Linde Technology 1 2004 En17 10192

Transcript

e_01_Titel_Umschlag_03 09.07.2004 11:17 Uhr Seite U1 Reports on Science and Technology Linde Technology Designing Biotechnology Plants Forklift Ergonomics Cracking with Oxygen Economic Ammonia Production LNG for Land and Sea Flexible Solutions for Wastewater Treatment 1/2004 e_01_03_Editorial_Inhalt_03 09.07.2004 11:33 Uhr Seite 1 Reports on Science and Technology Linde Technology Editorial Dear Readers, What are the processes involved in the modern concept development, design and construction of biotechnology plants – a rapidly growing industry, in which the time-to-market factor has extreme priority? Has the science of ergonomics already reached its limits in the design of forklift trucks? What are the options of chemical plant operators to respond to the ever-increasing economic challenges they face? These and other topical questions which are subject to the often conflicting influences of technology and economics are investigated in the newest issue of Linde Technology. The issue provides an overview of the wide spectrum of active research, which Linde considers to be just as important as the market-oriented further development of existing technical solutions. We hope to have put together another interesting mix of topics for a stimulating and enjoyable read and look forward to your feedback. Stefan Metz Head of Technical Press Linde AG e_01_03_Editorial_Inhalt_03 09.07.2004 11:33 Uhr Seite 2 Cover Photograph Innovative plant technology on the western coast of Norway. At this natural gas separation plant in Kolisnes we produce the environmentally friendly fuel LPG (Liquified Petroleum Gas). e_01_03_Editorial_Inhalt_03 12.07.2004 14:41 Uhr Seite 3 x 1/2004 Contents New Design Approach for Biotechnology Projects 4 Front-End Engineering for Pharmaceutical Plants Marc Reifferscheid and Dr. Karin Bronnenmeier Biotechnology Plant for Insulin Production 9 Planning and Construction of Complete Pharmaceutical Plants Jens-Peter Mendelsohn Swivel Seat Improves Ergonomics 15 How a Swivel Seat Affects the Ergonomics of Reverse Driving in Counterbalanced Forklifts Dr. Frank Schröder und Dr. Thomas Seitz Lantus Aventis: Tank station at the insulin plant in the Höchst Industrial Park near Frankfurt. Cracking with Oxygen 23 Approaches to Economic Solutions for Refineries Dr. Michael Heisel, Dr. Christer Morén, Prof. Dr. Alexander Reichhold, Andreas Krause und Antonio J. Berlanga-González The Location Makes the Difference 32 Economic Production of Ammonia Dr. Paul Kummann Offshore Plants for LNG Production 37 The Benefits of Cold Deep-Water for LNG FPSOs in Tropical Seas Eginhard Berger, Manfred Boelt und Bjørn Sparby LNG Travels Through China Linde Series 392 forklift with swivel seat. Using Available Resources Creatively Upgrading Industrial Wastewater Treatment Plants Dr. Manfred Morper Imprint Oxygen is used in one of Spain’s largest refineries, near Gibraltar. 44 LNG Baseload Plant in Remote North-West of China Max Bräutigam 50 e_04_08_Bronnemeier_11 09.07.2004 11:37 Uhr Seite 4 Marc Reifferscheid and Dr. Karin Bronnenmeier Front-End Engineering for Pharmaceutical Plants New Design Approach for Biotechnology Projects “Products come from technologies”. This sentence from the 2003 Biotechnology Report by Ernst & Young characterizes the current status of “red” (medically oriented) biotechnology in Germany and throughout the world. Biopharmaceuticals have already surpassed the classical chemically-synthesized active pharmaceutical ingredients [APIs] as newly approved medications. The Front-End Engineering Approach for Biotech Projects from Linde-KCA takes this development into account According to a new study by Frost & Sullivan, biopharmaceuticals, medications produced using biotechnological processes, reached sales of 41.3 billion dollars in 2002. That amounts to about 10% of worldwide pharmaceutical sales. It is noteworthy that a significant portion of these sales (about 6 billion dollars) is from medications based on EPO [erythropoietin], which have gained the top rank in the “hit list” of the most-often-sold medications. EPO, the hormone erythropoietin, is used to treat anemia. It is one of the best-known biotech products for the public. The potential for biopharmaceuticals becomes even more apparent from a glance at the product development portfolios of the pharmaceutical and biotech companies. This class of APIs has been dominant there for some time, with the result that biopharmaceuticals surpassed the classical chemically-synthesized APIs as approved medications for the first time in 2002. Construction of production plants has to keep up with this development, and must not become the limiting factor for introducing new drugs to the market. “Time to Market” is a critical factor for the commercial success of innovative biopharmaceuticals. The consequence for the plant designer is that the engineering activities for biotech projects must – in contrast with conventional plant construction projects – usually start while the client’s product and process development are still going on, so that they suffer from particularly great uncertainties. The engineering conversion of the client’s particular production process is extremely complex, and must comply with strict regulatory requirements. Because of that, the plant designer must have a thorough understanding of the process, coupled with biotechnological knowledge, to lead interdisciplinary cooperation between engineers and scientists. That is best accomplished by teams into which the client’s experts are integrated. 4 Linde-KCA’s Front-End Engineering Approach was developed specifically to make reliable evaluation of status, risks, costs and schedule possible in the early phase of biotechnological high-tech projects with the goal of fast-track realization. That can minimize risks and optimize the schedule for planning and construction of plants. The result is reliability of plans and decisions for both the client and for the plant designer. The need for front-end engineering services has been identified both on the route from the laboratory to biopharmaceutical production in biotech companies, as well as in established pharmaceutical companies. This has been confirmed in reference projects. Front-End Engineering Figure 1 presents a survey of the major services of front-end engineering by Linde-KCA: After getting the existing process and project information from the client at the beginning of the project, a technology concept is generated and then a block layout is developed. The technology concept together with the block layout make up the basis for an early estimate of the capital investment. In parallel with those activities, a process risk analysis can be carried out in close cooperation with the pharmaceutical company. The results of this analysis flow back into the process development. Development of a preliminary project schedule completes the documentation of the front-end study. Depending on the size and complexity of the project, such a study can be completed within three to six weeks. This gives the pharmaceutical company early information about costs, process and project status, and the schedule for the project. The documentation also makes up a reliable basis for the subsequent project phases, i. e. for the conceptual and basic design. Linde Technology I 1/ 2004 e_04_08_Bronnemeier_11 09.07.2004 11:37 Uhr Seite 5 Generation of the Technology Concept Know-how concentrated in the plant design company is a prerequisite for the fast and efficient generation of a technology concept, shown in Figure 2. To assure quick availability of the information required, this know-how is assembled in a library of plant unit models, e.g. for bioreactors or for centrifuges. This library contains functional descriptions as well as information about interfaces, procurement times, and costs of the various units. It makes up the basis for generating a technology concept. Project-specific plant units are defined on the basis of the process information from the pharmaceutical manufacturer. With the help of identification parameters for each plant unit, such as the working volume of a bioreactor, a plant unit model can be selected from the library and adapted to the specific project. In this way, the entire process is described with project-specific plant units. This serves as the basis for developing the block layout and the cost estimate. Additional Services for Transatlantic Technology Transfer Client’s Process & Project Information Technology Concept Review & Block Layout Development Investment Cost Estimate Process Risk Analysis for Scale up Preliminary Project Schedule Process Development Conceptual Design Basic Design Figure 1: Linde-KCA’s Front-End Engineering Approach. Linde Experts Linde-KCA‘s Sources of Know-how External Consultants Completed Projects Equipment Suppliers Library of Plant Unit Models (Functional Descriptions, Interfaces, Delivery Times & Costs) Linde-KCA Tools Block Layout Development Project A block layout brings together the preliminary space requirements for all the engineering disciplines such as process, process infrastructure, HVAC (heating, ventilation, air conditioning), electrical engineering and automation. Figure 3 shows the Linde-KCA concept for development of a block layout. Specific tools such as a layout-typical library for plant units and a layout planning handbook have been developed to assure an efficient design approach. A layout-typical for a plant unit shows the arrangement of the equipment and all the other space requirements for the unit, such as those for handling and logistics, in a plan view and side view. Every project-specific plant unit is assigned a layouttypical from the library, and is adapted to the specific project. At the same time, a functional program and an initial layout arrangement are worked out on the basis of the layout planning handbook. With this information, the layout-typicals can be assembled into a block layout. Estimation of the Capital Investment The total investment cost [TIC] can be estimated from the technology concept and the block layout by making use of benchmarking factors (Figure 4). The hardware costs for package units and other equipment are determined from the project-specific plant units. The hardware costs Linde Technology I 1/ 2004 Identification Parameters Technologietransfer-Processes Conversion into Unit Dokumentation Process Information from the Client Project-specific Plant Units for Block Layout and Cost Estimation Figure 2: Generation of the technology concept. Linde-KCA Tools Project Documents Linde-KCA Activities Project-specific Plant Units Assignment / Adaptation Layout-Typicals Layout-typical Library for Plant Units Functional Program Layout Arrangement Linde-KCA Layout Planning Handbook Project-specific Layout-Typicals Block Layout Figure 3: Block layout development. 5 e_04_08_Bronnemeier_11 11:37 Uhr Seite 6 Technology Facility Project-specific Plant Units Block Layout Project Scope for HVAC Project Scope for Cleanrooms Project Scope for Buildings Bulk and Constriction Factors Volume Factors Area Factors Area Factors Total Technology Costs HVAC Costs Cleanroom Costs Building Costs Eng. Factor for HVAC Eng.Factor for Cleanrooms Eng. Factor for Buildings Costs of Package Units Equipment Costs Hardware Costs Engineering Costs 09.07.2004 Eng. Factor for Package Units Eng. Factor for Equipment PCS-Costs Eng. Factor for PCS Engineering Costs for Technology Engineering Costs for HVAC/Cleanrooms Eng. Costs for Buildings for the process control system are determined likewise, with the complexity taken into consideration. The complete technology costs are obtained by adding the bulk and construction costs, estimated using benchmarking factors. The engineering costs can also be determined by using specific benchmarking factors for package units, equipment and process control systems. The block layout also serves as the basis for the facility costs. It is used to determine the scope of the project for HVAC, cleanrooms and buildings. Then the hardware costs can be determined from the HVAC volume factors and from the area factors for cleanrooms and buildings. The engineering costs for the facility are also determined by means of benchmarking factors. As a rule, this route for an early estimate of the total investment cost allows an accuracy of ± 30%. Total Investment Cost ± 30% Figure 4: Estimation of total investment cost using benchmarking factors. Process Risk Analysis Check requirements of the Linde standard against the client‘s process status Process steps Fermentation Cell Separation Concentration Cell Disintegration Process Risk Analysis Product Isolation High Resolution Purification Completed To be done Conceptual Design Process Risk Analysis ■ Review of development reports and process documentation ■ Verification of the scale-up requirements ■ Review design data ■ Establishment of the qualification requirements Test Program Process Development Figure 5: Process risk analysis No. Description 11 1 Kickoff 2 Conceptual Design 3 Milestone 4 Basic Design 5 Milestone 6 Extended Basic Design 7 Construction 8 Detailed Design 9 Procurement 2002 01 03 05 10 Installation 11 Mechanical Completion 12 IQ 13 Start-up / OQ 14 Handover Figure 6: Preliminary project schedule 6 07 09 11 2003 01 03 05 07 09 11 2004 01 Critical process steps and critical scale-up steps must be identified at an early stage for the plant design to be reliable. To accomplish that, an analysis of the developmental results and of the process documentation is carried out with experienced biotechnologists and bioprocess engineers. That includes a comparison of target and actual design data for the plant and an initial determination of the principal requirements for qualification. In this way, it is possible to determine which stages of the process have attained the status required for starting conceptual design, and which require deeper analysis. For the latter case, shown in orange in Figure 5, the measures necessary to reduce risk and to make sure of the scale-up are derived. They are documented in the form of a test program for process development, synchronized with the requirements for further plant design. Close interaction with the pharmaceutical manufacturer’s experts is critical for the process risk analysis described. Only close cooperation can assure feedback of results to process development and assure that the test program is translated into action. This is the responsibility of the client. Preliminary Project Schedule The last step of a front-end study is working out a preliminary project schedule. Figure 6 shows an example of such a schedule with reduced project time, under the boundary conditions of a fast-track project for a small plant with modular design. The major project phases and their durations are stated, from conceptual design and basic design to start-up and qualification. The finished Linde Technology I 1/ 2004 e_04_08_Bronnemeier_11 09.07.2004 11:37 Uhr Seite 7 front-end study forms a reliable basis for the fastest and technically best performance of the next project phases; the conceptual and basic design. Linde-KCA’s Phased Approach for Pharma & Biotech Projects versus Client’s Project Status Planning Phases Prozess & Utilities Basic Data Handling & Logistics International Technology Transfer Demands of the global market, cost advantages and regulatory aspects have become decisive criteria for the selection of a production site for pharmaceutical and biotechnical companies, too. As a result, new production plants are often built far from the centers of excelllence for research and development and for engineering. This requires international technology transfer, which often has a transatlantic dimension because of the leading position of the USA in medically oriented biotechnology, with all its consequences for the internal resources of the companies affected. Linde-KCA supports such technology transfer projects with know-how from front-end engineering, with particular in European plant design and construction, and with particular knowledge about authority Linde Technology I 1/ 2004 GMP & Safety Building & Layout Status of Delivered Documents: Basic Data Conceptual Design and Basic Design Completed Client Specialists and Linde-KCA Key Team Evaluate Project Status Requirements Concerning Completion of the Concept Design Conceptual Design Status of Documents: Conceptual Design Basic Design Focus: Process Figure 7: Conceptual design and basic design Investment Mature Biotech & Pharamceutical Companies Support The conceptual design should be strictly coordinated over all engineering disciplines simultaneously from the very beginning. That achieves high reliability for the technology, building size, and cost estimate, and assures the fastest possible development of all the necessary planning documents. To comply with those requirements, Linde-KCA has developed a phase model for pharmaceutical and biotechnical projects which predefines the work and document flow for each discipline and for the interfaces among the disciplines (Figure 7). At the beginning of the conceptual design, a core team, together with the experts from the pharmaceutical manufacturer, checks the documents provided and evaluates the status of the project for the various engineering disciplines. This provides an early focus on the planning jobs that are critical for completing the conceptual design. The conceptual design, worked out iteratively, should include all the relevant documents for the major disciplines, a detailed project schedule and a cost estimate that serves as a basis for management decision on the investment. The conceptual design makes up the reliable base for beginning the basic design. In the latter planning phase, the phase model is also applied as well as engineering tools tailored for the application to assure error-free know-how transfer from the conceptual design and to allow the fastest and best development of the basic design. Electrical Instrument. PCS HVAC Preferred Regions LINDE Technology-Transfer Process Project Evaluation Technology Transfer Documentation Engineering/ Contracting Site Evaluation Figure 8: International technology transfer Financing Models engineering and commercial aspects for future production sites in Europe (Figure 8). Front-end engineering, with analysis of the project and process status, risk, costs and schedules is of particular importance in technology transfer projects. The status analysis of a project, and especially of the process development, must also be performed at the site of the process development, as must the additional required development missions that are identified. However, on-site analysis at the selected site of the investment is preferred for costs and scheduling due to the decisive influence of regional factors and requirements. Complete and well-founded technology transfer documentation is the best basis for successful execution of the project at the selected site utilizing the proven 7 e_04_08_Bronnemeier_11 09.07.2004 11:37 Uhr Seite 8 engineering and contracting service spectrum from conceptual design to start-up and qualification. Success of the project can be assured by additional support in the search for potential financing and by cooperation in site evaluation. The Authors Marc Reifferscheid Summary Increased requirements on the pharmaceutical and biotechnological industries in past years have fundamentally changed the demands for pharmaceutical plant design and construction. Linde-KCA has developed know-how in front-end engineering and in international technology transfer to support early-phase decisions about investments, to minimize design risks, and to optimize the overall project schedule. This involves the following points in particular: ■ Experience in assembling integrated project teams involving pharmaceutical companies and Linde-KCA, as well as on-site planning ■ Established methods for early and accurate estimation of capital investment ■ Established capabilities and methods for process risk analysis ■ Tools for tailored adaptation of the work flow to a specific project ■ International experience with highly varied projects ■ Knowledge of the relevant laws and regulations of quite different countries These capabilities and experiences facilitate very close cooperation with the experts from the pharmaceutical company in the earliest possible phase of the project. The objective is to offer capabilities tailored to the specific project and thus to assure the fastest and best planning and execution of high-tech projects. Dipl.-Ing. Marc Reifferscheid has studied bioprocessing technology and has since 1999 been a Senior Process Engineer at LindeKCA-Dresden GmbH. In this position, he has been involved in planning and building of numerous pharmaceutical plants in Germany, Denmark and Hungary. Before joining Linde-KCA, he worked at the Instituto Superior Técnico, Lisbon, Portugal from 1997 to 1998 and, in 1996, at TIBRAS Titânio do Brasil in Camacarí, Brazil. Dr. Karin Bronnenmeier Dr. rer. nat. habil. Karin Bronnenmeier has been Senior Process Biologist at Linde-KCA-Dresden GmbH since 2001. She works there in business development for pharmaceutical plants emphasizing biotechnology. Before joining Linde-KCA, Ms. Bronnenmeier was active in fundamental biotechnology research as Research Group Leader for molecular enzymology at the Technical University, Munich, where she qualified as a lecturer. Abstract Biotechnology is continuing to develop at break-neck speed. Biopharmaceuticals have already overtaken the conventional chemically synthesized medications in new medical approvals. Construction of production plants must keep up with this development, and must not become the limiting factor in the introduction of innovative medications to the market. With the FrontEnd Engineering approach of Linde-KCA we now have a concept that assures the fastest and best planning and execution of high-tech projects in biotechnology. 