Hydrocarbon Processing Copying and distributing are prohibited without permission of the publisher
Email a friend
  • Please enter a maximum of 5 recipients. Use ; to separate more than one email address.

Use a systematized approach of good practices in pygas hydrogenation via APC

10.01.2012  |  Bader, J.-M.,  Axens, Rueil-Malmaison, FranceRolland, G. ,  Axens, Rueil-Malmaison, France

APC with inferential modeling was successfully applied to pygas hydrogenation units. The lengthening of the catalyst run length limits downtime for both the pygas unit and upstream units such as the steam cracker.

Keywords: [pygas] [steamcrack] [aromatics] [styrene] [cataylst] [hydrogenation] [hydrogen] [control] [control models]

As illustrated in Fig. 1, steam crackers produce many basic building blocks for the polymer industry, along with aromatic-rich gasoline (pygas). When naphtha is used as the feed for cracking furnaces, pygas yields increase significantly. Table 1 shows the typical pygas yield and composition for naphtha cracking. Pygas is a large contributor to benzene production capacities.


  Fig. 1. Pygas production and other products
  from a steam cracker.

Before the pygas can be routed to downstream units (aromatics extraction, etc.), unstable compounds such as diolefins and styrenics must be removed. Also, olefins and sulfur must be eliminated to ensure that final products will meet their specifications. This pygas treatment is achieved through hydrogenation steps. However, if the pygas treating is not optimized, then other undesired processing operations—including hydrogen flaring, reactor channeling, poor use of the second beds, and other issues—have a greater possibility of occurring.

Optimizing control

Advanced process control (APC) offers a solution to systematize implementing best practices, avoid mis-operation, and generate substantial benefits. The following case study describes the operational improvements and steps necessary when applying APC. This case study will apply actual plant data to demonstrate and quantify the benefits attainable from APC installations.


A case study will describe the industrial results obtained with the two-stage pygas hydrogenation process (PGH), as shown in Fig. 2. This PGH unit includes a first-stage process (GHU-1) to improve the stability of the raw pyrolysis gasoline by selectively hydrogenating diolefins and alkenyl compounds, thus making it suitable for further processing in the second stage. The reaction is carried out mainly in the liquid phase on a specific catalyst in a fixed-bed reactor.


  Fig. 2. Pygas hydrogenation flow scheme.

The selected operating conditions maximize conversion of the diolefins and alkenyl aromatics, while minimizing the formation of heavy products by polymerization. These operating conditions minimize aromatics loss.

In the second stage of PGH (GHU-2), the C6–C8 heart cut is further processed to prepare a feedstock suitable for aromatics recovery, by selectively hydrogenating the olefins and removing sulfur via hydrodesulfurization (HDS). The reactions are conducted through a series of specific catalysts in fixed-bed reactors. The operating conditions are selected to prevent aromatics losses by hydrogenation and to minimize heavy product generated by polymerization.


For APC of the hydrogenation processes, the hydrogen network plays a major role, and it needs to be studied carefully. Both pygas-treating stages consume hydrogen. Several configurations are possible to supply hydrogen to the pygas reactors, as shown in Fig. 3:

  • High-purity hydrogen option
  • Low-purity hydrogen option
  • First-stage purge in the second stage.

One concern that cannot be ignored is that other facility processes are also hydrogen users, such as selective hydrogenation of C2, C3 and C4 streams. In a situation of low-hydrogen availability, these processes have priority. As a consequence, pygas can reach a situation in which the first stage is temporarily operated with insufficient hydrogen, which has negative consequences on process performance and catalyst life. When excess hydrogen is available, it is important to reduce wasteful hydrogen flaring and to improve pygas operation by utilizing all available hydrogen. Efficient pygas operation can ensure that the best use is made of available hydrogen.


  Fig. 3. Hydrogen network summary.


There are four key areas that have the potential to deliver operational improvements. These areas are: improving first-stage product quality, reducing the risk of channeling, maximizing second-bed usage, and optimizing global hydrogen usage.

