Liability or Asset

Reactions and Separations Hydrogen: Liability or Asset? In a petroleum refinery, viewing hydrogen as an asset rather than a liability can lead to increased profits. The key is in the mindset. A typical investor uses money to make money. This article shows you how to use hydrogen to make money. What does this have to do with hydrogen? This article shows that refiners will never get the most value from their hydrogen unless they have the correct view of it. For the most part, refiners tend to view hydrogen as a utility that has to be supplied for them to operate. It is a necessary evil, the cost of which must simply be borne, just like that of fuel, electricity and water. Supplying hydrogen takes money out of their pockets, and so is a liability. Wrong. Hydrogen, if properly managed, can be an asset. Just as there are good and bad investments, there are good and bad ways to use hydrogen in a refinery. The secret is finding the proper ways, which requires a willingness to question conventional wisdom and to take a wider view of the issues. Here are the tools needed to make such an analysis, as well as some of the lessons learned from industrial projects carried out by the authors’ companies (2). Nick Hallale and Ian Moore, AspenTech Ltd. Dennis Vauk, Air Liquide hese days, you can’t be involved in petroleum refining without coming across something on hydrogen management in regard to the stricter fuel specifications on sulfur and aromatics, as well as the changing product markets. Engineers say that more hydrotreating and hydroprocessing will be required as a result. The conclusion? Refineries are going to need more hydrogen. Some vendors suggest that refiners will need to import hydrogen or perhaps build a hydrogen plant. Coincidentally, these are often the same vendors who sell hydrogen gas and hydrogen plants. This article takes a different tack. Instead of focusing on the problems, it looks at the opportunities, so engineers can think of hydrogen as an asset, not a liability. T Assets and liabilities In “Rich Dad, Poor Dad” (1), Kiyosaki and Lechter explain that most people never become wealthy because they do not know the difference between assets and liabilities. The authors’ definition of assets and liabilities is simple enough: “an asset is something that puts money into your pocket, while a liability is something that takes money out of your pocket.” Many people view items such as their homes as assets, but do these put any money in their pocket? They are, in fact, liabilities. True assets are those investments that make money for their owners, such as real estate, businesses and shares. Recognizing the difference between the two is vital. Perform a proper mass balance Just as in accounting, in a mass balance, what goes in must be accounted for. When we are dealing with hydrogen systems, the total amount of hydrogen produced and/or supplied must equal the total that is chemically consumed, exported, burned as fuel or flared. Unfortunately, it is rare to find a refinery where all of the hydrogen is accounted for. There is usually a poor hydrogen balance to begin with. Stream flowrates and compositions are often not measured and, when they are, there are often con- 66 www.cepmagazine.org September 2002 CEP flicting data. In many cases, large imbalances are often accepted and H2 23.50 45.00 attributed to leaks, distribution Import CCR 75.00% 92.00% losses, measurement errors or un0.65 10.66 44.35 accounted flows. Significant benFlows in MM scfd 92.00% efits — valued at hundreds of (standard ft3/d) 5.57 thousands of dollars or more per 38.78 0.04 11.31 year — can be achieved by exam75.97% ining the hydrogen balance and HC finding cost-free housekeeping 2.64 DHT IS4 11.29 improvements. Simple as this 8.21 75.00% 8.61 86.53% may sound, it can only be realized CCR Catalytic Reformer 8.65 70.00% HC Hydrocracker by understanding the overall hyKHT CNHT DHT Diesel Hydrotreater drogen system which, in turn, re3.47 KHT Kerosene 3.47 4.32 12.08 quires a systematic analysis and Hydrotreater 75.00% 65.00% 71.44% 12.80 the capability to model the sysCNHT Cracked Naphtha NHT Hydrotreater tem. NHT Naphtha 6.55 As an example, the authors’ Hydrotreater 60.00% companies have formed an alFuel IS4 Isomerization liance, PRO-EN, which recently Fuel Fuel Gas System performed a hydrogen-system study for a U.S. refinery. The refinery had been supplementing s Figure 1. A completed hydrogen balance may turn up hidden loses and opportunities. the hydrogen generated by its catalytic reformer with