Flow level pressure
One of the most consequential shifts in petrochemical monitoring over the past few years has not been a breakthrough sensor or analyser but an architectural rethink.
Across mature, brownfield sites, operators are increasingly decoupling monitoring and optimisation from core process control, allowing new analytical capabilities to be layered onto existing plants without disturbing safety-critical systems.
This approach, strongly associated with NAMUR Open Architecture thinking, is reshaping how monitoring technologies are deployed.
Historically, petrochemical plants faced a stark choice. Either monitoring remained limited to what the distributed control system could reasonably handle, or operators undertook expensive and risky DCS upgrades to unlock higher-resolution data and advanced analytics.
Both options carried significant downsides. Control systems are rightly conservative environments, prioritising determinism and stability over experimentation.
As a result, many promising monitoring ideas, like high-frequency condition data or machine-learning-based anomaly detection, never progressed beyond pilot studies.
In sidecar architectures, instead of embedding monitoring logic inside the control system, process signals and instrument diagnostics are exported through a carefully governed interface into parallel monitoring environments.
These environments may include edge analytics platforms, cloud-based condition monitoring systems, or specialised asset health tools. Crucially, the data flow is typically one-way and read-only, ensuring that monitoring enhancements cannot interfere with control logic.
For petrochemical sites with ageing automation, this is more than a technical convenience. It provides a viable route to modernisation that avoids long shutdowns, revalidation exercises, and the organisational disruption that accompanies major control projects.
Monitoring teams can deploy new capabilities incrementally, focusing on high-value assets such as compressors, furnaces, reactors, and critical utilities, rather than attempting plant-wide transformation in one step.
The impact on instrumentation practice is significant. Instruments are no longer specified solely on the basis of accuracy and reliability for control. Diagnostic richness and metadata quality are becoming just as important.
Smart transmitters that expose drift indicators or signal quality metrics suddenly have greater operational value because those diagnostics can be consumed by monitoring platforms designed to look for subtle degradation patterns over time.
Edge computing has accelerated this trend. Rather than streaming raw data to central systems, many sidecar deployments now perform preprocessing close to the source, extracting features or running lightweight models before forwarding results.
This reduces bandwidth demands and allows faster local response while still preserving the separation between monitoring and control. For geographically distributed petrochemical assets, such as tank farms or pipeline networks, this architecture is particularly attractive.
Perhaps the most important consequence, however, is cultural. Side-car architectures legitimise experimentation in monitoring. Because new analytics do not threaten process stability, teams can trial soft sensors, advanced fault detection, or novel visualisation approaches with far less resistance.
Over time, successful experiments can be formalised, while failures remain contained.
PIN 27.2 Apr/May 2026
Product directory
