New and existing plants use new automation to solve longstanding operational issues across multiple domains. For existing plants we speak of digital transformation. for new plants we speak of “born digital”. In both cases ‘digital’ simply means automation which has industrial software at its core including industrial artificial intelligence (AI).

Jonas Berge, Emerson

The Importance of Automation

Plant automation, which includes industrial AI, is key when building new plants. An advanced automation strategy for a high degree of automation, a plant autonomous to a greater degree, makes that project more viable thus accelerating the project final investment decision (FID) by giving credibility to meeting the plant operational excellence objectives set by government and investors around safety, sustainability, reliability, and production.

Similarly, existing plants with aging automation need to keep up with the new plants coming online so there is a need to raise the level of automation in existing plants by deploying more automation for these same operational excellence reasons.

Many companies operate a global fleet of plants and want many functions centralized such as management of safety, sustainability, reliability, production, as well as technical support for their automation systems. This drives a new level of “enterprise OT” automation. This also includes a new level of automation for the workflow between operational departments within the plant and between plants within the company.

Safety

For safety, health, and environment (SH&E) plants have an aspirational goal like “zero incidents”. The operational risk is incidents. Fortunately, automation improves plant safety by supporting the SH&E team. The main methodology is real-time threat monitoring to become more responsive. These automation solutions make the plant a safer place to work. Results include fewer injuries, fewer incidents, reduced clean-up cost, and reduced fines.

Operator performance

Operators practice on a virtual plant, a digital twin of the plant, in the safety of a classroom where they can make mistakes without repercussions. Console operators practice tasks like startup and shutdown of a unit, feed or grade change, as well as response to process upsets or abnormal situations. This is enabled by a digital twin process model and operator training simulator (OTS) software. Similarly, field operators practice tasks like startup and shutdown of a unit, loading and unloading, batch charging and dispensing, and cleaning. This is made possible by life-like 3D plant model immersive simulator software, digital twin process model, and virtual reality (VR) goggles.

People out of harm’s way

Automatic data collection is fundamental to modern plant automation. Both to enable industrial AI and to keep employees out of harm’s way by reducing work in the field. For the maintenance team this means automated inspection rounds enabled by non-intrusive wireless temperature sensors and many other types of sensors for vibration, ultrasonic noise, and ultrasonic thickness. For field operators this means automated operator rounds. This is made possible by wireless sensors for pressure, process temperature, level, and flow.

Situational awareness

Automation improves personnel situational awareness. For instance, first responders receive automatic notifications with location when someone activates a safety shower. In plants that don’t already have this capability, this is enabled by proximity switches and wireless transmitters.

Automatic notifications are provided on H2S or CO leaks, or oxygen depletion. This is made possible by permanent sensing by wireless gas concentration sensors throughout the plant. Similarly, automatic notification of inadequate hydrant pressure for fire prevention is made possible by wireless pressure gauges.

Sustainability

For sustainability, including energy and emissions, plants have an aspirational goal like “zero emissions”. The operational risk is emissions non-compliance. Fortunately automation improves plant sustainability by supporting sustainability/energy and process teams. The main methodology is real-time performance and emissions monitoring to become more responsive. Results include reduced energy consumption and losses, lower energy cost, reduced fugitive emissions, carbon footprint, venting, and flaring.

Emissions reduction

Automation supports greenhouse gas (GHG) emissions reduction efforts. Automatic notification of thief hatch left open and on emergency relief vent release. Both are enabled by proximity switches with wireless transmitters. Automatic notification on methane leaks such as from seals and flanges. This is made possible by gas concentration sensor with optional wireless adapter.

Energy efficiency

Automation helps improve energy efficiency. Automatically detect control valve underperformance like stick-slip action and hunting. This is made possible using control valve positioners with embedded causal AI valve performance monitoring. Note that smart valve positioners must be integrated with a valve monitoring app.

Automatically detect steam trap failures like blowing steam and trapping condensate. This is enabled by a 2-in-1 wireless non-intrusive acoustic noise and temperature sensor measuring once a minute in conjunction with a causal AI steam trap monitoring app interpreting the data.

Automatic energy flow submetering with finer granularity of plant areas, units, and individual pieces of equipment to balance energy for energy accounting. This is made possible with wireless DP flow meters measuring once a minute or non-intrusive ultrasonic flow meters.

