Articles by Sheldon Brown and Others

12 High-End Frames in the
EFBe Fatigue Test

Text and photos by Robert Kühnen, from TOUR magazine 10/1997
Translation by John Allen
Original-language version:
German German flag

find us on FB

Spoke Divider Spoke Divider

This ground-breaking article, first published in a German magazine, compared the resistance of of aluminum, carbon-fiber and steel frames to fatigue failure. Damon Rinard considered the article so important that he struggled mightily to prepare a translation of it. It deserves a new translation, which appears here.

What is so important about the testing this article describes? Even the testers were surprised to find that the high-quality lightweight aluminum and carbon-fiber frames they tested were more resistant to fatigue failure than heavier, high-quality steel frames. A caveat, though, is appropriate: the results hold for a frame which is ridden hard but not damaged or abused: corrosion, nicks, dents, scratches and stresses due to incorrect installation of components could alter the results, as the article itself notes.

The company which developed the test, EFBe (Engineering for Bikes) has developed a number of additional bicycle durability tests and established a certification program. Many bicycle manufacturers have subscribed to EFBE's program. The work being done by EFBe is of prime importance and deserves wide attention.

-- John Allen

What do a blade of grass, an airplane and a bicycle have in common? Design: all three are lightweight structures, and two embody the same basic concept. The “idea” of lightweight construction is a principle of natural evolution: to achieve the best mechanical properties with a minimum of material, and so of resources. Engineers have borrowed this principle from nature, and apply it in a wide variety of projects. Desirable results follow: for example, air travel and elegant, user-friendly sporting equipment. And, in this connection, an apparently paradoxical result has revealed itself: lighter can also be more robust. Does this sound crazy? It isn’t. Light weight in connection with greater durability, however, requires a greater design effort.

Light Alone Isn’t Enough

There’s nothing special about building a bicycle frame which is merely light. Building a frame which is both light and durable is more difficult. The actual requirement, however, is to combine the lightest possible weight with sufficient stiffness and durability. Only in this case is it accurate to speak of successful lightweight construction. Also, it is necessary to clarify what is meant by “sufficient.”

No frame should fail to meet certain stiffness requirements, as, for example, have been measured in tests published in TOUR Magazine -- or else the riding qualities suffer. Rating the stiffness of a frame in relation to its weight results in a parameter for the quality of the lightweight construction, which TOUR calls the STW (stiffness-to-weight) quotient.

Materials differ widely in their suitability for lightweight construction. The ratio of stiffness to weight of aluminum exceeds that of all other materials which are used in bicycle frames. That statement, however, says nothing about the expected service life.

How Long Is Long Enough?

Professional racers may find a frame which is reliable only for a single race to have a sufficiently long service life, if there are compensating advantages. Touring cyclists, on the other hand, prefer a frame which carries them problem-free for years, and never leaves them stranded. Technically, both want a “serviceable” frame which holds up for the intended period of use.

Aircraft, automobiles and bicycles must be reliable in service. A contrasting approach to reliability, though, is in that of durable construction, with which no failure occurs, even in an “infinite” service life. The small difference between definitions is of great importance in actual design applications. Attempts at durable construction quickly lead to the use of a variety of materials, and so not everything which travels by air, land or water can have an infinite service life. Aircraft with an infinite service life, for example, would not be serviceable at all: they would never get off the ground.

Spoke Divider

Spoke Divider

Reality-Based Testing

Observations about the durability of bicycle frames may be derived from actual experience with them – above all, in the claims departments of manufacturers. Theoretical knowledge, on the other hand, is rather meager, as durability tests are not standard in the bicycle industry.  This deficiency is at last becoming apparent with new frame designs and the use of new materials. Without machine testing, no reliable statements about the durability of new designs can be made. Not a few manufacturers turn the tables on responsibility for this task, which they perceive as burdensome: they wait to see how their products hold up in service – but that should not be so.

For that reason, TOUR went looking for a test procedure – and found one with a specialty firm, the Bochum testing company EFBe Prüftechnik. Manfred Otto, the CEO, is a pioneer in destructive testing of bicycle frames. The durability-testing procedure which he developed is one of the main elements of the new DIN 79100 bicycle-testing standard, which will go into effect at the end of this year. Compared with other test procedures (see for example the TOUR front-fork test in the January, 1997 issue), this grading test is simple, but effective. What is simulated is not single instances of loading, but pedaling while standing: the most challenging demand which must be met to achieve long service life. The best endorsement for the procedure, which has been tried for a year, is that of real life: according to Manfred Otto, failures on the test stand are like those in actual use. In this way, one of the most important prerequisites for a realistic test procedure is met. There is only one prediction which even this procedure cannot achieve: durability in kilometers of use. Statements such as “this frame will last for 20,000 kilometers” cannot be made reliably on the basis of this test. A much more extensive service-life test would be necessary to substantiate such statements.

