September 1999

Process Technology

Overhaul process reactors

Several examples utilize a general model to optimize and speedup process upgrades

Gualy, R., Dutta, S., GTC Technology US, LLC

Process reactors do not function well due to several reasons – normal aging and improper design or operating conditions. Very little can be done to reverse the effects of aging. However, action can be taken to correct poor design. Several case studies demonstrate the scope and opportunity for revamps and upgrades using a general reactor model.

To remain competitive, HPI operating companies must consider revamping under-performing reactors. Reactors are the heart of processing plants. A revamp can boost the processing facility's yield and productivity. Due to more restrictions from environmental regulations, revamps also facilitate reducing and/or eliminating hazardous wastes from effluent streams.

Candidate reactors for revamps fall into two categories – units that are under-performing or malfunctioning and the others that are performing normally. Catalytic gas-solid reactors will be the primary focus. Several examples illustrate the benefits of using a general model for two key commercial reaction systems – acrylonitrile and maleic anhydride.

A general reactor model should predict the performance of a variety of reactor configurations, and design and operational options for any given reaction or catalyst system. Conversely, the user should be able to estimate performance for any known reaction system or catalytic activity for a given reactor with the aid of a general model. Such modeling is thus applicable to any combination between the reaction system and configuration. Thus, the user can determine the optimal design or revamp option for an existing reactor with these tools. Details regarding such a model and how to use it is available in an earlier paper by the authors.1

Under-performing or malfunctioning reactors. In processing units, under-performing or malfunctioning reactors fall into two subcategories – ones due to normal aging and others due to improper design or operating conditions. Examples of aging reactors are:

  • Falling yield or productivity due to expected decay of catalyst activity over time
  • Circulation line capacity loss due to sticky fines buildup in line segments
  • Plugging of fluid distributors with fine solids, sticky byproducts or trace polymers
  • Pressure build-up due to fines accumulating in packed beds, distributors and flow lines
  • Expected erosion and/or corrosion of reactor internals and cooling coils.

Besides non-optimized operating conditions, Table 1 lists examples of poorly designed reactors.

   TABLE 1. Examples of poor design of process reactors.   
   Gas maldistribution and hot spots   
   Tendency of temperature runaway   
   Poor hydrodynamics   
   Control and operational problems, production loss and safety concerns due to reactor instability   
   Control and operational difficulties and production loss due to circulation instability or choking in solids flow lines   
   Low productivity or poor product quality due to   
   ü Unacceptable temperature profiles   
   ü Inadequate heat transfer   
   ü Wrong locations of feed, discharge and recycle lines on reactor   
   ü Improper design of the feed and discharge ports   
   ü Improper or inadequate design of feed distributors   
   ü Inadequate mixing   
   ü Excessive back-mixing   
   ü Wrong design of baffles and internals   
   ü Gas maldistribution and/or channeling   
   ü Improper bed voidage profiles or low bed density   
   ü Inadequate solids circulation rates   
   Inability to boost capacity and yield due to narrow design margins of key variables, such as   
   Temperature   
   Pressure   
   Feedrate and feed composition   
   Heat transfer capacity   
   Circulation rates   
   Deterioration of any or all conditions faster than expected from the normal aging process   

Revamp of under-performing reactors. Under-performance of reactors from aging is correctable, as required, during routine maintenance and shutdown periods. Poorly designed reactors may need trouble-shooting and revamping. In most cases by directly applying a reactor model, the source for failing performance can be identified. These models should handle both reaction and hydrodynamic aspects (including flow in the circulation systems) of the system. Thus, the best possible design and operating conditions to revamp the reactor can be determined. Four examples portray the benefits of redesigning existing process reactors.

Example 1 – Packed-bed catalytic reactor with gas maldistribution. Gas maldistribution and hot spot formations are typical problems associated with deep catalytic packed-bed reactors. Wrong design of feed gas nozzle, and its location with respect to the bed surface and design of the feed section may aggravate the problem. Feed gas velocity and property of the packed-bed material may also be contributing factors. The hot spots may lead to bed agglomeration, product degradation, and in the worst case, reactor temperature runaway. Quantification of such problems and their magnitudes may not always be possible. However, such problems can be diagnosed qualitatively by comparing observed behaviors with those expected from the reactor based on known catalyst performance data.

