Like any other risk, development risk is determined by the probability of problems occurring, and the extent of these problems. The extent problems, i.e. damages, depend especially on the effort and consequential costs involved, which are determined by factors such as not being able to complete delivery contracts (Ill. 14-17).
The estimation of the probability of success by the responsible persons is of paramount importance for decisions regarding development. Reaching decisions on this subject demands a very informed, objective analysis. This, in turn, requires sufficient technical knowledge, a comprehensive understanding of the state of technology, as well as appropriate judgment and experience. As always, the size of the acceptable risk depends on the probability of success and the resources available.
In order to minimize risk, it is advantageous to know the influencing factors (the following order does not indicate importance, nor is there any claim to completeness):
Causes for the development risk:
One can see that the most serious errors can be made in the decision-determination phase for a development. For this reason, the technical preparation of the decision-making process is of vital importance, and must be used as the basis for the estimation of costs and time requirements. Only then can those responsible for the development make decisions with a definable amount of risk. Many additional influences act on decision. Some, such as control, workload, personal future goals, and tendencies, are subjective. This realization seems self-evident, but experience has shown that it is not. In this way, corporate culture affects the value of technical understanding in the hierarchies, and therefore influences the risks of a development.
Fundamental rules need to be observed in order to minimize the development risk. These are included as notes is this chapter (text boxes with gray backgrounds), although with no claim to completeness.
Fiber-technical (Ill. 14-28 to 14-32) and monolithic ceramic (Ill. 14-34) turbine engine technologies are used as examples of the development and implementation of new technologies.
Illustration 14-17: The top diagram shows a pragmatic approach (Ref. 14-20) for the estimation of time requirements. The complexity of the problem often precludes accurate results, even with extensive resources and effort. The depicted principle shows that roughly 80% of a product`s maturity only requires 20% of development costs and work. The greatest costs are are contained in the last 20% of maturity. This means that:
A largely completed development still contains considerable risks before it reaches the maturity necessary for serial implementation.
The risks (i.e. costs) for resolving problems in a knowledge-based approach (Ref. 14-19) are shown in the middle diagram. Implementation is computer-aided. The use of simulations and animations makes it more likely to discover hidden flaws early in the design and development phases, and to improve the FMEA (Failure Mode Effects Analysis, Ref. 14-28) This increases the probability that flaws will be discovered early, i.e. in the development phase, and reduces risk considerably.
Costs increase exponentially with the progression of development to serial implementation.
The typical progression of costs over the life cycle (life cycle costs = LCC) can be seen in the bottom diagram (Ref. 14-18). Experience has shown that
LCC can be controlled most effectively at the beginning of a project.
The possibility for influencing LCC decreases considerably as development progresses. The Life cycle costs are decisively determined by the initial phase. This is understandable, since the greatest portion of costs (time and effort, top diagram) occurs late in the development cycle.
Example14-4.1 (Ref. 14-30):
Excerpt: “…The main change entailed raising the cooling tube section from 4×4 mm to 4×6 mm, allowing the number of tubes to be cut from 456 to 288 and production time to be sliced from 13 weeks to five…the conjunction of costcutting and performance enhancement requirements had forced designers `to flirtwith the limits of the state of the art.'…although tubewall thickness was increased from 0.4 mm to 0.6 mm to compensate for the increase in tube size, the design changes virtually doubled inflight aerodynamic loads…'the changes…were dictated by the market and…the limitation of finite-element modeling systems used to validate the design…made it impossible to fully predict the effect of thermomechanical stresses in flight. It's part of the normal learning curve.'
…The inquiry panel recommended that engineers change the nozzle design, `drawing on experience obtained with Vulcain 1'…Ariane 4 has seen years of service without incident….(comment about former finite-element calculations) 'It's obvious we underestimated the amount of margin required for existing procedures.'
Comments: This incident concerns the crash of a large civilian rocket (Ariane 5). The damage costs were several billion Euros and were accompanied by major loss of prestige and high consequential costs (e.g. insurance premiums, temporal delays). Apparently, the horizon of experience from the previous version was only crossed to a very small degree (Ariane 4). The computer-aided calculations do not seem to have been able to ensure sufficiently safe conclusions. It is questionable whether this type of disaster can still be considered to be part of a “normal” learning curve.
Illustration 14-18: An applied technological development must meet certain requirements in order to minimize the development risk. The table contains typical examples with no claim to completeness.