Manfred otto protrait
Idealist: Manfred Otto struggles doggedly and persistently for better quality in the design and manufacture of bicycles. He carried out the TOUR frame test.

For all types of bicycles used in street traffic, ranging from shopping bicycles to road-racing bicycles, the DIN test stipulates a minimum load of 850 Newtons (just under 85 kilograms or 190 pounds) alternating 100,000 times on the left and right pedals. For common bicycles, this new, enhanced DIN is a valid test. Sporting applications, however, confront the materials with entirely different loadings. For this reason, TOUR subjects the frames to substantially higher loadings: first, 100,000 stress cycles at 1,200 Newtons, and then continuing at the next higher load level, 1,300 Newtons. If the frame withstands the total of 200,000 cycles without breaking, the testing is ended.

Twelve frames, of four different materials, underwent the test. The results – though without statistical verification – allow for conclusions about the potential of materials for bicycle frames. At this time, aluminum is the material with the best ratio of stiffness to weight. If smartly designed and built, aluminum frames are also very durable – see Cannondale and Principia. Carbon also comes out exceptionally well among the frames tested: it is a material for the future. Titanium frames can be light, very durable, but not very stiff. Steel is the material with the worst potential for lightness.

The Testing Process

In order for frames to be evaluated reliably, the test procedure must meet high standards of quality. The EFBe test makes this possible by stressing the frame in a way similar to that in actual use, with computerized measurement and control. The higher loadings of the frame in the TOUR test than those in the DIN standard guarantee that the high-performance frames are confronted with forces consistent with sporting use. Test standThe front of the frame is held rigidly by a special measuring front fork; the rear by an elastically-hinged support which simulates the rear wheel. In this way, conditions of pedaling out of the saddle which impose the greatest loading on the frame are simulated. Two pneumatic cylinders alternately apply the pedaling forces to the cranks at an angle of 7.5 degrees (corresponding to the leaning of the bicycle when pedaling out of the saddle), and a lever mechanism conveys chain tension to the rear of the frame. The cranks are at a 45 degree angle below the horizontal, the angle where the cyclist applies the greatest force. Alternating between sides, first 100,000 cycles at 1200 N, then another 100,000 at 1,300 N are applied. The measuring equipment keeps track of the deflection of the frame in response to the applied force, and shuts down the test when a substantial crack develops.

Spoke Divider

Spoke Divider

The Candidates

TOUR entered eight light frames weighing between 1,200 and 1,500 grams in the running. These included products of well-known manufacturers – Trek, Klein, Cannondale, Merlin and Principia – as well as frames from less-familiar brands, but just as light. The starting lineup for the test was interesting in that the range of prices was very great, from 5,199 marks for the expensive Merlin titanium frame down to 1,299 marks for the moderately-priced Stevens aluminum frame. Steel frames do not compete in this weight category: the Columbus Nemo tube set allows a weight below 1,500 grams, if small tube diameters are used. (Various outside diameters may be chosen with the Nemo concept.) Such a frame is not stiff enough at the frame size tested, 58 centimeters, and for this reason was not tested.

Steel still did have to be tested – for purposes of comparison. Two lugged and two welded frames served as references. The two lugged frames were from De Rosa and Barellia, and used the most successful steel tube set of the last ten years: Columbus SLX. These frames are representative of classic steel frames with small-diameter tubes. A very light, welded Fondreist frame with thin-walled, large-diameter tubing, and a medium-weight, simpler Nishiki frame, also welded, represented modern steel-frame design.

Light and Strong

“Pfffffft, pfff, pfffffft .....” For two weeks, the computer-controlled pneumatic cylinders in the EFBe laboratory did their hard work, hissing, and giving frames the beating of their lives. The best of the best survived two days of this, without failing. Three lightweight frames endured without visible signs of damage. After 200,000 cycles which visibly flexed even the stiffest frames, the computer shut down the testing for the Cannondale, Trek and Principia frames, according to plan.

This is a sensational result, considering the lightweight tubes and enormous test loads. Even Manfred Otto found the result surprising: “These are the lightest frames which have ever wandered onto the EFBE test stand, but also the most durable.” The Time carbon frame failed the test by only a little, with a broken chainstay after 182,000 cycles. The second-lightest frame in the test, the low-priced Schmolke titanium frame, made in Russia, exceeded all expectations at 160,000 cycles. Klein’s Quantum Race failed more quickly: the down tube broke after 132,000 cycles. Next was the Merlin Team Road titanium showpiece, struck down at a relatively low 106,000 cycles – the greatest disappointment, considering its price. The last among the light frames was the Stevens RPR4, which, however, was also the least expensive.

