Cost Effective Strategy for Ultra Low Sulfur Diesel Technology. Klaus-Dieter Rost BAYERNOIL Ingolstadt, Germany. and

Transcription

1 Cost Effective Strategy for Ultra Low Sulfur Diesel Technology Klaus-Dieter Rost BAYERNOIL Ingolstadt, Germany and Tai-Sheng Chou H2Advance 4 Jordan Court Cherry Hill, New Jersey Key Words: Hydrotreaters, Ultra Low Sulfur Diesel, Reactor, ULSD Revamp Prepared for presentation at the AIChE Spring National Meeting 2003 Session 129-Process Industries Expo User Conference-I March 30-April 3, 2003, New Orleans, Louisiana Unpublished AIChE shall not be responsible for statements or opinions contained in the paper or printed in its publications Abstract Environmental Protection Agency (EPA) regulations require that sulfur levels in highway diesel should be decreased from the present 500 ppm to less than 15 ppm by June The production of ultra low sulfur diesel (ULSD) may require major capital investment by the refining industry. The process schemes considered include grassroots diesel hydrotreaters or revamps for existing diesel hydrotreaters. An industry-wide survey indicated that the cost for diesel hydrotreater revamps is about 60% of that for grassroots diesel hydrotreaters. With the current low profit margins in the refining industry, the capital burden even for hydrotreater revamps may be excessive. Therefore, cost-effective hydrotreater revamps are essential for producing ULSD. Based on a recently completed diesel hydrotreater reactor project in Germany, H2Advance demonstrated that the project cost for implementing ULSD with less than 10 ppm sulfur can be substantially below what was perceived revamp cost, provided that the project is properly executed. This paper details the installation of a new diesel hydrotreater reactor utilizing trickle bed technology. The 18-ft inside diameter reactor, a world scale diameter trickle bed reactor, uses three beds to satisfy a diesel product specification of less than 10 ppm sulfur. The efficiency of the trickle bed reactor depends on high quality feed distribution and quench zone design. The significance of the reactor internals design increases when producing ULSD due to the requirement of minimal flow channeling in the reactor. Based on the temperature spread observed in the reactor and its hydrodesulfurization (HDS) performance, this state-of-the-art reactor design demonstrates that it is entirely possible to design a trickle bed reactor with an inside diameter of 18-ft or larger that still provides minimal channeling for a ULSD project. Details of the reactor internals design philosophy are also discussed.

2 Introduction The United States EPA mandated a clean diesel program that will become effective in the US starting June The program requires sulfur in highway diesel to be reduced from a current specification of 500 to 15 ppm. This 97% reduction in the sulfur concentration challenges the refining industry with potentially heavy capital investments. Innovative schemes that alleviate the financial burden while satisfying the environmental regulations are essential to provide sufficient low cost transportation fuel while reducing emissions from cars and trucks. H2Advance ( recently assisted Bayernoil (1), Germany, with an ultra low sulfur diesel (ULSD) project by providing a state-of-the-art reactor design. The hydrotreater unit started up in May 2002 and met all design specifications for producing less than 10 ppm sulfur in diesel. The impetus for producing ULSD in Germany prior to 2005 is through tax incentives. H2Advance was awarded the reactor design project, because of their reactor design techniques, which give up to 10% lower capital cost than alternatives with an improved catalyst utilization efficiency. Diesel Hydrotreater The hydrotreater reactor provides the catalytic sites for removing sulfur and nitrogen from the diesel stream. Sulfur and nitrogen in diesel not only generate environmental pollutants that cause health problems, they also render future emission control devices ineffective, due to sulfur poisoning. The newer generation hydrotreating catalysts (2, 3) provide higher catalytic activities than catalysts available several years ago. However, catalyst replacement alone is not sufficient to provide ULSD production. Bayernoil s 53,000 barrel-per-stream-day (BPSD) diesel hydrotreater has an operating pressure of 900 psi and temperature of 670 to 720 F. Catalyst volume, a measure of available catalytic sites, is often expressed as liquid hourly space velocity (LHSV). LHSV is defined as the ratio of the hourly feed volume to the catalyst volume in the reactor. Prior to the ULSD program, Bayernoil s hydrotreater reactor had an LHSV of 2.8. With the addition of a new reactor having three times the catalyst volume of the old reactor, the two reactors provide an LHSV of 0.7. The new reactor at Bayernoil has an inside diameter of 18-ft, a world scale diameter trickle bed reactor, with three catalytic beds and a total reactor tangent-to-tangent height of 70 ft. The total weight of the reactor shell exceeds 450 metric tonnes. It is commonly known in the reaction engineering field that providing uniform flow distribution in a large diameter trickle bed reactor is a challenging task (4, 5). Liquid flow distribution in a trickle bed reactor depends on the gravitational flow of the liquid, which trickles down along the catalyst surface. The random walk nature of trickling liquid in a deep trickle bed may reduce the catalyst and liquid contacting efficiency. Therefore liquid redistribution, via quench zones in the trickle bed reactor, becomes a critical issue to ensure high catalyst utilization efficiency. Liquid redistribution is particularly critical in ULSD projects due to the requirement of minimal channeling for the ULSD production. Cost Effective Reactor Design Due to the large catalyst volume required for the HDS to attain less than 10 ppm sulfur in diesel product, the catalyst in Bayernoil s new diesel hydrotreater reactor was partitioned into three beds with two quench zones. This allows for liquid redistribution to improve catalyst utilization efficiency. The strategy for using 18-ft inside diameter reactor results from consideration of the catalyst volume required, loop pressure drop for the recycle gas (or compressor head), and the feasibility of transporting the new reactor to Bayernoil. The estimated pressure drop of the new reactor is 23 psi. Overall hydraulics were accommodated by a minor modification to the feed preheat train, avoiding a costly revamp to the recycle compressor. 2