Realistic goals are a prerequisite for a successful development. This means that, in the case of a user, serial application of the technology must be the ultimate development goal. For this reason, development of semi-finished parts and fundamentals (Ill. 14-21) should be assigned to “natural” partners such as universities and suppliers of semi-finished parts. The user concentrates on tasks such as:
One should always assume that obvious, innovative ideas have already occurred to other competitors who are equally capable of their realization.
A high potential development risk accompanies tasks where activity has been reported in the field in question, but was evidently stopped for some reason. This can indicate imminent success, but experience has shown that it is more likely that an (at least temporarily) unsolvable problem occurred (in extreme cases, a “show stopper” Ill. 14-5).
Therefore, the most important question is:
Why was this idea no realized (yet)?
Only after this question has been plausibly answered should a decision regarding development be made.
Fiber-technical containment rings have long been successfully used in civilian engines. The depicted example explains why there has not yet been a known serial application of fiber-technical containment rings in fighter aircraft engines. A lack of space to make use of the required expansion of the ring prevents weight-optimized functioning. There are indications, that an introduction of this technology will soon take place. However, one can only speculate about which conditions have changed to make this possible.
Illustration 14-20: In order to recognize the potential of a technology and better estimate the development risks, there are certain tests that can be used. These tests must convincingly prove the concept or function, and therefore also demonstrate the advantages and disadvantages of the technology to the decision-makers (proof of concept). However, experience has shown that this very sensible method of approach harbors large risks. The difficulty lies in selecting a suitable test for verifying the concept in question.
A prerequisite is that the life-determining damage mechanism is also activated by the demonstration test.
These demonstration tests often resemble medieval heavenly judgements in which, for example, a torn rope would prove the innocence of the criminal being hung. A typical example is verifying the potential increase in the cyclical life span of a highly stressed rotor disk in a rotor burst test. In this case, cyclical fatigue is replaced by violent tearing apart. These two damage processes are fundamentally different and are controlled by different material properties. The result of the fatigue test (dynamic strength) depends decisively on a certain present flaw size. In the burst test, the fracture strength and fracture strain (stress transfer due to plastic deformation) are tested. The result usually only has a minor dependency on the relatively small flaws. It is entirely possible that a special, less plastically deformable material (e.g. with fiber-technical reinforcements) would not permit any noticeable improvement in the rotor burst RPM relative to the original version. Stress concentrations can be balanced out by a material with high ductility. However, at the lower operating RPM, a less plastically deformable, fiber-reinforced material may make considerably longer cyclical life spans possible. In this case, the test would not be able to show this advantage. The opposite, more dangerous situation is also possible, if the test indicates advantageous behavior that cannot be realized later.
Therefore, one must always be certain which advantage or risk is to be demonstrated, and the demonstration test must be selected accordingly.
There is an extremely high development risk if a necessary step for realization of serial implementation of a technology is missing.
This can be a semi-finished product or a procedure.
In order to realize a monolithic turbine disk, for example, a material is required that is suitable for the specific part geometry. Requirements are high strength and little scattering. While these properties seem attainable in light of the medium-term trends, they have not yet been realized.
Experience has shown that waiting for a missing technological step, even if deadlines have been agreed upon, frequently leads to failure of the entire project. In questionable cases, the new technology should be verified
Illustration 14-22: Technologies and designs with growth-capable flaws that cannot be sufficiently safely found through non-destructive inspection present a high risk. Therefore:
A realistic and sufficiently secured, verifiably serially implementable quality assurance system must be in place, which guarantees that flaws greater than a maximum flaw size determined by the design cannot occur.
The maximum allowable weak point is defined by the zone responsible for part strength. The designed life span is determined by operating requirements.
Quality assurance must be ensured with suitable non-destructive testing and/or process controls (Ref. 14-21). However, one must remember that
non-destructive testing is preferable to process controls.
The prerequisite for a sufficiently safe process control is that all relevant procedural parameters and effects are controlled and that no unnoticed deviation occur. These include the failure of monitoring equipment, as well as human error.
Example 14-4.2 (Ref. 14-22):
Excerpt: ”…According to …(the OEM) the …(engine) burns some 6 percent more fuel than original estimates, a discovery that will likely force the company to redesign the high pressure compressor…The source of the fuel-efficiency problem appears to stem from …(the OEM's) aim to make the engine simpler and lighter by minimizing the number of stages in the high-pressure compressor. But the…five-stage design may place too much strain on each stage of the compressor…The …(OEM) has considered, adding another stage and using lighter blades and discs throughout the compressor to compensate for the resulting weight increase…“
Comments: Evidently the horizon of experience of the compressor design was exceeded to such a great degree that a redesign became unavoidable in the advanced development stage. This threatened the success of the entire project.