8 Linde Technology I 1/ 2004 e_09_14_Mendels_11 09.07.2004 11:41 Uhr Seite 9 Jens-Peter Mendelsohn Planning and Construction of Complete Pharmaceutical Plants Biotechnology Plant for Insulin Production About 800,000 people in Germany alone must use insulin every day to control their blood sugar level. Because of the speedy progress of genetic technology in the past 30 years, it has become possible to make almost unlimited amounts of human insulin and insulin analogs with consistently high quality using genetically modified microorganisms. The most modern biotechnological production plants are needed to meet the demand for innovative analogs, such as glargin insulin. Planning and construction of pharmaceutical plants is a relatively young field of business for Linde Plant Construction, located in Dresden at Linde-KCA (LKCA). Even so, LKCA has, over the past ten years, established itself as a leading bidder in the design, planning and construction of pharmaceutical and biotechnological plants. The latest reports are among others which confirm that: the Bayer AG pharmaceutical pilot plant in Elberfeld, which started up in October 2000, and the new F. Hoffmann-La Roche AG “Kilolabor” in Basel, are both plants for synthesis of innovative pharmaceuticals. LKCA obtained its first contract, as general planner for construction of a new biotechnology plant, in April, 2000. The plant, for Aventis Pharma, Frankfurt, is to produce glargin insulin, a genetically engineered insulin derivative with depot action. LKCA turned the plant over to Aventis only 29 months later, after successful completion of the start-up activities, for continuation of special performance tests by Aventis, and for the beginning of Validation (Figures 1 and 2). A stand-alone plant on the ‘green-field’ The plant is an independent complex consisting of four buildings among which the production and auxiliary operations are distributed on the basis of their requirements for classified cleanrooms and for areas and processes with explosion protection. All the subplants are supplied through a central connecting road. The road serves the flow of both personnel and materials. It also connects the central main building to the offices, conference rooms, laboratories and the real heart of the plant, the central control room. There are also storage tanks as well as the entire infrastructure with lines from the plant to medium supply and disposal at the Höchst Industrial Park (Figures 2 and 5). The plant produces about 1,700 kg of glargin insulin every year. The total investment is around 200 million euro. Because its amino acid sequence has been altered by genetic engineering, this insulin derivative has depot action for 24 hours in the body. Glargin insulin was developed independently by Aventis, which was the client for this project. The product is made in the subplants for fermentation, processing, purification and end-product treatment. The production complex also includes supply and disposal installations (propanol distillation plant, auxiliary material preparation, wastewater disposal, tank storage and a warehouse for packaged goods). Figure 1: Lantus Aventis at the Höchst Industrial Park. Linde Technology I 1/ 2004 9 e_09_14_Mendels_11 09.07.2004 11:41 Uhr Seite 10 Production in five steps The plant for producing biosynthetic glargin insulin is organized into these five processing sections: ■ Fermentation ■ Processing 1 ■ Processing 2 ■ Purification, and ■ End product treatment. The steps in the process for producing glargin insulin differ from those for producing rapid-acting insulins essentially in purification and end product treatment. The fermentation and processing stages use similar production technology. In the fermentation stage genetically modified Escherichia coli bacteria are cultured and grown in successive fermenters with increasing capacities. They are stimulated to produce a fusion protein by adding an inducer. When the fusion protein has been formed, the bacteria are killed with a disinfectant. This part of the process – culture and killing of genetically modified bacteria – is covered by the Genetic Technology Act, and is the genetic engineering part of the process. After killing, the bacterial suspension is concentrated by centrifugation during processing, and then disintegrated. After disintegration, the specific heavier fusion protein is separated from the other cell components by continuous centrifugation, washed twice with water, and isolated. The “folding” of the molecule, in which the fusion protein folds into a “native spatial structure” takes place in the next step of the process. That is accomplished by dissolving the fusion protein in an aqueous urea solution in the presence of cysteine. Byproducts are then precipitated by a shift in the pH and separated by centrifugation. In the subsequent purification process, the glargin insulin is freed of urea and inorganic salts by adsorption-desorption steps and simultaneously concentrated. The desorption is done with an aqueous solution of propanol. The crude intermediate product is then purified chromatographically in two process steps. Then it is crystallized at high purity, filtered off by suction, and stored temporarily as moist crystals. The high-purity glargin insulin crystals are dissolved again, crystallized and freeze-dried to get larger homogeneous batches. The end product treatment is done in Class C cleanroom conditions because there are limits for the bacterial counts and endotoxins (“pyrogens”) in the dried material. The plant, which only produces the solid pharmaceutical, not the finished medicinal product, had to be approved not only under the Genetic Technology Act but also under the Federal Environmental Protection Act. Figure 2: View of the tank storage, with the propanol distillation plant and the fermentation buildings (right). 10 Linde Technology I 1/ 2004 e_09_14_Mendels_11 09.07.2004 11:41 Uhr Seite 11 Diabetes mellitus Diabetes mellitus is one of the most common and significant metabolic diseases in humans. It can be caused by malfunction of the pancreas, which produces too little insulin, or by insulin resistance of the body cells. The result is that the concentration of sugar in the blood is no longer normal and must be corrected by medications. Although the pancreas of a healthy person produces 2 mg of insulin a day, a diabetic must get about 1.5 mg of insulin per day externally. In Germany alone, about 800,000 people must take insulin daily to adjust their blood sugar levels. According to an estimate by the World Health Organization (WHO) there are at least 120 million diabetics in the world today, including undiagnosed cases. More than Exacting production process All the steps of the process use deionized and purified water. Its microbiological and physical-chemical quality are monitored routinely (“Purified water”). Ozone is removed from the water with UV light before it reaches the points where it is used. The consumers are supplied with process water through a ring piping system. “Pyrogen-free purified water” produced by ultrafiltration is used In the last steps of the process (purification and end product treatment). In addition, all the raw materials and auxiliary materials (solvents, acids, bases, salts, buffers, etc.) are of pharmaceutical quality or meet internal plant specifications, and are checked regularly. The solvent, propanol, is distilled after use and returned to the process. The nitrogen used for inerting is purified through a HEPA (high-efficiency particulate air) filter before use to prevent contamination by particles. As a rule, all the stages of the process are carried out in closed tanks with solid pipe connections. Manual handling of the product is reduced to a minimum. The material used for the tanks and connecting pipes are corrosion-resistant stainless steel with defined surface roughness (internally electropolished for critical process steps). This assures good and complete cleaning. These cleaning processes have a critical part in the planning of pharmaceutical plants to assure the required sterility and constant product quality. They are generally established by the client in the corresponding specifications. Linde Technology I 1/ 2004 8 million patients are being treated with insulin in the industrialized countries alone. The worldwide insulin consumption is around five to six tons per year. It is predictable, too, that this requirement will increase substantially in the future, in view of the worldwide increase in population and changing social structures. Process monitoring and documentation The heart of the glargin insulin production plant is a high-capacity process control system, which processes about 18,000 items of input and output information and monitors and controls the entire process. All quality-related instruments were calibrated when they were put into service. This calibration is repeated regularly at appropriately established intervals. In-process samples of the intermediate products are also tested in the laboratory to make sure that their quality meets specifications (Figure 4). All the process and analytical data, either from routine operations or from deviations, are stored in the process control system or in the laboratory information management system (LIMS). Then the specific process data are supplemented with the general operating data and printed out either in partial lot records or a full lot record, along with the operational data. Conceptual design – the key to success This project began with a process which Aventis developed and tested at pilot plant level, which had to be converted to a full-scale plant. Utilization of the structure of a similar project for insulin production, which had already been accomplished, was a significant boundary condition. In this phase, the plant designers were expected to apply their experience and know-how to work out the technological problem and to participate actively in the design process, especially with respect to process optimization, scaling up and the defined production concept. 11 e_09_14_Mendels_11 09.07.2004 11:41 Uhr Seite 12 Figure 3: The 3D design model and the structural plan of the plant for producing a genetically engineered insulin derivative with depot action. Tank Storage Main Building Laboratory and Administration Processing Connecting Structure – Central Supply Route High Purification Purification and End Product Treatment The major emphasis in planning is that of making sure that the product is produced under reproducible conditions (with respect to product quality and validation). LKCA set up a team of 12 employees specifically for the project. They worked together with the Aventis project team for three months following April 2000, under a preliminary contract, to develop the initial plant concept. In this planning phase, it was important to bring the process technology into harmony with these requirements: ■ building architecture ■ cleanroom classification and HVAC (heating, ventilation, air conditioning) ■ logistics and storage ■ information technology and process control ■ functional descriptions as the basis for the formulating operation ■ product quality and GMP (Good Manufacturing Practice) concepts. Plant-oriented design planning was established, starting from the basic data determined and specifications of process-oriented design principles (Figure 3): 12 Fermentation Warehouse and Supply ■ functional layout ■ building concept ■ master plans for beginning-to-end quality assurance, from design to start-up ■ plant design, connected to the existing infrastructure. This process was essential so that the documents needed for the construction contract and for the Federal Environmental Protection Act contract could be produced as quickly as possible on an assured design basis and submitted to the authorities responsible. The conceptual design produced in that manner was based on: ■ Layout studies of critical plant areas (e. g., recrystallization) with respect to structure, GMP requirements, material handling, cleaning, and serviceability ■ layout designs for medium supply and disposal, electrical engineering, measurement, and control technology, as well as HVAC ■ logistical concepts with respect to material, personnel and product flow. Linde Technology I 1/ 2004 e_09_14_Mendels_11 09.07.2004 11:41 Uhr Seite 13 Figure 4: Quality monitoring by offline-testing in the analytical laboratory in the main building. Insulin glargin (Lantus®) Glargin insulin is a peptide hormone analog of insulin, produced by genetic engineering. Like human insulin (NPH insulin), it controls the blood sugar balance in humans, but has a long-term action profile. Because of that, a diabetic needs only one insulin injection per day. That is a definite improvement from the patient’s viewpoint. This makes glargin insulin a product that is tailor-made for diabetics. Together with the classical human insulin Serum insulin level (with a relatively fast action profile), it rounds out the patient’s requirement. The active pharmaceutical agent, glargin insulin (solid, nonsterile) is dissolved in the finished medicinal production and supplemental materials are added. The product is filled into cylindrical ampules, and marketed under the tradename Lantus®. Glargin insulin, like human insulin, is a protein molecule, which reduces the blood sugar content to the normal value of about 70 to 120 mg/100 ml. In current treatment of diabetes, human insulins play an important role in the basal supply of insulin for the body. NPH insulins exhibit a distinct maximum action after about 4 to 7 hours, and have an effective duration of about 12 to 16 hours. In contrast, glargin insulin does not have any distinct maximum activity, instead having a constant flat action profile over 24 hours. Previous NPH insulin Now: LANTUS® The difference in the action profile between glargin insulin and NPH insulin. Linde Technology I 1/ 2004 Time 13 e_09_14_Mendels_11 09.07.2004 11:41 Uhr Seite 14 Figure 5: View to the southeast of the complete Aventis Lantus plant at the Höchst Industrial Park, with the central office building in the foreground. Successful plant start-up Aventis received approval for construction promptly in December 2000, so that the groundwork could be started in January 2001. Laying of the cornerstone was celebrated jointly on 4 April 2001. Then it was possible to complete the rough construction work, an important project milestone, in August 2001. The mechanical preparation was done building-by-building, oriented to the course of the process, until August 2002. After finishing the assembly, we could then begin the start-up in the same sequence. The plant is distinguished by a high degree of complexity. It is a large-scale computer-controlled plant. Just one single person at the control room computer screen could manage it. It does require about 6,500 control points (connections to the process control system). The start-up of the entire process control system ran smoothly. By February 2002 the technology was available to start operations of the individual sections of the plant. Abstract The total investment is about 200 million euro. The centerpiece of the glargin insulin production plant is a high-capacity process control system, with which about 18,000 pieces of input and output information are processed, and which monitors and controls the entire course of the process. The Author Jens-Peter Mendelsohn Jens-Peter Mendelsohn, a graduate engineer, has been manager of the Development Department in the area of Pharmaceutical Plants since 2003. He started at Linde in 1980, as manager of plant design, and, after 1987, he coordinated use of CAD technology for pharmaceutical plants. Since 1990 he has been responsible for the Central Engineering Technology Department. From 1995 to 2002 he directed various major projects, including the construction of the new Aventis insulin plant. Within ten years, Linde-KCA-Dresden GmbH has established itself as a leading bidder in the area of pharmaceutical plants. In just 29 months, under a contract with Aventis, Linde built a plant to produce genetically engineered insulin (glargin insulin). The plant produces bout 1,700 kg of glargin insulin annually. 14 Linde Technology I 1/ 2004 e_15_22_SchroeNEU_11 12.07.2004 14:43 Uhr Seite 15 Dr. Frank Schröder, Dr. Thomas Seitz and Jörg Hudelmaier How a Swivel Seat Affects the Ergonomics of Reverse Driving in Counterbalanced Forklifts Swivel Seat Improves Ergonomics Frequent reverse driving places a significant strain on the operators of counterbalanced forklifts. A joint study conducted by Linde AG in Germany and the Institute of Ergonomics at the Technical University of Munich shows how the ergonomics of forklifts with double-pedal control can be improved by the use of swivel seats. When moving bulky loads or traveling down an incline, a forklift operator is often forced to drive in reverse for relatively long stretches. This comes in addition to other common instances when driving in reverse is required, such as loading and unloading goods. The position that the operator is forced to assume is uncomfortable when held for longer periods of time and his field of vision is also greatly restricted during these maneuvers. Surveys have found that, depending on specific circumstances, forklift operators spend between 6% and 73% of their time driving in reverse with counterbalanced forklifts. Out of 162 cases studied, an average of one third of the total driving time was spent going in reverse [2]. Medical studies have documented increased muscle tension as rates of reverse driving increased. Drivers with approximately ten years of experience or more under such conditions also had an increased rate of back-related complaints [3]. Several design-based solutions already exist for improving the posture of forklift drivers when driving in reverse. For example, the field of vision to the rear can be increased using various rear-view mirror arrangements; however, as in street traffic, a look to the rear at the beginning of each turning operation remains essential. Swivel seats Another solution is the driver’s seat on the Linde 336 series of electric forklifts, which turns 45° on the “Panorama” version. On this model, when the driver needs to go in reverse, he rotates the seat, including the Linde Load Control (LLC) control lever (located on the arm supports attached to the chair), 45° to the right. An additional set of pedals is available in this swiveled position for driving and braking. The steering wheel remains fixed in place for both seat positions, where it continues to meet ergonomic requirements in the swiveled position as well. The proven benefits of the “Panorama” solution are accompanied by high design costs. In addition, the increased space required by this concept for the second set of pedals and the seat’s wide turning range is not practical for more compact vehicles and current forklifts powered by internal combustion engines. Figure 1: Linde forklift series 392 H25D and swivel seat for Linde forklift series 392 with a rotational range of 17° to the right. Linde Technology I 1/ 2004 15 e_15_22_SchroeNEU_11 09.07.2004 11:46 Uhr Seite 16 For this reason, a second variant of the swivel seat was designed that could forgo an additional set of pedals (Figure 1). The driver uses the fixed-position pedals even when the seat is turned. A maximum angle of 17° proved to be most suitable for the swivel seat, providing the best compromise between the ability to comfortably reach the pedals, an improved rear field of vision and compatibility with available special equipment. The angle also reduces the amount of static twisting to which a driver’s spinal column is subjected. The swivel seat’s point of rotation was chosen so that the distance between the edge of the seat and the reverse driving pedal remains nearly constant both in the non-swiveled and in the swiveled position. To rotate the seat, the driver activates a lever (conveniently integrated into the front area of the arm support) that disengages the seat’s position lock and permits the seat to turn. The ergonomics of different variants of the swivel seat were studied jointly by Linde AG and the Institute of Ergonomics at the Technical University of Munich. Driving tests and three-dimensional human models provided detailed information on the positions assumed by individual joints in the human body. Some of the results from this study are described below. Vehicle variants and the test course In driving trials, three vehicle-mounted video cameras were used to observe the test subjects (who had different body and waist sizes) while they performed their driving tasks. In addition, the drivers were questioned as to their impressions after each test drive, and the drivers’ individual seat adjustments were noted in order to statistically record how drivers used the available turning range. Forklift Lifting mast control Driving control Type of seat FS 392 LLC (“joystick”+arm support) Linde double-pedal Standard seat LLC (“joystick”+arm support) Linde double-pedal Swivel seat (17°) LLC (“joystick”+arm support) Single-pedal with direction of travel switch Swivel seat (17°) Linde double-pedal Standard seat FS 392 FS 392 FS 351 Mechanical actuation of control valve, no arm support The investigators compared three different versions of the forklift series (FS) 39X. Compared to their predecessor (FS 351), the FS 39X vehicles offer numerous improvements to ergonomics and operator comfort in addition to a standard electronic lift-mast controller that takes the form of a “joystick” integrated into the right arm support. This arrangement reduces the actuation forces and paths required by the driver while simultaneously improving the sensitivity of the controls [4]. The test vehicles differed in their control pedal arrangements and in the types of driver’s seats (standard fixed seat or swivel seat that can be turned to the right to make reverse driving easier). For comparison purposes, the driving maneuvers were also performed with a forklift from the earlier BR 351 series. From the driver’s perspective, it is characteristic of a hydrostatic drive that the vehicle’s driving and braking functions are controlled using a single pedal. The driver accelerates the vehicle both forward and backward up to its maximum speed by altering the position of the pedal. Unlike with other vehicles, there is no need to switch the foot between drive and brake pedals; hydrostatic braking is activated simply by the driver removing his foot from the pedal. Linde combustion engine forklifts come with Linde’s double-pedal system as standard equipment: there is one pedal each for forward and reverse driving, with the two pedals mechanically linked to one another. The driver uses the right foot to accelerate and brake in forward driving while the left foot controls these functions when driving in reverse. As a result, the driver’s feet can rest continuously on the pedals and he