Catalyst Deactivation Model Used to Select Laboratory Procedures for FCC Catalyst Testing

Volume 5 Issue 3


Laboratory testing and evaluation of fresh Fluid Catalytic Cracking (FCC) catalysts involves two steps - a deactivation step and a catalytic testing step. Much attention has historically been given to improved methods for laboratory performance evaluation in Microactivity Testing (MAT) and Pilot Plant testing, particularly with respect to simulating the reaction conditions of a commercial FCC riser. Proper simulation of the deactivation mechanisms which occur commercially is equally important.

A deactivation model has been developed which combines cataIyst and unit parameters to generate predicted equilibrium catalyst property distributions. These distributions can be approximated by blending catalyst steamed at varying severities. Alternatively, the effort required for steaming can be minimized by proper blending of fresh and steamed catalyst prior to testing. A case study is presented, along with test results comparing two BASF catalyst types, which shows that rnimicking the FCC Catalyst age distribution is essential to accurately predict performance differences.

Laboratory Deactivation

Laboratory deactivation of fresh FCC catalyst is accomplished by steam treatment at elevated temperatures to accelerate the hydrothermal aging which occurs in a commercial regenerator. Steaming conditions are typically chosen to achieve a steamed property such as MAT activity, zeolite unit cell size, or surface area which targets typical equilibrium cataIyst properties. A wide range of steaming conditions is used in the industry. However, in order to reduce the time required to achieve the target property these conditions all deviate substantially from the commercial conditions.

BASF laboratory studies have shown that relative laboratory performance rankings of cataIysts are most influenced by the severity of the steaming treatment (Ref. 1). In particular, coke selectivity rankings show a large dependence on the steamed catalyst properties tested. Therefore, it is very important to properly reflect anticipated equilibrium properties in the catalysts tested. One area which is commonly neglected in laboratory testing is the effect of catalyst age distribution. Because the nature of FCC unit operation dictates that a fraction of the catalyst inventory be replaced with fresh catalyst on a continual basis, a distribution of properties reflecting the age distribution will exist in any equilibrium sample. Measured bulk properties are, therefore, averages. Laboratory steamed samples which appear to match these bulk average properties may still perform differently if they do not match the property distributions.

FCC Catalyst Deactivation Mechanisms

There are two types of catalyst deactivation which occur in FCC systems - reversible and irreversible. Reversible deactivation occurs due to coke deposition each time the catalyst passes through the reactor, and is reversed by coke burning in the unit's regenerator. What is addressed here is the longer term, irreversible deactivation of the catalyst as it ages in the unit. This can be viewed as a combination of four separate, but interrelated, mechanisms:

  1. 1. Zeolite dealumination

    2. Zeolite decomposition

    3. Matrix surface collapse

    4. Contaminant effects

Zeolite dealumination, as measured by unit cell size reduction, reduces the acid site density and, hence, the inherent activity per unit of zeolite. Zeolite decomposition, measured by crystallinity or micropore surface area loss, also reduces activity. Both processes occur simultaneously in the hydrothermal atmosphere of an FCC regenerator. As will be discussed later, they are not independent. Matrix surface area collapse reduces activity by reducing catalytically active matrix sites, as well as by reducing porosity of the particle, which can restrict accessibility of the zeolite. Contaminants such as vanadium and sodium also contribute to deactivation in various ways. The model presented here considers only the first two zeolite-related mechanisms.

First Order Deactivation Models

The catalyst deactivation model most widely used for predicting equilibrium activities is a first order decay type:

  1. A = Ao exp(-kt)


  1. A = Activity (Second Order, Conv. / (100-Conv. ) )

    Ao = Fresh Catalyst Activity

    k = Deactivation Rate Constant

    t = Age in Unit, Days

This model was originally published by Mobil (Ref. 2) and was refined to include vanadium effects by BASF (Ref. 3). While useful as a first approach, the model contains several deficiencies. Because it only models activity, it can not be used to predict property and selectivity changes. Also, commercial data are required to determine k, and laboratory measurement of Ao is difficult. Extended time laboratory steaming data have indicated that a third order equation fits catalyst aging behavior better. As will be shown later, the third order fit can be represented by a combination of interactive, first order property changes.

