Jo Kliebers 8 golden rules for safety tests at bike components.
1. Realistic, accurate transmission of force
is needed along with an extensive assimilation of data showing the actual load impact and stress on the product during riding is necessary (e.g. the DMS-measuring rides carried out by “Tour”, “Bike”, Velotech and the TU Hamburg-Harburg). Then you need a test machine, which is able to transform that data to precise, powerful load impact testing. If these two factors cannot be achieved, you should not bother to test, because you will not learn anything!
2. Like in real life
strength relevant components, such as frames, tubing, bars, etc, need to be tested in several steps in combination with small (frequently occurring) and huge forces (once in a blue moon). The product being tested must be mounted and tested in a way that reflects reality: both situations which result from manufacturer recommended use and misuse of the product. Overweight or overforceful, one-step testing does not reflect actual use of the component and is useless. In simple terms, cheaper one step tests** are good for testing value springs in motors, for example, but not for base frames used for riding or flying!
accurate evaluation of the examinee’s reliability and safety doesn’t depend on the fact that it breaks, but when it breaks. The aim of development is service strength - a compromise between low weight and high safety. For each application (e.g. competition-MTB, racing cycle for road, city bike) this compromise looks different. Therefore each application requires it’s own standard test, which is subjected to the real conditions (e.g. DH 2002.2 for handlebars and stems of competition-MTBs). Good test machines are in fact “time machines”. With their help you are able to simulate the lifetime of a bike within only a few days. And afterwards you can release the product for selling. Without these “time machines” it would take developers of bike parts hundreds of test rides and about 10 years time to feel really confident…
4. One is not enough
Different models of the same prototype often diverge in their durability about more than 50%, in extreme cases even up to 20-fold. For that reason, several examinees must be tested (Absolute minimum: 3+1!). Equally important is the allowance of a safety factor--the forces measured in reality are increased by a defined factor for testing purposes (Rule-of-thumb: the less random the tests become: the higher safety factor is needed). The weakest model of a prototype shows what purpose the respective product can be used with an acceptable risk.
5. No projection
Defining a product’s lifetime based on the least number of strokes the part can withstand on the machine is primarily relative to realistic “benchmarking” (reproducible direct comparison with reliable reference models). The reality data used for testing is likely to vary more than the actual difference between a good and bad product. Therefore, the data must, over time, be compared and varied based on actual field performance.
6. There are several bike life stories
…and in fact very different ones. A really useful test decision has to cope with this fact. The “only good/only bad logic” (theory of John Wayne) beclouds more than it illuminates. Instead of that, the test results of the examinees in merging, practically defined utilisation ranges is very important. Only these results allow each customer to find his or her ideal component. (For example, an MTB-bike used for shopping doesn’t necessarily need a downhill handlebar which can withstand downhill reality testing).
A serious test report includes clear documentation about how the test has been carried out. This means the publication of all important parameters of the entire testing procedure in a clearly arranged, easily understandable format as well as at least one significant photo of the testing arrangement.
8. Once is not enough
Producers and distributors of strength relevant bike parts have to be aware of the fact that playing blind man’s bluff is irresponsible and exceedingly dangerous. Therefore: we should keep the eyes peeled and check each production run by arranging spot tests on the test machine. There is always the right time for learning. After all, being stupid is no shame for bike developers (and other people), but staying stupid is.
P.S.: He who is interested in deeper details concerning the VR-3 test standards (DH, Road, etc.) is willingly allowed to ask for more information. Ideas and suggestions for improvement are explicitly desired and affectionately welcome. GoldenRules@syntace.com
*Technically speaking: Working load tests with safety relevant components.
**”Zero step tests” (i.e. no tests at all) are indeed still worse than one step tests…