Fig. 1 shows such a reactor typical performance diagnostic by a general model and a revamped reactor configuration. The redesigned reactor uses three catalyst support plates, instead of one. Each plate carrys a third of the original catalyst packing. Due to reduced bed depth of each segment and gas redistribution between two segments, the maldistribution and associated problems are minimized in the revamped reactor. The performance diagnostic shows a comparison of the conversions and selectivities before and after the revamp.

 Fig. 1    

Revamp of a packed-bed catalytic reactor with gas maldistribution.

Example 2 – Tubular fixed-bed catalytic reactor with temperature runaway. Temperature control is crucial for tubular fixed-bed catalytic reactors involving exothermic reactions. Achieving the highest possible operating temperature level and using the minimum number of tubes of acceptable diameter and length is a challenge when designing these temperature-controlled reactors (TCR). For TCR reactors, a slight increase in temperature above a threshold value, either of the reactor feed or coolant, may cause a temperature runaway. Therefore, the reactor must operate at a temperature level below the threshold limit. However, there are cases when an operation at such temperatures is also unacceptable due to poor conversion.

For example, Fig. 2 shows the original design of a reactor used to produce acrylonitrile from propylene. (The kinetics are not presented due to proprietary reason.) For a feed temperature at or above 332°C, a temperature runaway occurs as shown in Fig. 2. At a lower feed temperature, the conversion is very poor.

 Fig. 2    

Revamp of a tubular fixed-bed reactor with temperature runaway problems.

Fig. 3 shows two revamp options for this reactor. In option 1 (Fig. 3), a smaller tube diameter is used; it improves heat transfer and hence, allows the reactor to operate at a higher temperature level. However, a larger number of tubes must be used to achieve the desired production rate. In option 2 (Fig. 3), the tube length of the original reactor is extended by adding an extra section. This arrangement compensates for the lower reaction rate from lower operating temperatures. Temperature profiles of the two revamp options (1 and 2) are shown along with that of the original design (Fig. 2). The choice between these two and other possible options will depend on mechanical and other considerations, and total revamp cost.

 Fig. 3    

Retrofit options for a tubular fixed-bed reactor with temperature runaway problems.

Example 3 – Bubbling fluidized-bed catalytic reactor with poor hydrodynamics. Hydrodynamics of bubbling/ turbulent fluidized-bed reactors depend primarily on the physical characteristics of the solid particles. Heavier and/or larger particles usually form larger bubbles that cause excessive gas bypassing thus, leading to poor conversion and yield. Conversely, very fine particles stick together and impede uniform fluidization. Gas channeling or rat-hole phenomena are often the result of using such particles, which also causes inferior reactor performance. In Fig. 4, the schematic of a fluidized-bed reactor operating with large bubbles is illustrated. Simulation of this reactor, for an actual commercial reaction system, using three average bubble sizes is shown in Fig. 4. The conversions and selectivities are the normalized values obtained by dividing the absolute values by arbitrary numbers, for proprietary reason. Reactor performance improves with decreased bubble size and is quite evident. To optimize this reactor will involve the design and installation of appropriate measures/ devices to control bubble growth within the reactor as shown schematically in Fig. 4.

 Fig. 4    

Revamp of bubbling fluidized-bed reactor with poor hydrodynamics.

Example 4 – Liquid-phase reactor with instability and control problems. Many commercial liquid-phase reactors, including hydrotreaters in refineries, operate with an external cooling loop through which part of the reactor product circulates at a high recycle rate. The arrangement is shown in Fig. 5. The heat generation (QG) and heat removal (QR1) rates of this reactor-heat exchanger loop are typically represented by the curves as also shown in Fig. 5. The points of intersection of these curves represent the stable operating conditions. The stable operating temperature of this reactor is, therefore, either at TA or TB. However, a slight disturbance or change in operating condition either from the reactor or heat exchanger could move the two curves. This will cause the reactor temperature to march quickly either to the higher (TA) or lower (TB) stable point. This is particularly true when the two curves are very close and almost parallel to each other near the desired operating temperature. This is a typical phenomenon of reactor instability.

 Fig. 5    

Retrofit of a liquid-phase reactor with instability and poor control.

Control and operation of the reactor often becomes very difficult in such a situation. Furthermore, a shift of the operating point toward the higher stable temperature may lead to severe product degradation and polymer formation, line plugging or cooling tubes fouling. In the worst-case situation, a temperature runaway may occur leading to catastrophic explosions.