Illustration 14-23: The horizon of experience describes the limits of operation in which the part, technology, design, and configuration have proven themselves. The horizon of experience should be defined in a specified location (e.g. technical design manual).
There are many examples from turbine engineering in which an extrapolation only exceeded the horizon of experience by a very small degree. As a result, despite the empirical development, realization was no longer possible and/or high damage rates resulted. This type of situation can occur in various fields, such as strength, process engineering, aerodynamics (Example 14-4.2), and operating loads (Volume 2, Ill. 6.1-15.1, Example 14-4.1).
For example, a design-conforming surface pressure that exceeds the limits that have proven safe in actual operation may lead to dynamic fatigue fractures after unacceptably short run times (diagram). This is the case if the dynamic loads exceed the fatigue strength due to load changes and/or a decrease in dynamic strength. A dangerous increase in the dynamic loads may be traced back to a rise in the friction forces at the contact surfaces over the period of operation. Even if the loads do not increase, a decrease of the utilizable dynamic strength due to influences such as fretting or mean stress increases can contribute to serious life span reduction.
This means that:
Even seemingly minor extrapolations within the framework of an empirical development beyond the horizon of experience can be extremely risky. If in doubt, the necessary experience must be acquired sufficiently early.
Illustration 14-24: The development risk is especially dependent on the consideration of a technology in the design.
A technology that is a component of the life span design demands a much greater development effort and cost, and therefore has a considerably greater development risk than a technology that is merely intended for damage minimization.
Typical examples include the use of thermal barrier coatings (top diagram; also see Ill. 12.1-5) in order to increase the gas temperature, and the use of shot peening to increase fatigue life (bottom diagram; Ill. 14-7). As long as these technologies are used to ensure a part life that can also be attained without them, they contribute to safe operation. However, if the design requires these technologies in order to reach a certain life span, the failure risk rises considerably.
Illustrations 14-25 and 14-26: The prerequisite for the success of a revolutionary technology is that it reaches serial application. However, certain conditions must be fulfilled for this to happen:
This means that the development of technologies must be a long-term process (Ill. 14-36). It must be completed even before the project phase. In the initial phase, technological development must be conducted as a long-term strategy, independent of any special project. The following examples show the importance of this requirement:
Example 1: The repair and logistic problems of a blisk demand a very low FOD damage rate. This can be achieved through suitable shaping of the fan area and HPC inlet. The HPC inlet, which is set at an angle to the bypass duct, redirects foreign objects (also see Volume 1, Chapter 184.108.40.206).
Example 2: In order to be able to use compressor blades and other parts (housings, spinners) made from fiber-reinforced synthetics in fighter aircraft engines, the bird strike risk must be minimized (Volume 1, Ill. 220.127.116.11-5). Suitable angling of the inlet duct prevents birds from striking
the fan area directly. If the bird first strikes the wall of the inlet duct, it desintegrates and the blade stress is minimized due to the small size of the bird parts.
Example 3: Extensive investigations have shown that fiber-reinforced synthetic fan rotor blades can only be safely used below a certain circumferential speed (impact speed of the bird; top right diagram, Ref. 14-27). This must be taken into consideration when designing and configuring the fan (Volume 2, Ill. 8.2-18).
Example 4: Weight-optimized fiber-technical containment requires sufficient radial expansion. This must not be restricted by surrounding components. Therefore, the mounting of the auxiliary components must already consider the containment concept.
Illustration 14-27: Weaknesses of a technology can have very different importance, depending on the application. Characteristics that prevent implementation in one case may be acceptable or advantageous in another. A typical example is technologies for the engines of unmanned fighter aircraft. These aircraft probably require considerably lower life spans than manned fighter aircraft. Therefore, the oxidation-dependent limited life span of fiber-reinforced ceramics could be tolerated, and their benefits could be used. The intense application of fiber-reinforced synthetics in the compressor would no longer be prevented by erosion problems and the threat of dangerous bird strikes. The cost and effort required for life span verification should be relatively low.
This scenario also presents a rare situation in which revolutionary development steps may be possible, which would be far too risky in manned fighter aircraft.
The use of a technology demands the optimal utilization of its advantages. In addition, special design characteristics must reduce the specific drawbacks to an acceptable level.
