Automation Intensity – How to Elevate Your Plant’s Level

• Plants have amazing control systems, but you still see lots of manual activity in the field
• By increasing the plant automation intensity fewer manual tasks are required
• The recommendation is to deploy a second layer of automation, for a higher degree of automation
• The result of more automation is greater productivity and a new level of operational excellence

Plants have sophisticated process control systems. So why are there so many persons still working in the field? Well, when you look at all duties around the plant, beyond the production of product, the degree of automation is not as high as we’d like to think it is. The automation intensity is relatively low. This impacts the plant’s ability to achieve higher operational excellence targets. To solve this we must look to the expanded role of automation going beyond the P&ID with a second layer of automation. Sure, there is no ‘absolute zero’ for manual tasks but many manual tasks can be automated, with particular focus on the tasks in the field, so work now transforms from the field to indoors. Remote sites operate unmanned for long periods of time, but all plants can benefit. And sure, some tasks like repairs still cannot be automated today. So, what is the recommendation for the new automation that makes this possible and what is the role of industrial AI? Here are my personal thoughts:

Automation Intensity

We could think of an automation intensity metric as a percentage of tasks which are automated as opposed to done manually. Hard to calculate – what data would you base it on? But 0% would mean everything is done manually. 100% means it’s completely automatic, autonomous. We want to get closer to 100%, fully automatic. Indeed even when a function is supervised remotely by a human (manually) from a center of operations, such that the site itself is unmanned, that is considered an autonomous site. Or you could think of automation intensity in terms of a specific kind of task like data collection; what percentage of data points are collected automatically vs manually. This is easier to calculate because data is easily obtained. You may think most of your data is collected automatically by sensors on your control system like the DCS or PLC but think again. Think of all the sight level glasses, variable area flow meters, and pressure gauges inspected manually. Think of all the corrosion test points, erosion test points, vibration test points, temperature test points, steam trap, and relief valves checked manually with various portable testers. Think of all the manual valves and other things visually inspected. And manual grab sampling for lab testing. These are thousands of missing measurements. So when you think about it, the automation intensity might be less than 50%?

Automation intensity could be used as a metric to explain to management why more automation is required in plants. And don’t think of data collected at these manual positions as only once a year, once month, a week, or per day as it is done now – instead think of how often you should be checking them to become more situationally aware, responsive, predictive, and productive and you realize data collection must be automated. The same is true for data interpretation. And maybe even valve actuation. The recommendation is to start a program to increase the automation intensity. And I’m not shaming anybody for not already having done all this because until a few years ago practical sensors to automate the data collection were not available. Practical industrial AI apps to automate the data interpretation were not available. And open protocol valve remote control was not yet a thing.

Low Automation Intensity

The reason why plants today have a lower automation intensity than they might think, is because although their process control system is very sophisticated, it is exactly that; a system dedicated to control the process. It does core process control (CPC), and the associated process safety system helps keeping the process safe. But there are many other functions with tasks which is not done by the control system, or any other system, and which therefore are very manual. This includes functions like occupational safety, health, and environment management, sustainability and energy management, reliability, maintenance, and integrity management, and quality management. So even though the core process control is highly automated – has a high degree of automation – which is usually the case in a continuous process plant, the other tasks have many manual elements, such as data collection, data interpretation, and setting valves – zero automation. Thus, overall the plant has a low degree of automation – the automation intensity for the plant is low. Sure, there is a machinery protection system which may include a vibration monitoring system to help reliability management, but only deployed on large turbomachinery. Other machines like pumps, fans, blowers, gearboxes, and smaller compressors rely on manual data collection – and there are many of those.

Because it is hard to keep up with all this manual inspection and data interpretation, plants have challenges like incidents and slow first response, overconsumption, flaring, emissions, equipment failure, process downtime, loss of containment, off-spec product, and production interruption. Many are avoidable.

Higher Automation Intensity

Traditionally automation as found on the P&ID is for process control and safety. The vision now is to automate ‘everything’ meaning to increase the level of automation in all the operational departments, in the functions they perform, reducing their manual workload with the added benefit of real-time information. This means providing workers in many domains with information they did not have before, in many domains that had little or no automation in the past. So this is a new expanded role of automation. This higher degree of automation includes data collection and interpretation for the domains mentioned above: safety, sustainability, reliability, and production. This is automation beyond the P&ID as we know it. Adding this much needed automation means higher automation intensity. For office automation Gartner has a concept they call ‘hyperautomation’ – so ?for plant automation you could call this higher automation intensity as industrial hyper-automation if you will.

Increase The Plant’s Automation Intensity

There are two parts to the new automation:

The core process control (CPC): in the plant is usually mostly already taken care of with existing sensors, DCS, and automated valves. The gray pyramid in the illustration below. But there still are routine field operator rounds and ad-hoc callouts. The recommendation is to add new automation to tackle this. It’s taking the automation from good to great.

Monitoring & Optimization (M+O): refers to the auxiliary safety, sustainability, reliability, and production functions. The pink sliver in the illustration below. This is very manual. The recommendation is to add new automation to tackle this.

That is, automate as much as you can; on both the CPC and M+O sides of the NAMUR Open Architecture (NOA), these two security zones are independent yet integrated. Start with easy to automate tasks such as data collection and data interpretation. There are tasks that cannot be automated today, that rely on people, such as repairs, replenish coolant, oil and filter change, and lubrication. Although robots now exist, they are incapable of such tasks.