Automatically monitor heat exchanger duty and fouling factor to optimize the time of cleaning. This is enabled by a quad wireless temperature transmitter and wireless DP flow meters measuring once a minute and causal AI heat exchanger monitoring app interpreting data. Similar solutions are used for cooling towers and air-cooled heat exchangers.

Automatically optimize APC setpoints to the crude assay and product slate to maximize energy efficiency. This is enabled by real-time optimization (RTO) software that updates APC setpoints for multiple process units.

Automatically optimize light brightness depending on natural light and occupancy. This is made possible by wireless illuminance and motion sensors with causal AI lighting monitoring app interpreting the data.

Loss control

Automation supports loss control. Automatically detect pressure relief valve (PRV) release to flare and issues like seat passing, simmering, bellows rupture or pinhole venting, releasing at too low or high pressure, stuck open, rupture disc burst or pinhole leak. This is made possible with wireless non-intrusive acoustic noise sensor measuring once a minute in conjunction with causal AI PRV monitoring app interpreting the data. Optional wireless pressure sensor for pilot operated PRV and bellows.

Automatic mass flow submetering with finer granularity of plant areas and process units to balance mass for production accounting. This is enabled by Coriolis mass flow meters and production reconciliation and accounting software.

Reliability

For reliability including maintenance and integrity, plants have an aspirational goal like “zero downtime”. The operational risk is production downtime. Fortunately automation improves plant availability by supporting the reliability/inspection, maintenance, and integrity/corrosion teams. The main methodology is broader condition monitoring to become more predictive. Results include reduced downtime (greater availability), lower maintenance cost, reduced loss of containment, and greater asset utilization.

Integrity

Automation is critical to support plant integrity tackling corrosion and erosion. Automatically predict piping and vessel loss-of-containment, monitor corrosion rate and estimated remaining useful life (RUL) to optimize pipe section replacement and corrosion inhibitor injection. This is enabled by non-intrusive wireless ultrasonic thickness (UT) sensors measuring wall thinning twice a day and causal AI corrosion monitoring app interpreting data. There is also an important safety element to corrosion monitoring.

Predictive Maintenance

Automation is fundamental to predictive maintenance. Automatically predict pump problems like bearing failures, cavitation, strainer plugging, mechanical seal failure, and motor winding insulation breakdown. This is made possible by non-intrusive wireless vibration sensor measuring hourly, and other sensors, used in conjunction with a causal AI pump condition monitoring app interpreting data. Similarly, monitor air-cooled heat exchanger fans and cooling tower gearboxes. Blowers, fans, and compressors are also monitored in similar fashion.

Automatically detect control valve problems like air leaks, regulator failure and filter plugging, as well as packing wear increasing friction. This is made possible by a control valve positioner with embedded in-service causal AI valve condition monitoring. Note that smart valve positioners must be integrated with a valve monitoring app.

Automatically detect tank farm storage tank floating roof tilt as well as water or product pooling on top of the floating roof. This is done using wireless level sensors, wireless level switches, and tank management software.

Automatic data collection helps plants keep up with inspection: automated maintenance inspection rounds. This is made possible by non-intrusive wireless temperature sensors and many other types of sensors for vibration, ultrasonic noise, and ultrasonic thickness.

Production

For production including quality, plants have an aspirational goal like “zero off-spec” product. The operational risk is product recall. Fortunately automation improves plant production by supporting the production and quality teams. The main methodology is balance of process monitoring. Results include reduced off-spec product (higher quality, reduced scrap/waste and rework), reduced quality giveaway, greater throughput, improved yield, greater feedstock flexibility, and greater finished product flexibility.

Agility

Automation makes plants like refineries more flexible in selecting feedstock and final product slate to quickly respond to customer demand and market opportunities. Automatically monitor piping and vessel corrosion rate and estimate remaining useful life (RUL) to optimize refinery crude blend such as incorporating high-TAN opportunity crudes. This is made possible by non-intrusive wireless ultrasonic thickness (UT) transmitter measuring wall thinning twice a day and causal AI corrosion monitoring app interpreting data.

Automatically monitor heat exchanger duty and fouling factor to optimize refinery crude blend such as incorporating heavy opportunity crude with asphaltene. This is enabled by quad wireless temperature transmitters and wireless DP flow meters measuring once a minute and causal AI app interpreting data.