Improving first-stage product quality

The product entering the second stage must be hydrogenated to the correct level to prevent polymerization of any remaining diolefins or alkenyl aromatics in the second-stage reactor, which is operated at a higher temperature and in the vapor phase. If the hydrogenation process lacks sufficient hydrogen or has a low temperature profile, there will be a high tendency to form gums at the inlet of the second-stage reactor, thus generating unacceptable pressure drop and performance reduction.

A good indicator of the hydrogenation of diolefins or alkenyl aromatics is the styrene content of the first-stage reactor product. Normally, the styrene specification is set at 1,500 ppm to efficiently protect the second-stage catalyst. In addition, reasonable catalyst cycles are followed.

Fig. 4 presents a statistical distribution of the styrene content at the outlet of the first-stage reactor without APC. The histogram is characterized by a small number of off-specification values that could have damaged the second-stage catalyst. Also, a large proportion of product was well below the required specification. This over-quality is translated as a cost or give-away due to the unnecessarily high reactor temperature in the first stage that would reduce the catalyst cycle.


  Fig. 4. Typical styrene statistical distribution in
  a first-stage reactor outlet (ppmwt) from an
  online analyzer.

Reducing channeling by appropriate diluent flow

In the first-stage reactor, the flow entering the reactor is mainly n liquid phase, and it is constituted of fresh pygas feed, diluent cooled and recycled from the first-stage reactor outlet and hydrogen makeup, as shown in Fig. 5. If the diluent flow is too low, then the hydraulic loading of the catalytic bed may become too small, and thus possibly cause channeling.


  Fig. 5. Diluent flow adjustment in first-stage reactor.

If the diluent flow is too high, then the velocity in the reactor may be excessive. More importantly, it will lower the average bed temperature, thus a higher inlet temperature will be required to maintain performance. The higher operating temperature will negatively impact the stability of the catalyst.

The total flow to the first-stage reactor (fresh feed + diluent), also called “liquid load,” must be adjusted to an optimal target value close to the design value to produce the most continuous temperature profile. This condition is illustrated in
Fig. 6. With an inappropriate liquid load, the irregular temperature profile reveals the occurrence of channeling.


  Fig. 6. Improving from bumpy to smooth
  temperature profile with APC.

Maximizing second-bed usage by quench flow

In both pygas reactors (first and second stage), there is usually a quench injection between the first and second beds, to control the reactor temperature profile. The quench flow is often kept too high by panel operators to prevent temperature runaways.

In the first-stage reactor, as illustrated in Fig. 7, the consequences of excessive quench flow are a lower bed ΔT, which results in lower hydrogenation levels in the second bed. This leads to increased styrene content in the product. To compensate for this case, the first-bed temperature is frequently increased, which is detrimental to catalyst life cycles. APC objectives for the first-stage reactor include the balance of the hydrogenation between the first and second reactor beds.


  Fig. 7. Effect of quench flow changes (during
  14 hours) in the second bed of first-stage

Optimizing hydrogen usage by minimizing flaring

This principle is described in Fig. 8. The hydrogen-network purge to the flare is piloted by a hydrogen-network pressure controller. If the valve of this pressure controller is not fully closed, then hydrogen is wasted and sent to flare. In this case, it is possible to increase hydrogen flow to the pygas unit, until the pressure controller valve is fully closed, thus optimizing usage of all available hydrogen

In reality, the hydrogen-network pressure-control strategy implemented in the distributed control system (DCS) can be much more complex than presented in Fig. 8. Using in-depth knowledge of DCS capabilities, a new method was developed to minimize hydrogen loss and further increase the hydrogen makeup for the pygas unit without affecting the network pressure. This approach uses pressure controller parameters (setpoint, process value, valve opening, etc.) and dynamic models derived from step-test data. The benefits from this approach are to deliver more hydrogen to the pygas unit and thus improve hydrogenation performance.


  Fig. 8. Using hydrogen network information to
  maximize hydrogen supply to pygas.


