operator can see when a heat exchanger no longer trans- fers its typical thermal output under certain flow conditions. In such cases, the operator has three options available to handle the situation: increasing the flow rate to increase the thermal output, reducing the flow rate to achieve the required outlet temperatures of the fluids, or doing nothing. In the first alternative, the heat transfer rate of the heat exchanger can be raised to its former level, but the outlet temperatures of the fluids do not attain their former levels. In the second alternative, the outlet temperatures of the flu - ids can be returned to approximately the starting position at the expense of the fluid flow rates, which may slow down production. In the third alternative, both the outlet tem - peratures of the fluids and the heat transfer rate are driven further away from their original operating points. It is clear, nevertheless, that in such a situation, the heat exchanger’s performance has deteriorated, and the previous operating level is no longer possible. Many parameters vary greatly depending on fluid flow rates. For the parameters to be comparable, they must, therefore, be proportioned to the prevailing flow conditions. Appropriate efficiency monitoring methods are limited by the available process measurements. However, diverse analyses can also be conducted with very few measurements. The efficiency of a heat exchanger can be examined, inter alia, with the help of the following measurements shown in Case 1 and Case 2 .
• Expert systems software programs that use knowledge- based rules to predict fouling rates based on input variables such as fluid properties and operating conditions. Expert sys - tems can incorporate knowledge from experienced operators and engineers to provide accurate predictions of fouling rates in heat exchangers. Overall, a combination of these frameworks can be used to provide accurate and reliable predictions of fouling rates in heat exchangers in refinery and petrochemical operations. Q To what extent do you see the expansion of digitalisa- tion for improving regulatory compliance, equipment reli- ability and operability, and reduced maintenance costs? A Philippe Mège, philippe.mege@axens.net, Marie Duverne, marie.duverne@axens.net, and Pierre-Yves Le Goff, pierre-yves.le-goff@axens.net, Axens Digitalisation of process operations has developed thanks to data transfer technologies, sensors and soft sensors gen- erating data, and IA, especially with machine learning tools, building systems that learn from data. This leverages process expertise, resulting in digital twins designed to optimise asset operation, thus securing the business decision. That ensures reliable operation and allows us to anticipate maintenance, which reduces its cost, and compare actual performances vs forecasted ones. This applies to catalysts and also equipment like heaters, compressors, and heat exchangers. Such tools are available in the Axens Connect’In digital application. Several machine learning tools have already been developed and used by operators for octane prediction, recycle gas purity estima- tion, heat exchanger network optimisation, or generating soft sensors to get streams analysis not directly measured. A Richard Evans, Head of Solutions and Innovation, Delaware UK, richard.evans@delaware.co.uk Digitalisation works on so many levels within the oil and gas sector. For example, it makes it much easier to create a uni - fied experience among employees, which helps to secure the right people with the appropriate qualifications and experience to implement work. This means it is a lot easier to remain compliant with safety regulations, for example. On another front, technology like digital twins has also matured and is now very accurate and reflective of the operating environment. This technology means planners can remain onshore, therefore reducing the risk of accidents while also saving carbon on unnecessary helicopter flights to rigs. More recent developments have been around AI and IoT and the collection and analysis of time-series data. These data first have to be normalised and tagged – does the source represent pressure, temperature or even acoustic sound- waves? But once collected, the data must be used, and this is where intelligence is applied. The data need to be referenced against example datasets to be able to spot anomalous sig- nals. An AI model can be trained to spot irregularities and build a picture of what is really happening. Whether there is excess wear in a pipe or gas is not being sufficiently heated, the AI will be able to accurately discover this and suggest preventative maintenance. It is a very granular approach that
Case 1: Temperature tags available
Sr. No.
Sensors
1 2 3 4 5
Feed temperature in Feed temperature out
Cooling medium temperature in Cooling medium temperature out
Feed flow rate
Case 2: Pressure tags available
Sr. No.
Sensors
1 2 3
Feed pressure in Feed pressure out
feed flow rate (good to have)
Some heat exchanger fouling prediction frameworks that can benefit refinery and petrochemical operations include: • Empirical correlation equations derived from experimen- tal data that can be used to estimate fouling rates based on operating conditions. These correlations are often based on heat transfer coefficients and can be used to estimate foul - ing rates in a variety of heat exchanger types • Fouling indices metrics that can be used to quantify the tendency of fluids to foul heat exchangers. These indices are based on fluid properties such as viscosity, density, and thermal conductivity, and can be used to estimate fouling rates based on fluid composition and operating conditions • Artificial neural network (ANN) machine learning algo - rithms that can be used to predict fouling rates in heat exchangers. ANNs can learn from historical data to identify patterns and predict future fouling rates based on input vari- ables, such as temperature, flow rate, and fluid properties
10
PTQ Q3 2023
www.digitalrefining.com
Powered by FlippingBook