A published paper by Ashland (Ref. 4) presented a first order approach with a more fundamental basis. They considered zeolite dealumination and decomposition to be separate, independent deactivation effects. A laboratory separation procedure was applied to equilibrium catalyst samples to develop relationships of catalyst properties as a function of in-unit age. One key conclusion from their study was that the dominant zeolite deactivation mechanism was catalyst property-dependent. Thus changes in fresh catalyst properties influence the way in which performance changes as a catalyst deactivates commercially.

Improved Model Development

Using the same approach taken by Ashland as a framework, a new model was developed which included both zeolite dealumination and decomposition features but also incorporated several modifications, including:

  1. - Zeolite dealumination was modified to a first order, approach to equilibrium mechanism. The "equilibrium" unit cell size is determined by catalyst rare earth and sodium content and unit severity parameters.

    - Zeolite decomposition was maintained as a first order mechanism, but with a changing rate constant which is itself a function of the degree of dealumination. Thus, as the unit cell size is reduced, the zeolite becomes increasingly stable toward further hydrothermal decomposition. This interaction is observed in both laboratory and commercial deactivation studies.

    - Activity per unit of zeolite is a function of the unit cell size (UCS), which is a measure of catalytically active acid site concentration. Laboratory studies were used to define the nature of this relationship, which is linear at low concentrations (low UCS) and flattens out at higher levels, which are contributed by the fraction of the inventory at relatively fresh ages.

The activity of a catalyst as a function of age is then determined by multiplying the activity per unit of zeolite by the zeolite content. Adjustments for matrix differences can also be included. For a given unit's replacement rate, an age distribution can be calculated assuming random replacement. Catalyst properties and activity are then integrated to determine the bulk "equilibrium" properties as well as their distributions. Two deactivation constants are required - for dealumination and zeolite loss. For a given equilibrium situation, the relative values of the two constants will determine the dominant deactivation mechanism and the resulting catalyst performance.

Case Comparison - Effect of Fresh Catalyst UCS

A hypothetical comparison of two catalysts which differ primarily in their fresh UCS is used to demonstrate the effects predicted by the model. Catalyst A is a typical USY octane catalyst, while Catalyst B has been further stabilized to a lower fresh unit cell size. Both are considered to be zero rare earth catalysts in this comparison. The model parameters for a typical commercial unit are shown in Table 1. The equilibrium activity and the addition rate are the same, as is the final unit cell size. The key difference is a reduced fresh unit cell size for Catalyst B, along with a slightly lower fresh zeolite surface area to reflect the additional processing involved. The dealumination rate constants are I assumed to be the same, while the decomposition rate constant is lower for Catalyst B due to its enhanced stability.

Figure 1 shows the predicted deactivation curves for each catalyst. The shapes are quite similar, with Catalyst B showing a more stable activity response. These curves more closely resemble a third order deactivation than a single constant first order model, which as noted earlier provides a good fit to extended time laboratory steaming data. However, the combination of first order processes considered here gives more insight into the deactivation mechanisms and catalyst property changes.

Figure 2 shows the integrated or cumulative activity response. While both catalysts reach the same equilibrium activity, Catalyst A derives a higher contribution of its total activity from lower age fractions of the inventory. If activity were the only concern, the analysis would stop here. However, the model also predicts cataIyst property distributions, which can be used to address selectivity differences as well. Figure 3 shows the response of zeolite unit cell size, indicating a measurably lower value for Catalyst B for up to about 20 days of in unit age. Referring back to Figure 2, well over half of the total activity is contributed by the catalyst within less than 20 days age, even though this fraction represents a much smaller fraction of the total inventory. The effect of this difference is illustrated in Figure 4, a cross-plot of cumulative activity vs. unit cell size. While both catalysts have the same final activity and UCS, the distributions are quite different.

The impacts of zeolite unit cell size distributions on selectivities are summarized in Table 2. Based on the distributions reflected in Figure 4, Catalyst B would be expected to produce a more favorable product distribution in commercial use. In particular, lower coke, higher octane, and less gasoline overcracking to LPG would be expected due to the elimination of the non-selective front end cracking by the > 24.45 Angstrom zeolite. Total gasoline and LPG yields would involve some compensating trade-offs, but both would be expected to be more olefinic. Octane-barrels would be expected to be higher due to improved octane without sacrificing gasoline yield.

