By David R. Thornton, P.E.

Many pressure vessels in the petrochemical industry contain internals that, like the pressure boundary components, are designed for a given set of conditions. Examples include fixed bed supports in hydrofining reactors, tray supports in fractionation towers, and grids and cyclones in fluidized catalytic cracking (FCC) regenerators. The internal’s design involves stipulating the desired operating life, a corrosion rate if applicable, and the design temperatures, and usually a design differential pressures. The sizing of the internals then involves using this information along with an allowable stress that accounts for the given operating environment and desired life.

For FCC components operating in the creep regime, such as regenerator grids or cyclone systems, the design of the components involves several operating scenarios each with its own predicted duration, temperature and differential pressure or other load. The engineer designing the components then typically uses a life fraction approach where each combination of stress (a function of load and component thickness or dimension), operating temperature, and duration consumes a portion of the available creep life. Typically in the design the load, temperature, and operating duration are fixed and component thickness or other dimensions determined so that the stress at the various operating conditions consumes the total desired component life (i.e. the sum of the life fractions equals 1.0).

Often the actual operating conditions vary from those assumed during the design. In such cases, the equipment owner may want to determine the amount of the predicted life used to date by past operating conditions and how various future operations will affect the predicted component life. Having this information allows the equipment owner to make economic decisions regarding the future operating conditions, e.g., by comparing the potential additional revenue from operating at a higher temperature or with at higher loading versus the cost of component replacement. Conducting such an assessment involves the following steps:

- Determine the history of the past operating conditions (temperature and loading). The history may be constructed from the unit operator’s recollection of past conditions but preferably will come from recorded process information such as daily average temperature and differential pressure.
- Divide the history into periods of substantially uniform temperature and loading.
- Estimate the desired future operating loading (e.g., differential pressure or catalyst loading).
- Determine the component stresses for each past or future loading condition.
- Calculate the life fraction consumed by past operations from the stresses, the operating temperature, and the duration using the Larson-Miller Parameter (LMP) approach of API RP 530 (!insert title here!). The permitted future life fraction is then 1.0 minus the life fraction consumed to date.
- Determine the allowable life versus operating temperature for each desired future loading condition, again using the LMP approach of API RP 530.
- Calculate the life fraction consumed by each future operating condition (loading and temperature) by dividing the expected duration of the condition by allowable life at the condition determined in Step 6.
- Adjust the future operating conditions and durations such that the total of the life fractions equals the permitted future life fraction determined in Step 5.

Using these steps along with information regarding equipment replacement costs and the increased revenues possible by the future operating scenarios permits the owner to make informed economic decisions.

As an example of this type of assessment, Carmagen Engineering, Inc. conducted a study for a large Gulf Coast refinery to evaluate the effect of past operating history on the creep life of a FCC regenerator grid and determine the life under various operating scenarios. The refinery’s operations support department supplied the operating temperatures below and above the grid and the grid differential pressure data for the evaluation of the effects of past operating history.

The study showed that, while the grid had been in service for close to twenty years, past operating conditions used approximately twenty-three percent (23%) of the creep life. Based on this finding, we advised the refinery that the regenerator grid did not require replacement at the next turnaround.

The creep life curves developed as part of the study and (shown in Figures 1 and 2) provided the permissible duration in hours for a particular combination grid differential pressure and process temperature above the grid.

A summary of the study methodology includes the following:

- We statistically analyzed historical operating data that included process temperatures below and above the regenerator grid and differential pressure across the grid, to determine values to use in the study.
- We calculated grid stresses as a function of grid differential pressure by conducting a series of axisymmetric finite element analyses. The finite element model, shown in Figure 3, included the grid, grid supports, and part of the regenerator vessel. The analyses considered the regenerator operating pressure and regenerator design pressure.
- Based on past heat transfer analyses of the regenerator grid, we equated the grid metal temperature to the average of the process temperatures above and below the grid.
- We obtained an estimate of the creep life consumed to date and the generated the creep life curves included in this memorandum using the time for stress to rupture versus temperature approach in API Recommended Practice 530.

The creep life curves for use in evaluating the effects of future operating conditions appear in Figures 1 and 2. Figures 1 provides the predicted regenerator grid life, in hours, for a limited range of grid differential pressures, 2 psi to 3 psi, and operating temperatures above the grid of 1300°F to 1400°F (five 10°F increments from 1300°F to 1350°F and two 25°F increments from 1350°F to 1400°F). Figure 2 provides the predicted regenerator life, in hours, for a slumped catalyst bed as a function of the process operating temperature above the grid just prior to the loss of fluidization in the regenerator.

The basis for use of the grid life curves shown in Figures 1 and 2 comes from API RP 530. API RP 530 contains a set of material based curves that relate the stress in a component to a value called the Larson-Miller Parameter (LMP). Two values of the LMP actually exist, one for the average stress to rupture and one for the minimum stress to rupture, with the later usually controlling. For a given material, and component stress along the ordinate (i.e., Y) axis, the curves determine the LMP along the abscisa (i.e., X) axis. Once the value of the LMP is known API RP 530 provides an equation that relates the LMP, the component temperature and the component predicted life:

LMP = (T + 460)(C + log_{10}(L))10^{-3}

where:

T = component metal temperature,°F

C = a constant that depends on the component material

L = predicted component life, hours

This equation can be rearranged to provide the predicted component life as a function of metal temperature, material, and the LMP determined from the component stress:

The portion of life used then depends on the duration at the given stress and temperature:

The example below shows how to use the curves to assess the effects varying conditions. It assumes that past operations have consumed 0.23 ( 23 %) of the component life. Thus future operations may account for 0.77 ( 77 %) of the life.

Table 1 provides the conditions we will assume for the example illustrating how to use the curves. The total duration consists of a four year FCC run between turnarounds. We have divided the four years into three combinations of operating conditions and one upset where the catalyst “slumps” on the top of the regenerator grid.

Examining Figure 1 for the grid delta P and process temperature conditions of Operating Condition 1 we see that the permitted life equals about 116,700 hours. The duration of this condition equals 980 days times 24 hours or 23,520 hours. Operating Condition 1 thus consumes a percentage of the grid life equal to:

Examining Figure 1 for the grid delta P and process temperature conditions of Operating Condition 2 we see that the permitted life equals about 69,800 hours. The duration of this condition equals 230 days times 24 hours or 5,520 hours. Operating Condition 2 thus consumes a percentage of the grid life equal to:

Examining Figure 1 for the grid delta P and process temperature conditions of Operating Condition 3 we see that the permitted life equals about 31,900 hours. The duration of this condition equals 248 days times 24 hours or 5,952 hours. Operating Condition 3 thus consumes a percentage of the grid life equal to:

Examining Figure 2 for the process temperature condition just prior to losing fluidization, we see that the permitted life equals about 1,660 hours. The duration of this condition equals 3 days times 24 hours or 72 hours. The slumped catalyst condition thus consumes a percentage of the grid life equal to:

To evaluate the acceptability of the proposed operating conditions we add the percent life consumed by each condition to the 23% consumed by the past operating conditions). This value must be less than 100%.

Since this value is less than 100%, the proposed operating conditions would be considered acceptable from a life standpoint. However, the proposed conditions consumed about 51% of the grids creep life in a four year time period. Future operations of a similar nature for a subsequent four year run would consume more than 100% of the predicted life.