can very quickly change driving directions. If desired by the customer, these forklifts can also be delivered in a single-pedal version with a supplementary directional switch in the arm support. During the driving trials described here, both pedal variations were tested in order to assess their suitability for use with the swivel seat. The driving task consisted of transporting a load on a defined course. In addition to a straight section to be covered in forward and reverse, the course also consisted of two right-angle curves, and at the end of the course, the driver had to operate the vehicle’s lifting hydraulics. The straight section was narrow enough that the drivers could only go straight ahead or in reverse. In order to bring about comparable head positions, a pylon placed directly behind the vehicle was used as a visible target on this section of the course. Video recordings were taken of this section. Stills from the video recordings were used to create three-dimensional assessments of the drivers’ postures when traveling forward and in reverse. Table 1: Vehicle designs for ergonomic comparisons 16 Linde Technology I 1/ 2004 e_15_22_SchroeNEU_11 09.07.2004 11:46 Uhr Seite 17 The test drivers and the human model The group of test drivers consisted of 13 persons with different body sizes and waist circumferences in order to cover typical segments of the population. All test subjects were measured digitally using still images. The “human model” (RAMSIS) for each test driver was individually adapted to the anthropometrical data obtained from these images. As a result, each of the human models corresponds, in all dimensions, to the extremities of the respective test driver and can be displayed and moved three-dimensionally on a computer with the aid of PCMAN software [6]. In doing so, all joints can be rotated according to their actual physiology. In connection with the tests, the human models were adjusted three-dimensionally to conform to the actual driver positions recorded on video. During this adjustment, all of the models’ joints were adapted translationally and rotationally to the various camera perspectives so that every position assumed by the driver (during reverse driving, for example) could be recreated (Figure 2). In this manner, it was possible to spatially measure the driver’s positions while he was driving without impairing his normal range of motion with measurement devices attached to his body. Using measurement functions, it was then possible to use PCMAN to evaluate the angular positions of each human joint depicted in the human model and compare these to known comfort angles and critical angles for that joint. The body positions calculated in this manner for all of the test drivers were stored in a database of positions. The database contains the driver’s position in the form of 29 joint angles as well as the coordinates of the hip reference point for each time point evaluated during the test. Using these data as a foundation, conclusions can be drawn as to the comfort of a given body position and the accessibility of operational elements, as well as lineof-sight analyses. Influence of the arm supports The technical literature describes numerous requirements for seated workstations [e.g., 5]. Under different conditions, an arm support is seen as mandatory in the following situations: ■ When the task is fulfilled primarily in a sitting position with one's back against the backrest. ■ During precision fine-motor work when the elbows or the lower arm need support independent of the working surface. ■ When working in dynamic systems (ground, air, and water vehicles), the arm supports can contribute to minimizing the translational and/or rotational accelerations affecting the body. Linde Technology I 1/ 2004 From an ergonomic perspective, an arm support design must meet the following criteria: ■ If there is no variable height adjustment, the height (h) between the supporting surface of the arm support and the sitting reference point must be 180 mm < h < 230 mm. ■ The width of the supporting surface may not be less than 50 mm. ■ Arm supports that are coupled to angle-adjustable backrests must be designed so that they can be brought into a horizontal position independent of the angle assumed by the backrest. ■ Arm supports should be padded in order to minimize the surface pressure, especially in the elbow area. These points show that it is reasonable, from an ergonomic standpoint, to expect significantly improved sitting comfort and a reduction of muscle fatigue using an arm support in forklifts. The arm supports used on Linde forklifts meet these requirements. Figure 2: Adapting the human model to the actual position during reverse driving. There are two typical positions associated with driving in reverse. The majority of the test drivers generally retain the arm position used during forward driving and turn themselves using only the pelvis and the spinal cord. However, in about 10% of the driving instances, it was observed that the driver reaches with the right arm behind the B-pillar, using his arm as a supporting lever. It was also revealed that in vehicles without an arm support, about one third of the test drivers moved their right hand significantly away from the normal position during reverse driving. In other words, they released their grip on the hydraulics control lever. With only one exception, however, drivers operating an FS 392 (equipped with an arm support) kept their arm on the arm support so that the right hand always remained in the immediate area of the operational controls. 17 e_15_22_SchroeNEU_11 09.07.2004 11:46 Uhr Seite 18 ❾ ❽ ❼ ❻ ❺ ❹ ❸ ❷ ❶ FS 351 FS 392 standard seat FS 392 swivel seat, double-pedal FS 392 swivel seat, single-pedal Figure 3: Average positions of all drivers during reverse driving. 9 Rotation of neck joints and head 8 7 ■ Relief of the spinal column 6 5 Rotation of the spinal column 4 3 2 Pelvis rotation on the seat Figure 4: Spinal column rotation in the individual spinal column joints of the human model, average values of all drivers. 1 Seat swiveling range -40 -30 -20 -10 0 FS 392 standard seat, double-pedal Under practical conditions, this hand position represents an increased level of safety because the driver has faster access to the operating lever in critical situations. Furthermore, the arm support significantly relieves the arm and shoulder muscles, which contributes to a noticeable stress reduction for the driver both when driving forward and in reverse. Body position when driving in reverse When the angular data of each body joint (available in the position database described above) are averaged individually, the result is an “average position” of all drivers for each vehicle in a given driving situation. RAMSIS was used to convert these positions into a visual 18 10 20 30 40 FS 392 swivel seat, double-pedal 50 60 70 80 90 100 110 Rotational angle [°] FS 392 swivel seat, single-pedal depiction from which the influence of the various vehicle designs can be inferred. Figure 3 shows the average positions during reverse driving. The positional data also permit conclusions regarding the stress placed on individual regions of the body while driving a forklift. The rotation of the spinal column is of special interest. Figure 4 shows the spinal column joints that were incorporated into the model. Of these, the ribcage (7) should be perceived as rigid, while the neck and head joints (8 and 9) exhibit high degrees of movement and are subject to major positional changes depending on the line of sight. A rotation of the pelvis (1) means that the driver is rotating with his entire body relative to the seat surface. In other words, the person is changing his position in space. Therefore, the rotational angles in the spinal column area between the lumbar sacrum (2) and the lower cervical vertebrae (7) are definitive for the analysis of the sitting Linde Technology I 1/ 2004 e_15_22_SchroeNEU_11 09.07.2004 11:46 Uhr Seite 19 Forklift Seat swiveling range Rotation of the spinal column (2 through 7) Rotation of the lower lumbar spinal column region (2 through 4) Total rotation between the driver’s head and the vehicle (1 through 9) FS 392 standard seat, double-pedal 0° 31° 19° 124° FS 392 swiveling seat, double-pedal 17° 21° (-32%) 12° (-37%) 130° (+5%) FS 392 swiveling seat, double-pedal 17° 23° (-26%) 20° (+5%) 129° (+4%) Percentage values indicate changes relative to standard seat Table 2: Total rotation between head and vehicle, average values of all drivers Seat/pedal type Driver model Angle of line of sight [°] Field of vision [°] Change relative to standard seat [°] Standard seat, double-pedal, left pedal depressed Very tall, corpulent Average Very short, corpulent 117 116 114 57 - 177 56 - 176 54 - 174 - Swiveling seat 17°, double-pedal, left pedal depressed Very tall, corpulent Average Very short, corpulent 123 124 126 63 - 183 64 - 184 66 - 186 +6 +8 +12 Swiveling seat 17°, single-pedal, pedal (right) depressed Very tall, corpulent Average Very short, corpulent 114 116 109 54 - 174 56 - 176 49 - 169 -3 0 -5 Table 3: Line-of-sight angle for various body models position. When a person drives a vehicle, the joints in this area are exerted primarily in a static posture, which causes the muscles to tire rapidly. From an ergonomic point of view, small rotational angles in the joints are better than large ones. In addition, rotating the pelvis on the seat so that the driver sits at a slight angle to the driving direction is better than rotating the joints of the spinal column relative to one another. The forces and therefore the individual strain increase very disproportionately as the body approaches its physiological limits of movement. The result is that the driver fatigues significantly faster when assuming larger angles of rotation. In addition to other physiological and biomechanical effects such as fatigue, agitation, pain, etc., the strain of muscle groups, in particular, contributes to perceived discomfort. It is also important to note that as a person ages the maximum possible rotation of the person’s joints decreases greatly. Therefore, reducing joint rotation is beneficial, especially for older people. Linde Technology I 1/ 2004 The lower section of the lumbar spinal column (depicted in the model as joints 2 through 4) is especially important for assessing posture. This area is well known as the starting point for common lower back pain [1]. As described above, it was possible to use the drivers’ body postures (measured during the driving tests) to obtain a realistic driver posture during reverse driving in the form of a human model and depict this through the use of the RAMSIS software. By altering the body proportions in RAMSIS, it was also possible to reach conclusions about the probable sitting positions and resulting fields of vision for people with extreme body proportions that were not represented in the group of test drivers. For these advanced position analyses, two extreme body types (corpulent and very tall or corpulent and very short) were used in addition to an average body type. 19 e_15_22_SchroeNEU_11 09.07.2004 11:46 Uhr Physiological Limits Joint Seite 20 Minimum Maximum Hip joint (A) -36° 60° Knee joint (B) -20° 40° Ankle (C) -24° 23° Figure 5: Critical rotations of the leg joints, physiological limiting angles. 20° Reverse driving, standard seat, double-pedal Reverse driving, swivel seat, double-pedal Rotation around the X-axis [°] 15° Reverse driving, swivel seat, single-pedal Forward driving 10° 5° 0° 5° Hip joint Knee joint Ankle joint Figure 6: Rotation of the right leg during reverse driving (“average man”). The disadvantages of single-pedal control The results of the analyses indicated that when operators drive in reverse with a non-swiveling seat, they turn in the seat by an average of 22°. In addition, they rotate themselves approximately 102° around the spinal column and the head joint. With a non-swiveling seat, the total rotation of the head relative to the vehicle averages 124°. These results are shown in Figure 4 and table 2. The rotational angles in the individual joints show how the swivel seat, when combined with the doublepedal control, relieves the spinal column by approximately one third in the region between the lumbar sacrum (2) and the rigid ribcage (7). When compared to a standard seat, a seat that rotates 17° can reduce the rotational burden by more than one third (37%), especially in the important lower lumbar spinal column region (joints 2 through 4). On double-pedal vehicles, the extra angle of 20 rotation increases the total rotation of the driver’s head relative to the vehicle’s longitudinal axis to 130°, thereby increasing the field of vision. In this vehicle variant, the pelvis is rotated a distance relative to the seat surface that is similar on average to a standard seat, which does not place a burden on the spinal column. In contrast, on a vehicle with a swivel seat and single-pedal control, the driver is able to rotate his pelvis considerably less (compared to the other vehicle variants) relative to the seat because he must use his right foot to operate the pedal, which on this vehicle variation is located on the right, when driving in reverse. The driver must compensate for this lack of pelvic rotation by rotating the lower lumbar region of his spinal column in a manner similar to that encountered with a standard seat. Thus the tested combination of a swivel seat and single-pedal control did not prove to help relieve this area of the back, which is associated with lower back pain. In fact, reduced rotational strains were recorded with the single-pedal vehicle only in the upper section between the upper lumbar vertebra and the ribcage (4 through 7), an area where the spinal column rotates almost not at all. Instead, a greater rotation of the head in the region of the cervical vertebrae is observed, which resulted in the highest recorded values in a comparison of the vehicle variants. The recorded values range up to the physiological limits of these joints. Consequently, a vehicle equipped with single-pedal control provides the driver only incomplete relief with regard to rotation of the spinal column. The relief that is provided is distributed unevenly along the back, the cervical vertebrae reach the limits of their flexibility and there is no improvement in the lower section of the lumbar spinal column. When driving in reverse, it is not only the torso that rotates, but also the area of the hips and legs. The advanced positional analysis evaluated leg rotations in the hip joint, knee joint, and ankle joint for typical body proportions. The rotations of the joints shown in Figure 5 were used as the critical measure of discomfort because these joints move closer than the others to their physiological limits of flexibility. Due to the location of the seat’s point of rotation relative to the vehicle frame, the left leg has nearly a constant distance from the pedal during the swiveling motion, so that the most clearly noticeable angular changes appear on the right side of the body. Compared to forward driving with double-pedal control, only small differences to the ergonomically favorable leg position appear with the standard seat and the swivel seat with double pedals, Figure 6. With double-pedal equipment, the driver uses his left foot for both acceleration and braking when driving in reverse. On a single-pedal vehicle, however, the driver must stretch the right side of his body to a great degree in order to operate the pedal, which is located on the right. This results in a rotation (particularly in the right hip joint) compromising the driver’s comfort in the length of time. Therefore, with regard to the position of the legs, the combination of a swivel seat and double-pedal control produces the best sitting comfort Linde Technology I 1/ 2004 e_15_22_SchroeNEU_11 09.07.2004 11:46 Uhr Seite 21 Standard seat Left eye Increased field of vision Right eye Reduced rotation of the spinal column Adjustable range of the swivel seat Swivel seat Left eye Right eye Figure 8: Overlay of the reverse position on the 392 forklift series, comparison between rotated and non-rotated sitting positions. Figure 7: The average man’s fields of vision when driving in reverse, unmoving eyes, comparison between the standard seat and the swivel seat, both with doublepedal control. Visibility to the rear Besides to the driver’s posture, the combination of seat and pedal arrangement also affects the driver’s line of vision. In addition to RAMSIS, the analysis functions from pro/ ENGINEER were used to analyze the visibility. In these tests, the RAMSIS model was configured with various lines of sight, for example: in the direction of the driving path behind the vehicle or in the direction of a possible obstacle behind the vehicle at a medium height. The line of sight was different from the driving task for the test drivers, for whom a fixed visual target was provided during reverse driving. Therefore, while the test drivers’ visual target was constant and they had to adjust their body positions accordingly, the RAMSIS simulation calculation described here assumes a uniform strain on the body’s joints and compares the resulting lines of sight for the different vehicles. As a result, the calculated lines of sight are different from the measured head positions of the test driver group, depicted above. Figure 7 shows an example of the lines of sight of both eyes for the “average man” body stature as determined by the RAMSIS visual analysis; the rotation of the eyes is not reflected in this depiction (“fixed gaze”). The fields of view visually depict the increased vision to the rear that is achieved by the swivel seat in combination with double-pedal control. To evaluate the actual vision in the vehicle, it is necessary to also take into account the rotational range of the eyes. Eyes typically have a horizontal range of movement from –60° to +60°. Assuming typical body Linde Technology I 1/ 2004 rotation with a standard seat, the driver has a field of vision of approximately 180° to the direction of travel, so that the driver can just discern objects that are located in a straight line behind the vehicle. In contrast, the swivel seat increases this field of vision (as already determined in the examinations of body position) into the region to the rear and diagonally to the left while simultaneously reducing the rotation of the upper body. In order to reach quantitative conclusions in addition to this qualitative evaluation, pro/ENGINEER was used to calculate the angle of the line of sight with respect to the vehicle median level for the calculated body positions. As shown in table 3, equipping a vehicle with a swivel seat and double-pedal control increases the field of vision to the rear, depending on the size of the driver, by +6° to +12° compared to a fixed seat. As a result, the field of vision includes much more than 180° for drivers of all sizes. However, a swivel seat combined with single-pedal control does not result in improved vision during reverse driving due to the need for the driver to stretch the right half of his body. The total rotation of the spinal column is also reduced only slightly (see table 2). This becomes even more apparent for drivers with extreme body measurements (corpulent and very tall or corpulent and very short; in other words: in both cases with relatively short leg lengths). In these cases, the need to operate the driving pedal with the right foot on vehicles equipped in this manner even leads to a slight limitation of the field of vision (up to –5°, see table 3) when driving in reverse. In summary, it can be shown that vehicles equipped with a swivel seat and the Linde double-pedal control also represent the best variant ergonomically as regards the resulting field of vision. 