purchased molecular weight of hydrogen is so low, small changes hydrogen. Flowrate and composition data were collected, in stream composition can significantly affect the molecand a flowsheet balance was made with surprising reular weight. For example, a mixture consisting of 99% sults. The analysis showed that hydrogen worth $2 milhydrogen and only 1% ethane has a molecular weight lion/yr was being lost to flaring and fuel gas through that is 14% greater than that of pure hydrogen. three specific means, one of which was a leaky valve. Flowmeter readings are typically corrected for temOther opportunities for operating improvements included perature and pressure, but not for composition. Proper backing off the amount of hydrogen purged for comfort software can automatically account for this. It can also or buffer reasons, as well as avoiding compressing gas carry out a data reconciliation, whereby the user can streams only to drop their pressure somewhere else in enter the available data and then specify how much conthe plant. What is needed to set up a hydrogen balance? fidence he or she has in each value. For example, a 1. Flowrates, compositions and pressures must be deflowrate can be an accurate measurement (within ± 5%), termined at key points in the hydrogen system. Flow diaan accurate estimate (± 20%), a rough estimate (± 50%) grams of the hydrogen-consuming processes are also or a guess (± 500%). The software will then reconcile the needed, so that reactor and separator configurations, as data to achieve a balance, while staying within the confiwell as recycle locations, can be determined. dence bounds. Figure 1 shows what a completed hydro2. Models of the hydrogen consumers are required. gen system balance might look like. These need not be totally rigorous because this would take too long during the early stages of a project. HowPut needs ahead of wants ever, the models should be sufficiently detailed to cap“Rich Dad, Poor Dad” advises that people should not ture the important operating features of the units, as well spend money on unnecessary items, while trying to build as to predict their performance. Modeling software can a foundation for wealth. By delaying gratification now, be customized so that it allows fit-for-purpose simplified people will be able to use their money to buy assets that models for reactors, flash tanks and separation columns build up enough real wealth so that they can afford anyto be used as building blocks. thing they want later. The trick is to put one’s needs 3. Data correction and reconciliation tools are reahead of one’s wants. quired. One difficulty with hydrogen systems is that The same applies when trying to make more money stream compositions can greatly affect flowmeter readfrom hydrogen in refineries. The authors suggest that reings. Plant personnel often rely on flowmeters calibrated finers question whether certain units actually need to be for a gas with a molecular weight different from the fed with hydrogen at high purity. If there is a supply of stream that is currently being monitored. Because the CEP September 2002 www.cepmagazine.org 67 Reactions and Separations Makeup Hydrogen 40 MM scfd 99% Hydrogen Consumer Liquid Feed HP Purge 15 MM scfd 80% LP Purge 5 MM scfd 40% Liquid Product Makeup Hydrogen 40 MM scfd 99% Hydrogen to Reactor Only 83% Recycle Hydrogen 200 MM scfd 80% Cooler HP Purge 15 MM scfd 80% HP Separator LP Purge s Figure 2a. A black-box view of a hydrogen consumer can mislead one into thinking that it needs 99%-pure hydrogen ... 99%-purity hydrogen available, chances are that engineers or operators will claim that their unit needs to be fed with this purity. This mindset needs to be challenged. Once one breaks away from this, the scope for reusing and recycling hydrogen becomes greater and the benefits can be powerful. One European refinery used a significant amount of highly pure hydrogen for drying. When questioned about this, the engineers replied that this was the way it had always been done. That made sense because, until recently, the refinery had a surplus of hydrogen. Now, however, that flow is worth several hundred thousand dollars per year. A more-sensible alternative is using the offgas from one of the other hydrogen consumers to accomplish the same job for free. Admittedly, this was an exceptional case. Most refineries cannot find such obvious savings, since their hydrogen consumers require a certain flowrate and purity to operate properly. The flowrate is needed to maintain a gas-to-oil ratio high enough to prevent