3 Multi-bed reactor design requires quench zones to provide proper temperature control and fluid redistribution. However, many conventional quench zone designs require excessive reactor height, resulting in high capital investment. H2Advance s quench zone design reduces the capital requirement of the reactor up to 10%. It utilizes a novel quench nozzle and mixing chamber design to reduce the quench zone height while improving thermal equilibrium achievable between the quench gas and the higher temperature fluids coming from the bed above. The liquid distributor is an integral part of the quench zone, and it contains the essential features for determining the uniformity of liquid flow distribution in the bed below the quench zone. Use of the correct distributor type in the quench zone is critical to ensure a successful ULSD project. Quench Zone Design Features Novel Quench Nozzle Quench gas is introduced into the quench zone through quench nozzles located between the catalyst support beams. Conventional quench nozzle designs often require a quench gas distributor located below the catalyst support beams, which can add up to 12 inches reactor height for each quench zone. Consequently, the quench gas nozzle of H2Advance design requires less reactor height, but still maximizes thermal equilibrium achievable between the quench gas and the higher temperature fluids coming from the bed above. The H2Advance quench gas distribution assembly is located away from the centerline of the reactor. This novel design eliminates the removal of quench gas assembly during unit turnarounds, greatly improving unit turnaround schedules and unit reliability. Novel Mixing Chamber Design An innovative design for the mixing chamber provides thorough mixing of fluid coming from different quadrants from the bed above to attain uniformity of temperature at the quench zone exit. Drip tubes are used to calm the liquid flow out of the mixing chamber. Therefore, the H2Advance design requires only a single liquid distributor below the mixing chamber (Figure 1). Alternative designs for the quench zone often require two trays below the mixing chamber to break the momentum of the fluid leaving the mixing chamber. The ability to use a single distribution tray below the mixing chamber reduces the quench zone height by another 12 inches. Additional considerations for the Quench zone Design Several other features were incorporated in the design of the new Bayernoil s reactor to improve its catalyst utilization efficiency. The essential features are The use of Gayesco Flex-R for temperature measurements in the reactor The use of external catalyst dump nozzles for all three beds Gayesco Flex-R thermocouples provide accurate temperature measurement in the reactor (Figure 2) at reduced capital cost. One reactor nozzle of 3 inches diameter can accommodate up to about 40 measurement points per bed (Bayernoil reactor has eighteen temperature measuring points per bed with six points at bed inlet and twelve points at bed outlet). The use of Gayesco Flex-R thermocouples also eliminates penetration of the distributor tray. If vertical thermowells were to be used, seal problems could occur in the areas where the thermowell penetrates the quench zone tray and distributor tray. External catalyst dump nozzles were incorporated in the reactor design to completely eliminate any distributor tray penetration. 3

