Apr-2025

Hybrid digital twin for DHDT unit performance monitoring

High-fidelity models with collation of ‘theory and plant data’ are essential to track hydrotreating unit performance.

SK Shabina, Ranjith Kumar Bojja, Indranil Roy Choudhury and Sarvesh Kumar
Research & Development Centre, Indian Oil Corporation limited

Article Summary

In today’s pursuit of net-zero emissions, refineries aim to increase efficiency and flexibility for processing a crude mix along with optimum hydrogen (H2) consumption. Diesel hydrotreatment (DHDT) units are one such fundamental treating process in the refinery to produce marketable fuel from heavier and refractory feeds, such as straight-run gasoil (SRGO), straight-run vacuum diesel (SRVD), light cycle oil (LCO), and light coker gasoil (LCGO).

Recent developments for wider options in coprocessing renewable feedstock to produce lower-carbon-intensive diesel while extending catalyst life present challenges for refineries to operate units optimally, meeting diesel product specifications. Digital tools and their solutions play a key role in optimisation. They proactively take actions such as operating units at optimum temperatures, extending catalyst life by reducing throughput, scaling back refractory feed streams, and scheduling procurement for the next catalyst charge. Such diligent solutions from digital tools are  made possible by integrated models built on both:
• Fundamental kinetics.
• ‘High-quality process data’ of inferring units’ operation.

Conventional refinery process kinetic models based on fundamental kinetics use mathematical equations to describe the process, hence relying less on data. This type of modelling approach is successful when there is a deep process understanding and feed/product stream characterisation. They are developed using programming platforms such as Fortran, C++, and MATLAB.

However, these models cannot adequately represent the complexity involved in deactivation mechanisms and non-ideal phenomena. Hence it becomes difficult to use them for monitoring catalyst health throughout its lifecycle. On the other hand, only data-driven models based on ML algorithms use a huge reams of datasets to learn and identify the patterns and associations between process operations, stream properties, and yield variables, enabling them to make predictions or decisions. They are developed using programming platform such as MATLAB and Python.

These models become helpful for describing complex phenomena where explicit mathematical equation formulations become difficult. Also, these models adapt to new scenarios whenever retrained with updated data. However, commercial process data availability with sufficient variability limits this approach. Also, lab-scale data generation that includes variability requires cost, time, and resources. Thus, hybrid models – by combining fundamental principles and process data – power the strength of both kinetic and data-driven model approaches, improving the model’s robustness for continuous application.

DHDT kinetic model
Hydrotreating is a catalytic process typically operated with pressures ranging from 7 to 11 MPa for feeds with a boiling range of up to 400°C in the presence of hydrogen. The objectives of the process are:
•    Removing sulphur compounds, including refractory compounds like dibenzothiophene (hydrodesulphurisation [HDS]).
•    Removing nitrogen (N) compounds such as porphyrins and quinolines (hydrodenitrogenation [HDN]).
•    Saturating mono, di, and polyaromatics (hydrodearomatisation [HDA]).
•    To improve the quality of fuel in terms of density, Cetane Index (CI), and T95.

Slight thermal cracking also takes place in this operating regime. The main chemical reactions associated with the hydrotreating process can be seen in Figure 1.

IOCL proprietary DHDT model is developed based on rigorous structure-oriented kinetics of desulphurisation, dearomatisation, denitrogenation, olefin saturation, and cracking reactions. Desulphurisation kinetics is based on the detailed chemistry of different sulphur species analysed through GC-SCD ASTM D 5623 that are present in DHDT feed ranging from thiophene to 4,6-dimethyldibenzothiophene molecules. Suitable sulphur and nitrogen lumps have been considered in the model to capture accurate chemistry of desulphurisation and denitrogenation reactions, including inhibitions due to the presence of H2S and nitrogen compounds in the system.

The aromatic mono, di, and poly compounds characterised through HPLC ASTM D 6591 have been used for their saturation chemistry and accordingly incorporated into the model along with reversibility reactions considering both kinetic and thermodynamic regimes. As the olefin saturation reaction is very fast at DHDT process conditions, the single reaction for olefins saturation is considered in the model. Also, cracking reactions have been considered in the model to capture naphtha and gas formation from diesel.

Considering the reactor as plug flow, the whole bed length is uniformly divided into small zones over which ordinary differential equations for mass and energy balance are applied for each reacting lump:

- mass balance for species i

- Energy balance equation

These ordinary differential equations are solved simultaneously to estimate the sulphur, aromatic, nitrogen, hydrogen sulphide (H2S), ammonia (NH3), H2 consumption, diesel yield, and temperature across the length of each bed. The intrinsic kinetics are generated from pilot-scale experiments, and diesel product properties (density, CI, and distillation curve) are estimated as a function of feed properties and saturation level of different aromatics lumps. The process input variables considered in the model are the total pressure, hydrogen purity, recycle gas rate, reactor inlet temperature, and space velocity (liquid hourly space velocity [lhsv]). The lhsv and weighted average bed temperature (WABT) are the key hydrotreating unit parameters for unit optimisation and control of the catalyst lifecycle. However, the process kinetic model configured based on the start-of-run (SOR) kinetic parameters. Hence it is required to update with catalyst deactivation profiles to predict the unit’s performance throughout the catalyst lifecycle.