coking, while the purity is required to maintain the required hydrogen partial pressure for effective kinetics. With fixed flowrate and purity demands, it may seem as if there is no chance for improvement. However, the opportunities are still there if one knows where to look for them. The secret is simple: look at reactor inlets and not makeup streams. Figure 2a shows a hydrogen consumer with the makeup hydrogen and the high-pressure (HP) and low-pressure (LP) purges. Viewing it this way, it is easy to be misled into thinking that the unit requires 40 million standard ft3/d (MM scfd) of hydrogen at 99% purity. However, this is wrong. To find the real requirement, look at the internal workings of the consumer (Figure 2b). The hydrogen makeup stream is mixed with hydrogen recycle before being fed to the reactor inlet. Therefore, the purity that the reactor actually sees is only 83%. Recognizing this opens up the possibility for reusing other gases in the makeup and using less of the 99% hydrogen. As long as the flowrate and purity of the hydrogen going into the reactor do not change, the makeup purity is not that important. Put another way, the consumer may want 99% purity in the makeup, but really needs 83% purity at the reactor inlet. Liquid Feed 5 MM scfd 40% Reactor LP Separator Liquid Product s Figure 2b. ... while the true requirement is 83% purity. Consider the example in Figures 3a and 3b. There are two consumers, both taking makeup hydrogen from an external supplier at 99% purity, with a total demand of 40 MM scfd. If we say that the makeup purity has to be fixed, then there is no chance for reusing hydrogen (Figure 3a). However, by focusing on the reactor inlet, it is possible to reuse the purge from consumer A as part of the makeup to consumer B (Figure 3b). This allows the demand from the external supplier to be reduced by 3.4 MM scfd, or 8.5%. Using typical hydrogen costs, this will save between a half a million and a million dollars per year. The flows pressurized by consumer B’s makeup and recycle compressors are lower, reducing power costs, as well. These savings are achieved for the cost of a new pipe. Consumer B still has the same flowrate and hydrogen purity at the reactor inlet as before. All that has changed are the makeup and recycle flowrates, and, of course, the way of thinking. Hydrogen pinch analysis The above method may be suitable when there are only two consumers, but real refineries have a lot more than two. Should hydrocracker offgas be fed to the diesel hydrotreater or rather be sent to the naphtha hydrotreater? Perhaps this offgas should be used as fuel instead? Or could it be purified in via pressure-swing adsorption (PSA) or with a membrane? Given so many alternatives, the problem can seem confusing and overwhelming. What is needed is a quick and systematic way of knowing what the maximum hydrogen recovery achievable is. Enter hydrogen pinchanalysis (HPA). The method is similar to the well-known energy pinch-analysis used for designing heat-exchanger networks (3). Instead of looking at enthalpy and temper- 68 www.cepmagazine.org September 2002 CEP gen-containing streams that could potentially be used and include catalyticreformer hydrogen, “on-purpose” hydrogen, as well as the overhead gases from high- and low-pressure separaPurge Consumer tors in the various consumers. When A Reactor Inlet 8 MM scfd plotting the hydrogen demands, re80 MM scfd 91% 92.8% member to follow needs vs. wants. Do 18 MM scfd 99% not make the mistake of using makeup purities. The area enclosed between Hydrogen Fuel the source and sink composites yields Plant the hydrogen surplus diagram (Figure Recycle Production 4b). This is analogous to the grand 40 MM scfd 98 MM scfd composite curve in heat-exchanger85% 22 MM scfd network synthesis and shows the sur99% Purge Consumer plus hydrogen available at each purity B 2 MM scfd Reactor Inlet level. If the hydrogen surplus is posi85% 120 MM scfd tive throughout the diagram (as in Fig87.6% ure 4b), then there is some slack in the system. The hydrogen-generation s Figure 3a. With fixed makeup purities, the import flowrate cannot be reduced. flowrate can be reduced until there is a zero surplus (Figure 4c). The purity at which this occurs is the hydrogen pinch, and is the theoretical bottleneck Recycle for how much hydrogen can be recov62 MM scfd ered from the sources and be fed to the 91% sinks. The on-purpose hydrogen Consumer flowrate that results in a pinch is the A Reactor Inlet minimum target and is determined be80 MM scfd fore any network design. 