Conversely, a shift toward the lower stable temperature may lead to "chilling" or severe production loss of the reactor. A revamp of this unstable reactor system is shown in Fig. 5. The reactor retrofit involves increasing the system's heat removal capacity. Another heat exchanger is installed in the circulation loop; the dual exchanger is operated at a higher feed temperature T2. With the additional capacity (higher heat transfer surface), the exchanger can now be operated with a lower D T, i.e., (TCT2), and thus, the increased slope of the heat removal curve QR2. The reactor will be stable, and all problems associated with the unstable conditions should be minimized. A general reactor model can be used to pinpoint such problems and identify the appropriate solution(s).

Revamp of reactors performing soundly. A primary driver for revamping many solid-catalyzed reactors is the availability of improved catalysts. A general reactor model can be used to do these revamps. Once catalyst performance data is known, an existing reactor can be easily used to accommodate the new catalyst and predict the optimum operating conditions for the catalyst. If required, a general model can be used to fine-tune modifications for the existing reactor design to heighten catalyst and reactor performance.

Implementing design modifications alone – without an improved catalyst – can increase reactor yield and productivity. Scope and opportunities for these revamp types can be assessed by applying a general reactor model. Applications for two commercially important reaction systems – acrylonitrile and maleic anhydride – are presented. They are based on reaction mechanisms and kinetics that closely represent the behaviors of typical commercial reactors.

Criteria for revamps. Many criteria are necessary on which revamp decisions are based. Four such criteria are used here, in addition to usual safety considerations: conversion (X), selectivity (S), yield (Y) and mass time yield (MTY). MTY represents tons of the desired product(s) obtained per ton of catalyst used in solid catalyzed reactors. Yield is defined as the percent of the primary reactant converted to the desired product. For example, in acrylonitrile application, propylene is defined as the primary reactant, and acrylonitrile as the desired product. It is the product of conversion and selectivity, i.e., Y = (X ´ S)/ 100, where X and S are also expressed as a percent basis. Mechanical and detailed economics are not considered as additional criteria for this analysis.

For single-pass reactors, i.e., reactors without recycle, a reactor is designed to maximize both conversion and yield. In many reactions, the selectivity for the desired product decreases as conversion increases. Thus, yield may approach a plateau or reach a maximum as conversion increases. Therefore, an optimum conversion is the design target for these cases.

For a recycle reactor, however, a higher selectivity may be a deciding factor when evaluating the optimum design. This is particularly true for an expensive feedstock, where maximum utilization of the feed to the desired product is the primary goal. This is also true when an optimum integration of the whole process, i.e., the reactor-separator-recovery integrated system, indicates that a higher selectivity is preferred over a higher yield.

MTY is a key criterion for all solid-catalyzed reactor systems, since it is the measure of reactor productivity per unit weight of catalyst used. Catalyst is often a significant fraction of total production costs. Therefore, the reactor should be designed to achieve the maximum possible MTY. However, MTY or catalyst productivity drops with conversion. Furthermore as conversion increases, this drop-per-unit conversion becomes larger. As the reactor conversion approaches a plateau, MTY drops dramatically with minimum increase in reactor yield or productivity. A choice between high yield and high MTY also determines the reactor configuration or its modification during revamp decision.

Example 5. Tubular fixed-bed reactor to produce acrylonitrile. Design conditions used for this reactor are listed in Table 2. Reaction mechanism for the design is illustrated in Fig. 6. Based on the reaction kinetics used, it is estimated that 48,000 tubes will be required for the desired capacity. The reaction mechanism and kinetics are different from a previously presented example.1 Fig. 7 and Fig.8 show the temperature profile and the XSYMTY profiles of this reactor as a function of catalyst bed height (or tube length), along with the other profiles of the revamped reactor designs. Due to heat transfer limitation typical of this TCR reactor, the projected conversion (X) is 73.6%, which is far short of near-complete conversion expected for commercial production. The projected selectivity and yield are 80.3% and 59.1% respectively. The MTY is 0.231.

 Fig. 6    

Reaction mechanism for propylene-to-acrylonitrile reaction system.

 Fig. 7    

Revamp of a fixed-bed acrylonitrile reactor.

 Fig. 8    

Performance curves for fixed-bed acrylonitrile reactor.