Steel was in crisis: The De Rosa SLX broke after only 57,000 cycles, only half as many as Brügelmann’s Barellia frame with the same tube set. Interestingly, they both broke in the same characteristic way, just above the lower head-tube lug – a type of failure, by the way, which TOUR’s testers had already seen in on bicycles in use. The Fondriest frame, very light for a steel frame, did not last through the first testing cycle, and, like the much heavier Nishiki frame, broke after only 80,000 cycles.

Summary of Results


Thousands of load cycles

The results in detail: there are two parts to the test, the first with 100,000 load cycles at 1,200 N and the second with another 100,000 at 1,300 N. The potential for damage in the second part is significantly greater than in the first. A frame which withstands the complete test is therefore much more robust than one which breaks after 100,000 load cycles. All of the tested bicycles are durable compared with “common” recreational bicycles. Cannondale, Principia and Trek withstood the test without failure, and the machine was turned off.

Spoke Divider

Spoke Divider

Representative Examples

Cannondale CAAD3 Aluminum used optimally: Very stiff, light and very durable – Cannondale’s  CAAD3 is exemplary.
High-en carbon frame Trek: carbon at a high level: The lightest frame in the test was one of the most durable. Excellent engineering.
Schmolke titanium frame Titanium for the people: Schmolke’s lightweight titanium frame surprises with exceptional durability.
Barellia steel frame Traditional steel: Barellia’s SLX is not light, but offers reasonable durability at an attractive price.


The clear result of the test is surprising, and gratifying, because it shows that a fundamental rethinking of design can work to the advantage of all of the desired positive characteristics: durability, stiffness and minimal weight are not mutually exclusive. There is no question that this result will determine many future customer choices.

Still, the result needs interpretation, to avoid misunderstandings. We do not consider any of the tested frames to be worrisome, or even dangerous. Even with the “worse” frames in this test, most cyclists who ride racing bicycles will become unhappy with the color of a bicycle before it reaches the end of its expected lifetime. Strong and heavy riders should, however, seek out the frames which have been shown to be more durable.

Because only single samples were tested, no statement is possible about quality of a series as a whole, or about the statistical distribution. With welding and brazing, considerable deviations from the results of sample testing are not unusual. For these reasons, it makes little sense to make comparisons between frames for which results were close.  The more general ordering, rather than a difference of 10,000 or 20,000 stress cycles, is important. The Time frame, for example (182,000 stress cycles), and the Russian titanium frame (160,000), are in the same league. And, aside from all of the imponderables of the single-sample testing, the test as a whole validates the concept of durable lightweight construction. The results with lightweight and intensively designed frames cluster together at the upper end of the scale of results, those with the inexpensive and heavier frames, at the bottom.

That aluminum and carbon frames lasted longer than steel frames in this test is in our opinion an issue not of materials, but of the design and construction effort. Not the material, but rather, the sophistication of its application, leads to the outcome. Logically, manufacturers concentrate their efforts on frames with good potential for light weight – and these are of aluminum or carbon; only as an exception (because of lower stiffness) of titanium.

Spoke Divider

Spoke Divider

Learning from Failures

The photos of breakages with this article show that the cause of failure is often something small: drillings and steps should be avoided on the highly-stressed down tube, and on the chainstays. Small welds for components also are critical. Aluminum, especially, is sensitive to stress-risers, unforgiving of such “prior damage,” but titanium also is sensitive. It is indicative of the quality of the lightweight frames than none of them failed at a load-bearing welded or glued joint. Klein’s Quantum Race broke at the shift-cable entry hole in the down tube; Schmolke’s titanium frame, at a drilling for a bottle cage; Merlin’s Team Road, starting at the weld for a shift-lever boss. It follows that the dimensioning of the main tubes and their joints was appropriate.

Minor modifications could presumably eliminate the small weak spots which the TOUR test revealed.

With carbon fiber, joints with other materials are often the weak spots. The Time frame offers an example: the failure (though very late in the test) was of a massive aluminum internal lug which extends from the metal bottom-bracket shell. The Trek frame with its carbon-fiber lugs presumably transmitted forces more evenly, and so lasted longer.