92.8% Reuse Using the appropriate software, a 18 MM scfd 8 MM scfd task that would have taken days, 99% Hydrogen 91% weeks or months can be accomplished Fuel Plant in hours. If the gap between the target Production Recycle and the current use is large, then it is 36.6 MM scfd 18.6 MM scfd 93.4 MM scfd worth spending time to reconfigure the 99% 85% hydrogen network. The key is to avoid cross-pinch hydrogen-transfer and to 8.5% Reduction Consumer Purge be careful above the pinch purity, B Reactor Inlet 6.6 MM scfd since this is the region that is most 120 MM scfd 85% Hydrogen to 87.6% tightly constrained. Hydrogen streams Reactor Unchanged with a purity greater than the pinch purity should not be used to feed consumers requiring hydrogen below that s Figure 3b. Allowing the makeup purity to change reduces the hydrogen requirement by 8.5%, but purity. Also, hydrogen streams above does not affect the reactor inlet. the pinch purity should not be sent to ature, the concerns here are gas flowrates and hydrogen the fuel system or flared. If the gap is not large, time purities. In its original form, this method aims to maxiwould be better spent looking at other improvement opmize the inplant reuse and recycling of hydrogen to mintions. One of these is purification. imize the “on-purpose” or “utility” hydrogen production Purifying hydrogen (4). (Later, this article will address whether this should be minimized in all cases.) There are three main options for purification. But In HPA, the first step is plotting hydrogen composite these options are more fundamental than PSAs, memcurves (Figure 4a) of purity vs. flowrate for all sources branes or cryogenic systems. All of these do essentially and all demands in the refinery. Sources are any hydrothe same thing — they split a feed into a product stream Recycle 62 MM scfd 91% CEP September 2002 www.cepmagazine.org 69 Reactions and Separations with a high purity and residue with a low purity. Here, the options relate to the placement of the purifier relative to the pinch. As Figure 5 shows, there are three possible placements: above the pinch, below it and across it. Placing a purifier above the pinch means that the feed has a purity greater than the pinch purity. This placement can have benefits, but there are some risks involved. The increase in purity between the feed and product streams will, of course, be beneficial. However, that is only half of the picture. If the hydrogen recovery of the unit is low, a good deal of hydrogen above the pinch is just going to be lost in the residue and will wind up as fuel gas. The effects fight each other and a minimum recovery must be achieved to make the purifier worthwhile. Placing the purifier below the pinch is not wise. This simply takes hydrogen from a region where it is in excess and purifies it before putting it back in the same region. In essence, this would be akin to buying a purifier to make purer hydrogen for burning. One might argue that this is just common sense; no one would design a purifier to give a low product purity. That is true, but what about situations where the pinch purity is high, say 98%. Many membranes do not readily give product purities higher than 97%. So what does this tell about using a membrane for this case? The best option is to place the purifier across the pinch because it moves hydrogen from a region of excess to a region that is tightly constrained in hydrogen. It will free up hydrogen from the on-purpose source. Any lost hydrogen to the residue would have ended up in the fuel system anyway. 1 0.8 0.6 1 Source Composite Curve 0.8 0.6 Hydrogen Pinch 0.4 0.2 0 Purity Demand Composite Curve 0.4 0.2 0 0 50 100 150 200 250 300 Purity 0 10 20 30 40 50 Flowrate, MM scfd Hydrogen Surplus, MM scfd s Figure 4a. These hydrogen composite curves are the first step in identifying the pinch point. s Figure 4c. The “on purpose” hydrogen supply target is found by reducing its flow until a pinch appears. 1 + 0.8 0.6 0.4 0.2 0 - Purity 0 50 100 150 200 250 300 0 10 20 30 40 50 Flowrate, MM scfd Hydrogen Surplus, MM scfd s Figure 4b. Plotting the purity vs. the area between the composite curves gives the surplus diagram. 