   TABLE 2. Reactor design conditions to produce acrylonitrile.   
   Capacity 100 tpd of acrylonitrile   
   Feed composition   C3H6 4.16%, NH3 9.56%,
 Moisture 1.54% Air balance
  
   Feed pressure   1.72 psia   
   Feed gas velocity   50 cm/s (superficial velocity at inlet conditions)   
   Catalyst particle size     1/8 in.   
   (Other conditions are shown in Fig. 7)   

Revamp options for TCR reactors are limited when compared to fluidized-bed reactors. Fig.7 shows one revamp option (case 2) in which the tubes are packed with the catalyst with varying concentration to modify the temperature profile. One way to accomplish this is to physically mix the original catalyst with an inert material of the same size, shape and density. Within this arrangement, the various combinations of catalyst concentration and bed segmentation are possible. The optimum design can be determined by modeling and then followed by validation with a test rig. The modified temperature profile of the revamped reactor along with its XSYMTY performance curves, as predicted by the model, are compared with the original reactor design. Using this retrofit design, reactor conversion can be raised to as high as 98.0% with a selectivity of 81.6% and yield of 80.0%. Due to catalyst dilution, the MTY (based on active catalyst content only, and excluding inserts) also increases to 0.493. Thus, the catalyst consumption is also reduced by more than one half. Remember, the benefits from such a revamp depend primarily on the reaction system and its kinetics.

This revamped reactor performance is compared with two other design options. One is a shallow fluidized bed with a booster fixed-bed reactor, which is shown as case 3 in Fig. 9. The other is a single deep fluidized-bed depicted as case 4 in Fig. 9. Comparing both designs illustrates that although a comparable yield may not be achieved by these designs, the MTY achievable is significantly higher in a fluidized bed (case 4) and even more in a fluidized fixed-bed combination (case 3). The higher MTY's are due to operation with both higher propylene feed concentration (8.32%) and higher temperature level possible in these designs. Of course, we assumed that both fluidized- and fixed-bed catalysts will have the same activity and selectivity.

 Fig. 9    

Retrofit designs for fixed-bed acrylonitrile reactor.

Example 6 – Bubbling fluidized-bed reactor to produce maleic anhydride. Compared to a fixed-bed reactor, a fluidized-bed reactor is more flexible not only in control and operation, but also in terms of available options for revamps and modernization. This example illustrates a few such possibilities using a commercial reaction system to produce maleic anhydride (MA) by catalytic air oxidation of n-butane. Proprietary design and operating conditions and absolute values of performance projections for this reaction system are not presented due to proprietary reasons. However, a general model application to evaluate several revamp options can be demonstrated by comparing the results on a relative basis. The reaction mechanism used for this demonstration is shown in Fig. 10. The rate expressions used are those provided in an earlier publication.2

 Fig. 10    

Reaction mechanism for n-butane to maleic anhydride reaction systems.

Fig. 11 shows three revamp options of an original fluidized-bed reactor. In option 1, part of the feed air is injected through one or more injection ports along the bed. In this design, pure oxygen as a replacement for air also is explored. In option 2, the feed concentration is raised and unreacted n-butane is recycled after separation and recovery. In option 3, a higher feed concentration and recycle is used as in option 2 together with reduced reactor temperatures. In these figures, TBC represents the original or base-case reactor temperature.

 Fig. 11    

Retrofit options for a fluidized-bed maleic anhydride reactor.

Various alternatives under each design option are again possible. Under option 1, various combinations of: number of injection points (for air/ oxygen), height of each injection point and the amount of air/ oxygen injection at each point can be considered. Various combinations between the three revamp options 1, 2 and 3 are also possible. Model projections of XSYMTY values for several possibilities are compared in Table 3. The table shows the percentage changes over the original design.

   TABLE 3. Revamp options of fluidized bed maleic anhydride reactor – comparison of projected performance   
  
Revamp option 1 – Multistage air/ O2 injection
  
     Percent change in   
     Conversion Selectivity Yield MTY   
   4-stage air 3.22 –0.17 3.02 –9.23   
   2-stage air 3.85 –0.43 3.43 –6.92   
   4-stage O2 11.25 –2.44 8.57 –14.61   
   2-stage O2 13.77 –3.40 9.94 –13.85   
  
Revamp option 2 – Higher feed concentration with n-butane recycle
  
     Percent change in   
     Conversion Selectivity Yield MTY   
   50% higher n-butane concentration –9.99 2.96 –7.33 39.23   
   100% higher n-butane concentration –25.65 6.36 –20.90 58.46   
  