Automatically optimize APC setpoints to a new crude assay or to a new product slate. This is enabled by real-time optimization (RTO) software that updates APC setpoints for multiple process units.

Lean

Automation is critical for plants to be able to continue operation in the face of skilled workforce shortage, to become more autonomous. The vision is no hand-operated valves but instead valve remote control at a click of a button. This is made possible by pneumatic, electric, and hydraulic actuators on a shared communication network.

Automating data collection with balance process monitoring enables people to be reassigned from collecting data to acting on that data. For field operators this means automated operator rounds. This is made possible by wireless sensors for pressure, temperature, level, and flow.

Inferential sensing models are built and updated semi-automatically based on process measurements history and lab results. This is made possible by deep learning AI built on Recurrent Neural Network (RNN) tensor model.

Quality

Automation improves quality by reducing process variability and human factors. Automatically detect control valve underperformance like stick-slip action and hunting. This is made possible by control valve positioners with embedded causal AI valve performance monitoring. Note that smart valve positioners must be integrated with a valve monitoring app.

Console operators practice in the safety of a classroom where they can practice tasks like startup and shutdown of a unit, feed or grade change, as well as response to process upsets or abnormal situations and make mistakes without repercussions. This is enabled by a digital twin process model and operator training simulator (OTS) software.

Automatically detect steam trap failures causing loss of process heat. This is made possible with wireless non-intrusive acoustic noise sensor measuring once a minute with causal AI steam trap monitoring app interpreting the data.

Automatically optimize APC setpoints to the crude assay and product slate to minimize quality giveaway. This is made possible by real-time optimization (RTO) software that updates APC setpoints for multiple process units.

Throughput

Automation improves throughput by reducing process variability, giving operators the confidence to operate with smaller comfort margins closer to operational constraints without exceeding those constraints. Automatically detect control valve underperformance like stick-slip action and hunting. This is made possible by control valve positioners with embedded causal AI valve performance monitoring. Note that smart valve positioners must be integrated with a valve monitoring app.

Automatically find setpoints for advanced process controller (APC) software to meet a new target for throughput, quality, or another parameter. This is made possible by a “co-pilot” with natural language-like user interface that in just a few seconds runs simulation iterations finding a few ranked setpoint options, presenting these for the operator to choose from.

Real-time optimization (RTO) software models are automatically created for process units for which there is no time to create first principles models. This is enabled by machine learning used to create process unit regression models.

Automatically optimize APC setpoints to the crude assay and product slate to maximize throughput. This is made possible by real-time optimization (RTO) software that updates APC setpoints for multiple process units.

System Engineering

AI also helps system engineering. Automatically explain legacy control system control strategies from legacy code like control language and ladder diagram to generate (translate to) modern function blocks, structured text, and sequential function charts (SFC). This is made possible by rule-based AI, machine learning, and GenAI ‘trained’ on prior legacy systems database migrations.

A “co-pilot” automatically retrieves information from system documentation based on natural language questions (prompts) and generates answers. This is enabled by large language model (LLM) with retrieval augmented generation (RAG) from manufacturer curated system manuals and knowledge base articles (KBA).

New Automation

Over and above the specific positive results from additional automation in each domain, labor productivity cuts across all domains. But implementation is key to success.

For instance, these automation examples fall into two main buckets as per the NAMUR Open Architecture (NOA). Some are core process control (CPC) while others are monitoring & optimization (M+O). Architecting the automation system this way gives greater flexibility to add new industry 4.0 or industrial internet of things (IIoT) applications on the M+O side while protecting the production and safety critical CPC side.

Wireless sensors are used extensively as part of this automation expansion. All these sensors use WirelessHART protocol technology such that only a single wireless sensor network infrastructure is required to support all these wireless sensors.

Similarly, applications use a common AI software framework to simplify gradual app deployment.

Lastly, the foundational instrumentation must be good. Flow meters and level gauges must be accurate. Valves and positioners must be high performance and so on. Without a base layer of good field instrumentation, the PID, APC, RTO, and AI above cannot function as intended.

Automation Partnership

Companies need a I&C plant automation partner for project success be it a greenfield construction or brownfield modernization and transformation. Such plant automation including industrial AI which from automation companies that have the hardware, software, and expertise – not from office IT system integrators or management consultants.