21 e_15_22_SchroeNEU_11 09.07.2004 11:46 Uhr Seite 22 Summary As depicted in Figure 8, assuming the same general conditions, the swivel seat with an adjustment range of 17° offers the following advantages over a non-swiveling seat: ■ The operating elements swiveling together with the arm support (a necessary prerequisite for a swivelseat vehicle) increase the driver’s comfort by reducing muscle fatigue during both forward and reverse driving. ■ Compared to a non-swiveling seat, the sitting comfort increases significantly, which can be demonstrated with approximately one-third (double-pedal) or onequarter (single-pedal) reductions in strain, measured as the spinal column’s decreased angle of rotation. ■ On vehicles with double-pedal control, in which the right foot does not perform any driving or braking function in reverse, but can instead be positioned as desired, the vast majority of the relief occurs in the lower section of the lumbar spinal column (relevant for lower back pain), which must be rotated by 7° less. ■ In addition, on vehicles with double-pedal control the swivel seat also improves visibility during reverse driving for drivers of all sizes compared to the standard seat, resulting in improved safety. Abstract Linde AG and the Institute of Ergonomics at the Technical University of Munich worked together to assess the ergonomic design of the new 39X series of Linde forklifts with various levels of equipment. The analysis focused primarily on the seated workstation with a newly designed swivel seat, which permits the operator to rotate his body to the right in order to relieve strain when driving in reverse. The data obtained during measured driving stretches with various drivers served as the basis for a CAD-supported analysis of the sitting workstation with regard to the importance of arm supports and body position and the driver’s field of vision when driving in reverse. The results demonstrate that the swivel seat, especially when combined with the Linde double-pedal control, considerably reduces torso and head rotation. In addition, this vehicle configuration permits the driver’s legs to assume a more comfortable position and improves the driver’s sight conditions during reverse driving. Literature [1]: Drug Commission of the German Medical Association (publisher): Kreuzschmerzen, AVP-Sonderheft Therapieempfehlungen, 2nd edition, Cologne, 2000 The Authors Dr. Frank Schröder Before joining Linde AG, Dr. Frank Schröder was a scientist in the Department of Automotive Engineering at Darmstadt University of Technology (Germany). In Darmstadt, he led studies on driver behavior, emissions and fuel consumption for various types of motorized vehicles. He has been employed since 1999 in the forklift development department of Linde AG where he contributes to the development process by conducting vehicle trials. In 2001, Dr. Schröder took over responsibility for the product acceptance department in the development of industrial trucks; since that time, he has concentrated on balancing vehicle functions with customer needs, compliance with conformity regulations, and questions regarding the ergonomics of forklifts. Dr. Thomas Seitz 22 Dr. Thomas Seitz is a certified physicist and a scientist at the Institute of Ergonomics at Technical University of Munich where he focuses on the development of the RAMSIS PC-supported human model in various applications. His has researched automotive design, the ergonomics of work stations (including medical work stations), and the modeling of people as operators of technical systems. [2]: Gebhardt, H.; Müller, B.H.; Meissner, K.: Komplexe Arbeitssysteme – Herausforderung für Analyse und Gestaltung. In: Bericht zum 46. Arbeitswissenschaftlichen Kongress der Gesellschaft für Arbeitswissenschaft, GfA-Press, Dortmund, 2000 [3]: Meissner, K.; Gebhardt, H.; Küpper, T.: Belastungen von Gabelstaplerfahrern. In: Die BG 10/1998, 1998 [4]: Roth, J.: Verbesserte Flurförderzeug-Funktionen durch elektrische Verarbeitung der Bedienvorgänge. In: VDI reports 1590, Düsseldorf, 2001 [5]: Schmidtke, H. (publisher): Ergonomie, 3rd edition, C. Hanser Verlag, Munich, Vienna, 1993 [6]: Seitz, T.: PCMAN – Ein Messsystem nicht nur zur Analyse von Fahrerarbeitsplätzen in Gabelstaplern. In: REFA-Nachrichten issue 6/2002, Darmstadt, 2002 Linde Technology I 1/ 2004 e_23_31_Heisel_11 09.07.2004 11:56 Uhr Seite 23 Dr. Michael Heisel, Dr. Christer Morén, Prof. Dr. Alexander Reichhold, Andreas Krause, and Antonio J. Berlanga-González Approaches to Economic Solutions for Refineries Cracking with Oxygen Refineries are confronted with a major challenge due to the worldwide, increasing demand for high quality fuels, such as diesel and kerosene, and shrinking markets for heating oil and heavy fuel oil. Consequently, it is an important goal to increase the production of middle distillates and simultaneously reduce the fraction of lower value residues. In terms of technology, chemistry and economics, the admixture of residue oils from atmospheric or vacuum distillation to the crude oil used in FCC (Fluid Catalytic Cracking) plants is a suitable measure. Refineries are facing a number of major challenges these days, some new and some older, in part enforced ones: ■ Refinery operators have to comply with the new “clean fuels” regulations, while ■ the pressure persists to improve economy and margins, and ■ the demand for refinery products is shifting, e.g. towards more diesel and kerosene in Europe, while the demand for heating oil and heavy fuel oil is decreasing The remaining refineries are being operated at higher load. Essentially that means less spare capacity is available to respond flexibly to shifting markets. The example of shifting diesel demand versus gasoline shows that there are substantial changes in the offing (Figure 4). Compounding the problem is the economic situation of the refineries. In the period 1993 through 1999, the return on capital employed (ROCE) of the refining operation was typically a low 4 to 6%. The easiest solution to this economic pinch is the operation of the refineries at a higher load as shown in Figure 3 illustrating US and German refineries. Obviously, there are limits to this option. Once the max. operational capacity is reached, the load can be increased only by making changes in the existing equipment. This often involves major investment. However, for a few processes, solutions are available which allow the refiners to expand capacity substantially at a low cost. The most economic of these processes is the FCC, in which oxygen enrichment can be used to increase throughput by approx. 15% and conversion by up to 3% with only a small investment of capital. The economic pressure resulted in a reduction of the number of refineries in Western Europe at slightly increased capacity (see Figure 2). Both the capacity and the number of refineries decreased in the US (Figure 3). Figure 1: FCC plant in a refinery at Ingolstadt (Germany). Linde Technology I 1/ 2004 23 e_23_31_Heisel_11 09.07.2004 11:56 Uhr Seite 24 130 125 850 120 115 800 110 750 105 No. of refineries Capacity [million t/a] 900 100 700 1998 1999 2000 2002 2001 Figure 2: Development of the refining industry in Western Europe. 800 80 600 60 400 40 200 20 0 0 Capacity [mio t/a] no. refineries 100 Utilization (%) Changes in US Refinieries 1000 Capacity and utilization in German Refineries 100 120 80 100 60 80 60 1985 1990 1995 2000 Figure 5 shows a schematic diagram of an FCC plant. In the oxygen enrichment process, oxygen O2 (blue arrow) is admixed to the air for regenerating the catalyst. More than 30 FCC plants throughout the world apply oxygen in the regeneration process so that this technology can be considered “mature”. Experience has shown that the conversion to oxygen enrichment is not associated with any undue problems. The hardware required for conversion is straightforward and simple (Figure 6): Oxygen from a liquid oxygen tank, a dedicated on-site air separation unit or from a pipeline is metered via a control unit into the air duct leading to the FCC regeneration. Preferably, O2 is added downstream of the air blower in order for this unit to need no approval for operation with oxygen. The piping of the air duct is usually made of carbon steel and does not need to be changed for adding oxygen. However, certain restrictions apply in other areas, such as maximum allowable gas velocities in elbows. For safety it is advisable to have a block-andbleed installation during the shut-down of the oxygen addition to ensure that there are no undetected creeping gas flows from the FCC back to the O2 source. 2005 Figure 3: Development in the number, capacity and load in US refineries, and capacity and load in German refineries. Testing at Vienna University of Technology Diesel versus Gasoline 2.50 China Ratio Diesel/Gasoline consumptions for Oxygen Enrichment 40 1980 2.00 Germany Europe 1.50 1.00 0.50 USA 0.00 1995 2000 2005 2010 2015 Figure 4: Diesel versus Gasoline consumption by geographical region. 24 Utilization (%) Capycity [million t/a] 140 Hardware Needs 2020 The technology of oxygen enrichment in FCC plants is straightforward and not spectacular. But since all reactions in an FCC riser are limited by kinetics, the results are difficult to calculate. Therefore, a test program was developed in cooperation with the Institute of Chemical Engineering, Fuel and Environmental Technology of Vienna University of Technology to quantify the effects of oxygen enrichment on throughput, conversion, and product composition in an FCC pilot plant. The feeds to the FCC pilot plant were to be varied: not only the typical vacuum gas oil, but also atmospheric and vacuum residue was to be admixed. Since these feeds accelerate the heavy metal poisoning of the catalyst, the accumulation of heavy metals was measured also. The same equilibrium catalyst from an FCC plant in a nearby refinery was applied in all tests. The composition of the catalyst and especially the load of heavy metals was analyzed before and after the experiments. The oxygen concentrations in the regenerator fluidization gas was varied. In each test run, the composition of the cracking products and the conversion were analyzed and the composition of the off-gas from the regenerator was measured. The feed load of the system was varied between 100 and 135%. The temperature in the riser and regenerator was kept as constant as possible. Linde Technology I 1/ 2004 e_23_31_Heisel_11 09.07.2004 11:56 Uhr Seite 25 Figure 5: Schematic diagram of a Fluid Catalytic Cracker (FCC) plant. ➀ ➁ ➂ ➃ ➄ FCC reactor Regenerator Steam boiler Fractionation Cycle oil Figure 6: Process principle of oxygen enrichment for FCC regeneration. Result: Enhanced Catalyst Activity The experiments showed that oxygen enrichment resulted in improved regeneration of the FCC catalyst leading to higher catalyst activity. As shown in Figure 7, oxygen enrichment facilitated an increase in plant load at constant conversion by approx. 10% (dotted blue line). Alternatively, it was possible to increase the conversion at a constant load by approximately 2 to 3% (continuous red line). When the increase in catalyst capacity was used to increase the conversion, less residue was generated: as shown in Figure 8 3.5 to 5 % by weight less residue was generated with oxygen enrichment as compared to air alone. This can help the refinery to reduce the low value residues to produce more gasoline and middle distillates. Linde Technology I 1/ 2004 The tests showed that oxygen enrichment in FCC units allows the refinery operator to: ■ Increase the FCC capacity at constant conversion, or ■ Increase conversion at constant throughput ■ Utilize the improved conversion to admix heavier residues, such as atmospheric residue ■ Get more flexibility in the choice of feedstocks ■ Process heavier feedstocks, and ■ Reduce the quantity of residues produced by the FCC unit. 25 e_23_31_Heisel_11 74 09.07.2004 Seite 26 72.5 Increase of Load 72 Conversion [weight %] 11:56 Uhr Conversion 27 Vol.-% O2 70 68 Increase of coversion 68.3 72.5 68.8 66 65.8 64 62.5 Conversion 21 Vol.-% O2 62 61.6 60 58 56 90 95 100 105 110 115 120 125 130 135 140 Feed Amount [%] Figure 7: Increased conversion and capacity with oxygen enrichment in FCC regeneration. 40 35.45 Conversion [weight %] 36 33.72 Residue at 21 Vol.- % O2 32 30.88 27.53 28 26.02 Residue at 27 Vol.- % O2 27.13 24 21.73 20 2.8 3.0 3.2 3.4 3.6 3.8 4.0 4.2 Feed amount [l/h] Figure 8: Reduced production of residues with oxygen enrichment in FCC regeneration. This test plant also allows to process feed oils from customers applying their equilibrium catalyst to predict the effects of oxygen enrichment. However, the small-scale pilot plant generates trends only, rather than numbers that can be applied directly to large-size units. Operating results of the CEPSA Refinery The CEPSA FCC in San Roque (Spain) is a UOP side-by-side design with complete combustion of CO in the regenerator. The original design capacity of 4,200 m3/d has since been increased to an actual capacity of 6,000 m3/d. Since the existing air blower was not designed for this increased demand it was decided to employ oxygen enrichment. 26 CEPSA’s experience at the San Roque refinery was that oxygen enrichment increases the flexibility of the FCC with regard to varying feedstocks. This allows to increase conversion, or alternatively to increase throughput at constant conversion. This was of particular interest to CEPSA since the feedstock for the FCC plant changed daily. Because the feed changed on a day-to-day basis, the decision concerning the use of oxygen was also made daily on the basis of the production demand. The highest oxygen concentration used at the CEPSA plant was 22.4 vol %. The resulting potential was exploited in three ways: ■ the unit was able to cope with rapid changes in feed composition; ■ conversion could be increased for a given feed; ■ as another option, throughput could be increased at constant conversion. A frequently asked question is whether oxygen enrichment does not raise the temperature in the regenerator beyond tolerable levels. A simple heat balance showed that oxygen enrichment resulted only in a negligible reduction in heat sink capacity due to the reduced amount of nitrogen. The main heat sink is the catalyst whose heat capacity is not changed by the addition of oxygen. Accordingly, there is basically no correlation between regenerator temperature and oxygen concentration of the regenerator air feed. In contrast, the coke burn-off from the catalyst has a much more significant impact on the regenerator temperature. The operating results of the CEPSA FCC at San Roque can be summarized as follows: ■ increasing the oxygen content by 1% resulted in an increase in regenerator capacity of 6%. ■ increasing the oxygen content by 1% increased the regenerator temperature by less than 2 ºC. Coke burn-off, rather than O2, defines the temperature rise. ■ Normal operating conditions were easily obtained with enriched air. ■ Oxygen increases the ability to handle high coke formation. ■ This, in turn, allows the refiner to treat heavier feeds, especially by adding residues to the feed. ■ Oxygen enrichment allows to increase conversion and/or throughput capacity of the FCC. ■ Oxygen reduces the formation of residues in the FCC. Costs of Oxygen Enrichment The modification of an FCC unit for oxygen enrichment requires relatively small investments, typically of the order of 100 000 to 300 000 US$. An oxygen source must also be provided, though often a simple liquid oxygen tank is sufficient. The costs of this depend primarily on the tank size, i.e. ultimately on oxygen needs and size of the FCC. The length of the O2 duct between O2 source and FCC also contributes to the costs. Linde Technology I 1/ 2004 e_23_31_Heisel_11 09.07.2004 11:56 Uhr Seite 27 °C Before tests First test run Second test run After tests Riser temperature 499 507* 508 499 Feed temperature 235 235 224* 233 Additional feed due to oxygen enrichment [t/d] – 240* 317* – Additional feed [% of original load] – 4.5 6.0 – Table 1: Test conditions (CEPSA refinery) * Conditions changed for the test, while the other conditions were kept constant. Before tests First test run Second test run After tests Energy savings [€/d] --- 148.00 331.00 --- Daily turn-over [€/d] 92,009 102,346 102,124 90,663 Net profit due to oxygen enrichment during test [€/d]* --- 9,099.32 12,272.67 ---- Calculated annual net profit due to oxygen [€/a]* --- 3,030,074 4,086,798 --- * For calculation of the net profit the cost of oxygen has been deducted from the gross profit Table 2: Economic evaluation of the tests (CEPSA refinery) If large amounts of O2, i.e. in excess of 1 000 m3/h, are consumed continuously, an air separation unit may become economically feasible. Many industrial gas suppliers offer lease options for such units or over-thefence delivery sparing the refiner the investment. Economic Results of the CEPSA refinery Two test runs with oxygen enrichment were carried out at the San Roque refinery within a short period of time keeping the quality of the feed consistent. Oxygen enrichment was tested at two different levels. The main test parameters are listed in Table 1. In the first test run, the temperature of the feed oil was kept constant at 235°C. Oxygen was added to the regenerator air. The feed inlet was raised until the output of the wet gas compressor became limiting. The temperature of the riser was allowed to increase. After seven hours, samples were taken and then the second test run was initiated. In the second run, the oxygen concentration was increased to 22.4%, the feed preheat was lowered by 11°C to 224°C, and the riser temperature was kept Linde Technology I 1/ 2004 almost constant except for an increase by 1°C. The feed amount was increased until the wet gas compressor capacity again became limiting. After four hours, samples were taken and the addition of oxygen was discontinued. As planned, the quality of the feed was basically constant during the tests. Only the aniline point increased by 2.2°C. The 90% distillation end point for gasoline was 158°C before the tests, 156°C during the first test run, 157°C during the second test run, and 162°C after the tests. The octane indices were not measured. The economic data shown in Table 2 were obtained in the analysis. CEPSA drew the following conclusions from these test runs: ■ The use of oxygen is profitable. ■ It allows to use HCO (heavy cycle oil) as a normal feed component. ■ The limitation posed by the air blower capacity can be eliminated. Subsequently, the oxygen enrichment test installation was converted to a permanent supply and the FCC was successfully operated in this mode for several years. 27 e_23_31_Heisel_11 09.07.2004 11:56 Uhr Seite 28 Economy of Oxygen enrichment in FCC, “gasoline” FCC % op. cost O2 at 110% load % op. cost O2 at 100% load Products Gas 1.75 % 1.77 % LPG 15.33 % 16.24 % Naphtha/Gasoline 61.25 % 61.99 % LCO/Diesel 12.01 % 11.13 % 9.68 % 8.87 % Coke - 2.18 % - 2.19 % Feed oil 94.39 % 95.03 % Catalyst 1.89 % 1.90 % Energy 5.23 % 5.26 % Cost of oxygen 0.67 % 0.00 % 105.93 % 100.00 % Decanted Oil Feeds Net profit Figure 9: Comparison of operating cost air operation versus oxygen enriched operation. Economic Results of a Refinery in Brazil A refinery operator in Brazil permitted the use of his proprietary valuation numbers to calculate the effects of oxygen enrichment. This operation was a fairly typical case in that the production of gasoline was the primary goal of this FCC. All other products are less valuable, though the value of LPG comes very close to that of gasoline. The data included the change in conversion and ensuing change in product spectrum that resulted from the increased load and oxygen enrichment. Based on this these data, we calculated the respective turn-over using this refineries’ internal rating numbers. The results for oxygen enrichment to 22.9% are summarized in Figure 9. Oxygen enrichment increases the capacity of the plant. Due to the influence of oxygen there is a slight shift in product composition in favor of LCO/Diesel and decanted oil. This was paralleled by a shift in costs and revenues. Altogether, the refinery increased its profits by 5.93% by oxygen enrichment as compared to the operation with air. However, this these data is are dependent on the oxygen price. Oxygen enrichment necessitates a relatively small investment only, but increases operating costs, which is equivalent to low fixed costs, but high variable costs. Consequently, this option becomes less attractive 28 with increasing oxygen prices. The O2 price is very site-specific and different between individual FCCs. By comparison, the installation of a new air blower is associated with higher investment costs that need to be depreciated over several years. This is equivalent to high fixed costs, but low variable costs. Economic Results of the HOLBORN Refinery Various refiners confirmed that the payback of the oxygen enrichment installation in their FCCs also was on the order of a few weeks only. A more detailed insight was granted by HOLBORN refinery in Hamburg. HOLBORN is especially interesting because they use their FCC primarily to produce middle distillates rather than gasoline. Therefore, the FCC is operated at a low cat/oil ratio of less than 5 and the riser operating temperature is a mild 500°C. HOLBORN has two main incentives to consider oxygen enrichment: 1. The air blower of the regenerator often reaches its limits leading to an insufficient amount of air being available to burn-off all the coke on the catalyst. This limits flexibility in the choice of feed oils with higher ConCarbon content. Linde Technology I 1/ 2004 e_23_31_Heisel_11 09.07.2004 11:56 Uhr Seite 29 Figure 10: The Spanish mineral oil company, CEPSA, operates one of the largest refineries in Spain near the Rock of Gibraltar. 2. As the FCC riser is operated for middle distillates at low severity it generates barely enough coke to keep the temperature of the system up. Oxygen enrichment can improve this situation by burning more coke off the catalyst and at the same time reducing the amount of inert nitrogen needing to be heated up without use. decision on whether or not to convert to permanent oxygen use. A test installation requires: ■ An O2 dosing station, a so-called FLOWTRAIN® ■ A trailer tank for supply of liquid O2 ■ Piping or pressure hoses connecting the O2 tank and the FCC unit. We calculated the profit for the HOLBORN case based on the internal valuation numbers using the procedure described above. The result evidenced an increase of the net profit by approx. 10% with oxygen accounting for only 0.22% of the combined feed cost. Comparing the gasoline-FCC in Brazil and the HOLBORN middle distillate-FCC, substantial differences in the effects of oxygen addition are apparent. These differences are due to the difference in cracking severity: more severe cracking in the gasoline-FCC is associated with more extensive coke formation and an ensuing higher air demand for regeneration. The much higher air demand in the gasoline-FCC reduces the economic effect of a given amount of oxygen added. FLOWTRAINS® and liquid oxygen supply tanks can be leased from Linde. Only the installation of the piping between tank and FCC is associated with investment costs, though often pressure hoses can be used for connecting tank and FCC. These hoses can also be obtained on a rental basis. The subsequent test runs usually take between 4 and 6 weeks depending on whether or not the effects of strongly varying compositions of feedstocks and/or the addition of residue oil is to be tested. Usually, oxygen is added in the test runs to the predetermined level. Alternatively, the oxygen content of the regenerator air can be slowly increased while monitoring the temperatures in the plant. The oxygen flow can be interrupted at any time without difficulties. However, since it takes the equilibrium catalyst several hours to adjust to the new conditions it is not advisable to interrupt the tests. The interruption of oxygen addition does not adversely affect plant safety. Equipment Needs for Test Runs Typically, refinery operators want to see the effects of oxygen enrichment first hand and, therefore, tests are desired. The test runs usually require only little investment. The majority of the equipment required for the tests is available for rent. The tests generate reliable data for the Linde Technology I 1/ 2004 29 e_23_31_Heisel_11 09.07.2004 11:56 Uhr Seite 30 Switching to Permanent Use The rented FLOWTRAIN® can be converted from a lease to a buy option. Often, a tailor-made device may better serve the purpose at hand. For the supply of oxygen, either a tank installation, an on-site air separator or overthe-fence deliveries from a pipeline may be considered. Which of these supply options is best suited depends on the amount of oxygen required, the fraction of time in the supply is needed, and the range of fluctuation anticipated. The control of the FLOWTRAIN® has to be integrated into the FCC unit’s control system including safety interlocks, etc. While alarms may be adequate during the test period, an automatic switch-off may be required in the permanent installation. The details have to be discussed for each case and adapted to the refinery’s overall control and safety system. Safety of Oxygen The FCC unit is guarded by FLOWTRAINS® against unplanned admission of O2 by a number of safety installations. 