February 2018

Process Optimization

Use online analyzer technology to optimize CCR unit operations

First, we must consider the background: Why is optimization of refinery process units so common and so necessary, and what analytical tools exist to help?

First, we must consider the background: Why is optimization of refinery process units so common and so necessary, and what analytical tools exist to help? The key problem in refining is that, although crude oil refining is a continuous and high-volume process with very significant raw material and energy costs, it is not steady-state. Crude oil feedstocks vary continuously in quality, availability and cost. At the same time, refinery products and their markets are very dynamic in terms of demand, specifications and pricing. This scenario leads to the use of relatively complex linear programming (LP) models to manage these changes. Underneath these models, individual process unit advanced process control (APC) packages must keep the units on target—even though these targets will change—and under control.

Refinery naphtha complex with CCR and integrated petrochemical units

If we look at one specific area within the refinery—the so-called naphtha complex, or naphtha conversion area—we can see the interaction between many different process units and streams. Central to these process units and streams is the catalytic reforming (CCR) unit. This unit takes low-value, heavy naphtha from the crude distillation unit (CDU) and converts it, after hydrotreating, into a higher-value, high-aromatics, high-octane feedstock.

In this article, CCR is used as a generic term to indicate a catalytic reforming unit. The arguments presented apply particularly to continuous catalytic regeneration reformers, but can also be applied to fixed-bed units. The questions are: What alternatives might there be for naphtha processing or sources of CCR naphtha feeds, and what alternative uses exist for the various unit products?

FIG. 1 shows a simplified, idealized view of these scenarios. For example, the reformate product from the CCR is frequently directed to the gasoline blending pool as a useful high-octane blend component, but the high-octane value of reformate derives from high aromatics (i.e., butane, toluene and xylene, or BTX) content. This has alternative uses and, depending on the price breaks between blended gasoline product and the aromatics unit, diversion as an aromatics unit feed might be determined.

FIG. 1. Cost, capacity, quality, value and demand continuously change.
FIG. 1. Cost, capacity, quality, value and demand continuously change.

Similarly, the straight-run naphtha from the CDU, usually hydrotreated as CCR feed, might be better employed as raw material for the naphtha steam cracker olefins unit. This depends on the relative instantaneous profitability of gasoline, aromatics and olefins products.

Measurement at some level is key to process optimization (FIG. 2). Measurement yields information, which allows for the possibility of control. What form this measurement takes is a slightly more open question, and one subject of considerable debate between those (mainly engineers) who like statistics and dislike analyzers, and those (mainly chemists) who do not trust anything that is not a directly traceable analytical result. This scenario leads to various approaches to APC:

  • APC based on inferential models: Uses many basic, mass-flow, pressure and temperature transmitters; requires a chemical engineered model of the unit; and requires lab test data to calibrate and maintain the inferential quality estimator.
  • APC based on physical analyzers: Uses many single-property, physical analyzers for direct measurement; and requires extensive maintenance, calibration, training and spares stockholding.
  • APC based on advanced analyzers: Uses a small number of multi-stream, multi-property analyzers; requires calibration or calibration model development; and normally offers a significant improvement in speed, precision and reliability.
FIG. 2. The cycle of measurement and optimization.
FIG. 2. The cycle of measurement and optimization.

APC based on actual process stream quality measurements from real analyzers is superficially attractive, but fraught with risk. Historically, this approach has been hindered by:

  • High capital costs
  • Limited reliability and high lifecycle costs
  • Large infrastructure requirements for installation
  • Complex operational requirements (e.g., calibration and validation).

However, technological advances have led to:

  • A wider range of available technologies
  • Simpler, more robust, lower-cost analyzers
  • A significant reduction in installation and operational demands.

The following are two examples of modern, robust analyzer technologies that have enabled easier and more reliable implementation of APC strategies based on real-time process analytical measurement. Fourier-transform near infrared (FT-NIR) analyzers with long maintenance intervals and low lifecycle costs have offered a solution to part of the problem. They offer space technology levels of reliability and uptime (the technology is routinely used in climate sensing satellites). Online FT-NIR analyzers have a proven track record in reliable hydrocarbon stream property measurement [in this case, research octane number (RON) and BTX in reformate product and paraffins, iso-paraffins, naphthenes and aromatics (PINA) in heavy naphtha feed]. The second technology is a solid-state, electrochemical sensor-based method for monitoring the hydrogen (H2) recycle/net gas stream, which is also critical to CCR operations.

FIG. 3. An example of operating parameter trade-offs in CCR operations.
FIG. 3. An example of operating parameter trade-offs in CCR operations.