70 www.cepmagazine.org September 2002 CEP HPA applications and enhancements Pinch analysis is a powerful way for getting an immediate overview of the system, setting targets, as well as for initial screening of ideas. However, there are a number of areas where it has limitations. It does not fully cope with all the complexities of network design. The two-dimensional representation only considers flowrates and purities, but does not incorporate other important practical constraints such as pressure, layout, safety, piping, operability and, of course, capital cost. One of the more-important constraints is stream pressure. The targeting method assumes that any stream containing hydrogen can be sent to any consumer, regardless of its pressure. Of course, in reality, a stream may be reused only if it is at a sufficiently high pressure. Thus, the targets generated may be too optimistic and unachievable without installing new compressors. Unfortunately, compressors are among the most expensive capital items in a process plant. Thus, a retrofit should aim to make the best reuse of the existing compression equipment. Often, the true limit on hydrogen recovery will be set by bottlenecks in the existing compressors and not by purity and flowrate constraints alone. To account for pressure (and other constraints), a mathematical programming or optimization approach is required (5). Before moving on to the mathematical aspects of the method, the relevant physical, engineering issues will be considered. Obviously, direct reuse of hydrogen between consumers is only possible if the pressure is sufficient. However, it is possible to reuse a hydrogen stream indirectly, i.e., by routing through an existing compressor, provided that certain conditions are met. There has to be sufficient capacity in a compressor to accommodate the stream. Reusing hydrogen will change the makeup and recycle flowrates throughout the system, consequently, capacity may be freed up in one or more compressors. Also, the initial pressure of the reused stream needs to be high enough to be fed to the compressor, since these machines are designed for a specific inlet pressure. In addition, the compressor should be able to compress the stream to a pressure high enough so that it can be used by the consumer that needs it. To address these issues, first set up a superstructure that connects every sink with every source in which the source pressure is greater than or equal to the sink pressure (Figure 6). Note that compressors are included as sources and sinks. The basic constraints are then formulated, such as balances on the total flowrate and hydrogen flowrate, as well as any compressor limitations, such as maximum power or maximum throughput. A host of other constraints can also be incorporated, including: space limitations; the maxim that no new compressors are allowed; and the old favorite of not spending any capital. 1 Above Pinch Across Pinch Purity 0.6 0.4 0.2 0 Where? Below Pinch 0.8 0 10 20 30 40 50 Hydrogen Surplus, MM scfd s Figure 5. A purifier is generally best placed across the pinch, not above or below it. 600 psi Fresh H2 A A 1,600 psi 1,500 psi B 2,200 psi 1,700 psi B Fuel 1,600 psi 200 psi 2,200 psi 200 psi 1,600 psi 1,500 psi 2,200 psi 1,700 psi s Figure 6. Superstructure of all feasible connections among sources and sinks is a key to finding possible revamps. Next, the superstructure is subjected to mathematical programming that eliminates all of the undesirable features to satisfy an objective function, which could be minimum hydrogen generation. There is no requirement to use an algorithm to minimize hydrogen generation. The algorithm could for example, minimize operating or total annual costs. All relevant costs can be considered, including the hydrogen cost, compressor-power cost, fuel-gas credits and the capital cost of new equipment. The mathematical details of this step are found in Ref. 5. CEP September 2002 www.cepmagazine.org 71 Reactions and Separations ference in cost between the two solutions indicates how much not enabling 28.61 23.50 the connection will cost. If it is too CCR 92.00% 75.00% costly, then the engineer may decide 2.72 not to veto this option. 31.25 The other major limitation of 5.33 pinch analysis is that, while it gives 0.04 37.15 fundamental guidelines about purifi0.35 cation placement, it does not always HC help answer questions such as, IS4 1.01 11.84 9.21 75.00% which particular streams to purify 8.65 DHT and whether to use a PSA, memCNHT KHT brane, cryogenic separation or an al70.00% 9.14 1.03 ternative process. This is the prove65.00% 4.32 nance of an engineer familiar with 3.68 2.69 such unit operations. 