Study compares methods that measure hydrogen use in diesel hydrotreaters

Oct. 13, 2008
A recent study found that the more accurate methods for analyzing hydrogen consumption are the feed-total liquid product (TLP) H-content analysis and chemical analysis methods.

A recent study found that the more accurate methods for analyzing hydrogen consumption are the feed-total liquid product (TLP) H-content analysis and chemical analysis methods. These comparable methods were more accurate than using gas analyses alone.

Due to variability in the various methods used to determine chemical hydrogen consumption in pilot-plant testing, however, all three methods discussed in this article should be used.

The chemical analysis method can provide some insight into where and how much H2 is consumed in the various reaction classes and the effects of processing conditions such as pressure, temperature, and Space velocity. A properly designed and executed pilot-plant testing program can determine hydrogen consumption within ±2 normal cu m/cu m (±12 scf/bbl).

Obtaining an accurate estimate of the hydrogen requirement for the design of a grassroots or revamped diesel hydrodesulfurization (HDS) unit is important given the current global refining environment. There have been substantial increases in the capacity and operating severity of diesel HDS units worldwide in recent years.

In reSponse to tightening of sulfur and other Specifications of diesel fuels, many refiners have built or plan to build grassroots diesel HDS units and almost all refiners have revamped or plan to revamp existing units (OGJ, Oct. 23, 2006, p. 28). Consequently, the consumption of hydrogen has increased substantially.

With today’s natural gas prices, hydrogen consumption can account for more than 80% of the variable operating costs of a diesel HDS unit. It is therefore important to obtain an accurate estimate of the hydrogen requirement when designing a grassroots or revamped diesel HDS unit.

And because the hydrogen requirement is strongly dependent on the feedstock and design conditions, conducting pilot-plant tests using the design feedstock is important.

Several different methods can determine the chemical hydrogen consumption in pilot-plant testing. This article compares chemical H2 consumption calculated by various methods and discusses advantages and shortcomings of each and their applicability to commercial diesel HDS and other hydroprocessing units.

Measuring H2 consumption

In a pilot-scale diesel HDS test unit, as in a commercial HDS unit, Equation 1 (see attached Equation box) can represent the hydrogen balance. Two commonly used methods for determining chemical hydrogen consumption, in mass per unit time, are represented in Equations 2 and 3, which derive from Equation 1.

Click here to enlarge image

The first method (Equation 2) requires measuring flow rates and hydrogen content in the feed and TLP as well as flow rates and compositions, such as light hydrocarbons, H2S, H2O, and NH3, in all offgas streams. The second method (Equation 3) requires only the flow rates and H2 analyses of the makeup gas and all offgas streams.

Because hydrogen consumption is commonly expressed as N cu m/cu m (normal cu m of hydrogen/cu m of oil, at 0° C. and 1 atm) or scf/bbl, Equations 2 and 3 can be rearranged to calculate the volumetric H2 consumption per unit volume of oil feed.

For example, Equations 4 and 5 can determine hydrogen consumption in N cu m/cu m.

In addition to these two methods, one can determine total chemical hydrogen consumption by calculating the sum of the stoichiometric hydrogen consumption for each class of hydrogen-consuming chemical reactions, such as HDS (for sulfur conversion), hydrodenitrogenation (HDN), hydrodeoxygenation (HDO), olefins saturation, and aromatics saturation (Equation 6).

Experimental program

Petrobras plans to install a grassroots diesel HDS unit to produce ultralow-sulfur diesel (ULSD) in its Alberto Pasqualini Refinery, Canoas, RS, Brazil (REFAP SA).

REFAP solicited and received design proposals from several process licensors. To assist in evaluating the various process design proposals, REFAP contracted PetroTech Consultants LLC to plan and execute a catalyst testing program to compare the catalyst systems proposed by the various vendors and licensors.

The pilot plant and lab facility of Intertek-PARC Technical Services, Pittsburgh, were used to carry out the testing program.

Click here to enlarge image

Table 1 shows the feed properties. The design feed for the new diesel HDS unit consists of 62 vol % straight-run heavy gas oil, 10 vol % coker light gas oil, and 28 vol % FCC light cycle oil.

The test program encompassed running multiple reactors, each loaded with a different catalyst system proposed by one of the vendors and licensors, for a range of processing conditions for treating the design feed. Test results include yields and product qualities (sulfur, cetane, aromatics, etc.) as well as hydrogen consumption at the various processing conditions for each of the catalyst systems.

This article discusses only the results for hydrogen consumption.