Laboratory Catalyst Comparisons

To properly reflect the performance differences predicted by the model, proper laboratory test procedures need to be employed. In particular, laboratory steam deactivation procedures which address age distribution effects are required. Traditional laboratory testing techniques involve homogeneous batch steaming, which can lead to misleading results. It has been proposed (Ref. 5) that equilibrium catalyst be approximated by blends of catalyst steamed at varying severities. While this is an excellent concept, it requires considerably more steaming effort and is probably not practical on a routine basis. As an alternative, proper amounts of fresh (unsteamed) and severely steamed catalyst can be blended as a first approximation to an age distribution. This is illustrated in Figure 5, where the commercial model curves are shown for the two catalysts compared along with mixtures of 95% steamed (to a UCS of 24.25 A) and a 5% fresh. While the 5% addition does overstate the effect of fresh catalyst in the commercial inventory, it significantly understates the total non-equilibrated (>24.25 A UCS) zeolite and is, therefore, useful as a first approximation.

MAT test results are compared in Table 3 for the two catalysts tested both ways. When 100% steamed catalyst was tested, the product selectivities obtained were essentially identical. With fresh catalyst addition, both catalysts show substantial shifts due to the fresh activity. Major increases in LPG (particularly saturates) and coke are accompanied by a loss of gasoline, indicating overcracking and excessive hydrogen transfer activity in the high UCS fresh zeolite. All of these effects are much less pronounced for CataIyst B due to its higher degree of stabilization. Interestingly, only minor differences in activity boost have been noted in this type of testing even though the more highly stabilized Catalyst B has a lower acid site concentration. This indicates that above a certain site density, diffusion limitations will restrict further gas oil cracking and additional sites are used only for overcracking of the lighter gasoline product. Also noteworthy are lower dry gas and better bottoms conversion for Catalyst B when tested in this fashion.

The MAT experiments were repeated in a circulating pilot unit as shown in Table 4. Because the absolute conversion increases obtained were not the same, the selectivities are shown as normalized deltas in this case. All of' the same effects as noted in the MAT were observed in the pilot unit, but to a reduced extent. Engine octanes measured on pilot unit products indicated that RON decreased on fresh addition but MON actually increased. Further analysis of the gasolines indicated that the RON decrease was due to decreased olefin content, as expected with the high UCS fresh zeolite. The MON increase was due to increased aromatics and iso-compounds, particularly isopentane, which result from increased hydrogen transfer. These increases were higher for Catalyst B, indicating that its fresh UCS is in the optimum range for promoting these MON selective reactions.

Similar experiments were done at both MAT and pilot scale for the rare-earth-exchanged analogs of these products, with similar relative results. The benefits noted for the low fresh UCS catalysts have been verified in field performance tests. While none of these benefits would be predicted from traditional steamed catalyst testing, they are when test procedures are modified to include age distribution effects.


Selectivity effects noted in laboratory testing of fresh FCC catalysts are most dependent on the properties of the steamed catalyst tested. A deactivation model was developed which predicts the behavior of catalyst property distributions, and hence selectivities, due to aging in a commercial unit. Traditional testing procedures using uniformly steam-deactivated catalysts can lead to misleading results because they do not reflect the effects of age distribution. Experiments using blends of fresh and steamed catalysts were shown to be a more accurate representation of commercial performance for a particular comparison of two catalysts which differ primarily in their fresh properties. Specifically, the use of a low fresh unit cell size catalyst results in improved product selectivities compared with a standard USY zeolite catalyst.


1. E.L. Moorehead, M.J. Margolis, and J.B. McLean; ACS Svmposium 411, Catalyst Characterization and Development,1989, p. 120. (1989).

2. W. Lee, Ind. Eng. Chem. Proc. Des. Dev., 9, No. 1, (1970).

3. "Method Predicts Activity of Vanadium Contaminated FCC Catalysts", BASF The Catalyst Report.

4. J. L. Palmer and E.B. Cornelius. Applied Catalysts, 35, 217-235, (1987).

5. D.A. Keyworth et al, Oil & Gas Journal, 65-68, (1988).