Revamp option 3 – Higher feed concentration with n-butane recycle and lower bed temperatures
  
     Percent change in   
     Conversion Selectivity Yield MTY   
   100% higher n-butane concentration with T= TBC–10°C –26.12 12.02 –17.27 64.61   
   100% higher n-butane concentration with T= TBC–20°C –28.64 17.94 –15.83 65.38   

   Revamp option 1 – Multistage air/ O2 injection. Four possibilities – air injection in 2 and 4 stages and oxygen injection in 2 and 4 stages – are examined. In all cases, the total quantity of oxygen injected is kept the same as in the original reactor where it is injected as air. The number of indicated stages include the first injection stage near the bottom of the reactor in all cases. Table 3 lists the results; they indicate that both conversion and yield can be improved by this revamp, with maximum benefit seen from two-stage oxygen injection. However, it also indicates that this is achieved at the expense of lower selectivities. Furthermore, the benefits of higher conversion and yield may also be partially offset due to a lower MTY or a higher catalyst demand.

   Revamp Option 2 – Higher feed concentration with n-butane recycle. Two feed n-butane concentrations (fresh feed + recycle) –50% and 100% higher than the original are examined, assuming that these higher concentrations are acceptable from a safety point of view. In Table 3, the results show that although both conversion and yield decline with this revamp, selectivity is significantly improved. Furthermore, the reactor throughput or MA productivity increases dramatically due to higher feed concentrations, as reflected by the increase in MTY values.

   Revamp Option 3 – Higher feed concentration with n-butane recycle and lower bed temperature. Two bed temperatures – one 10°C and the other 20°C lower than the original value (TBC) – are used in this option along with twice the original (base case) n-butane feed concentration. The results shown in Table 3 indicate that although the conversions are decreased due to lower temperatures, they are compensated by a substantial increase in both selectivities and MTY. Lower operating temperatures are also preferable from the reactor safety point of view. Therefore, option 3 is a clearly a better choice over option 2.

Model validation. As shown by previous illustrations, a general reactor model can examine various reactor designs or design modifications before final choices are made. However, the model projections must be validated through testing in appropriate test units before implementing any revamp option. This is particularly true for reaction systems for which the mechanism and kinetics have not been established. Mechanical, detailed safety analysis and measures, and economics are obviously the final factors in the revamp decision.

Many choices. There are many choices when it comes to revamp reactors – whether to boost the capacity of the current reactor to meet increased demand, to make up for out-dated design or to correct for design mistakes made originally. A general reactor model can help discriminate between these choices, explore hidden options and guide one to the best revamp possible. In many cases, the cost and effort involved is minimal when compared to the boost in productivity and profit that can be achieved.

LITERATURE CITED

  1. Dutta, S. and R. Gualy, "General Reactor Model Improves HPI Applications," Hydrocarbon Processing, July 1999.
  2. Dutta, S. and G. D. Suciu, "Unified Model Applied to the Scale-up of Catalytic Fluid Bed Reactors of Commercial Importance," Fluidization VI, J. R. Grace et al., Eds., Engineering Foundation, NY, pp. 311–318, 1989.
 

Dr. Dutta is a senior staff consultant for GTC Technology Corp., specializing in designing and implementing reactor models to industry. He has published more than 30 papers and a book chapter describing work in reactor modeling. He has served numerous times on advisory boards and research teams across the globe, addressing aspects of chemical engineering and process modeling. Dr. Dutta has a long history of industry experience, spanning over 30 years, having held senior positions for Simulations Sciences, SABIC Research and Development, Engelhard, Fluor Daniel and CE Lummus Crest.

 

Ron Gualy is the general manager of the Chemicals Group for GTC Technology. His responsibilities include business unit marketing, sales and licensing of the technologies. He also manages business development and strategy, coordination and negotiation of strategic alliances, definition of research activities, and growth of the business. Mr. Gualy holds a BS degree in chemical engineering from Texas A&M University and is a registered professional engineer. He is a member of AIChE and the American Management Association. Mr. Gualy holds four patents concerning carboxylic acid recovery and has other patents pending. He has been with GTC Technology since its beginning six years ago.

The Authors

Related Articles

From the Archive

Comments

Comments

{{ error }}
{{ comment.comment.Name }} • {{ comment.timeAgo }}
{{ comment.comment.Text }}