4 Penetration of the liquid distributor tray to accommodate vertical thermowells or internal catalyst dump pipes may induce liquid channeling in the catalyst bed below due to the obstruction of liquid distribution elements (downcomer or bubble-cap) by thermowell pipes or internal catalyst dump pipes. Liquid Distributor Design Considerations A well designed liquid distributor should have the following characteristics: Low sensitivity to tray out-of-levelness Low pressure drop Large spray angle High distribution element density Turndown flexibility Ease of cleaning There are two major types of liquid distributor designs available for hydroprocessing reactors that provide low pressure drop. One is the downcomer type (6, 7) (8, 9, 10). The second type is the bubble-cap like distributor design. Bubble-Cap Like Liquid Distributor Bubble-Cap like liquid distributor designs utilize the vapor phase for lifting liquid over the center riser of the bubble-cap. The key design parameters for bubble-cap distributors are the vapor phase flow rate and its physical properties. Therefore, bubble caps distributor design depends on the vapor lifting effort to balance tray out-of-levelness (Table 1). It is understandable that the bubble-cap like distributors may have applications for processes with high gas circulation rate, e. g., hydrocrackers or grassroots hydrotreaters processing a high percentage of cracked feedstock. When attempting a revamp of a diesel hydrotreater to attain ULSD specs, the economically attainable treat gas rate may only be about 1000 SCF/B or less. The lower treat gas rates in revamp scenarios coupled with higher activity HDS catalysts (i.e. lower operating temperatures) increase the liquid to vapor ratio in the reactor. ULSD also may require hydrogen purity increases in the reactor to enhance catalyst activity and cycle length. All of these factors (limited gas circulation, lower reactor temperature, and lower vapor density) make the bubble-cap like distributor design less desirable. Downcomer Liquid Distributor Downcomer type liquid distributors use orifices to control liquid level on the distributor tray. Therefore, the design of the downcomers is strongly dependent on the liquid rate in the reactor, making the downcomer distributor design an ideal candidate for diesel hydrotreater revamps. The square root dependence of the liquid rate on the liquid height above the downcomer orifice reduces the sensitivity of the liquid distribution on tray out-of-levelness. Table 1 Downcomer vs. Bubble-Cap Like Distributor Downcomer vs. Bubble-Cap Like Distributor Reactor Riser Riser/Downcomer Sensitivity of liq to Type ID,ft count Density, #/ft2 Pitch Spray Tray Unlevelness Downcomer " Square fine mist cone Square root of liq height Bubble-Cap like /4" triangular ring spray Depends on gas rate Sensitivity of Flow Distribution of Liquid on Tray Levelness The specification for distributor tray out-of-levelness is normally set at 10 mm maximum deviation from high to low point on the tray. Therefore, field installation (Figure 3) of reactor internals requires critical supervision for internals hardware adjustments to satisfy tray levelness requirements. 4

5 Levelness measurements during installation of the 18-ft diameter Bayernoil reactor trays showed maximum out-of-levelness values of 13 mm (Table 2). Even though the measured out-of-levelness exceeded the reactor internals spec of 10 mm, reliance was made on the square root dependence of liquid distribution uniformity on the liquid height above the downcomer orifice to minimize the impact of this out-oflevelness. Table 2 Measured Out-Of-Levelness for Distributor Tray and Tray Ring Distributor Tray Out-Of-Levelness at Bayernoil: Tray Ring Distributor Tray Out-Of-Levelness Out-Of-Levelness Prior to Adjustment Post Adjustment Prior to Adjustment Post Adjustment Tray mm mm mm mm Inlet Bed Bed Out-Of-Levelness spec for reactor internals is 10 mm. The above data demonstrate that for 18-ft inside diameter or larger diameter reactor, good supervision during the field installation for the reactor internals provides the opportunity to minimize the tray out-oflevelness. Test run results (1) strongly support that good distributor tray with downcomer design can provide liquid distribution uniformity sufficient for ULSD in a world scale trickle bed reactor even if tray levelness is modestly off-spec. Test Run Results The process test run at Bayernoil was conducted in June 2002 at near design rate and feed slate. Less than 10 ppm sulfur diesel product was achieved indicating minimal flow channeling in the reactor. At 99.8% desulfurization of the feed, any appreciable channeling (or bypassing) would cause failure to satisfy the product sulfur specification. Radial temperature profiles observed at the bed inlet indicated that the maximal temperature difference (peak-to-peak) was less than 3 F in the new reactor during the test run (1), indicating an effective quench zone design. It is commonly recognized in the reaction engineering field that large diameter trickle bed reactors are somewhat challenging to design with minimum flow channeling (reactors for the processes involving 100 percent vapor phase reactions, e. g. naphtha hydrotreaters, have no limitation on reactor diameter). The test run results clearly demonstrate that for Bayernoil s new 18-ft ID reactor, flow channeling is minimized. Revamp Cost Reduced The required revamp cost for Bayernoil diesel hydrotreater included a new reactor, two new shells for the feed/effluent exchangers, and piping modifications for the recycle and make-up gas compressors. Several items included in the project scope were added to improve the process reliability. 5