Results comparison

In pilot-plant testing, the most common method used to determine chemical H2 consumption is Equation 4, which requires measuring H2 in the feed and TLP. This method is accurate, for any single data point, to about ±10 N cu m/cu m or 60 scf/bbl, due to analytical test accuracy of measuring hydrogen in the feed or TLP of ±0.1 wt %. Variability in mass flow measurement of feed and liquid and gas products, however, can add to variability in the H2 consumption determination.

To assess the variability of hydrogen consumption results calculated by other methods, we compared results of Method 2 (Equation 5) and Method 3 (Equations 6-11) against those from Method 1 (Figs. 1a and 1b).

Click here to enlarge image

Fig. 1a compares H2 consumption by gas analysis (Equation 5) with that of feed-TLP-H-content method (Equation 4). Although the bias (represented by the slope of the curve of 1.03 vs. the parity line) is good, H2 consumption from gas analysis varies significantly vs. that of the feed-TLP-H-content method.

The variance, represented by an R2 of 0.723, is poor and the standard deviation of 20.7 N cu m/cu m (123 scf/bbl) means that the gas analysis method is unacceptable as a replacement for the feed-TLP-H-content method.

This is understandable because H2 consumption calculated from Equation 5 is based on the small difference between mass flow rates in the makeup gas and offgas streams. Accuracy of the mass flow measurement of the offgas stream is usually poorer relative to flow measurements of feed or makeup gas.

This problem is exacerbated due to the varying composition of the offgas streams, which may contain various amounts of light hydrocarbons (C1-C5) as well as H2S, NH3, and N2 obtained at different treating severities. Calibrating the feed hydrogen and offgas meters at the beginning of the run eliminates any bias.

Fig. 1b compares H2 consumption by chemical analysis method, Equations 6-11, with that of the feed-TLP method. Agreement between the two methods, represented by a curve slope of 0.999, is excellent and the variance (R2= 0.926 with a standard deviation of 8.7 N cu m/cu m or 52 scf/bbl) is reasonable. The H2 consumption by the chemical analysis method is therefore an acceptable alternative to the feed-TLP method.

One major benefit of the chemical analysis method is that it allows an estimate of where the H2 is consumed amongst the various reaction classes.

Hydrogen consumption

Equations 7-11 estimate the amount of H2 consumed in the various reaction classes. For example, Table 2 shows hydrogen consumption via the various reaction classes to produce ULSD from the feedstock in Table 1 at 82 atm.

Click here to enlarge image

Table 2 shows that only 30% of all hydrogen consumption is due to all nonaromatics, which are consumed roughly in equal amounts by HDS, HDN, and olefin saturation (for the feed in Table 1). For comparison, monoaromatics consume almost 30% and polyaromatics consume about 40% of the total chemical hydrogen.

The amount of H2 consumption by nonaromatics stays relatively constant at various product sulfur levels, while H2 consumption by aromatics continues to increase significantly with decreasing product sulfur level resulting from increasing hydrotreating severity.

Click here to enlarge image

Figs. 2a and 2b show this trend. Fig. 3 shows that the percent of H2 consumption by aromatics increases with decreasing product sulfur and increasing hydrogen partial pressure.

Accuracy of hydrogen analyses

The accuracy of any single hydrogen analysis is about ±10 N cu m/cu m, or 60 scf/bbl given normal laboratory conditions. This accuracy can improve, however, if one averages multiple analyses.

A tYpical pilot study will often include many different catalysts and operating conditions giving 20 or more determinations of hydrogen consumption. If the errors in hydrogen consumption determination are random (unbiased), which they usually are, then value-averaging reduces the error in determining hydrogen consumption by a factor of 1/N1/2 where N is the number of hydrogen consumption determinations.

Twenty experiments would improve the accuracy to ±2 N cu m/cu m or 12 scf/bbl.

Product quality improvement

Although the increase in hydrogen consumption does not seem to dramatically reduce the product sulfur level to less than 100 ppm (wt) (Figs. 2a, 2b, and 3), it does significantly improve other diesel product qualities, such as increased cetane index and reduced density (Fig. 4). This improvement in diesel quality primarily results from increasing conversion of polyaromatics and monoaromatics (Fig. 5).

Click here to enlarge image

Hydrogen consumption is a strong function of feed quality and operating pressure (OGJ, May 15, 2006, p. 48). The extent of aromatic saturation is much greater at a higher pressure and lower Space velocity of ULSD units designed for aromatic feeds.

Click here to enlarge image

The combination of more-severe conditions and higher feed aromatic content gives much higher hydrogen consumption. Depending on the relative values of hydrogen and diesel, the volume swell that results from polynuclear aromatic and aromatic saturation during the production of ULSD might not fully offset the cost of hydrogen that is required for aromatic saturation.