Core Process Control (CPC)

Routine field operator rounds to read mechanical instrumentation can be automated with wireless sensors. The recommendation is to deploy a plant-wide WirelessHART sensor network and use wireless sensors. WirelessHART sensors provide an update once per minute which is 1440 times faster and therefore more predictive and more productive than daily manual checks. At one update per minute most WirelessHART sensors will have a 10-year battery life or longer. Other wireless solutions like 3G/4G/LTE/5G cannot provide a comparable result. WirelessHART also provides automatic conversion to OPC-UA and Modbus/TCP without coding making integration to the plant historian or control system easy. That is, replace mechanical instruments with wireless sensors. This reduces the workload on field operators and improves situational awareness for console operators.

Grab sampling for lab testing of density and viscosity can be automated. The recommendation is to deploy vibrating fork sensors to measure density and viscosity continuously. These meters are wired, but with a WirelessHART adapter the signal can be transmitted wirelessly, but wired power is still required. As a result of real-time sensors plants can reduce off-spec product.

Actions which cannot be made directly on the operator console are associated with high stress and workload and poor health. High levels of stress lead to errors. For example, hand operated valves using handwheel operator. This was acceptable in the past but not anymore. The recommendation is to upgrade handwheel operated valves by adding remotely controlled (centrally) actuators. Actuators can be pneumatic or electric. The actuators can be networked to reduce the amount of wiring required for the control and feedback signals. The recommendation is to use a standard bus protocol, there is no need to get locked in with a proprietary protocol. As a result of operation from the operator console instead of the field, workload and stress are reduced.

Monitoring & Optimization (M+O)

The key part of the new automation is the second layer of automation, the pink sliver in NOA. Routine inspection rounds for weekly sustainability and energy management data collection, monthly reliability data collection, and yearly integrity (corrosion and erosion) data collection with portable testers and reading mechanical gauges and meter registers can be automated with wireless sensors, often referred as M+O sensors or IIoT sensors. Some of this data collection is done by plant staff, but a lot of it may be done by various paid external contractors. The recommendation is to deploy a plant-wide WirelessHART sensor network and use wireless sensors for real-time data collection. WirelessHART sensors provide an update once per minute which is more than 10,000 times faster than once a week, or once per hour which is 720 times faster than once a month, or twice a day which is 730 times faster than once a year. This is more predictive and more productive than manual checks. Just imagine the vast number of positions manually inspected in your plant. Most WirelessHART sensors will have a 10-year battery life. Other wireless solutions like mobile (cellular) cannot provide a comparable result, and not all the required types of sensors are available. WirelessHART also provides automatic conversion to OPC-UA and Modbus/TCP without coding making integration to various systems for asset performance monitoring (APM) easy. That is, replace portable testers and clipboards with wireless sensors. This reduces the workload on inspection teams.

Automating manual data collection is valuable in any plant, but even more so in remote sites such as offshore installations or oil & gas fields in rural or uninhabited areas. Companies have a vision of autonomous sites meaning they are normally not manned, meaning they are operated from a central location and there are no scheduled visits for months on end.

Manual data interpretation to determine pipe section corrosion or erosion rate and remaining useful life, predicting equipment failure, determining steam trap failure and estimating losses, relief valve release volume or passing, equipment overheating, rate of fouling and optimum time of cleaning equipment with heat transfer surfaces, seal failures, energy and utilities overconsumption can be automated with industrial AI. Some of this data interpretation is done by plant staff, but a lot of it may be done by various paid external contractors. The recommendation is to deploy industrial AI apps for corrosion and erosion monitoring, equipment condition monitoring, control valve, steam trap, relief valve, and equipment performance monitoring, and energy management etc. These drive maintenance activities including cleaning and replacement. The other recommendation is for these apps be based on causal AI using first principles models for performance monitoring and cause & effect agents for condition monitoring. Causal AI is easy to use because it doesn’t require years of historical data and meticulous maintenance logs, no data cleansing, no data science, and no algorithm training, or periodic retraining etc. Causal AI is very robust because it is based on established knowledge, not statistical correlations so avoids associated false alarms and misses. These apps are part of various asset performance monitoring (APM) systems used by the various departments in the plant to do their job better. AI is nothing without data, so the recommendation is to use industrial AI apps that support OPC-UA to tie in with plant systems and wireless sensor network gateways. As a result the plant runs more sustainably, reliability, and producing on-spec.

Action Plan: Transformation by Automation

Use automation intensity as a metric to justify to management why more automation is required in your plant. What is the automation intensity in your plant? Review with the operational teams how much data is collected and interpreted manually. The below list may not be a perfect fit for the organization of your plant, but it gives a general idea of how to go about it:

• Number of data points collected manually by field operators
• Number of data points collected manually by the inspection team
• Number of data points collected manually by external contractors
• Number of data points which are reviewed manually by production and quality teams
• Number of assets for which data is reviewed manually by the reliability, sustainability, and integrity teams
• Number of assets for which data is reviewed manually by external contractors

To transform this manual work, additional automation is required in the plant. To enable this transformation the I&C department must be given a larger share of the technology budget to pay for the required automation such as the wireless sensors and IA apps. The I&C engineers know what sensors and apps to use for each type of equipment. They are experts in this sort of thing. The recommendations are found in the paragraphs above.

And remember, always ask vendor for product data sheet to make sure the software is proven and pay close attention to software screen captures in it to see if it does what is promised without expensive customization.

Well, that’s my personal opinion. If you are interested in digital transformation in the process industries click “Follow” by my photo to not miss future updates. Click “Like” if you found this useful to you and to make sure you keep receiving updates in your feed and “Share” it with others if you think it would be useful to them. Save the link in case you need to refer in the future.