE Summit 2024 | Paris
- Kasper Hesberg serves as CEO of Aveva and member of the executive committee at Schneider Electric.
- Approximately 90% of Fortune 500 companies utilize Aveva software for SCADA, operations management, design, or time-series data historians.
- Aveva operates across key industrial sectors including energy, power, food, BEV, chemicals, and infrastructure.
- The core objective of industrial AI implementation is to reduce energy consumption and improve process efficiency by 20% to 50%.
- Global industrial data creation has reached one zettabyte, representing an exponential growth over the last 30–50 years.
- Despite the data volume, more than 70% (two-thirds) of industrial time-series data remains unused due to operational silos.
- Only 40% of global companies successfully access and utilize data across their supply chains to improve operational processes.
- Companies that effectively leverage data ecosystems demonstrate higher profitability and growth compared to those that do not.
- Hesberg identifies human reluctance to share data across silos as a primary barrier, driven by a need for job and management protection.
- Industrial AI faces challenges regarding inconsistent data formats and the necessity for human insight to distinguish between AI-generated errors and valid insights.
- Aveva has utilized predictive analytics on time-series data for 12+ years, specifically to predict maintenance failures in industrial assets.
- A case study with Duke Energy indicates that accurately predicting a single turbine failure can save up to $90 million per event.
- Wind turbine maintenance faces critical bottlenecks due to sparse maintenance crews and a generational shift reducing the willingness of workers to climb towers.
- A key strategic goal is to support decision-making for engineers who lack 12+ years of specialized training by integrating asset data into a single interface.
- The demonstrated solution combines predictive analytics (time-series AI) with Large Language Models (LLMs) to create an industrial assistant.
- The industrial assistant features the ability to:
- Summarize information from 50–70 technical manuals per plant.
- Answer "what" and "why" questions regarding output failures.
- Guide operators through a 3D model of the asset to visualize faults.
- Facilitate remote collaboration with other personnel via avatars for parts replacement.
- Aveva plans to launch this integrated AI product in June, positioning it as one of the first practical AI applications in the industrial sector.
- Hesberg notes the solution achieves zero faults in its current application because the data scope is strictly limited to verified industrial telemetry and documentation, mitigating hallucination risks.