1. Surplus oxygen: The FLOWTRAIN”contains a high alarm in the flow controller which reacts when the target value for oxygen is exceeded by 0.2 % (v/v). This alarm allows the operator to re-adjust the oxygen content by appropriate means. If the oxygen content continues to rise there is a high-high alarm plus safety switch at 0.35 % (v/v) of excess oxygen and the O2 flow is stopped completely. 2. Failure of oxygen injection: A failure of the oxygen supply is irrelevant for plant safety, means that coke accumulates on the catalyst over time leading to reversible deactivation of the catalyst. If the oxygen supply cannot be re-established within 1/2 hour, the amount of feed oil must be adapted to operating conditions with air only. 3. Failure of instrument air: upon failure of the instrument air, the block-and-bleed valves automatically switch into safe position, i.e. the block valves are closed and the bleed valve is opened. 4. Low temperature switch for oxygen: If the oxygen temperature drops below –5°C (23°F) an alarm rings. This type of failure can happen when feeding the plant from an liquid oxygen tank. If the temperature drops below –20°C (-2°F), the oxygen flow is stopped to ensure that no liquid oxygen enters the air duct of the regenerator where it might cause thermal stress. This measure also effectively prevents the instrument air in any of the actuators from freezing. 30 Safety controls and switches of the DCS system of the FCC: If the FCC is switched off by the DCS of the refinery, the air flow is switched off also. Since the oxygen flow is coupled to the air flow, switching off the air necessarily also switches off the O2. That means that the O2 control is indirectly connected to the safety switches of the FCC. Summary Oxygen enrichment in FCCs is a mature technology applied in more than 30 units world-wide. The procedure is safe to use, yields high returns, and is associated with only small investments. To quantify its effects, a test program was carried out at Vienna University of Technology, in which the chemical and physical effects of oxygen enrichment in a small-scale FCC plant were measured. The main effect of oxygen enrichment is that throughput and/or conversion can be increased, while residue is reduced. These features enable a refiner to respond flexibly to fluctuations in throughput, feed composition, and market trends. Heavier feeds, especially residues, can be treated such that only a smaller amount of residues needs to be marketed. Oxygen enrichment can contribute to the production of more middle distillate. The economic impact was calculated on the basis of data from three commercial refinery FCCs. The payback time of oxygen enrichment proved to be on the order of a few weeks and no more than several months. The profits were up to 25% higher as compared to the air-blown FCC operation. Therefore, the application of oxygen enrichment can be expected to increase as the refineries are forced to further improve their economic efficiency. Abstract New regulations for “clean fuels” and changes in demand are some of the major challenges faced by refineries. An efficient response to this situation is the increase of the use of residues in the FCC in order to enhance the production of middle distillate and reduce the fraction of lower-value residues. To quantify the effects of this procedure, test runs with a small-scale FCC unit have been performed at Vienna University of Technology. The additional capacity required for the use of more residues was supplied by oxygen enrichment in the FCC. The test results were evaluated with regard to their economic impact on a number of refineries. Linde Technology I 1/ 2004 e_23_31_Heisel_11 09.07.2004 11:56 Uhr Seite 31 Literature [1] L. Cabra, “Refinery Challenges in the Current Decade”, ERTC 6th Annual Meeting, Madrid, Nov. 2001 [2] 1999 Annual Report of the Mineralölwirtschaftsverband (Petroleum Economic Union) [registered association], Hamburg, 1999, pp. 48. [3] Oil & Gas Journal, Worldwide Refining Survey, issues 1996, 1999, 2000, 2001 [4] T.E. Swaty et al., “What are the options to meet Tier II sulfur requirements?”, Hydrocarbon Processing, Feb 2001, p 62 ff [5] Hydrocarbon Processing, Jan 2002, p 21 [6] R. Sadeghbeigi, “Fluid Catalytic Cracking Handbook”, Gulf Publishing, Houston, Texas, 1995. [7] D. Farshid et al., “Hydroprocessing Solutions to Euro Diesel Specifications”, Petroleum Technology Quarterly, Winter 1999/2000, pages 29 ff. The Authors Dr. Michael Heisel Dr. Michael Heisel is Project Manager of Linde Gas and Engineering Germany. He is responsible for the area of gas applications in refineries and, in particular, for new applications for process intensification. In his previous position with Linde Engineering and Contracting, several of his projects inventions received international awards. He has published various papers and owns patents that have been applied in process plants. Michael Heisel received his Doctorate from the Technical University of Munich. Dr. Christer Morén Dr. Christer Morén is a member of Linde Gas and Engineering’s working group on refineries and petrochemistry. He joined the AGA group, now owned by Linde, in 1976 being responsible at its affiliate, Tudor AB, for improved battery systems. He has held various positions with AGA AB and AGA Gas AB since 1981 in the following areas: process technology, cryo-technology, the food and health industries, welding and cutting technology, and acquisition. He majored in process technology at the Royal Institute of Technology in Stockholm and is a member of ACS and AIChe. Prof. Dr. Alexander Reichhold Prof. Dipl.-Ing. Dr. techn. Alexander Reichhold received his Doctorate in Process Technology from the Vienna University of Technology. He currently holds a position as an Assistant Professor at the same university’s Institute of Process Technology and is the director of the research group, “Refinery Technology & Fluidized Bed Systems.” Alexander Reichhold is a member of the board of the Austrian Society for Petroleum Sciences. Andreas Krause Alexander Krause is a process engineer with HOLBORN Europe Refinery GmbH where he is responsible for the FCC Department. He looks back on more than 13 years of experience with process control applications and petrochemical plant technologies. He majored in Process Technology at Hamburg (Germany) and then worked for seven years as a process Control Engineer for Shell Chemie Köln. His area of expertise includes processes, technical support, and process analysis. He designs and calculates the technical components and equipment for the improvement of the plant. Antonio J. Berlanga-González Dipl-Ing. Antonio J. Berlanga-González graduated from the University of Málagá. Since joining CEPSA, he has accumulated more than 30 years of experience in petrochemistry. He has held a position in the FCC Process and Operations Department since 1982. Linde Technology I 1/ 2004 31 e_32_36_Kummann_11 12.07.2004 14:43 Uhr Seite 32 Dr. Paul Kummann Economic Production of Ammonia The Location Makes the Difference Ammonia is a base product of the chemical industry. A variety of feedstocks is used to make the production of ammonia economical. While natural gas is preferred as a starting product in countries with natural gas resources, countries depending on imports use cheap refinery residues. Despite large price fluctuations (between 90 and 250 USD/t), the average price for ammonia has essentially remained unchanged over the past 20 years. This is also evident from the historical price trends in the Caribbean and Middle East, as the major export regions, and NorthWest Europe as a typical import region. The average FOB prices derived from these charts are 146 and 145 USD/t for the Caribbean and Middle East respectively, as compared to a CIF N.W. Europe of 169 USD/t. Wellhead-sited locations with low NG prices are typical export regions for ammonia. The price difference to major import regions, like Europe, India and East Asia, is mainly due to transport costs. Since the costs of shipment from the Middle East to Shanghai are in the range of 40 USD/t, market-sited average ammonia cost CIF East Asia (excluding import duties) of approx. 190 USD/t have to be expected for coastal areas in East Asia, and even higher prices further inland, when further transshipment and/or import taxes have to be considered. Based on this background it can be investigated, under which conditions new projects become economical. Including a satisfactory internal rate of return (IRR) export plants in wellhead-sited regions must not exceed ammonia production costs of approx. 145 USD/t. At market-sited locations at some distance from the coast, even ammonia production costs of up to 250 USD/t may be tolerated, in particular if this production is protected by distant transport and/or import duties. Three feedstocks, one product Ammonia, a compound containing hydrogen and oxygen, can be produced by a number of different process technologies. The characteristic feature of these technologies is how the hydrogen is obtained from the feedstock. The sequence of process steps is basically the same and includes: 32 Figure 1: Heavy oil gasification-based ammonia plant in Jilin (China) with a capacity of 1,000 t/day. ■ ■ ■ ■ ■ Generation of synthesis gas Heat recovery and CO conversion Purification of synthesis gas Ammonia synthesis Steam/driver system This article analyses in more detail three different plant concepts, which make use of either natural gas, refinery residues or raw hydrogen. ■ Steam reforming of natural gas or naphtha according to the Linde Ammonia Concept (LAC™). ■ Partial Oxidation of refinery residues by Texaco high pressure gasification ■ Ammonia synthesis from raw hydrogen and nitrogen Most ammonia plants use light hydrocarbons, especially natural gas, as a feedstock. The high hydrogen content of methane and the high purity of NG favor this use, since relatively little energy is required for the generation and purification of synthesis gas as compared to other feedstocks. With only small additional investment costs, natural gas-based ammonia plants can be operated continuously, temporarily or partly with naphtha or other light hydrocarbon feed streams. The supplementary costs are in the range of a few percent and include additional costs for the vaporization and pre-reforming of naphtha. The Linde Ammonia Concept (LAC™) is a process for the production of ammonia from natural gas. This concept involves the process steps illustrated in Figure 2. Linde Technology I 1/ 2004 e_32_36_Kummann_11 12.07.2004 14:44 Uhr Seite 33 Another feedstock basis for the production of ammonia are refinery residues, such as vacuum residue, visbreaker tar or asphalt. The driving force behind the use of these heavy and sulfurous feeds is their general availability at moderate cost in regions where NG or naphtha are not available in sufficient quantity and at an attractive cost. The production of synthesis gas from heavy hydrocarbons is based on partial oxidation of these types of feedstocks with oxygen. Further design features of ammonia plants of this type are high operating pressures during synthesis gas generation, Rectisol wash, nitrogen wash, and ammonia synthesis in the absence of inert components. A typical plant is represented by the process description of Figure 3. Ammonia plants with much lower investment costs can be designed if raw hydrogen and nitrogen are available. A hydrogen surplus may be available for instance from a refinery, methanol plant or ethylene plant. Provided the available quantity is sufficient, hydrogen can be converted to ammonia following the production scheme shown in Figure 4. In this process scheme, the hydrogen can be purified either by pressure swing adsorption (PSA) or by a liquid nitrogen wash. PSA is advantageous if the raw gas pressure is below 50 bar. At higher pressure, though, depending on the composition of the gas, the liquid nitrogen wash may be more beneficial due to its high hydrogen recovery rate above 99 %. Long-term profitability The economic evaluation of the three basic plant concepts is to take into consideration the investment costs, operating costs, and profit expectations. The investments costs of a new facility include the price of the turn-key process plant and additional project costs of the buyer. The turn-key price for a process plant is determined on the basis of the required technology and plant capacity. Buyer's additional costs for project development, approval, and in house works are added as lump sums. This type of cost may differ significantly from one project to another, depending on available infrastructure and plant integration with existing facilities. For the purpose of this investigation, buyer's additional project costs were rated to be 30% of the turn-key price for the process plant and added. A common commercial evaluation method was used to compare the ammonia production costs of the three different technologies. Firstly, the annual net cash flow (NCF) was determined for the construction period and 15 subsequent years of plant service. The net cash flow is obtained by subtracting investment and operating costs from revenues. Applying a market-typical interest rate (cut-off rate), the NCF values are then Linde Technology I 1/ 2004 1. Natural gas desulfurization 2. Feedstock preheating 3. Reformer 4. Reformed gas cooler 5. CO conversion in isothermal reactor 6. Heat recovery 7. Hydrogen purification by PSA 8. Fuel gas recycling 9. Fuel gas heat recovery 10. Nitrogen production by air separation 11. Syngas compression 12. Heat exchanger 13. Ammonia synthesis 14. Steam production 15. Ammonia separation 16. Cooling with refrigerant 17. Steam/driver system 18. Condensate system 19. Refrigeration unit Figure 2: Flow diagram of the Linde Ammonia Concept (LACTM). Figure 3: Block diagram showing the production of ammonia from residue oil. 1. Drier station 2-4. Heat exchanger and gas cooling 5. Nitrogen wash 7. Cold box 10. Nitrogen production by air separation 11. Syngas compression 12. Heat exchanger 13. Ammonia synthesis 14. Steam production 15. Ammonia separation 16. Cooling with refrigerant 19. Refrigeration unit Figure 4: Schematic diagram showing the production of ammonia from raw hydro- 33 e_32_36_Kummann_11 12.07.2004 14:44 Uhr Seite 34 Figure 5: Linde LAC™ Ammonia plant in Vadodara (Spain) with a capacity of 1,350 t/day. converted into a series of discounted cash flows (DCF), such that the sum of DCF values corresponds to the net present value (NPV) of the plant after a certain period of operation. Finally, the NPV after 18 years is set equal to zero (0) and the required max. interest rate is determined. This interest rate is the internal rate of return (IRR) and has to cover all interest and risks. The ammonia production cost determined with this procedure correspond to the desired product prices leading to a desired IRR of 20% and a net present value of the plant of zero after 18 years (construction period plus service period). The plant operating costs are an influential element in the calculation of the annual net cash flows. Common parameters were taken into account in the calculation of the operating costs at fixed rates (static method). The calculation includes as fixed costs constant feedstock prices (0.7/1.7/3.0 USD/MMBTU, million British thermal units = approx. 293 kWh), maintenance and insurance costs (2% each per year), cooling water prices (0.04 USD/m3), and personnel costs according to European standards (40,000 USD/year). The plants were assumed to be operated for 8,000 hours per year and a construction period of 30 months was assumed. Three prices for one product taking the capital requirements and operating conditions of the plant into consideration. In a first approximation, the result obtained is also valid for naphtha as the feedstock and for steam reformer-based technologies offered by other contractors. Obviously, the ammonia production costs depend mainly on the feedstock cost and plant capacity. Analyzing historical price trends since 1993, it is evident that, at a NG price of 0.7 USD/MMBTU, new plants with capacities of 1,350 t/day and above are an attractive investment in export regions where the average ammonia price is 145 USD/t. The operation of market-side ammonia plants remote from gas wells and closer to consumers can tolerate higher gas prices and/or smaller capacities and still be economical. To give an example, a NG-based ammonia plant with a capacity of 600 t/d serving a phosphate complex some distance from the ocean and waterways is an attractive investment. In this case, the acceptable NG price level depends on the transport surcharges of imported ammonia. On the other hand, it is also evident from Figure 6a why ammonia producers in market regions with excellent transport infrastructure, e.g. near rivers and canals in Central Europe, face severe competition by imported material. NG prices in Central Europe are too high to support the construction of new ammonia plants. The price of ammonia in certain market regions is mainly determined by the cost of imported material. As a result, only large and fully depreciated ammonia plants can be operated economically in Europe. Refinery residues are advantageous in that they are available at moderate cost in industrial regions throughout the world. Depending on the capacity, type of crude oil, and process configuration of the respective refinery, the surplus of residue is frequently sufficient to serve as the feedstock for one (1) ammonia plant. A typical residue price of 60 USD/t corresponds to approx. 1.6 USD/MMBTU was used as a fixed value in this evaluation. Compared to the NG price at the wellhead this may still seem high. However, for the comparison to be valid it must be related to NG prices of 3.5 – 4.5 USD/MMBTU in the markets at the end of a NG pipeline or LNG supply chain. For example, only small amounts of NG are available in countries like China and India. Consequently, the ammonia and urea production in Asia is largely based on refinery residues, or even on coal. The construction of oil-based plants in regions with a high fertilizer demand provides the additional advantage that the regional transport infrastructure is not burdened by these transportation loads. This may explain why many ammonia/urea projects using refinery residues were built in Asia. Many of these projects were also supported by soft loans from countries like Japan, Germany and Italy in The costs of ammonia production from NG under these conditions are illustrated in Figure 6a. The production costs listed therein were evaluated for the LAC™ process, 34 Linde Technology I 1/ 2004 14:44 Uhr Seite 35 order to promote a certain degree of self-sufficiency. According to Figure 6b, production costs of 240 - 285 USD/t are obtained for partial oxidation (POX)-based ammonia plants with capacities of 1,000 – 1,350 t/d. Since approx. 0.8 t of residues are required to produce 1 t of ammonia, the product price contains approx. 45 USD/t of feedstock costs. This explains the strong dependence of the production cost on the capital costs required for the generation and purification of synthesis gas from a chemically less optimal feedstock. Any economic analysis has to take into account that many plants of this type were built at lower cost than today's costs of a turn-key plant. Plant costs increased moderately, while the market prices of ammonia remained fairly constant. Some partial oxidation plants have been in operation for more than 15 years and are essentially paid off. In several plants, the plant capacity was enlarged or the customers integrated additional CO-based downstream production facilities, e.g. for acetic acid, toluene diisocyanate (TDI), and methanol, to further improve plant efficiency. Subject to the condition that the raw hydrogen is available from other sources, the investment costs of ammonia production can be significantly lower. This is because an investment in syngas generation is not required under these conditions. The resulting costs of ammonia production from raw hydrogen are illustrated in Figure 6c as a function of plant size and hydrogen cost. For the production of 600 t/d ammonia, approx. 50,000 Nm3/h hydrogen are required. For stoichiometric reasons, this amount of hydrogen may in principle be available as purge gas from older methanol plants lacking an auto-thermal reformer. If this hydrogen is used as fuel gas, its value is basically only 0.7 USD/MMBTU, resulting in attractive ammonia production costs. However, more realistically one has to assume that rather less hydrogen is available as free capacity. Considering, for example, a market region with average ammonia prices of 200 to 250 USD/t, the surplus hydrogen (e.g. 25,000 Nm3/h) is supplied to an external ammonia synthesis loop of 300 t/d. The associated investment is attractive for hydrogen costs of up to approx. 3 USD/MMBTU (LHV). The LAC™ process provides for the production of several by-products aside from ammonia. For example, in an LAC™ plant hydrogen and nitrogen are available as pure intermediate products prior to their processing into synthesis gas. Consequently, these two by-products can be exported for other applications without a need for additional equipment and changes in the process configuration. The production of these by-products is associated with an attractive economy of scale resulting from the often larger scale of ammonia production. Further products, such as CO2 (from conversion), oxygen and argon can be produced after some 300 Ammonia production cost [USD/t] 12.07.2004 250 200 150 ● NG Cost: 3.0 USD/MMBTU 100 ● NG Cost: 1.7 USD/MMBTU ● NG Cost: 0.7 USD/MMBTU 50 0 0 500 1000 1500 2000 Plant capacity [t/day] Figure 6a: Ammonia production costs as a function of plant size and natural gas prices. Ammonia production cost [USD/t] e_32_36_Kummann_11 300 280 260 240 ● Oil Feed 1.6 USD/GJ 220 200 1000 1350 1800 Plant capacity [t/day] Figure 6b: Costs of ammonia production from residue oil. Integrated Facilities Further opportunities to enhance the economic efficiency result if ammonia production is integrated into existing facilities rather than newly erected on the green field. These opportunities, though, largely depend on the individual circumstances of the project and are difficult to present in the form of charts and diagrams. Generally, improvement in economic efficiency is due to the utilization of suitable by-products and cost savings from integration into existing facilities. Linde Technology I 1/ 2004 Ammonia production cost [USD/t] 300 250 200 150 ● H2: 3.0 $/MMBTU 100 ● H2: 1.7 $/MMBTU ● H2: 0.7 $/MMBTU 50 0 0 100 200 300 400 500 600 700 Plant capacity [t/day] Figure 6c: Costs of ammonia production from raw hydrogen. 