This novel approach on APC strategy was successfully applied to optimize industrial pygas process operations. It incorporated several key control methods to improve the hydrogenation process:

Maximize feed

Common practice is to place an intermediate product tank between the steam cracker and the pygas unit. The volume of this tank can usually absorb one day’s production of pygas. The inventory of this tank should be minimized to reduce the risk of polymerization of unsaturated components present in the raw pygas, until downstream constraints have been saturated.

Use all available hydrogen

Optimize the global hydrogen management.

First-stage reactor

The first target is to do ultra-deep hydrogenation of diolefins and alkenyl aromatics, by controlling the styrene content, as measured by an online analyzer, in the first-stage product. The next step is to stabilize reactor operation by controlling the reactor liquid load at an ultimate level to avoid channeling.

Hydrogen partial pressure is maximized to promote hydrogenation by increasing the reactor pressure while maximizing the dissolved hydrogen fraction in the liquid phase. The process is operated to ensure a minimal gas purge flow to prevent concentration of inert species in the hydrogen recycle gas. (This is applicable if the unit is equipped with a recycle-gas compressor.) The temperature profile is optimized, using reactor inlet temperature, diluent and quench flow to prevent temperature runaway, to balance reactor ΔT between the two beds, and to maximize catalyst cycle length.


APC needs to identify the right compromise between the quality of the separation and energy savings.

Second-stage reactor

The first target is to perform complete hydrogenation of olefins and sulfur removal by controlling the bromine index (BI) and sulfur content of the reactor effluent. The next step is to minimize hydrogenation of aromatics by avoiding unnecessarily high-temperature process conditions. Finally, stable reactor operation will be achieved by the control of reactor ΔT and hydrogen-recycle gas density.


The pygas inferential model proposed for APC, as shown in Fig. 9, is based on highly evolved kinetic models that enable online styrene content and BI estimation, and, consequently, reactor optimization. Laboratory analyses of the first-stage effluent are used to estimate the first-stage feed quality (styrene content, bromine number and density). The first-stage reactor model integrates the estimated feed quality and measured reactor operating conditions, continuously inferring the first-stage product quality: styrene, diolefins and bromine number. Fig. 10 illustrates the prediction of the styrene compared with online analyzer measurement.


  Fig. 9. Inferential model to maximize hydrogen
  management while minimizing styrene content.


  Fig. 10. Styrene estimation in first-stage 
  effluent by first-stage reactor model.

The second-stage reactor model integrates estimated feed quality and measured reactor operating conditions, continuously inferring the second-stage product quality. Fig. 11 illustrates the estimation of the BI in the effluent of the second-stage reactor. Using spot-detailed analyses and collection of operating conditions, APC users can generate the best tuning parameters to fit the current operation, thus allowing “real-time” control moves to improve performance.


  Fig. 11. BI estimation in second-stage effluent
  by second-stage reactor model.

Recommended APC architecture

All APC components are embedded in an APC server connected to the DCS architecture, as depicted in Fig. 12. The control and optimization application consists of these modules:

  • MVAC module: State space multivariable predictive controller (MVPC)
  • Pygas inference
  • First-stage optimizer.

The application provides one-minute cycles for MVAC (the MVPC) and 60-minute cycles for the optimizer. Controller execution time was determined by the process dynamics.


  Fig. 12. APC architecture to optimize pygas


Here is a simple example of APC potential, illustrated by the application to a real pygas unit optimization project. The control matrix components are presented in Fig. 13. The inputs or manipulated variables (MVs) are:

  • Feed flow to be maximized when available to reduce tank inventory
  • H2 flow used as long as available to prevent flaring
  • Reactor inlet temperature to control styrene content, but minimized when possible to lengthen catalyst cycles
  • Quench flow to control styrene content.


  Fig. 13. Simplified APC variables used for
  simulation example.

The outputs or controlled variables (CVs) are:

  • Styrene in product, which should stay below the maximum limit
  • Reactor second-bed ΔT, which should stay below the maximum limit.