The catalytic reforming unit, whether a CCR or a fixed-bed type, takes a heavy naphtha feed and, by catalytic conversion at reasonably high temperatures but low operating pressures, converts the paraffins and naphthenes to aromatics. The resulting product is an aromatics-rich reformate stream, and H2 net gas is generated within the unit and partially recycled.

What issues and choices exist for the operation of this unit? As previously indicated, the product of the CCR unit is more than a potential blendstock for gasoline blending. This is the traditional key product, but with varying markets and more complex refineries with extensive heavy oil up-conversion, what were previously seen as CCR byproducts now become significant and potentially attractive economic choices.

Reforming converts heavy naphtha into high-octane feedstock for gasoline blending, high-purity H2 suitable for use as hydrocracker makeup gas and high-aromatics (BTX) feed for petrochemicals. CCR unit operations offer numerous degrees of freedom, including severity vs. pressure vs. selectivity, all of which can be traded off to:

  • Run for maximum octane barrels
  • Run for maximum BTX yield
  • Run for maximum net gas
  • Run for maximum catalyst lifetime
  • Run for minimum energy usage.

The main operating parameters for the unit will be severity, pressure, and catalyst bed temperatures and profiles, which are interlinked and simultaneously affect yield, octane number, aromatics content, BTX spread and net H2 production.

The most basic measurement is octane number monitoring (usually RON) of the reformate product stream as an indicator of reactor severity. Personnel can use measurements to add chemical compositional parameters, such as total aromatics percentage; or discrete components, such as BTX percentages.

FIG. 4 illustrates a typical RON and aromatics modeling data set, as well as the resulting RON calibration model. Note: The model accuracy (vs. lab test) at approximately 0.2 RON at 1 sigma is better than the ASTM standard method reproducibility, due to good site laboratory precision. Therefore, the online FTIR does a better job than an online CFR engine, which would be significantly more expensive (FIG. 5). Advanced optical or solid-state devices for stream quality analysis are faster, better and more cost-effective.

FIG. 4. Calibration dataset, first derivative and partial least squares (PLS) regression plot for RON.
FIG. 4. Calibration dataset, first derivative and partial least squares (PLS) regression plot for RON.
FIG. 5. A validation plot of online FT-NIR RON data (Series 1) vs. lab test samples (Series 2).
FIG. 5. A validation plot of online FT-NIR RON data (Series 1) vs. lab test samples (Series 2).

The second stream analysis, which may be measured using the same FTIR unit as the one used for the reformate product, is the heavy naphtha feed. In this case, the target properties, which significantly affect the CCR unit yield and selectivity, are PINA and distillation (FIG. 6).

FIG. 6. PLS regression calibration plots for paraffins, isoparaffins, olefins, naphthenes and aromatics (PIONA) in the naphtha feed.
FIG. 6. PLS regression calibration plots for paraffins, isoparaffins, olefins, naphthenes and aromatics (PIONA) in the naphtha feed.

Naphtha quality variations can arise from varying CDU feedstocks and operation, but also from alternative naphtha feed sources. When CCR units are run with excess catalyst regeneration capacity, sub-optimum heavy naphtha feeds (e.g., from the FCC unit) can be run or mixed with conventional straight-run naphtha, resulting in a more dynamic unit envelope.

The final measurement is the net gas/H2 recycle stream. In this case, the key parameter is H2 mol%, but it must be measured in the context of a varying background of mixed light hydrocarbons content. The net gas recycle stream is not pure H2. It is mixed with other light gases recovered in the separator/recovery stages. This is a significant challenge for conventional technologies, such as thermal conductivity detection (TCD), that can handle only a limited number of interfering components (no more than two). The solid-state sensor is specific in response to H2, and is also protected against potential contaminants such as hydrogen sulfide (H2S) and carbon monoxide (CO) by a diffusion membrane, thereby allowing rapid H2 transport, but blocking larger contaminant species.

Takeways

We have reviewed the use of simple and robust advanced process analyzer technologies, specifically FT-NIR and solid-state, sensor-based H2 detection, to the most important process unit streams in the catalytic reforming unit. The octane, aromatics, PINA and H2 measurements can be made using these relatively straightforward analytical methods, and this data is reported in nearly real time (one-minute-stream cycle time), allowing close integration with unit APC. This allows better management of unit operational parameters, with a view to optimize the production of high-quality reformate and net gas/H2, with yields and composition better aligned with overall refinery and product market requirements. HP

The Author

From the Archive

Comments