0.35 11.14 99.00% For example, in a recent European NHT PSA 6.37 study, a refinery was facing a large in60.00% 7.28 crease in hydrogen demand to meet up17.33% coming sulfur specifications. Hydrogen Fuel recovery was more important than product purity, and so it was determined that, s Figure 7a. Retrofit design to maximize savings with an actual payback of 1.6 yr. for a certain stream, a membrane would be a better choice than a PSA unit. Another issue with purifiers is that different techIncorporating constraints at will to the model means nologies have different behaviors as far as pressure is that all practical considerations can be built in. The cost concerned. A PSA unit gives a product pressure close to of adding a constraint can also be determined. For examthe feed pressure, while the residue pressure is usually ple, an engineer may say he or she does not want to add extremely low. On the other hand, a membrane requires a long pipe to connect two units; so this option can be a large pressure drop to perform properly. Hence, the banned, and the optimization carried out again. The difproduct pressure is much lower than that of the feed, while the residue pressure is almost the same as the H2 feed pressure. These pressure issues 35.40 23.50 Import CCR 92.00% 75.00% need to be considered in the context 0.19 of the refinery, and can be handled using the superstructure approach de1.82 37.03 scribed here. The tradeoff is between 11.69 product purity, product pressure and 0.04 38.04 11.88 5.46 hydrogen recovery. A benefit of having purification HC 0.91 expertise on hand is the time saved. 0.99 IS4 75.00% Those familiar with HPA can work in DHT 8.65 parallel with engineers who are 70.00% CNHT KHT 10.17 knowledgeable in purification to 75.00% 65.00% rapidly assess the options and come up 1.63 6.14 with a flowsheet. This is more effec4.32 6.47 tive than generating a small number of 99.00% 3.81 options from the pinch analysis and PSA NHT then getting bids from a purifier ven3.70 6.56 dor, who works with this limited infor19.29% 60.00% Fuel mation to give size and cost quotations. With the parallel approach, both the pinch and purification engineers s Figure 7b. A lesser $5-million capital budget resulted in a different design. can see the entire picture. H2 Import 72 www.cepmagazine.org September 2002 CEP A case study illustrates the method. The existing hydrogen system in the refinery was that in Figure 1. The objective was to retrofit the network to minimize operating costs. All the process and cost data are given in Ref. 5. Several constraints were imposed by the refinery: • The existing compressors had 5% spare capacity. • There was space on-site for only one new compressor and one new purification unit. • A payback of more than 2 yr was not acceptable. The network retrofit was designed by setting the objective function to be the minimum operating cost, while constraining the payback time (capital cost divided by annual operating cost savings) to be 2 yr or less. Figure 7a shows the resulting design, with dotted lines indicating new equipment. (Note: PSA is pressureswing adsorption.) Both a new compressor and a PSA unit were used, and substantial repiping was done. The new compressor accommodates the increased recycle requirement for the naphtha hydrotreater (NHT), as well as compresses one of the feeds to the PSA so that its product can be used in the hydrocracker (HC). The total capital investment of the retrofit was $9.8 million and the operating cost savings was $6 million/yr, with a payback of 1.6 yr. Often, refineries have limited capital budgets, so these modifications might be too expensive. What would be the maximum savings achievable are with a fixed capital budget of, say, only $5 million? Adding the maximum capital expenditure as an additional constraint and reoptimizing gives the solution shown in Figure 7b. The best investment is in a PSA unit with no new compressor. Fewer new pipes need to be installed. The operating cost savings would be smaller (only $3.5 million/yr), but this is to be expected. Saving or investing? Up to now, the authors have addressed minimizing the hydrogen supply — in other words, saving hydrogen. This is a real problem for refiners. After all, hydrogen is expensive and a few percent saved can amount to millions of dollars per year. However, this is often not the most profitable course of action. As “Rich Dad, Poor Dad” puts it, no one has ever become rich by saving money. According to the book, the rich have their money work for them by investing it in assets that generate passive income. Thus, consider that hydrogen is money, and its consumers can be considered investors. Instead of merely saving hydrogen, why not consider reinvesting it where it will make money? How do we reinvest hydrogen? By feeding more and/or