The failures of the brazed frames point to temperature problems, and those of welded frames, to stress concentrations at the welds. Better process control should lead to better results. A positive feature of steel frames went unexamined in this test: steel is not so sensitive to stress risers, and is more tolerant of minor damage. Also, surprisingly, steel frames have less of a problem with corrosion than many aluminum frames. Certainly, steel rusts, but only slowly. Some aluminum alloys commonly used in bicycle frames, on the other hand, are rather vulnerable to so-called “grain-boundary corrosion,” which can work its way quickly through the material and so, lead to failure.

Failure Initiation: Small Causes, Large Effects.
Click on any of the images to see a larger version.
Failure of Merlin Team Super frame The myth of indestructibility collapses: Relatively early, considering the price and the high expectations, a crack spiraled around the down tube of the Merlin Team Road titanium frame. Point of origin: the small weld at the shift-lever boss.
Failure of Fondriest frame Classic fatigue failure with the Fondriest: both chainstays of the lightweight steel frame broke relatively early at the welded joints. The jump in stiffness from the thin, elastic tube to the massive bottom-bracket shell is presumably too great, and results in stress peaks.
Failure of the Stevens RPR4 frame Chainstays: the Stevens RPR4 frame also suffered a failure of the right chainstay. Flexed outward repeatedly by chain tension, the chainstay failed where two effects were superimposed: a stress riser due to the weld and the indenting of the tube to clear the tire.
Failure of the Schmolke titanium frame Weak point at a drilling: Schmolke's titanium frame cracked starting at a water-bottle-cage drilling. Stress peaks at drillings are classic crack-initiation points. In this case, however, the drilling was well reinforced. Presumably, the weld initiated the crack.
Failure of the De Rosa SLX steel frame It got too hot? De Rosa's SLX steel frame suffered a failure of the relatively massive head tube. One possible cause was excessive heating during brazing. This example shows that heavy construction does not necessarily help much, and that processing which is appropriate to the material plays a very important role.


Congratulations to the manufacturers: the best four of the eight lightweight frames are not only very light, but also extremely durable, and representative of lightweight construction at a high level. The myth of early failure of aluminum compared with always-superior titanium and steel can be set aside. Unfortunately, visual impressions cannot reliably identify which frames will be durable. Some results surprised even the TOUR testers. It would be desirable for top-level frames to receive certification from an independent source. It is high time for a certificate which would identify top-performing products.

Information and test results for the tested frames

Frame Weight, Kg Size, cm Price(1) Material and construction Cycles(2) Failure  location Frame stiffness(3) Reference/Info
Barellia SLX 2.080 58 798 lugged steel 119,316 head tube, lower lug 64.6 86.2 0 61 96 75 00 75
Cannondale CAAD3 1.520 60 1,990 welded alu 200,000 no failure 91.3 92.1 00 31 5 41 58 98 98
De Rosa SLX 1.895 57 1,650 lugged steel 56,690 head tube, lower lug 66.8 80.6 0 28 71 27 55 55 
Fondriest 1.630 61 2,600 welded steel 77,171 both chainstays 79.9 76.8 08 21 2 72 50 
Klein Quantum Race 1.415 57 2,700 welded alu 131,907 down tube, cable entry 89.0 94.5 0 61 03 5 07 00
Merlin Team Road 1.525 60 5,199 welded titanium 100,595 down tube, shift lever boss 65.9 81.3 0 40 4 80 60 40 
Nishiki Team 2.080 56 1,390 welded steel 78,206 bottom bracket/seat tube/down tube 67.2 93.5 0 28 71 27 55 55 
Principia RSL 1.460 60 1,895 welded alu 200,000 no failure 83.5 91.85 05 31 2 87 29 13
SchmolkeTitan 1.300 59 2,000 welded titanium 160,356 down tube, bottle cage boss 64.3 70.2 0 61 39 67 35
Stevens RPR4 1.515 57 1,299 welded alu 85,032 right chainstay 77.0 84.7 0 40 4 80 60 40
Time Helix HM 1.485 57 2,990 lugged carbon/alu 181,966 right chainstay 66.5 86.2 0 71 59 94 59 30
Trek OCLV 1.200 58 2,800 lugged carbon 200,000 no failure 75.3 94.5 0 61 03 5 07 00
  1. Price for frameset with fork in DM.
  2. tested for 100.000 cycles at 1,200 N, then increased to 1,300 N.
  3. Head tube stiffness / bottom bracket stiffness.

© 1997 Engineering for Bikes (EFBe), Bochum

Translated by John Allen from the original German-language version.
Used with permission.

Home back to Damon Rinard's Bicycle Tech Page

Copyright © 1995-2001 

Spoke Divider

Spoke Divider

Articles by Sheldon Brown and Others

If you would like to make a link or bookmark to this page, the URL is:

Last Updated: by Harriet Fell