The aspect of safety is especially important for the application of a technology. Fiber-reinforced synthetic compressor stator vanes in fighter aircraft are a good example of this.
For a long period, the requirements for acceptably minor damages in case of bird strikes prevented the use of fiber-reinforced synthetic blades in stators behind the first rotor stage. The filigreed titanium alloy stators in modern fighter aircraft, combined with the high flight speeds at low altitudes, show that in the case of a high-speed impact (Volume 1, Ill.18.104.22.168-7), even metallic blades can be expected to fracture. Given these conditions, fiber-reinforced synthetic blades are advantageous because they usually only cause minor consequential damages, giving them an advantage over metallic blades.
Changed safety philosophies demand sufficiently early coordination with the rating authorities. The verification of sufficiently good operating characteristics must also be determined.
If authorities are not contacted until late in the process, there will be a justified suspicion that the development goal can only be reached with concessions. This will make good arguments suspect or worthless.
Illustration 14-29: Fiber-reinfoced synthetic compressor blades have important advantages over metallic blades (Ill. 14-28). Their good performance relative to dynamic fatigue is especially important. As early as the design stage, the same blade profile can be made more resistant to resonances by changing the various oriented fiber layers, which influences the eigen frequencies of the blade.
The high damping of the fiber-reinforced synthetic material allows it to tolerate these vibration excitements, which would be dangerous for metallic blades.
The failure mechanism due to dynamic fatigue does not occur as uncontrollably fast crack growth, as it does in metals. In fiber-reinforced synthetics, the fatigue causes delamination (separation of the fiber layers, Ill. 14-30) with no dangerous fiber breakages. This process increases damping, and a sufficient tensile strength is still guaranteed. This results in a certain fail-safe behavior. This type of damage can be detected in time during inspections, considerably reducing the risk of damage due to fatigue.
Illustration 14-30 (Ref. 14-24): The “positive” failure mechanism of fiber-reinforced synthetic materials under dynamic loads (Ill. 14-29) is an important advantage over homogenous metallic materials. With fiber-reinforced synthetics, the low notch-sensitivity means that even crack initiation weak points are large relative to those in metals (top curve). Various damage modi can originate here (matrix cracks, delamination, fiber fractures). They distribute the applied dynamic energy across a greater volume. The ultimate failure of the part with accelerated damage growth only occurs after the damage is relatively large. The recognizability of damage such as delamination with considerable strength losses is correspondingly good (Ill. 14-31). In homogenous metallic materials, the flaw size at which cracking begins is relatively small (lower curve). The typical damage mode is crack growth. The crack remains small for a long time, and then grows rapidly. The period between the time critical crack size is reached and the time the cross-section fails is relatively short compared with fiber-reinforced synthetic materials. On the whole, the damage progress under equal loads takes longer in fiber-reinforced synthetics than in metallic materials. This means that the chances are greater that failures due to dynamic loads will be detected in time to prevent failure in fiber-reinforced synthetic parts.
Illustration 14-31: Different, material-specific failure mechanisms must be accounted for by the design. If, for example, a unilaterally fastened metallic beam with an thicker fastening section is placed under static flexural loads (top diagram), it will result in plastic deformation and cracking perpendicular to the surface. The crack leads to accelerated strength losses and fracture.
A fiber-reinforced synthetic bending beam (bottom diagram) will fail in same area as the metallic beam. However, the failure mechanism takes the shape of a separation of the layers near the surface, i.e. cracking parallel to the surface. At the same time, the strength losses are fairly moderate, because the delaminated fiber layers can still absorb tensile stresses. This means that the crack makes itself known sufficiently early and the problem can be dealt with in time.
Only sufficiently realistic tests that are close to operating conditions can reveal advantages and problems of a technology. Fundamentally:
Anything that one has not developed or at least verified oneself, is probably not sufficiently understood. This leads to a high risk for serial implementation.
In the depicted example, different fiber-reinforced synthetic spinners exhibit very different properties during bird strike tests. These tests show that expected advantages of “new fibers” and “new resins” could not be realized under impact loads. The best material for this part remains a glass fiber-reinforced epoxide resin that has been in use for many years.
Illustration 14-33: Material data are the basis for the design and are therefore vital for the safety of the part. For this reason,
the properties of a specimen must sufficiently correspond to those of the part that the specimen is intended to provide information for.
