PTQ Q4 2023 Issue

Consolidation layer

Inputs

Prediction models

Outputs

Simulation tool

Steam prediction models Fuel gas prediction models

Plant capacities

Real data from eld

An excel tool that is kind of a simulator Model inputs/Plant capacities can be changed by user Hourly result is also written to a PHD tag to track easily

Other parameters

Conversions with Energy Balance Equations

Plant capacities Other parameters

User inputs

Production plan report

Big data from sensors and user inputs are consolidated

AI based prediction models built to simulate the process using historical data

Steam and fuel gas unit conversions are done by energy balance equations

An excel le that contains the result is sent via e-mail to users as hourly basis

Figure 2 Flow of analytical study

Methodological approach Selection criteria for modelling production and consumption quantities

is produced in this plant by considering the complex’s steam demand. Steam is generated by firing fuel gas, a mixed gas consisting of side product fuel gas and natural gas. The steam demand of the complex is modelled as a func - tion of plant capacity and ambient temperature. As a result of detailed modelling explained in the modelling and simu - lation phase, overall steam demand is computed. In addi - tion to steam demand, plants’ fuel gas production capacity is analysed. Hourly fuel gas production and steam demand are estimated. The combustion of 1.0 ton/h natural gas is equivalent to producing an assumed amount of high-pressure steam, considering the efficiency of boilers. The overall natural gas demand of boilers can be calculated, and internally produced fuel gas routed to the boilers is subtracted from the total energy demand used to produce steam. While doing energy balance calculations, the enthalpy of steam and internally produced fuel gas are considered constant values. Aromatics plant The operation of the process furnaces within the aromatics plant is facilitated by employing a combination of internally generated fuel gas and natural gas. The formulation of the produced fuel gas depends on factors such as plant capacity and ambient temperature. In the initial phase, an analytical model is constructed to characterise the fuel gas consump - tion specific to the aromatics plant. Subsequently, an exam - ination is conducted to ascertain the potential correlation between fuel gas and natural gas demand. As anticipated, a discernible positive correlation emerges between natural gas consumption and fuel gas utilisation. Consequently, the amalgamation of fuel gas and natural gas is orchestrated and subsequently used as the composite fuel source for the process furnaces.

Production and consumption quantities were meticulously hand-picked in the quest for a comprehensive modelling approach. This selection specifically targeted quantities that either exhibited a pronounced consumption within the fuel gas or steam ring or showed noteworthy variability, including: • Determinants of choice: The potential quantity’s contri - bution within the ring was scrutinised. Only those that occu - pied a significant position were earmarked for modelling • Interrelation with other units: A thorough analysis was undertaken to examine the interplay of the considered quantity with other operational units • Consumption variability: If a particular quantity held importance within the ring and demonstrated substantial variability in its consumption, it was flagged for further modelling endeavours. Notable consumption patterns in the fuel gas ring The substantial consumption by the steam generation and Plant-2 factories makes them the primary focus for mod - elling efforts, given their significant impact on the overall distribution. The ethylene plant produces fuel gas and distributes it to all plants under Petkim. Any remaining fuel gas is chan - nelled to the boiler. Thus, modelling the fuel gas production within the ethylene plant is also imperative. Consumption patterns in the steam ring and gas turbine In the steam ring, four main types of steam are identified: XHS (extra high-pressure steam), HS (high-pressure steam), MS (medium-pressure Steam), and LS (low-pressure steam). While formulating models for the steam ring, the criteria outlined in the ‘Energy Balance’ section were employed. Moreover, modelling efforts were made for rare scenarios like steam transfers between factories and the amount of steam drawn by the equipment. As a result of analyses, more than 20 machine learning models were established to predict the relevant steam consumption/production values. Ultimately, models are consolidated with different

PTQ Q4 2023