35 e_32_36_Kummann_11 09.07.2004 11:57 Uhr Seite 36 equipment is added, however, with no substantial increase in the feedstock and operating costs. For example, the reboiler duty to separate CO2 by a gas scrubber can be covered by cool down of process gas. The production of oxygen or argon requires additional columns in the air separation cold box, but no increase in air compression. Basically, other side products, such as CO and methanol, can also be added, but their production increases the feedstock needs and requires additional equipment. In this case, the co-production of other products can only make use of a larger economy of scale. In view of these facts it is not surprising that most of the LAC™ plants contracted so far produce side products. Further opportunities for the improvement of the economic efficiency are provided by integrating the plant into existing facilities: e.g. the LAC™ plant in Daqing (China) does not need to have it's own nitrogen production facility. Instead, the customer owns an oxygen plant which was built to feed an oxygen blown secondary reformer in his methanol plant. Consequently, nitrogen was available on site and a nitrogen plant accounting for additional investment costs of approx. 15% of the total cost of the ammonia plant was dispensable. Moreover, the ammonia can be exported making use of the existing storage tanks and loading facilities. H2 and CO2 are exported for other applications. The LAC™ process typically includes a turbo generator for conversion of the steam surplus into plant power. In Phosphate Hill and Moura (Australia) steam surplus or the need for downstream facilities is balanced by the LAC™ plant. Moreover, the involved nitrogen plants produce the instrument air and plant nitrogen as well as power for other consumers. In the major industrial complex in Vadodara, India, hydrogen, carbon dioxide, oxygen and argon are produced as by-products. The storage and distribution of ammonia and the supply of utilities were integrated with other facilities of the complex. Conclusion The price of ammonia increases with the distance of a consumer from the NG well-heads. This is because large natural gas-based ammonia plants located in regions such as the Middle East and Caribbean determine the worldwide price structure. However, at some distance from the major export markets medium sized NG-based ammonia plants can also be an attractive investment, in particular if such plants serve a downstream consumer, such as a phosphate complex or ammonium nitrate plant. Refinery residues may be an alternative feed stock, if NG is not available in sufficient quantity and at an attractive cost. As bottom of the barrel product, refinery residues are available at moderate cost in refineries throughout the world and can be processed. 36 The economic efficiency of such partial oxidationbased ammonia plants is influenced mainly by plant capacity, integration with existing refinery facilities, and the co-production of by-products. In countries such as China and India, ammonia plants utilizing refinery residues as feedstock provide for a certain degree of self sufficiency in the production of ammonia and urea. Abstract The location of the plant is a crucial factor for the design of new ammonia production facilities. The economic efficiency depends mainly on the availability of the feedstocks and the local ammonia demand. Especially in the proximity of oil-producing countries, natural gas is the preferred source for the production of hydrogen and ensuing synthesis of ammonia. In contrast, oil refinery residues are often more economical in oil-import regions, such as the countries in Europe. In countries, such as China and India, the close proximity of ammonia and fertilizer plants reduces the burden on the local traffic infrastructure and is therefore subsidized by the governments. In addition, synergy effects with existing facilities can be utilized to save on investment and operating costs. The Author Dr. Paul Kummann Dr. Paul Kummann was awarded a doctorate in Physical Chemistry by the University of Münster (Germany). He taught Technical Chemistry at the Institute Algerienne des Petrol (Boumerdes, Algeria) and joined Linde in 1981 as a process engineer for natural gas plants. From 1989 he directed the Linde branch in Beijing (China) before focusing his attention on various marketing tasks at Linde between 1993 and 1998. He currently is Senior Project Engineer for Synthesis Gas Facilities. Linde Technology I 1/ 2004 e_37_43_Berger_11 12.07.2004 14:45 Uhr Seite 37 Eginhard Berger, Manfred Boelt and Bjørn Sparby The Benefits of Cold Deep-Water for LNG FPSOs in Tropical Seas Offshore Plants for LNG Production Major natural gas fields situated approx. 100 kilometers off the coast of Nigeria are waiting to be developed. Linde’s advanced floating production, storage, and offloading technology for liquefied natural gas offers an innovative solution for the commercial exploitation of these deposits. Extensive preliminary work has been performed as part of the development of the natural gas fields off the shore of Nigeria. The benefits of the floating LNG-FPSO concept (LNG = liquefied natural gas; FPSO = floating production, storage, and offloading) convinced the shareholders to further pursue this concept. As the next step, exploration wells will be drilled in order to demonstrate the resources available for exploitation. These efforts follow the lead of the Snøhvit project with 4.3 MTPA (million tonnes per annum) LNG production capacity which is currently under construction. Its execution includes the following steps: ■ Design and installation of the process plant on a barge ■ Yard construction of the barge ■ Yard construction of the process units ■ Integration and hook-up of barge and process units in a yard ■ Ocean transportation of the integrated process barge Deep-sea LNG-FPSO Water depth: 1,000 m Air temperature: 28°C Water temperature at a depth of 1,000 m: 5°C Shallow-water LNG-GBS Water depth: 23 m Air temperature: 35°C Water temperature: 28°C approx. 100 km NG Deposits These steps are in principle the same as required for LNG FPSOs, although the Snøhvit LNG process barge will eventually be grounded and integrated into a more or less conventional land-based LNG complex. Therefore, the Snøhvit LNG project can be considered as a first reference towards marinization of LNG baseload plants. Based on this experience, Linde has developed the topside process and utility facilities for two offshore LNG concepts: ■ An LNG FPSO (floating production storage and offloading) on a barge in deep-waters ■ A fixed LNG-GBS-plant in shallow-waters near the shore (GBS = gravity base structure = gravity platform grounded on the sea floor, usually made of concrete) Design Conditions Deep-water LNG FPSO Case A floating LNG plant concept in benign waters approx. 100 km offshore the West African coast had to be designed on the basis of the following main parameters: Ambient conditions: ■ Air temperature: 28 °C ■ Seawater temperature: 5 °C (at a depth of 1000 m) The process plant had to be designed to meet the following requirements: ■ The quality of the produced LNG, LPG (Liquefied Petroleum Gas) and gasoline must be suitable for international trade ■ No permanent flaring is permitted ■ International rules and regulations must be applied, e.g. API (American Petroleum Institute) and DNV (Det Norske Veritas). The plant had to be designed to handle feed streams from 3 fields with typical compositions. The pressure was 120 bar at plant inlet at a temperature of 28°C. The compositions of these three feed streams turned out to be ideally suited to LNG production, since there is only a very small fraction of nitrogen and the acid gas (CO2) content is comparatively low. H2S is not present. Any nitrogen would need to be removed, because nitrogen Figure 1: Alternative LNG concepts for the development of natural gas resources off the coast of Nigeria. Linde Technology I 1/ 2004 37 e_37_43_Berger_11 09.07.2004 11:59 Uhr Seite 38 Design Conditions Shallow-water LNG GBS Case The task was to design a fixed LNG plant on a GBS in 23 m water depth for the same feed gas compositions as in the LNG FPSO case. However, the location in shallow-water, relatively close to the shore, but remote from the production wells, required the following design parameter changes: ■ Design air temperature 35°C ■ Seawater temperature 28°C Figure 2: The Statoil-Linde MFC® Process. In addition, the gas pressure of one feed gas stream drops in the sub-sea pipeline to 60 bar by the time it reaches the plant on the GBS. Streams to be heated Streams to be cooled Process Selection Several processes are available for the large-scale liquefaction of natural gas: ■ SFMR process (Single Flow Mixed Refrigerant) ■ CC process (Classical Cascade) ■ C3/MR process (Propane Pre-cooled Mixed Refrigerant) ■ DMR process (Dual Mixed Refrigerant) ■ MFC® process (Mixed Fluid Cascade) ■ Nitrogen Expander Process Figure 3: Typical Cooling Curves in the MFC® Process. has no calorific value and excessive amounts of nitrogen fail to meet gas sales specifications. H2S, had it been present in addition to nitrogen, also needs to be removed, because this component would also freeze at low temperatures and plug the heat exchangers. The removal of these components would have been associated with considerable investment costs and power requirements. Both the high feed gas pressure of 120 bar and the low cooling water temperature turned out to be extremely beneficial for the process design. 38 The requirements on these processes are as follows: ■ Safety and reliability ■ Low weight and low area requirement ■ Low investment and operating costs ■ Low emissions ■ Large capacities in a single process train to obtain the economy of scale due to the lower specific investment costs At first glance, the safety criterion is met best by a nitrogen expander process. An inert gas, nitrogen is safer than all other processes, since these use hydrocarbons as refrigerants. These are flammable in case of a leakage. However, the power requirement of the nitrogen process exceeds drastically the power requirement of all other processes. For a given LNG production capacity this means that the prime movers (mostly gas turbines), the compressors and the cooling system, as well as all support systems have to be increased, which counterbalances the safety benefit to some extent and increases the investment costs to an unfavorable level. Linde and Statoil jointly investigated the features of all these processes in detail and concluded that the existing requirements are best met by the potentials of the MFC® Process (please refer to “The Snøhvit Project”, in Linde Technology 1/03). Linde Technology I 1/ 2004 e_37_43_Berger_11 09.07.2004 11:59 Uhr Seite 39 Figure 4: GE LM 6000 Power Output and Heat Rate. The MFC® Process is a proprietary process jointly developed and owned by the Statoil-Linde LNG Technology Alliance. The flow scheme is illustrated in Figure 1. The excellent efficiency of the MFC® Process is evidenced by an analysis of the cooling curve. Natural gas is usually liquefied at pressures above 40 bar in order to reach acceptable levels of compressor shaft power requirements. The typical cooling curve (heat over temperature) is S shaped as shown in Figure 2 and can be subdivided into the 3 sections: ■ Pre-cooling ■ Liquefaction ■ Sub-cooling The craftsmanship of the process designer is to match this cooling curve with refrigerant cycles as closely as possible, i.e. with minimum temperature differences at any enthalpy (heat) point. Linde “Optisim” software is a powerful tool to achieve optimized results taking into account relevant constraints like avoiding temperature overlapping (integrated pinch analysis) and compressor efficiencies, etc. Therefore, the MFC® Process was applied to the deep and shallow-water offshore LNG concepts. The cooling curve indicates that the cycle compressor shaft power requirements depend on the inlet temperature of the natural gas and the refrigerant cycles after cooling by the available cooling water. The effects of the cold deep sea water (5°C) and the warm shallow-water (28°C) as cooling water as well as the reduced feed gas pressure on the total refrigerant cycle compressor shaft power is shown below for an LNG production capacity of 5.1 MTPA: ■ LNG FPSO deep-water case (cooling water at 5°C): 132 MW ■ LNG GBS Shallow-water Case: (cooling water at 28°C): 176 MW Linde Technology I 1/ 2004 Prime Mover Selection and LNG Production Capacity The cycle compressors can be driven by ■ Steam turbines ■ Gas turbines ■ Electric motors ■ Combinations of the above drivers The compressors of the first LNG baseload plants were driven by steam turbines which was advantageous in that any required capacity was available. However, the lower efficiency, large size of the equipment, and the extensive cooling system caused the investors to decide in favor of a direct gas turbine drive. The gas turbines, however, are the most sensitive equipment of an LNG plant. Redundancy is required in order to obtain acceptable plant availability. Therefore, it is proposed to use an incremental number of highly efficient General Electric LM 6000 gas turbines to drive electric generators with at least one gas turbine/ generator set as back-up. The cycle compressors are driven by electric motors, operating on the plant’s power distribution system. This all-electric drive scheme is now being used at the Snøhvit LNG project (see Linde Technology 1/03). It is considered to be the most advanced system and provides the highest degree of availability for single train installations. The concept is expected to be adapted to further LNG plant projects both for land-based and for offshore installations. This concept increases the availability of the plant and thus the sales revenues considerably as compared to direct drive by gas turbines. An additional benefit is the decoupling of the train capacity from the size of the existing gas turbines. 39 e_37_43_Berger_11 09.07.2004 11:59 Uhr Seite 40 Figure 5: The MFC® Process with All-electric Drive and Waste Heat Recovery for Hot Oil Heating. Directly driving compressors by gas turbines leads to limitations since gas turbines are commercially available only at incremental capacities. The best option of a gas turbine both for onshore and offshore installations is the aero-derivative GE LM 6000 gas turbine made by Nuovo Pignone. It’s characteristic features are its low weight and high efficiency. All gas turbines are dependent on the outside air temperature, which is admitted and compressed and then used for fuel gas combustion. The dependence of the gas turbine power output and the heat rate is indicated in the diagram in Figure 4. As is evident from the diagram, the output of the LM 6000 gas turbine is approx. 46 MW at 10°C, compared to only approx. 33 MW at 35°C. Therefore, it is proposed to cool the air intake of the gas turbine with the 5°C cooling water available in the deep-water case. This air-cooling is performed in so-called quench coolers with a secondary demineralized water cycle, which is re-cooled counter-currently to the seawater. The cold seawater with its constant cooling temperature provides for very low fluctuations of turbine power output and consequently for stable production rates. This quench cooling corresponds to the state of the art in air separation plants operating under similar climatic conditions, where relatively cold cooling water (as compared to the ambient air) is available. The exhaust gas of the gas turbines is best used for hot oil heating in a waste heat recovery unit. A hot oil cycle is normally used for process heating mainly in the dehydration and acid gas removal units. The stated gas turbine output figures represent gross values. The net power that can be distributed to 40 the electric consumers in an LNG plant is calculated by subtracting the losses due to air intake cooling (deep-water LNG-FPSO), aging, waste heat recovery, and generators. The net gas turbine output for the two cases investigated are as follows: ■ Deep-water LNG-FPSO: 40.7 MW ■ Shallow-water LNG-GBS: 30.5 MW Multiple installations of these gas turbines are necessary to meet the power requirements of large LNG plants with production rates between 4 and 6 MTPA, which are currently envisaged for several baseload LNG projects. A capacity of 5.1 MTPA was chosen for the two cases for comparison reasons. The following number of GE LM 6000 gas turbines is required in each case: Deep-water LNG-FPSO: 4 +1 spare GE LM6000 Shallow-water LNG-GBS: 7 +1 spare GE LM6000 The spare unit is needed as a back-up to improve the plant availability, since the gas turbines require regular down time for maintenance. The combination of the MFC® Process and the all-electric drive system, waste heat recovery, and hot oil cycle for the deep-water LNG FPSO is illustrated in Figure 5. Power Requirements and Plant Efficiencies The all-electric drive concept with waste heat recovery presented above provides for excellent utilization of the fuel gas in excess of 70%. In addition, the all-electric drive configuration allows the compressor power loads Linde Technology I 1/ 2004 e_37_43_Berger_11 09.07.2004 11:59 Uhr Seite 41 Figure 7: CAD Model of a Shallow-Water LNG-GBS (Gravity Base Structure) for 5.1 MTPA. for the three refrigerant cycles to be adjusted freely such that the cooling curve can be optimally matched. As a consequence of this lowest possible total compressor shaft power requirement, the plant becomes highly efficient. This can be expressed as conversion factor or as internal consumption (hydrocarbon products / feed, or fuel gas / feed, all expressed in gross calorific values). In the two cases investigated here, these parameters are as follows: Conversion Factor Deep-water LNG FPSO: Internal Consumption 96.96% 3.04% Shallow-water LNG GBS: 96.00% 4.00% All products (LNG, LPG, and condensate) and process and utility consumers are taken into account in the above figures. Selection of Heat Exchangers Linde is in the unique position to fabricate the two types of cryogenic heat exchangers conventionally used in LNG baseload plants: the plate fin and the spiral wound heat exchangers. Each of these heat exchanger types has specific benefits and disadvantages. Among other things, the single plate fin heat exchanger core has a relatively competitive cost. However, large LNG plants need to have multiple, parallel plate fin heat exchanger installations, such that this advantage is counterbalanced to some degree by the complex interconnecting piping. The spiral Linde Technology I 1/ 2004 wound heat exchangers show extreme resistance to the thermal stress encountered in the low temperature sections during start-up or maloperation. A detailed comparison showed that each heat exchanger type has specific merits at the proper place. Therefore, it was decided to use the plate fin heat exchanger for the pre-cooling section and the spiral wound heat exchanger for the liquefaction section and the sub-cooling section. The Cold Box In most of the large LNG baseload plants, the cryogenic heat exchangers have individual insulation, which applies to both plate fin and spiral wound heat exchangers. This insulation mainly consists of polyurethane foam or of foam glass. However, an alternative insulation method was selected for the Snøhvit LNG project and in offshore LNG concepts: the cold box. Cold boxes are usually installed in cryogenic processes such as air separation and LNG peakshaving plants. It consists of a box of normal carbon steel plates enclosing the cryogenic equipment and piping. The void space is filled with the powdered mineral, Perlite, as the insulation material. The benefits of this type of cold box are evident: ■ The cryogenic process equipment and piping is all welded together and laid out as tight as possible resulting in minimized material and thermal losses and maximal safety. ■ The cold box is mechanically completed in workshops under optimized conditions ■ The cold box provides external mechanical protection during transportation and in the plant itself ■ Apart from the “all welded” principle of the cold box interior, which is considered as the safest installation mode, the cold box enables detection of possible leakages by control of a nitrogen purge stream in the Perlite-insulated space. 