1
Background and Problem Statement
The chilled water plants are traditionally regulated and monitored by a predetermined control strategy where energy efficiency is a major concern. In addition, emergency maintenance calls in general and summer season, in particular, have been a major concern in the industry. A chilled water plant owner located in Houston leveraged Ranial Systems’ platform to address these issues, gain energy efficiency and minimized ongoing maintenance overheads.
2
Approach
During the brainstorming sessions with the client, Ranial Systems identified the pain points of the traditional chiller plant control strategy. It was noted that while the operational model aims to fulfill the required cooling load demand, the predefined setpoint temperature defined within the PLC/SCADA system results in considerable energy wastage within the operating environment. Chillers are sequentially switched on one after another whenever the building cooling load demand increases, and often the part load of the operating chillers rises above a pre-set value.
Ranial has introduced its Cognitive edge computing and AI based condition monitoring platform to enable predictive operational control strategy. The aim was to improve the performance of chiller plants using real-time predictive operational intelligence and autonomous controls without any additional high-priced equipment procurement. The target-state solution should minimize the manual supervision and improve energy efficiency.
3
Solution
Ranial Systems collected historical data from the existing Historian and fed those inputs to the cognitive algorithm for the purpose of training the AI agent. It correlated the outdoor temperature, cooling demand, chilled water supply & return temperatures, chilled water flow rate, and energy consumption data speeded over the Summer, Spring & Fall usage patterns.
Upon testing the model performance, the models were deployed on the edge (site controller). The security requirements, specifications of the edge devices, and maintenance strategy were mutually discussed and agreed upon.
The offered CognitIoT™ solution supports multi-protocol integration (Modbus, BACnet, and OPC to communicate with the industrial assets and offers extensible communication/intelligence infrastructure through highly scalable AI-powered edge computing runtime. The interoperability and self-healing of endpoint IoT systems ensure resilient and low-latency interactions with PLC/SCADA systems. The cognitive edge runtime intercepts the sensory feeds to gain situational awareness to facilitate requisite control operations. The flexibility in integration with legacy control system infrastructure allowed Ranial systems to leverage such intelligence at the point of action with minimal hardware and software upgrades in the existing control systems. CognitIoT™ could integrate with multiple PLC/SCADA systems in a protocol-agnostic way and, offer real-time monitoring and site management functions through a centralized web-based console, while the edge controller performs the autonomous operations at the plant.
4
Implementation
The following workflow was designed to implement the predictive control strategy for the existing chilled water plant.
An edge controller preloaded with the AI agent was installed in the control room linked to the PLC/SCADA system over Modbus. Platform middleware and infrastructure tuning, AI/ ML modules, and monitoring dashboard customization were performed in line with the client’s requirement.
5
Result
In an implementation cycle of two months, Ranial systems rolled out the cost-effective solution by integrating existing PLC/SCADA systems, incrementally trained the model, and testing all possible scenarios prior to go-live.
The implemented system could improve up to ~8% energy efficiency of the chilled water plant. The edge native solution with cloud-based monitoring, Ranial Systems could achieve 30% saving on recurring maintenance costs.