purer hydrogen into the appropriate reactors. For example, the hydrogen freed up using the network-de- sign methods can be used to process cheaper feedstocks. The freed hydrogen could also be used to boost partial pressures to enhance reactor conversions, throughputs, yields and catalyst life. The key question is: Which units will be the most profitable ones in which to invest? The techniques for analyzing hydrogen investments are reactor modeling and refinery linear-programming (LP) modeling. Rigorous kinetic modeling will provide a good understanding of process operations under different hydrogen-feed conditions. There are also tools for modeling a variety of fixed-bed hydroprocessing units. These kinetic reactor models can be connected to rigorous fractionation models, thereby creating a fully integrated model of the entire hydrotreater, reformer or hydrocracker complex. Hydrocracking models can optimize tradeoffs between feed rate, conversion, catalyst cycle life, feedstock severity, operating conditions and costs. For recycle hydrocracking units, there are tradeoffs between fresh feed rate and conversion-per-pass in single-stage units, or between first- and second-stage conversion in two-stage units. It is not necessary to model every reactor and determine the benefits via trial-and-error. Rather, one should focus on a short list of key processes and potential changes. For example, an LP model can identify the bottlenecks to increasing refinery profit. If increasing the hydrogen partial-pressure could eliminate one of these bottlenecks, then such a unit would be a candidate for applying a rigorous reactor-modeling and subsequent process analysis. Hydrogen network analysis can also provide insight into process operations. Above a threshold hydrogen purity, hydrotreater operation is insensitive to purity. If reducing the hydrogen partial-pressure to a hydrotreating reactor results in a large saving in the overall refinery hydrogen target, then this unit can be also be selected for rigorous modeling to determine the true impact of reducing hydrogen partial pressure on its operation. All process changes need to be modeled, their impact on the hydrogen network evaluated, and the final benefit established through an LP model. Reactor models can be linked to an LP software package. The models can enhance detailed operations planning, economic evaluation and scheduling. In particular, they can help a refiner decide how best to apportion intermediate, heavy-distillate streams between different conversion units. Figure 8 depicts a hydrocracker model. In this model, the hydrogen purity of the makeup gas varied from (the current) 89% to 100%, and the effect of the purity increase on product yields was calculated. In moving to 100%, the following were noted: • Full-range distillate yield increased from 52.8% to 54.6% CEP September 2002 www.cepmagazine.org 73 Reactions and Separations drogen available? Using the simple marginal cost of hydrogen is not the way to answer this. 60 54.6 Figure 9 shows how network 52.8 50 Full-range Distillate design tools such as HPA and mathematical programming fit 40 into an overall profit optimization 30 study. These identify how much Heavy Naphtha 20 additional hydrogen can be made 17.8 19.4 16.6 available for certain costs. Typi11.3 10 Unconverted Bottoms cally, a few percent more hydro0 gen can be squeezed out with 0.85 0.9 0.95 1 simple modifications requiring Makeup H2 Purity, mole fraction little or no cost (e.g., via piping modifications). Then there will be s Figure 8. Reactor performance improves with increased hydrogen purity for a hydrocracker. a step-change in which getting further hydrogen will require a purification system. Finally, there • Heavy naphtha yield increased from 16.6% to 19.4%. will come a stage when purification reaches its limit and • Unconverted bottoms reduced from 17.8% to 11.3%. additional hydrogen will have to be supplied from exterThe LP model identified potential benefits of $2 milnal sources. Knowing the true cost of providing additionlion/yr. How much would it cost to make the needed hyal hydrogen allows the engineer to weigh it against the Yields vs. H2 Makeup Purity, Mild Hydrocracking Unit Yields, wt% fresh feed LP Modeling H2 Recovery/ Purifications Reactor Modeling Data Collection Hydrogen System Analysis H2 System Model H2 Pinch Analysis Roadmap for Future Implementation Strategy for Profit Enhancement Through Hydrogen Re-investment H2 Availability Current H2 Piping Modifications Purification/ Recovery H2 Plant/ External s Figure 9. The network design tools show how much it costs to make a certain amount of additional hydrogen available. 