Also see Ills. 22.214.171.124-6 and 12.6.2-23. Specimens that were produced separately often differ from later serial parts in their material and production conditions. Before a serial application can be considered, specific values must be determined using specimens of representative parts. The location of the specimen in the part must also be accounted for. If special technologies, such as joining techniques, are used, their influence on the design data must be determined under the named factors.
Illustration 14-34 (Refs. 14-25 and 14-31): Brittle materials such as ceramics are especially notch-sensitive and react to very even small flaws by cracking and fracturing. This behavior means that their strength is highly dependent on the probability of flaws occurring. The greater the critical, highly stressed volume or surface (depending on the flaw location), the smaller the usable strength. The measured values are influenced considerably by the size and stress distribution in the specimen (Ill. 12.3-7). Even if the specimen geometry is the same, a three-point test will result in higher values than a four-point test. The lowest values can be expected from the tensile specimen. Under a dominant surface influence, even small differences in finishing and treatment will affect strength. For this reason:
Strength values of brittle materials must be inspected especially critically with regard to the relevance of the specimen properties to the actual application. Brochure specifications with no exact specifications regarding the tests are unsuitable for use when designing.
Illustration 14-35: A technology includes
problem-free command of maintenance and assembly. This is also the responsibility of the design engineer (Ill. 13-28).
In order to keep the risk of special characteristics such as brittleness or low strength within acceptable limits, procedures and design characteristics must be developed, tested, and specified.
Illustration 14-36: The development of technologies demands strategic thinking and long-term planning. 5-year steps are to be expected. Periodic testing during these steps is recommended. At the end of a step, a revision must determine if and how to proceed with the project.
Serial maturity is only reached when all knowledge necessary for design, construction, finishing, and operation has been determined and specified.
Only then can the technology be considered for application in a serial project. A technological development, which is long-term by necessity, can therefore not be rejected with the argument that no serial product can be expected in the short term.
Illustration 14-37: Even comparatively short test runs can provide important information regarding the durability of components and technologies in engines (Ill. 126.96.36.199-10). The prerequisite is a very exact inspection of potential problem zones. Considerable experience is necessary to identify these areas. The initial signs of a worrying damage process should be recognized. Due to the damage-specific temporal progression, these damages may be very minor after the relatively short test run times (top right diagram). However, they could develop into dangerous damages within the overhaul intervals prescribed for they serial application. This creates a risk that the damages will be underestimated.
There are occasional attempts to minimize the considerable expense of sufficiently intensive follow-up inspections. These costs are then omitted from the budget. The test run then becomes a mere exercise to verify the functioning of the parts without typical operating damages. This increases the risk of long-term damages and high consequential costs.
In order to control these cost- and prestige-laden damage risks, the analysis of test run parts is extremely important. This must be planned as specifically as possible (e.g. weak points) and experienced technical personnel are required to conduct the tests, document them as exactly as possible, and analyze the results. After this, one must proceed in accordance with a specified process that ensures that necessary activities are completed.
14-1 S.F. Powel IV, “On the Leading Edge: Combining Maturity and Advanced Technology on the F404 Turbofan Engine”, Paper No. 40-GT-149 of the “35th International Gas Turbine and Aeroengine Congress and Exposition”, Brussels, Belgium, June 11-14, 1990, published in “Journal of Engineering for Gas Turbines and Power, January 1991, Vol. 113, pages 1-10.
14-2 J. Kumpfert, M. Peters, W.A. Kaysser, “The Potential of Advanced Materials on Structural Design of Future Aircraft Engines”, Proceedings of the NATO-RTO Symposium on “Design Priciples and Methods for Aircraft Engines” , 11-15 May 1998, Toulouse, France. pages 42-1 to 42-12.
14-3 S.W. Kandebo, “GE and Allison Probe Core Engine Failure”, periodical “Aviation Week & Space Technology” May 17, 1999, page 30.
14-4 C.H. Cook, C.E. Spaeth, D.T. Hunter, R.J. Hill, “Damage Tolerant Design of Turbine Disks”, ASME Proceedings Paper 82-GT-311, 1982, pages 1-8.
14-5 K.F. Mordoff, “General Electric Testing 404 Derivative Engine”, periodical “Aviation Week & Space Technology”, April 12, 1982, pages 59-61.
14-6 “F-18 Fighter Crashes in England”, periodical “Aviation Week & Space Technology”, September 15, 1980, page 20.
14-7 J.L. Bartos, “P/M Superalloys for Military Gas Turbine Applications”, Proceedings Paper of the “1979 P/M in Defense Technology Seminar” of the MPIF, Yuma, Arizona, pages 81-112.
