Print this article

Online big data chemical batch analytics

corresponding

MARTIN HOLLENDER1, MONCEF CHIOUA1, CHAOJUN XU2
1. ABB Corporate Research, Ladenburg, 
2. ABB Oil, Gas and Chemicals, Mannheim, Germany

Abstract

By allowing the manufacturing of multiple product types, batch processes represent a suitable configuration for changing markets that require agile operations. Still, batch processes are complex, dynamic, nonlinear processes, difficult to control and to monitor. The present article describes the development of an online operator assistant system and its validation in a batch chemical plant. The developed system, through the early detection of evolving process abnormal situations, and the isolation of relevant process variables, allows the process operator to timely implement corrective actions reducing therefore the amount of wasted material, energy and production time.


INTRODUCTION

Big data processing architectures like Hadoop or Spark enable new possibilities to analyse historical data generated by process plants (1)–(3). Predictive maintenance, operation support systems, soft sensing for monitoring and control, and integration of control and ERP layers are some of the envisioned applications of such technologies in the process industry sector.


The present article describes the development of new operator support systems able to detect and troubleshoot abnormalities of batches in real time. The developed online system helps operators to run plants in a smooth and trouble-free way as emerging process issues can be detected at an early stage and corrective actions taken while the batch is running.


The adopted approach uses historical data to learn the expected behaviour of the batch process under nominal conditions and builds a statistical “golden batch” model then used as a reference for the currently produced batch. Deviations from this “golden batch” model generate a warning to the operator.


APPLICATION SCENARIO BATCH ANALYSIS

More than 300 batch ...