Actual operation
Online model
Bene ts
Stages view
Maximum entrainment
Liquid
Stages
Vapour
80% Jet ood
1
Condensor
Operating point
Overhead
CS-1
Feed
15
Maximum weir load
Maximum weir load
0% Weep
Reboiler
Visually depicts how close your column to its operating limits
CS-2
50
Bottom
CS-3
HA column
60
Liquid mass ow (lb/hr)
Figure 2 Distillation column process digital twin
process models and the data analytics model for better visualisation and faster decision-making. Distillation column digital twin Producing the right products in the right quantities with minimum energy consumption from available feedstocks is a challenge for efficiency management at the site level. The distillation column is the most widely used unit operation in downstream facilities. Here, a mixture of hydrocarbons is separated into multiple streams based on their boiling points by utilising energy. Advanced process control (APC) may have been configured to minimise energy consumption. By using inferential models to predict the quality of the product streams, APC will tend to reduce the reflux ratio and reboiling duty within the prescribed limits. While this approach may bring conditions close to the optimum, there can still be gaps to address. This is where real-time process models come into play, helping to achieve the ideal reflux ratio (see Figure 2 ). The process model will also provide a stream summary of all the process streams, which could be used in
conditions at both the equipment level and the overall plant level. Data analytics evaluation The application of data analytics in process industries has improved significantly, thanks to the development of new-generation information technologies, such as the Internet of Things (IoT), big data, cloud-based computing, ML, and AI. It has evolved through several stages: from descriptive analytics, which focus on what happened, to diagnostic analytics, which focus on why it happened, then to predictive analytics, which focus on what will happen, and finally to prescriptive analytics, which focus on what should happen (see Figure 3 ).. In prescriptive analytics, decision aid is the open loop, whereas decision automation results in a closed loop. The goal is to achieve ‘automation wherever possible’ without the need for human intervention to enable timely actions. Data science/data analytics model development for process industries Data science is a field of technology that combines domain expertise, programming
downstream equipment and process units. For instance, the stream summary of compressor feed gas could be used to calculate the key performance indicators (KPIs) of the compressor. This kind of process model will enable us to reach and sustain the optimum
Decision automation Decision aid
Prescriptive analytics (Oversight) What should happen?
Predictive analytics (Foresight) What will / might / can happen? Diagnostic analytics (Insight) Why is it happening / why did it happen? Descriptive analytics (Hindsight) What is happening / happened? Data analysis
Decision Action
Manpower input
Figure 3 Data analytics evaluation in process industries ( Gao, 2022)
Refining India
61
Powered by FlippingBook