41 e_37_43_Berger_11 09.07.2004 11:59 Uhr Seite 42 Figure 6: CAD Model of a Deep-Water LNG-FPSO for 5.1 MTPA. ■ Fire resistance requirements can be met efficiently with a cold box. This is relevant for the compact Snøhvit plant layout as well as in general for offshore LNG plant concepts. The process and equipment configurations shown above have been used to design concepts for the floating deep-water case platform and the fixed shallow-water LNG-GBS. The respective CAD models are shown in Figures 6 and 7. All process and utility facilities are installed on the topsides. Deep Water LNG FPSO Shallow Water LNG GBS Area requirement 11700 m2 12 700 m2 Weight 35 000 t 38 000 t Fuel gas consumption 13.4 x 106 GJ/a 18.0 x 13.4 106 GJ/a CO2 emissions 1.02 Mill t/a 1.37 Mill t/a Compressor shaft power 132 176 Number of gas turbines 4 + 1 spare 7 + 1 spare Cooling water 26 000 t/h 36 000 t/h Investment cost 100 % 120 % Table 1: Comparison of deep-water-LNG-FPSO versus shallow-water-LNG-GBS 42 Overall Comparison of Deep-water LNG FPSO and Shallow-water LNG GBS The design conditions are obviously favoring the deep-water LNG FPSO concept in many respects, and penalize the shallow-water LNG GBS concept. The higher refrigerant cycle shaft power requirement and the lower power output of the gas turbines in the LNG GBS case impact the following items: ■ More gas turbines are required ■ Larger compressors are required ■ More cooling water is required ■ More piping, structural steel, instrumentation, electrical works, construction etc. ■ Higher capital and operating expenditures ■ More fuel gas required ■ Higher emissions The benefits of the deep-water LNG-FPSO concept as compared to the shallow-water LNG-GBS are summarized in Table 1. The deep-water concept is beneficial according to all criteria shown in the table and ultimately results in lower investment and operating costs. Moreover, the submarine pipelines between the wells and the plant are substantially shorter than in the shallow-water concept, in which they increase the overall costs substantially. The disadvantages of the LNG-GBS concept apply equally to a land-based installation near the shore. In general, both concepts are characterized in that, unlike onshore installations, there is no need for harbor facilities and complex cooling water systems, which may be associated with substantial costs in some coastal areas with shallow waters, in which a deep-water channel must be maintained for access of large LNG tankers. Some other beneficial aspects of offshore LNG concepts are attracting increasing attention: they involve no land use and the facility cannot be seen from the shore and does not give rise to concerns because of its environmental impact. Linde Technology I 1/ 2004 e_37_43_Berger_11 09.07.2004 11:59 Uhr Seite 43 Conclusions The design of the deep-water LNG-FPSO off the African shore features the highly efficient MFC® Process, cooling with cold deep seawater (5°C), and gas turbine air intake cooling. This enables the installation of LNG baseload projects offshore in tropical regions under conditions that are usually encountered only in northern regions like Norway or Alaska. The positive impact on the economic viability of such an LNG baseload project is enormous. References ■ Eginhard Berger, Wolfgang Förg, Roy Scott Heiersted, Pentti Paurola: The Snøhvit Project, Linde Technology, 1/2003 ■ Roy Scott Heiersted, Statoil: Snøhvit LNG ProjectConcept Selection for Hammerfest LNG Plant, GASTECH 2002, Qatar, Oktober 13-16, 2002 ■ W Förg, W Bach, R Stockmann, Linde and R S Heiersted, P Paurola, A O Fredheim, Statoil: A New LNG Baseload Process and Manufacturing of the Main Heat Exchangers. LNG 12 Conference, Perth, May 1998. Abstract The development of natural gas fields off the coast of Nigeria is being continued in 2004. The specific benefits of the floating LNG-FPSO concept convinced the shareholders of these deposits to further pursue this concept following the example of the Snøhvit project which is under construction. This paper focuses mainly on the advantages of deep-water LNG-FPSO concepts due to the availability of cold cooling water, whereas installations based on GBS plants are forced to utilize relatively warm surface water for process cooling. Linde Technology I 1/ 2004 The Authors Eginhard Berger Eginhard Berger graduated in mechanical engineering at the Technical University, Munich. He worked in the aerospace industry before joining the Linde AG, Engineering Division, Höllriegelskreuth, Munich, in 1969. He was first involved in computerizing the physical properties of natural gas components. As a project and sales manager he was later decisively involved with the LNG projects at Snøhvit (Norway) and Xinjiang (China). In addition, Eginhard Berger functions as project manager for the development of offshore-LNG plants. He is, among other things, a member of the European Technical Committee for the standardization of LNG plants. Manfred Boelt Manfred Boelt studied technical physics in Munich and joined Linde in 1980. He started with process engineering and was engaged in the design of plants for the treatment, separation and purification of hydrogen as well as natural gas. From the beginning, he participated in the development of natural gas separation plants and natural gas liquefaction plants. He has a wide experience in natural gas plant designs and holds several international patents in this field. Bjørn Sparby Bjørn K. Sparby, Project Manager, INT GEX Africa, Statoil, has spent more than 28 years in the upstream oil and gas business. He holds a BS and MS degree from the University of Wyoming in petroleum engineering. Most of his career has been in Stavanger, Norway, with various technical and commercial responsibilities for domestic and international activities. He has been in charge of developing Statoil’s floating LNG concept since 1990. 43 e_44_49_Braeutigam_05 09.07.2004 13:50 Uhr Seite 44 Max Bräutigam LNG Baseload Plant in Remote North-West of China LNG Travels Through China The first large-scale LNG baseload plant in China is being erected in Shan Shan in Xinjiang province. The nominal production capacity of the facility is approx. 430,000 metric tons per year. The liquid natural gas produced by the plant will be transported over several thousand kilometers to consumers at the East Coast of China. The technical and logistic challenges of the construction of the plant are enormous and the climatic conditions extreme: temperature differences of 70 °C between winter and summer are common. In March 2002, the decision was made to build an LNG (Liquefied Natural Gas) baseload plant in Shan Shan in Xinjiang province in the North-West of China. This LNG baseload plant will be the first large-scale baseload plant in China, and it will be associated with far-reaching changes in the development of China´s natural gas industry. Natural gas consists mainly of methane whose volume is reduced by the factor 1:600 during liquefaction. The temperature of the liquid is -162°C at atmospheric pressure. The facilities comprise units for gas processing and liquefaction, intermediate storage, and unloading of the LNG into tank wagons and road tankers. The LNG will be transported overland across several thousand kilometers to satellite stations in the East Coast Provinces of China, where the LNG is re-vaporized and distributed via local gas grids to industrial and household consumers. New Era in Natural Gas Supply in China This LNG plant commences a new era of meeting the demand for natural gas in China, which continued to rise even during the recent economic crisis in China. The introduction of this type of LNG plant combined with the respective transport infrastructure is the basis for dynamic opening and development of the natural gas markets. The natural gas processed in the LNG plant consists mainly of methane (> 90%), which is a very clean fuel since it consists only of carbon (one carbon atom) and hydrogen (four hydrogen atoms). It is evident that natural gas will play an increasingly important role in the primary energy mix, since it has a low environmental impact in terms of emission levels. The Shan Shan LNG plant is a valuable 44 contribution to increase the wealth of the country similar to the commercialization of natural gas and LNG in other countries. This LNG scheme is unique in the world with regard to plant type as well as plant and transport capacity. The market is comparatively small and, therefore, a pipeline would not be economical. The method applied in the facility will have a positive impact on the way of energy consumption and on the development of the market for clean natural gas. Therefore, the project has attracted wide national and international attention. It can be considered as an incentive for the commercialization of remote gas resources with similar market conditions worldwide and particularly in China. Some data of the LNG plant in Shan Shan: ■ The production capacity of the plant shall be the LNG equivalent of 1,500,000 m3(n)/d with an expected on-stream time of 330 days per year. This amounts to approx. 430,000 MTY. ■ LNG production capacity the Shan Shan LNG plant will be approx. 3 times larger than that of the largest existing peakshaving plants, and only one-third of that of existing small baseload plants. ■ The storage capacity will be 30,000 m3 of LNG (equivalent to approx. 18 million Nm3 natural gas) corresponding to the amount liquefied in 12 days. ■ The capacity of the LNG shipping and distribution system is designed for loading 100 trucks / train transport containers within the 16 working hours per day of the unit. The ratio of trucks to movable containers is approx. 30:70. ■ The specification of product LNG contains no special requirements, with the exception of the nitrogen content being limited to 1 mol% max. ■ The feed gas operating pressure can range from approx. 0.6 MPaG to 1.1 MPaG. The feed gas operating temperature can range from -15°C to 40°C. ■ Composition of the feed gas: nitrogen = 4 mol%, methane = 81 mol%, ethane = 10 mol%, propane = 4 mol%, butane = 1 mol%. ■ In addition, CO2 and traces of H2S and sulfur are present in the feed gas. ■ The ambient conditions at the site are unusual for LNG plants: The average ambient temperature is 37.1 °C in the hottest month; the extreme ground temperature varies between approx. 75 °C in the hottest and -15.6 °C in the coldest month. Linde Technology I 1/ 2004 e_44_49_Braeutigam_05 09.07.2004 13:50 Uhr Seite 45 Figure 1: The LNG is transported from Shan Shan to Shanghai where it is locally distributed. The process plant for the LNG production has been optimized with regard to power requirement, equipment cost and the inclusion of a maximal amount of engineering works and supplies from China. The main plant units include feed gas compression, acid gas removal, drying, liquefaction, storage, loading and utility facilities. The compressor-driving concept and the power supply from the grid were designed taking into account the relatively high specific cost of electric power and high cost of the feed or fuel gas. The main process of the Shan Shan LNG plant is illustrated in the diagram in (Figure 2). Natural gas treatment The pressure of the natural gas (feed gas) entering the plant is too low for the liquefaction process to be efficient. Therefore, the natural gas is compressed in 3 compressor stages after removal of solid and liquid particles in a separator. Air coolers provide for intermediate and after-cooling during the compression steps. After compression, the feed gas is routed to the CO2 wash unit for removal of CO2. A MEA (monoethanolamine) chemical wash process using an aqueous MEA solution as the solvent was selected for the removal of CO2 from the natural gas. The feed gas enters the MEA wash column and flows from bottom to top through valve trays. The lean amine flows in the opposite direction, forms a very weak bond to the alkali, and thus extracts the carbon dioxide. The clean gas exits the wash tower with its CO2 content being 50 ppm (v) water saturated. The loaded amine solution from the CO2 wash column is regenerated in a strip column, which requires hot oil heating and air cooling in order to separate the CO2 from the loaded amine. The purified amine is returned to the wash column. The sweet feed gas exiting from the CO2 wash column is then routed to the drier station. Linde Technology I 1/ 2004 The drier station is a 2-bed adsorber station with a cycle time of approx. 8 hours The natural gas flows downwards in the first adsorber bed. The water contained in the natural gas is adsorbed by the adsorbent and reduced to a level , at which no freezing can occur in the downstream liquefaction section. During this period, the second adsorber bed is heated and then cooled by the regeneration gas stream. The regeneration gas is heated by hot oil and cooled by ambient air, and then guided to a regeneration gas knockout drum, where the water is removed. The operation of the two vessels is switched periodically (Figure 3). Natural Gas Liquefaction The liquefaction process (Figure 4) uses a closed mixed refrigerant cycle utilizing nitrogen, ethylene, propane, and pentane as its components. Once the H2O and CO2 are removed, the natural gas is routed to the cold part (cold box) of the process, which contains three spiral wound heat exchangers, which are integrated in one shell (“rocket”), and several separation vessels. The natural gas is first cooled in the feed gas pre-cooler, potential off-spec heavy hydrocarbons are removed in a feed gas heavy hydrocarbon separator. The gas is then condensed in a feed gas liquefier and the liquefied natural gas (LNG) is then sub-cooled in a feed gas sub-cooler. The LNG from the bottom heat exchanger is guided to the storage tank where it is relaxed to 45 e_44_49_Braeutigam_05 09.07.2004 13:50 Uhr Seite 46 atmospheric pressure. The gas fraction thus produced is then returned to the heat exchanger for feed gas cooling and subsequently used as fuel gas in the gas turbine. The cooling energy required for liquefaction is mainly provided by a closed refrigerant cycle. This cycle provides cold temperatures by JouleThomson expansion at 3 different pressure and temperature levels. A special feature of the cryogenic section of the process plant is the spiral wound heat exchanger designed and manufactured by Linde. It is characterized by its operative robustness in the natural gas pre-cooling, liquefaction and sub-cooling process, in which the refrigerant cycle and product streams attain temperatures down to -160°C. Refrigerant System Figure 2: Block diagram of the Shan Shan LNG plant showing the process units. Figure 3: Natural gas (NG) treatment process. The refrigerant gas stream is withdrawn from the shell side of the pre-cooling section of the cryogenic spiral wound heat exchanger set. The refrigerant is slightly super-heated above its dew point. before its is compressed in the 3-stage cycle compressor and after-cooled in an air inter-cooler, in which it is not only cooled but also condensed to some extent. The liquid formed in the after-cooler is removed in the cycle compressor discharge drum. The liquid collected in the discharge drum is routed to the pre-cooling section of the cryogenic heat exchanger, in which it is sub-cooled, expanded in a Joule-Thomson expansion valve, and then used to pre-cool the natural gas. The cycle gas from the discharge drum is cooled in the pre-cooling section to the same temperature, condensed to some extent, and fed to the refrigerant separator. The liquid collected in this separator is subcooled in the cryogenic heat exchanger liquefier section to a temperature that is sufficiently low for the liquid to be as a refrigerant in the liquefier section after expansion in a Joule-Thomson expansion valve. The vapor from the refrigerant separator is condensed in the liquefier section and sub-cooled in the cryogenic heat exchanger sub-cooling section to a sufficiently low temperature and provides the final cold for the natural gas sub-cooling after expansion in a Joule-Thomson expansion valve. After expansion to the lower pressure, the cycle gas streams are warmed up in the common shell side of the cryogenic spiral wound heat exchangers and returned jointly to the suction side of the 1st stage of the cycle compressor. Figure 4: Natural gas liquefaction process. 46 Linde Technology I 1/ 2004 e_44_49_Braeutigam_05 09.07.2004 13:50 Uhr Seite 47 Figure 5: The LNG plant (top) with the process units (under construction) and the LNG storage tank (bottom). Compressor Driver Configuration A gas turbine is used as the prime driver for the cycle compressor. The compressed boil-off, flash, and displacement gas from the LNG storage tank is used first as regeneration gas and then as fuel gas for the gas turbine. Special challenges for the design of the gas turbine as the main compressor driver and for the design of the air coolers are provided by the climatic conditions at the Shan Shan site with high fluctuations of the air temperatures during summer, winter, day and night times. The conventional refrigerant for cooling, cooling water, is not available in Shan Shan. Therefore, ambient air has to be used as the cooling medium. The feed gas compressor is driven by an electric motor powered by the local grid. LNG Storage and Loading System The LNG is transported from the liquefaction unit to the storage tank via the tank filling line (Figure 6). Either the bottom or the top filling connection can be used to fill the tank. If large differences in LNG density are encountered, top filling will be selected. The storage tank is equipped with measurement instruments for filling level, pressure, and temperature. The protection system of the tank is connected via the safety control system to the distributed control system. The temperature and the density of the LNG in the tank are measured throughout the height of the tank to monitor the risk of a possible roll-over in the tank. The tank is equipped with a control valve relief to the flare, safety valves to the atmosphere, and vacuum breakers for under-pressure protection of the tank. The tank will be filled continuously during operation of the liquefaction system at a filling rate of approx. 111 m3/h. During 16 hours per day a discontinuous send-out operation to the truck and container filling is scheduled. For send-out operation, two submerged in-tank pumps with a nominal capacity of 320 m3/h will be installed, which is sufficient to cover the send-out capacity. The filling time of one container or one truck is estimated to approx. 1.2 hours including connection / disconnection time. The filling system consists of 6 loading stations for containers and 3 loading stations for trucks. Linde Technology I 1/ 2004 Utilities All of the flash, boil-off, and displacement gas from the LNG storage tank is compressed, cooled by ambient air, and used as regeneration gas in the drier section before it is guided to the gas turbine to serve as fuel gas. In order to enhance plant efficiency, the waste heat from the exhaust stack of the gas turbine is recovered by using it to heat hot oil at two different temperature levels, which covers the heating requirements of the process plant. The hot oil is heated to approx. 260°C to supply the heat for regeneration gas heating, operation of the CO2 wash unit and start-up in the winter. The system is heat-traced. The LNG plant is equipped with two flare headers: a warm gas flare header and a cold gas and liquid flare header including a blow-down vessel for the separation of cold liquid and vapor. The plant is designed for non– flaring during normal operation. Make-up for the refrigerant system is required mainly due to cycle gas losses via the gas seals of the refrigerant cycle compressor. The quantities for the individual components are adjusted as required according to the compositions as measured and the temperatures in the cold part. The nitrogen is stored as liquid nitrogen and vaporized, heated to near ambient temperature, and fed to a compressor suction drum when needed. Commercial ethylene is stored in gas bottles at high pressure. Commercial propane and pentane are stored in separate tanks and fed to the refrigerant cycle suction drum when needed. 47 e_44_49_Braeutigam_05 09.07.2004 13:51 Uhr Seite 48 Abstract Figure 6: LNG storage tank and loading system. In March 2002, Xinjiang Guanghui Industry and Commerce Group Co. Ltd. awarded a contract for the erection of a LNG baseload plant in Shan Shan in Xinjiang Province in China. This is an important step towards a more extensive utilization of the gas from the remote Xinjiang region of China. The facilities include units for gas processing and liquefaction, intermediate storage, and unloading of the LNG into containers and into road tankers. The LNG is then transported over several thousand kilometers to satellite stations located in some of the East Coast Provinces of China, where the LNG is re-vaporized and distributed in the gaseous state via short pipelines to industrial and household consumers. The LNG production rate of the Shan Shan LNG plant will be 1 500 000 m3(n)/day, which is equivalent to approx. 430,000 metric tons per year. This contribution focuses on the challenges that have to be faced during the construction of the new LNG baseload plant in China. The Author Figure 7: CAD model of the Shan Shan LNG plant. Max Bräutigam Make-up water for the closed cooling water cycle for machinery cooling and demineralized water for use as make-up water in the MEA in the CO2 wash unit will be obtained from sources outside the plant. is a graduate of Munich University of Science. He joined Linde Engineering several decades ago and gained extensive experience in cryogenic process engineering and the mechanical design of components. He is now a Senior Gas Plant Sales Manager and involved in several projects. Project Execution The plant is currently under construction and the majority of the work on the LNG storage tank has been completed. The construction work on the site was completed early in 2004. The commissioning is imminent. The final layout of the plant is shown in the CAD model in Figure 7. The model shows the compressor house, the pipe rack and air coolers, and the cryogenic spiral wound heat exchanger set included in a steel structure. The equipment and piping was arranged in compliance with all applicable safety regulations and aiming for the shortest possible pipeline length. The plant covers approx. 60 m x 130 m. The height of the cryogenic heat exchanger is approx. 