Click here to enlarge image

Cetane additives can improve the cetane of ULSD products that meet all Specifications except cetane number. Cetane increases of 5-10 numbers are possible at treatment costs of about 0.1 ¢/gal-cetane number.

Actual cetane uplift and treatment costs depend on the composition and base cetane of the diesel fuel; therefore, actual costs are site-Specific. Its effectiveness decreases and cost increases as dosage level is increased. The use of a cetane improver additive is often a more cost-effective way of increasing diesel cetane than aromatic saturation.

Commercial unit

In addition to chemical H2 consumption, some H2will be lost due to solution losses, purge losses, and mechanical losses in a commercial ULSD unit as in any other hydrotreating unit. Sufficient fresh hydrogen must be supplied to the ULSD unit to meet product-quality Specifications and achieve the catalyst cycle length target.

Solution loss is H2 that leaves the reactor circuit dissolved in liquid hydrocarbon leaving the high-pressure separators. Increased H2 partial pressure at the product separator will increase solution losses. TYpical ULSD solution losses are 2-10 N cu m/cu m or 12-60 scf/bbl.

Purge loss is H2 that leaves the reactor circuit in a recycle gas purge to improve recycle gas purity. Mechanical loss is H2 that is lost through mechanical leakage from the makeup and recycle gas compressors, i.e., packing vents and seals.

Depending on the unit’s design, solution and mechanical losses may be 10-20% of the total H2 consumption; purge losses are under operator control. Purging minimizes the buildup of light hydrocarbon gases as well as H2S for units without an amine absorber. TYpical losses for purging range from 3% to 20%, depending on makeup gas H2 purity, H2 consumption, and H2S levels in the recycle gas, with higher purge losses for units without an amine absorber.

Derivation of Equations 7-11

For sulfur-removal reactions, Equation 12 shows the generic HDS reaction. XHDS is the stoichiometric H2 consumption (molar ratio of H2 to hydrocarbon) of the generic HDS reaction.

For example, XHDS = 1 for mercaptans, 2 for sulfides, 3 for disulfides, 3 for benzothiophenes (or 6 if the adjacent benzo-ring is saturated), 4 for thiophenes, and 2 for di-benzothiophenes (or 5 if one of the adjacent di-benzo rings is saturated). For a tYpical diesel feedstock, an average molar ratio of 3.6 is assumed.

To calculate H2 consumption for all HDS reactions, these assumptions apply:

  • Basis: 1 cu m of feed: (1,000 * SGf) = kilograms of feed.
  • Total sulfur in feed, kg = 1,000 * (Sf *10-6) * SGf.
  • Total sulfur in total liquid product, kg = 1,000 * (Sp*10-6)* SGp * TLP yield.
  • Total sulfur removed by all HDS reactions, kg-mole = 1,000 * 10-6 * SGf * [(Sf) – (Sp * SGp/SGf * TLP Yield)] / 32.
  • Total H2 consumed, kg-mole = XHDS * total S removed, kg-mole.
  • Total H2 consumed, N cu m = 3.6 * 22.41 cu m/kg-mole * total H2 consumed, kg-mole.

The total H2 consumed by HDS in N cu m of H2/cu m of feed is calculated from Equation 7:

    CH2 HDS = 0.0252 * SG f * [(S f) – (S p * SG p / SG f * Y p)]

Similarly, for nitrogen-removal reactions, Equation 13 shows the generic HDN reaction. XHDN is the stoichiometric hydrogen consumption (molar ratio of H2 to hydrocarbon) of the generic HDN reaction.

For example, XHDN = 1 for primary amines, 2 for secondary amines, 3 for tertiary amines, 1 for anilines, 4 for pyroles, 6 for indoles, and 7 for quinolines or carbazoles. For a tYpical diesel feedstock, an average molar ratio of 5.0 is assumed.

Given a basis of 1 cu m of feed:

    1,000 * SG f = kilograms of feed.
  • Total nitrogen in feed, kg = 1,000 * (Nf *10-6)* SGf.
  • Total nitrogen in TLP, kg = 1,000 * (Np*10-6) * SGp * TLP yield.
  • Total nitrogen removed via all HDN reactions, kg-mole = 1,000 * 10-6 * SGf * [(SGf * Nf ) – (Np * SGp / SGf * TLP yield)] / 14.
  • Total H2 consumed, kg-mole = XHDN * total nitrogen removed, kg-mole.
  • Total H2 consumed, N cu m = 5.0 * 22.41 cu m/kg-mole * total H2 consumed, kg-mole.