6 Without unit reliability items, the required revamp cost for producing ULSD at Bayernoil is estimated at about $106 per barrel per SD (stream day). Energy Information Administration (EIA) provided cost analyses for producing ULSD in May The revamp cost for a diesel hydrotreater with operating pressure similar to that of Bayernoil is estimated at $600 per barrel per SD, based on 25,000 barrels per SD rate. At Bayernoil s 53,000 barrels per SD rate, the EIA estimate becomes $444 per barrel per SD. The capital estimate of EIA is for inside battery limits (ISBL) only without capital contingency. Based on the subject revamp project with the addition of a new reactor, it has been demonstrated that the project cost for implementing ULSD with a target of less than 10 ppm sulfur can be substantially below what was perceived revamp cost, provided that the project is properly executed. Conclusions The H2Advance design for the new reactor internals contributed to capital cost savings of up to 10%. In addition, significant reactor project savings at Bayernoil resulted from dedicated project group engineers who explored means to reduce capital cost without sacrificing technical integrity. The above clean fuels revamp project shows that ULSD can be produced in the US at substantial savings over published costs, if proper cost reduction protocols are followed. Technology Status A state-of-the-art reactor design for a world scale diameter trickle bed reactor in ULSD production service has been demonstrated successfully at Bayernoil. As demonstrated, ULSD projects can be executed in a cost effective manner to satisfy future clean fuels requirements. A US Patent has been filed at the U.S. Patent Office in April

7 References 1. Rost, K., R. Herold, C. K. Lee, and T. S. Chou, Producing 10 ppm Sulfur Diesel: Bayernoil s Commercial Experience,, AIChE Spring National Meeting, New Orleans, Tippett, T., K.G. Knudsen, and B.H. Cooper, Ultra Low Sulfur Diesel: Catalyst and Process Options, NPRA 1999 Annual Meeting, March 21-23, 1999, AM Torrisi, S., D. DiCamillo, R. Street, T. Remans, and J. Svendsen, Proven Best Practices for ULSD Production, NPRA 2002 Annual Meeting, March 17-19, 2002, AM Chou, T. S., F. L. Worley, and D. Luss, Local Particle-Liquid Mass Transfer Fluctuations in Mixed- Phase Cocurrent Downflow through A Fixed Bed in the Pulsing Regime, Ind Eng Chem Fundam., vol 18, no 3, 279, Chou, T. S., Liquid Distribution in A Trickle Bed with Redistribution Screens Placed in the Column, I&EC Process Design & Development, vol 23, 501, Grosboll, M. P., R. R. Edison, and T. Dresser, Apparatus and Process for Distributing A Mixed Phase Through Solids, US Patent 4,126, Campagnolo, J. F., T. S. Chou, W. F. Heaney, and J. D. Ruggles, Inlet Distributor for Fixed Bed Catalytic Reactor, US Patent 4,788, Ballard, J. H., and J. E. Hines, Vapor-Liquid Distribution Method and Apparatus for the Conversion of Hydrocarbons, US Patent 3,218, Gamborg, M. M., and B. N. Jensen, Two-Phase Downflow Liquid Distribution Device, US Patent 5,942, Jacobs, G. E., S. W. Stupin, R. W. Kuskie, and R. A. Logman, Reactor Distribution Apparatus and Quench Zone Mixing Apparatus, US Patent 6,098,965. 7

8 Figure 1 Bayernoil Reactor Quench Zone Assembly 8

9 Figure 2 Gayesco Flex-R Temperature Measuring Points Left-Bed Inlet Right-Bed Outlet 9

10 Figure 3 Bayernoil New Hydrotreater Reactor, Installation of Reactor Internals (photo with permission from Bayernoil) 10