43 m. 48 Linde Technology I 1/ 2004 e_44_49_Braeutigam_05 09.07.2004 13:51 Uhr Seite 49 Industrial gas plants installed directly at the customer’s site are used throughout the world. This synthetic gas plant supplies Celanese AG at its Oberhausen, Germany location. Linde Technology I 1/ 2004 49 e_50_56_Morper_05 12.07.2004 14:46 Uhr Seite 50 Dr. Manfred Morper Upgrading Industrial Wastewater Treatment Plants Using Available Resources Creatively Industrial wastewater treatment plants, once designed and built, must be adjusted to comply with changing requirements more frequently than municipal effluent treatment plants. Such upgrades are a good opportunity to reassess the resources available on the site. Frequently this can help to find a technically and economically more advantageous solution than a mere repetition of what already exists. in order to develop a solution with maximum benefit from synergies. Creative use of all available resources on a site can greatly ease the situation. Existing facilities and resources Just like municipal effluent treatment plants, industrial wastewater treatment plants are designed and built for certain flows and pollution loads. Any increase or change of production is likely to increase both flow and load and may even substantially change wastewater quality. In addition, changing effluent regulations in the context of environmental legislation may require higher degrees of purification or the removal of additional pollutants. It is therefore not uncommon that an existing effluent treatment plant, although technically in good shape, no longer serves its purpose. Those in charge of managing the challenge will often prefer to tear the plant down and substitute it by a more suitable one. However, this is only rarely feasible for many reasons. Typically, there is not enough space available on the premises and ongoing production will not allow longer periods of interrupted operation. The budget allocated for the measures to be taken is hardly ever generous enough for comfortable redundancies. Last but not least, demolition and scrapping of tanks, pipes and electro-mechanical equipment is always also a destruction of capital, i.e. an increase of investment costs. Mainly due to lack of better knowledge, production sites are not particularly inclined to share utilities with the local effluent treatment plant, apart from such commodities as electricity and service water. Conversely, central waste water treatment plants at larger production sites with numerous contributors do not readily accept what they regard as odd sub-streams, bound to disturb the steady function of the wastewater treatment plant. Instead, costly pre-treatment on the spot is a typical requirement, particularly if the discharger is not also the operator of the wastewater treatment plant. The necessity to upgrade an existing industrial effluent treatment plant should be an opportunity to check the suitability of available resources of a site 50 In addition to the bulk utilities normally supplied centrally, e.g. electricity, potable and cooling water, steam and heat, data processing and communications services, there are other, less obvious, resources worth evaluating for creative upgrading solutions for wastewater treatment plants. ■ Civil structures, particularly tanks and reactors, the capacities of which can be increased or the functions of which can be changed. ■ Gaseous (e.g. oxygen, carbon dioxide), liquid (e.g. acids and bases, organic solvents) and solid (e.g. lime, alum) bulk chemicals, either produced on site or purchased for production purposes, can be supplied to and shared by all the users. ■ By-products of wastewater treatment can be used as low-cost utilities. Most major industrial wastewater treatment plants use biological treatment methods supported, where necessary, by physico-chemical process stages. The examples described in the following therefore also focus on biological systems. Alternative uses for existing tanks Tanks used for wastewater treatment are either designed on the basis of hydraulic (e.g. buffer and equalization tanks, sedimentation tanks) or kinetic (e.g. aeration basins, digesters, chemical reactors) parameters. Once built, their volumes are fixed. For those of hydraulic design, increased flows or change of function generally results in a certain decrease in performance. This is acceptable, if it can be compensated by improved performance on the part of the associated facilities of the treatment chain. Linde Technology I 1/ 2004 e_50_56_Morper_05 12.07.2004 14:46 Uhr Seite 51 If, on the other hand, space is scarce and upgrading cannot be done without erection of new tankage, a change of function of existing tanks may be of advantage. This has been demonstrated for a pulp and viscose mill in India [1]. The existing treatment plant, consisting of two square tanks for equalization and neutralization, two circular primary clarifiers and an aerated lagoon (the biological stage) did not perform according to expectations and requirements. The lagoon was identified as the weak spot and had to be replaced by an efficient state-of-the-art bioreactor. Due to the lagoon’s large footprint, very little firm ground was left on the factory site for building new facilities. Although an oxygen aerated bioreactor of the LINDOX® type, known to have low space requirements, could be placed there, all other elements of the treatment chain had to be created from the existing tankage. For this purpose, one of the two equalization tanks was converted into five parallel primary clarifiers by constructing separation walls. Thus the two original primary clarifiers became available as secondary clarifiers for the new LINDOX® reactor. Figures 1 and 2 show the site plan of the plant before and after upgrading. Figure 1: Existing wastewater treatment plant at a pulp and viscose mill in India. Performance improvement of aeration tanks The activated sludge process is by far the most widespread method of biological wastewater treatment. The limiting design parameters (sludge age, sludge loading) are derived from the ratio of the daily pollution load to the quantity of active biomass in the bioreactor, also known as the F/M (food to micro-organism) ratio. The quantity of biomass is the product of the biomass concentration (MLSS) and the reactor volume. An overloaded plant can be defined as not having enough biomass for a given pollution load. Obviously, the required ratio must be established by increasing the biomass quantity. This can be done either by increasing the reactor volume, which means construction of new tankage, or by increasing the biomass concentration, which means no, or at least less, new tankage. The LINPOR® system, which can be applied for almost all kinds of common aeration tanks, substantially increases the biomass concentration without requiring major modifications to existing tankage. The effect is achieved by filling a certain quantity of highly porous plastic sponge cubes into the aeration tank, which serve as mobile carriers for the active biomass. Just like large biomass flocs they follow the hydraulic regime of an aeration tank. Contrary to conventional biomass flocs, they are retained in the aeration tanks by specially designed screens and thus do not contribute to the solids load of secondary clarifiers. Linde Technology I 1/ 2004 Figure 2: Extended wastewater treatment plant at the pulp and viscose mill showing changes of use for existing tanks. As all LINPOR® systems derive from the conventional activated sludge process, proven and readily available equipment for this established technology can be kept. LINPOR® is therefore particularly suitable for upgrading existing activated sludge plants. One of the numerous examples is the wastewater treatment plant of a waste paper processing mill in South Korea, where the overload situation of the existing plant was caused both by a production increase and poor sludge settling characteristics (bulking sludge). Space shortage demanded that performance be improved essentially by the appropriate modification of the existing facilities. The existing aeration tank was therefore converted into a LINPOR®-C reactor by the installation of effluent screens and an air-lift pump to distribute and 51 e_50_56_Morper_05 09.07.2004 12:00 Uhr Seite 52 Figure 3: Aeration tank of a paper mill in South Korea, converted to the LINPOR®-C process regenerate the carrier material and by replacing the surface aerators with a fine bubble aeration system, using standard membrane diffusers. In order to accommodate the full flow and load, a smaller new LINPOR®-C reactor and an additional secondary clarifier were erected on a narrow stretch of land next to the existing facilities. Figure 3 shows the aeration tank which was converted into the LINPOR®-C reactor. Change of function of existing aeration tanks Over the years, effluent quality requirements have become more and more stringent in most industrialized countries. Treatment plants, once designed for compliance with effluent limits for organic materials (BOD/COD) and SS now also have to eliminate additional pollutants such as N- and P-nutrients. TKN (ammonium and organic nitrogen) is biologically removed by nitrification and denitrification, which, chemically speaking, is oxidation to nitrate followed by reduction to the gaseous molecular nitrogen. Due to the slow growth rate of nitrifying bacteria, nitrification is the rate-limiting step. In order to provide favorable growth conditions for nitrifying bacteria, the respective aeration tank must be designed with a high sludge age or a low F/M ratio, which once again means a larger quantity of biomass than that needed for carbonaceous removal only. 52 For municipal and comparatively polluted industrial effluents, this biomass increase can be comfortably achieved by converting aeration tanks into LINPOR®-CN reactors for simultaneous carbonaceous and nitrogen removal [2]. However, for industrial effluents with both high organic and TKN pollution, the requirement of a low F/M ratio would result in extremely large aeration volumes. This can be avoided by separating carbonaceous removal and nitrification in a two-stage plant. If the organic load is removed in a first aeration tank with a comparatively high load, the requirement of a low F/M ratio for nitrification at the second stage is achieved in a comparatively small reactor volume. Unfortunately, nitrifying bacteria on their own do not readily form from settling sludge flocs, but they do attach easily to carrier surfaces. The LINPOR®-N process, where all the biomass is fixed on the mobile carrier material, again takes advantage of this fact. At a major coke works in Germany, total N effluent limits could be achieved by converting the existing activated sludge plant, once designed for carbonaceous removal, into a two-stage facility [3]. The new first stage consisting of an aerobic and an anoxic section both the organic load is removed and the nitrate, generated in the second stage and recycled to the first stage, is denitrified. The second stage is the aeration tank of the existing plant, now converted into a LINPOR®-N reactor. Even the existing sedimentation tank is integrated into the treatment chain for the polishing stage. Figure 4 shows an aerial view of this plant. Linde Technology I 1/ 2004 e_50_56_Morper_05 09.07.2004 12:01 Uhr Seite 53 Figure 4: Two-stage treatment plant for wastewater from a coking plant in Germany, showing the use of an existing aeration tank (circular tank left on the top) as a nitrification reactor according to the LINPOR®-N process Sharing of on-site utilities The upgraded coke oven effluent treatment plant described above is also a good example of how existing resources of a site can be of advantage for wastewater treatment. The new first stage uses pure oxygen instead of air for the aeration tank. Oxygen is produced in large quantities for the associated steel works. By diverting the comparatively minor quantity needed for effluent treatment, a host of technical and economical advantages are achieved: ■ Neither compressors nor large numbers of aerators are needed ■ Energy expenditure for oxygen dissolution is low ■ The aeration volume is considerably smaller than with air aeration ■ Foaming is no longer a problem ■ Adjustment to load variations is easy ■ No odor or corrosion problems from stripping effects occur For the elimination of recalcitrant organic substances, contained in particular sub-streams or in biologically pre-treated effluents, chemical oxidation by ozone is a potential solution. As ozone has to be produced on site from oxygen anyway, the use of oxygen for the biological treatment of the bulk wastewater is reasonable in order to avoid wasting oxygen. A covered LINDOX® reactor will also decompose non-reacted ozone without the costs associated with a chemical decomposer. The block diagram in Figure 5 shows the dual use of oxygen for both production and wastewater treatment. Other options for reducing investment and operation costs by the sharing of operational utilities in production and the treatment of wastewater include lime cycles [4], carbon dioxide from oxidation processes for neutralization or the use of organic solvents – even spent ones – as carbon sources for denitrification. For any industry bound to produce and use oxygen (or nitrogen) for its genuine production, using oxygen for wastewater treatment is logical. Potential candidates include steel works, pulp mills and major chemical and petrochemical factories. The availability of oxygen also facilitates other treatment options. Linde Technology I 1/ 2004 53 e_50_56_Morper_05 09.07.2004 12:01 Uhr Seite 54 neutralized with mineral acid and chemically oxidized with air or oxygen under pressure or oxidants such as H2O2. A suitably designed and operated closed oxygen aerated LINDOX® reactor does not require such costly pre-treatment. Due to its low exhaust gas flow, almost all of the carbon dioxide produced remains available for neutralization. Industrial Production Figure 5: The use of oxygen integrated into the production process for wastewater treatment Using treatment by-products as low-cost utilities Substances thought to be particularly toxic for microorganisms in wastewater treatment plants are generally eliminated by physico-chemical pre-treatment prior to biological treatment of the respective effluents. This is costly and often not necessary, if a suitably designed and operated biological wastewater treatment plant is used. Carbon dioxide for neutralization In aerobic wastewater treatment plants, the organic pollution load is partly converted into biomass (surplus sludge) and partly oxidized, as shown by the following equation: CxHyOz + ( x+ y/4 – z/2) O2 ➔ x CO2 + y/2 H2O In conventional air-aerated plants the carbon dioxide produced is mostly stripped off, i.e. driven off, with the waste air and lost. In petrochemical plants and refineries spent caustic is a very common wastewater, resulting from scrubbing acidic components (e.g. H2S, SO2) from gaseous products with NaOH. Central biological wastewater treatment plants generally do not accept untreated spent caustic because of the toxic effect of high pH and the content of reduced sulfur compounds. Traditionally, spent caustic is therefore 54 CO2 + NaOH ➔ NaHCO3 Normally, the amount of carbon dioxide produced exceeds the stoichiometric demands for spent caustic neutralization. Indeed, carbon dioxide consumption by NaOH neutralization improves the oxygen utilization rate, i.e. the economics of biological wastewater treatment, as lowering the partial pressure of carbon dioxide increases the partial pressure of oxygen in the gas phase and thus reduces the waste gas flow needed to maintain a set level of the latter. Reduced sulfur compounds at sub-toxic levels are readily oxidized biochemically in aeration tanks as shown by the following equations, provided that the oxygen concentration is maintained high enough to allow maximum reaction rates and to prevent subsequent reduction of sulphates by sulphate-reducing bacteria. S2- + 2 O2 ➔ SO422 SO32- + O2 ➔ 2 SO42In order to avoid the toxic effects of alkaline pH and respective concentrations of sulphur species, it is necessary to dose the spent caustic not into the bulk influent but into the different compartments of the biological reactor at controlled pH. This is shown in the diagram of Figure 6. An integrated spent caustic treatment of this type has been in continuous operation at a major petrochemical site in France for 25 years [5] [6]. Nitrate for sulfide oxidation Nitrate in industrial wastewaters, although sometimes due to nitric acid application for production purposes, is mainly due to TKN nitrification and therefore a wastewater treatment by-product. In order to comply with respective effluent quality requirements, it is biologically removed by denitrification, using either raw wastewater organics or external carbon sources, e.g. methanol, as reducing agents. But nitrate can also be regarded as an oxidant for suitable pollutants, such as reduced sulfur compounds, which are found in many industrial effluents. There are autotrophic bacteria that can oxidize sulphide by denitrification of nitrate, as shown by the following equation: 5 S2-+ 8 H+ + 8 NO3- ➔ 4 N2 + 5 SO42- + 4 H2O Linde Technology I 1/ 2004 e_50_56_Morper_05 12.07.2004 14:46 Uhr Seite 55 Waste Gas Spent caustic (NaOH, Na2S, Na2So3) Oxygen Wastewater Treated Wastewater Excess Sludge Returned Sludge Figure 6: Treating non-pretreated spent caustic in the oxygen-aerated activated sludge plant of a petrochemical works in France This reaction is used at a refinery in Germany as the first of three biological treatment stages [7] as shown in Figure 7. A recycle stream, taken from the clarifier effluent of the second nitrifying stage is fed into the denitrifying sulphide oxidizer, which is a combined mixing and settling tank. This flow pattern provides favorable growth conditions for the autotrophic denitrifying bacteria, separate from the heterotrophic denitrifiers in the second stage. Surplus sludge for heavy metal removal Aerobic and anaerobic surplus sludge from biomass growth is an unavoidable by-product of biological wastewater treatment. It is primarily a waste product, the disposal of which contributes to the overall costs of wastewater treatment. However, the physico-chemical properties and the metabolism of viable anaerobic sludge, as obtained from anaerobic digestion of primary and secondary sludges, make it a unique biosorbent for the removal of heavy metals [8]. Functional groups of bacterial cell walls act as chelating agents, while metabolites such as CO2 and H2S form insoluble metal precipitates The METEX® process, using an anaerobic sludge bed with an upward flow pattern, has been in operation at various industries with metal bearing wastewaters for more than a decade without it ever being necessary to change the sludge beds Linde Technology I 1/ 2004 Figure 7: Use of biochemically generated nitrate for sulphide oxidation in the treatment of wastewater at a refinery in Germany 55 e_50_56_Morper_05 09.07.2004 12:01 Uhr Seite 56 Conclusions There are several aspects to the upgrading of industrial wastewater treatment plants. Primarily, it refers to the measures to be taken in order to make an existing facility cope with flow and/or load changes or changing (i.e. generally more stringent) effluent quality requirements. Other aspects are the improvement of a plant’s performance and the economics involved. A thorough check of all available resources of a site can help to find a technically and economically more advantageous solution than a mere repetition of what already exists. The capacities or functions of existing tanks can be increased or changed by application of advanced process technologies. Utilities can be shared with other users. Even metabolites of biochemical reactions occurring with wastewater treatment can be used as low or no cost utilities. The potential is certainly larger than the few examples given. However, higher degrees of integration mean higher levels of interdependence, and therefore more mutual responsibility and less forgiving attitudes on a site. Hence efficient monitoring and control, highly qualified and communicative personnel and an overall management capable of enforcement are indispensable requirements for maximum synergistic benefits. Abstract The extension and conversion of effluent treatment plants on industrial sites is a job that recurs regularly in plant engineering and construction. The cost in time and money can be minimized by the innovative use of existing structures. The construction of new facilities is avoided. Making skilful use of resources and shared utilities leads to an improvement in working procedures and a saving in running costs. Conversely, products from effluent treatment can also be reused as utilities. Numerous projects around the world are already profiting from creative measures like these. 56 Literature [1] M. Morper, gwf Water Wastewater, No. 14/1999, pp 22 – 25, ISSN 0016-3651 B 5399 [2] M. Morper, Wat. Sci. Tech. Vol. 29, No. 12/ 1994, pp 167 – 176 [3] M. Morper, A. Jell, Linde Reports on Science and Technology, No. 62/2000, pp 20 – 26, ISSN 0942 – 5268 [4] M. Morper, Chemical Engineering, Volume 106, No. 8/1999, pp 66 – 70, ISSN 0009-2640 [5] C. Granger, Verdeil, Eaux de Rhone – Méditerranée – Corse, troisième trimestre 1981 [6] M.Morper, personal information on site Feb. 2003 [7] H.A. Joel, Th. Jenke, Erdöl Erdgas Kohle, Volume 110, No. 4/ 1994, pp 171 – 173 [8] T. Pümpel, K.M. Paknikar, Advances in Applied Microbiology, Vol. 48/2001, Academic Press, San Diego, p 135 – 171 The Author Dr. Manfred Morper Dr. rer. nat. Manfred Morper studied chemistry at the Technical University of Munich and was awarded his Ph.D. in that subject. Afterwards he worked at the Bavarian Regional Water Management Authority on the subject of industrial wastewater treatment plants, before coming to Linde in 1980. Here he worked in R&D until 1986. In 1987 he was appointed Head of Department with responsibility for the project planning, marketing and construction of waste-water treatment plants within the Plant Construction division. Since 1994 he has headed the Environmental Technology Department of Linde-KCA-Dresden GmbH on the company’s Höllriegelskreuth site in Munich. Manfred Morper has written over 50 articles for technical magazines and papers for international conferences on wastewater treatment. Linde Technology I 1/ 2004 e_01_Titel_Umschlag_03 09.07.2004 11:22 Uhr Seite U3 Imprint Reports on Science and Technology Linde Technology 1/2004 Designing Biotechnology Plants Publisher: Linde AG, Wiesbaden www.linde.com Editorial Staff: Editor-in-Chief: Stefan Metz, Linde AG; Science&Media, Büro für Wissenschaftsund Technikkommunikation, Munich Forklift Ergonomics Cracking with Oxygen Economic Ammonia Production LNG for Land and Sea Flexible Solutions for Wastewater Treatment Layout, lithography and production: D+K Horst Repschläger GmbH, Wiesbaden Translation: eurocom Translation Services GmbH, Vienna Printing: HMS Druckhaus GmbH, Dreieich Inquiries and orders: Linde AG, Communication P.O. Box 4020, D-65030 Wiesbaden or [email protected] No part of this publication may be reproduced or distributed electronically without the prior permission of the publisher. Unless expressly permitted by law (and, in such instances, only when full reference is given to the source), the use of “Linde Technology” reports is not permitted without the publisher’s consent. ISSN 1612-2232 Printed in Germany – July 2004 www This series and other technical reports can be downloaded at www.linde.com/Linde.Technology. e_01_Titel_Umschlag_03 09.07.2004 11:22 Uhr Seite U4 Linde AG Abraham-Lincoln-Straße 21 65189 Wiesbaden Tel.: ++49 (0) 611 770-0 Fax: ++49 (0) 611 770-269 www.linde.com ISSN 1612-2232