Equation 8 shows the total H2 consumed by HDN in N cu m of H2/cu m of feed.

A similar approach can estimate the H2 consumption by oxygen-removal reactions (Equation 9 and assuming a stoichiometric hydrogen molar ratio XHDO = 5). The organic oxygen content, however, in most diesel feeds is negligible; therefore the amount of H2 consumed by HDO reactions is negligible.

Olefins saturation

Bromine number is usually used to measure olefin content. Bromine will react with the amount of carbon double-bonds in the oil. This method is used to calculate H2 consumption, given a basis of 1 cu m of feed:

  • Feed (1,000 * SGf) = kg of feed.
  • Total double bonds in feed, kg = 1,000 * (BRf /100) * SGf.
  • Total double bonds in TLP, kg = 1,000 * (BRp /100) * SGp * TLP yield.
  • Total double bonds removed, kg-mole = 1,000 / 100 * SGf * [(BRf) – (BRp * SGp / SGf * TLP yield)] / 159.8.
  • Total H2 consumed, N cu m = 22.41 (cu m/kg-mole) * total H2 consumed, kg-mole.

Equation 10 is then used to calculate total H2 consumed by olefins in N cu m of H2/cu m of feed.

Aromatics saturation

Under tYpical hydrotreating conditions, polyaromatics will undergo a multi-step hydrogenation to monoaromatics, which are further hydrogenated to cyclic saturates: 4-ring aromatic + (1-2) H2 → 3-ring aromatic + (1-2) H2 → 2-ring aromatic + (1-2) H2 → 1-ring aromatic + 3 H2 → cyclic saturates.

The extent of the conversion from polyaromatics to monoaromatics depends strongly on the reactor’s operating pressure and catalyst tYpe. The single-ring aromatics are much more difficult to hydrogenate to cyclic saturates. All of the reaction steps are reversible.

For tYpical diesel feeds, multiring aromatics may contain up to 4-ring aromatics, depending on the cut point. To estimate H2 consumption for all polyaromatics to monoaromatics, we assumed a stoichiometric H2 consumption (molar ratio of H2 to hydrocarbon) of 3.3. For monoaromatics to cyclic saturates, we used a stoichiometric H2 consumption of 3.0.

Equation 11 calculates the total H2 consumption by aromatics saturation.

In pilot plant or commercial units in which the Specific gravity and yield of the TLP are not unavailable, Equations 7-11 can be further simplified to Equations 14-18 by assuming the term (SGp / SGf * TLP yield) = 1.

Click here to enlarge image

The H2 consumption results obtained from the simplified Equations 14-18 differ from the results from original Equations 7-11 by only 3.34% (Fig. 6).

The authors

Click here to enlarge image

C.K. Lee ([email protected]) is a principal of PetroTech Consultants, Mantua, NJ. His areas of expertise are hydroprocessing, catalytic reforming, and clean-fuel technologies. Previously, he worked for Mobil Technology Co. for more than 20 years. Lee holds a BS in chemical engineering from Cheng Kung University, Taiwan, and a PhD in chemical engineering from the University of Houston.

Click here to enlarge image

Steve McGovern ([email protected]) is a principal of PetroTech Consultants. His areas of expertise include hydroprocessing, catalytic cracking, clean fuel-technologies, and reactor design. Previously, he was a technology expert for Mobil Technology Co. where he worked for 27 years, primarily in process development and commercial operations support. McGovern holds a BS in chemical engineering from Drexel University, Philadelphia, and a PhD in chemical engineering from Princeton University, NJ. He is also a director of the Fuels and Petrochemicals Division of AIChE.

Click here to enlarge image

Luiz E. Magalhaes C. da Silva ([email protected]) is a process manager at Petrobras America Inc., Houston. He has worked for Petrobras for 30 years, previouly at Petrobras’ Refineria Alberto Pasqualini (REFAP SA), Canoas, Brazil. Da Silva has held positions in business development, technology, optimization, planning and scheduling, and as an operation manager. He holds a BS in chemical engineering from from Universidade Federal do Rio de Janeiro and a graduate degree in petroleum processing engineering from Petrobras, Rio de Janeiro.

Click here to enlarge image

Carlos A. Osowski (caosowski @petrobras.com.br) is a process engineer at Refineria Alberto Pasqualini (REFAP SA), Canoas, Brazil. He has worked for Petrobras for 38 years in the areas of optimization of existing units and design of new units. Osowski holds a BS in chemical engineering from Universidade Federal do Rio Grande do Sul, Porto Alegre, Brazil, and graduate degree in petroleum processing engineering from Petrobras, Rio de Janeiro.