74 www.cepmagazine.org September 2002 CEP Refinery Objectives Meet Small Increased and/or Hydrogen Demands Reduce Operating Cost Increased Cat. Ref. Throughput New H2 Plant 25 MM scfd Improve Refinery Economics Revamp HDS for Deeper Hydrodesulfurization Meet Future Standards Fuel Quality Now Reduce Hydrotreating Throughput H2 Network Modifications Optimize Unit Operations Add H2 Purification New H2 Plant 15 MM scfd New H2 Plant 23 MM scfd Time s Figure 10. Systematic tools help to construct a roadmap for the future. benefits found by reactor modeling and LP optimization. Figure 10 is a typical roadmap that shows where the refinery is now and where it wants to be at different points in the future (e.g., 1, 5 and 10 yr). (Note: HDS is hydrodesulfurization in Figure 10.) Conclusion If there is one conclusion you should take away with you, it should be that hydrogen is money. Stop thinking about it as merely an unglamorous utility and start lookCEP ing at ways to make more money from it. NICK HALLALE is a senior engineer with AspenTech’s Advanced Process Design (APD) group (Birkdale House, The Links, Kelvin Close, Birchwood, Warrington, WA3 7RB, U.K.; Phone: +44(0)1925 844 549; Fax: +44(0)1925 844 455; E-mail: nick.hallale@aspentech.com: He is the person to contact in regard to this article). He specializes in the application of process modeling and synthesis technologies. He has over six years’ experience in the development and industrial application of process design technology. He has acted as a consultant to many international companies, as well as to the British Government. Hydrogen management clients include refineries in Europe, the U.S., Canada and South Africa. Hallale graduated with a BSc and PhD in chemical engineering from the Univ. of Cape Town in South Africa, and then worked at the Dept. of Process Integration at the Univ. of Manchester Institute of Science and Technology (UMIST), where he developed new methods for refinery hydrogen management. His work has been published in many international journals and he has presented papers at international conferences, including several AIChE meetings. IAN MOORE is with AspenTech’s Advanced Process Design (APD) group (Phone: +44(0)1925 844 431; Fax: +44(0)1925 844 455; E-mail: ian.moore@aspentech.com). He has over 10 years’ experience of executing energy-reduction and hydrogen-management projects for refineries throughout the world. Previously, he was with Foster Wheeler’s refinery process group where he was involved in refinery process design and simulation. Moore has a BEng in chemical engineering from Exeter Univ., U.K., and an MEng in chemical engineering from McMaster Univ., Canada. DENNIS VAUK is a senior process consultant for Air Liquide’s PRO-EN Services group (Phone: +1-71-624-8343; Fax: +1-713-624-8350; E-mail: dennis.vauk@airliquide.com). Prior to joining Air Liquide, he worked at the Koch petroleum refinery in Pine Bend, MN, where he was responsible for developing plans to address low-sulfur gasoline and low-sulfur diesel regulations. Before moving to Pine Bend, he served for five years as a senior technical specialist in the Koch Technology Group, where he implemented projects to improve the hydrotreating and hydrocracking units in all of Koch’s refineries. He started his career in 1981 with Unocal’s Science & Technology Div. There, he received several awards for his technical contributions to the process development and process engineering groups. He has participated in the design and startup of 15 hydroprocessing units. Vauk received a BS in chemical engineering from the Univ. of Idaho. He holds six patents on petroleum refining processes. Literature Cited 1. Kiyosaki, R. T., and S. L. Lechter, “Rich Dad, Poor Dad,” Time Warner, New York (2000). 2. Cassidy, R. T., and E. Petela, “Life Cycle Utilities Management,” paper presented at the National Petroleum Refiners Association, 2001 Annual Meeting, New Orleans (March 2001). 3. Linnhoff, B., “Use Pinch Analysis to Knock Down Capital Costs and Emissions,” Chem. Eng. Progress, 90 (8), pp. 33–57 (Aug. 1994). 4. Alves, J., “Analysis and Design of Refinery Hydrogen Distribution Systems,” PhD Thesis, Univ. of Manchester Institute of Science and Technology (UMIST), Manchester, U.K. (1999). 5. Hallale, N., and F. Liu, “Refinery Hydrogen Management for Clean Fuels Production,” Advances in Environmental Research, 6, pp. 81–98 (2001). CEP September 2002 www.cepmagazine.org 75