When APC in turned ON, the styrene analyzer is at 1,300 ppm, below its 1,500-ppm maximum limit, and the reactor second-bed ΔT is at 65°C, below its 73°C maximum limit. As far as the operation is concerned, quench flow is too high, resulting in excessive cooling of the second bed. When the reactor inlet temperature is too high, a styrene giveaway occurs. More feed is available; the intermediate tank is not empty; and additional hydrogen is available, but is flared.

APC actions on the process, as plotted in Fig. 14, can be summarized as making use of all available hydrogen to reduce styrene at the first-stage reactor outlet and thus reducing quench flow as far as the second-bed ΔT allows. These conditions will also reduce styrene content. Simultaneously, the reactor inlet temperature is reduced and feed flow is maximized within the constraints of the maximum styrene content limits. By better operation of the second bed, and using the 10% additional hydrogen available, this APC system was able to increase production by 10%, while decreasing reactor inlet temperature by 4°C.


  Fig. 14. Eight-hour closed-loop APC simulation example.

APC with inferential modeling has been successfully applied to pygas hydrogenation units. The overall benefits are:

  • On-specification product without giveaway
  • Ability to treat more feed : +10%
  • Reduction of first-stage reactor inlet temp.: –4°C
  • Catalyst run length maximization: +4 months/current
  • Reduction of aromatics hydrogenation: –10%
  • Reduction of H2 waste to flare: –10%
  • Energy savings: 5%

The lengthening of the catalyst run length limits downtime for both the pygas unit and upstream units such as the steam cracker. An additional benefit observed by the operating staff was, with the ease of setting targets and the confidence that the APC system would meet these targets, they were free to concentrate on other plant activities. HP


Bader, J.-M. and S. Guesneux, “Advanced process control optimization increases MTBE plant productivity,” Hydrocarbon Processing, October 2005.

Bader, J.-M. and S. Guesneux, “Use real-time optimization for low-sulfur gasoline production,” Hydrocarbon Processing, February 2007.

Tona P. and J.-M. Bader. “Efficient System Identification for Model Predictive Control with the ISIAC Software,” (ICINCO), Sétubal, Portugal, 2004.

Grosdidier P. and J.-M. Bader, “Supervisory Control of an FCC unit through Sequential Manipulation,” Instrument Society of America, 1996.


The authors express their gratitude to Joël Chebassier, Orionde, for his contribution to this article and also to the customers for making this article possible.

This article is a revised and updated version from an earlier presentation at the American Fuels and Petrochemical Manufacturers (AFPM) Annual Meeting, March 11–13, 2012, San Diego, California.

The authors
Jean-Marc Bader is project manager at Axens’ Performance Programs Business Unit. His background is in energy engineering with over 23 years of experience in APC projects (design, development, implementation and maintenance) for refineries, petrochemicals and chemical plants (including ADU, FCC, CCR, alkylation, hydrotreating, ethylene, ammonia, blending), with several APC tools, on various DCS. He joined Axens in 2001, after several years with Elf and Total. His responsibilities include proposal and project management for APC projects. He graduated with honors from I.N.S.A. engineering school.

Gildas Rolland is a deputy product line manager—hydroprocessing and olefins for Axens. He started his career in 1998 at IFP Energies nouvelles as a process engineer in the R&D department. In 1999, he joined the process design department of the North American office in Princeton, New Jersey. In 2001, he moved back to Axens’ head-office where he served successively as start-up and tech service advisor, specialist in olefins technologies including R&D activities related to technology and catalyst improvement. Mr. Rolland was appointed to his current position in 2010. He is a graduate of the Ecole Centrale de Lille (E.C.Lille) and holds a master’s degree in refining and petrochemicals from the IFP School.


Have your say
  • All comments are subject to editorial review.
    All fields are compulsory.

Related articles


Sign-up for the Free Daily HP Enewsletter!

Boxscore Database

A searchable database of project activity in the global hydrocarbon processing industry


Is 2016 the peak for US gasoline demand?




View previous results

Popular Searches

Please read our Term and Conditions and Privacy Policy before using the site. All material subject to strictly enforced copyright laws.
© 2016 Hydrocarbon Processing. © 2016 Gulf Publishing Company.