14-8 “DMS Market Intelligence Report”, Gas Turbine Engines, 1982, page 5.
14-9 “Interavia Air Letter”, No. 9649, December 12, 1980, pages 1 and 2.
14-10 periodical “PM Powder Metallurgy”, 12, 1980, 4 page 507.
14-11 D.N. North, “Changes to F404 Engine Urged by Design Review”, periodical “Aviation Week & Space Technology”, December 8, 1980, page 24.
14-12 H.J. Coleman, “Rolls Studies Coatings to Cut Erosion of Carbon Fan Blades”, periodical “Aviation Week & Space Technology”, May 27, 1968, page 74.
14-13 G. Warwick, “GE confident of fan blade answer”, periodical “Flight International”, 21-27 June, 1995, page 5.
14-14 S.W. Kandebo, “New Engine Test Rules Contested”, periodical “Aviation Week & Space Technology”, October 10, 1994, page 33.
14-15 H.J. Coleman, “Rolls Nationalisation Clouds RB.211 Future”, periodical “Aviation Week & Space Technology”, February 8, 1971, page 26.
14-16 P. Middleton, “RB.211: transatlantic bridgehead”, periodical “Flight International”, 5. November, 1970, page 709.
14-17 J.M.S. Keen, “Development of the Rolls-Royce RB.211 turbofan for airline operation”, Proceedings Paper 700292, of the ASE “National Air Transportation Meeting”, New York, N.Y., April 20-23, 1970, pages 5 - 7.
14-18 E.M. Murman, M. Wallon, E. Rebentisch, “Challenges in the better, faster, cheaper era of aeronautical design, engineering and manufacturing”, periodical “The Aeronautical Journal”, October 2000, pages 481-489.
14-19 S. Vajna, “Approaches of Knowledge-based Design”, Proceedings of the “19th CAD-FEM Users' Meeting 2001, International Congress on FEM Technology”, 17-19 October 2001, Berlin, pages 1-9.
14-20 K-D. Thoben, F. Waber, M. Wunram, ” Pragmatic Approaches to Overcome Barriers in Knowledge Management”, Proceedings of the “19th CAD-FEM Users' Meeting 2001, International Congress on FEM Technology”, 17-19 October 2001, Berlin, pages 1-9.
14-21 D.R. Hinghton, W.J. Chrispin, “ Future Advanced Aero-Engines - the Materials Challenge”, Proceedings AGARD CP-449 of the conference “Application of Advanced Material for Turbomachinery and Rocket Propulsion”, pages 4-1 to 4-10.
14-22 G. Polek, “Engine woes cast pall over first flight of A318”, periodical “Aviation International News”, February 2002, page 57.
14-23 S.W. Kandebo, “Upcoming Blade Tests Aimed at Growth GE90”, periodical “Aviation Week & Space Technology”, May 31, 1999, page 40.
14-24 J.R. Schaff, “Fatigue and Life Prediction”, ASM Handbook, Volume 21, composites, pages 252-258.
14-25 A. Rossmann, “Schadenuntersuchung und Schadenverhütung an Bauteilen der Ingenieurkeramik”, book series Kontakt & Studium, Expert Verlag, Volume 308, H. Grosch, “Schadenskunde im Maschinenbau”, ISBN 3-8169-1202-8, 1990s, pages 76-96.
14-26 W.H. Rentzsch, G. Willmann, “Ein einfaches Hilfsmittel zum Konstruieren mit Keramik”, periodical “Materialwissenschaft u. Werkstofftechnik”, 33, 2002, pages 184-189.
14-27 “GE 90 Program Summary”, Lecture at the TU Munich, seminar of the Institut für Flugantriebe, November 1995.
14-28 W. Hansen, G.F. Kaminske, “Praxishandbuch , Techniken des Qualitätsmanagements” , Symposion Verlag, ISBN 3-933814-53-7, 2001, pages 236-240.
14-29 M. Phelan, “NASA's coating promises efficiency breakthrough” , periodical “Flight International”, 11-17 March 2003, page 30.
14-30 M.A. Taverna, “Ariane 5 EC-A Upgrade Faces Long, Costly Flight Delay” , periodical “Aviation Week & Space Technology”, January 13, 2003, pages 402-403.
14-31 D.R. Bush, “Designing Ceramic Components for Structural Applications”, periodical “Journal of Materials Engineering and Performance”, Volume 2 (6), December 1993, pages 851-862.