Business Putting Pedal to the Metal: The Ins and Outs of Bike Product Testing Marketgit TeamApril 21, 202403.3K views credit: canva.com Table of Contents Toggle 1. Introduction1.1 Purpose of Bike Product Testing1.2 Importance of Reliable Testing Methods1.3 Overview of the Blog2. Types of Bike Product Testing2.1 Performance Testing2.2 Safety Testing3. Testing Equipment and Tools3.1 Dynamometers and Power Meters3.2 Force and Torque Sensors3.3 Data Acquisition Systems4. Testing Procedures and Protocols4.1 Test Planning and Design4.2 Test Execution and Data Collection4.3 Data Analysis and Interpretation5. Challenges and Limitations of Bike Product Testing5.1 External Factors and Environmental Conditions5.2 Replicating Real-World Scenarios5.3 Cost and Time Constraints6. Advances in Bike Product Testing6.1 Virtual Testing and Simulation6.2 Artificial Intelligence and Machine Learning in Testing7. Conclusion7.1 The Role of Product Testing in Bike Development7.2 Continuous Improvement in Testing Practices 1. Introduction Overall, the main goal of the blog is to explain different testing methods used for bike products, discuss the strengths and limitations of each method, and suggest ways to test current and new products. This will ultimately enable product designers to select and develop the best test methods for their products to increase product reliability and safety. Considering the multitude of bike disciplines and product categories, the blog aims to provide a versatile resource amongst the bike industry. The blog will explore various types of product tests done in labs and the field, share real-life case studies of successful and failed product tests, and discuss the system development and test methods used for specific products. All of this content will be presented on a very technical level with the aim to be a resource for product designers and engineers. The primary purpose of developing reliable bike product testing methods is to prevent the occurrence of recurrent product failures amongst existing inventory or newly designed products. In addition, the requirement to increase product liability is strongly emphasized to minimize legal and warranty costs associated with product recalls and liability claims. The basic objective is to confirm product reliability, durability, performance, and safety for the consumer. A reliable product is one that can be used in good faith, with a reasonable expectation that the product will not fail and is free from safety hazards when used for the intended purpose. Exercising due diligence in the area of product testing and development will increase the quality and safety of bike products for an international market. 1.1 Purpose of Bike Product Testing With riders proving they have the drive and endurance to conquer the backcountry or sprint to the finish line, mountain and road bike technology is constantly pushed to new limits. Mountain bike racing is an arena of sport in which every athlete is striving for an edge on the competition, and performance equipment is the easiest means to obtain that edge. The testing of road and mountain bike equipment remains an essential part in the process of making equipment that is lighter, stronger, and of the highest quality. Although product testing is an essential part in the development of quality cycling equipment, it often lacks the acclaim and recognition that it rightfully deserves. This blog is an in-depth analysis of the testing methodologies used in cycling equipment, and how those methods are utilized to ensure the creation of reliable products effectively serving the needs of the cycling consumer. 1.2 Importance of Reliable Testing Methods Bike product testing can range from the mundane to the rigorous. The results of which can greatly affect a company and its future product. In today’s world, there is a wide availability of aftermarket bike parts to choose from, but seemingly few are worth the investment. Product testing, when done correctly, can influence the decision process for consumers and companies and aid in the growth of a given product and company. This is something that was certainly reflected upon when working with a recent frame testing with a fairly large frame company. Due to the nature of the frame, testing was overdue and when the test was completed and results sent to the company, it was found that the test had been done on the wrong model of frame. This was due to the fact that the model in question was no longer in production and was completely replaced by the new model (although the model name was kept the same). This caused quite a setback and frustration for peers and testers who had invested time into accumulating the testing data. This incident may have been neglected if more time had been invested in testing methods and the reliability of the frame identifier. 1.3 Overview of the Blog We will discuss the role of testing in the production process of bicycles and bicycle components. Testing is an extremely important phase of production which allows the engineer to validate their designs, expose manufacturing flaws, and prove the product is reliable. We will demonstrate this with real-world examples. We will also discuss methodologies and standards that are commonly used in the bicycle industry, with comparison to other industries. The tests often go unnoticed by the consumer because when performed correctly, the product fulfills its intended purpose and does not fail. In this case, the test was a success. An example where a test failure had a significant impact was in 2001 when Risse Racing tested a new design of a mountain bike suspension fork. An undetected flaw in the design allowed the fork to catastrophically fail while in use, causing injury to the consumer. The flaw was quickly identified and corrected, preventing a similar accident from happening. A common misconception is that a new product is extensively tested before it is released to the public. In reality, it is common for a company to prematurely release a product in order to receive financial return. The public becomes the test judges, and the first generation of the product becomes the test articles. We will discuss the advantages and disadvantages to the company, as well as the consumer, when products are released untested. This will then lead to the discussion of the testing life cycle of a product from conception to completion. 2. Types of Bike Product Testing Safety testing is conducted to ensure that a product will not be hazardous to the user or others in any way. This often involves putting products through legal standards in order to gain a safety accreditation. As standards are different all over the world, it may be necessary to conduct several different tests in order to gain accreditation for a global market. Bicycle related products generally fall into one of three test categories: helmet testing, mechanical testing, and material testing. Helmet testing is a specialized area that is looking to improve rider safety in an impact. This is generally a pass/fail testing and will involve procedures such as dropping a helmeted headform onto anvils and measuring the force applied to the head. Alternate tests may involve using a real person and measuring the impact by using accelerometers. Mechanical testing is much like performance testing but is aiming to evaluate whether a product is safe to use. This will often involve using the same tests as performance testing but will go on to intentionally break the product and then evaluate the nature of the failure. This is done to find out whether a product has weak points that may cause injury. Material testing involves putting individual constituents of a product through tests to gain an overall evaluation of a product’s safety. This can involve comparing a material to known hazardous materials or putting it through a variety of tests to evaluate its reliability. Performance testing is conducted in order to see how a product behaves in particular conditions. In our case, we want to find out the durability and efficiency of a bike or bike part. This is sub-categorized into three tests. The first is functionality testing. This evaluates how a product or part works compared to what is expected. For example, seeing how well a gear system changes on a newly designed rear derailleur. The second test is a wear and tear simulation. This test is an accelerated version of the product’s general use to find out how long it will last. An example of this would be putting a wheel onto a testing jig that applies weight and friction imitating riding up a hill. The wheel is run against the jig for a predetermined distance and then the wear is evaluated. The final test is determination of the product or parts strength. This involves finding out load limits and finding weak points within a product. An example of this test is identifying the point where a handlebar will snap after being put under different levels of pressure. The results from performance testing are used to identify any initial design faults in a product that may cause it to fail its intended use. This can be anything from finding a fault in the production process for said product, to finding a critical design flaw that may cause injury to the user. 2.1 Performance Testing Performance testing involves comparison of a component against another, a newer version of the same component or against a known benchmark on a test rig or using any variety of road tests. It also can involve comparing a product against other similar products in the form of a user trial. The intention is to show how well a product performs its function. It’s not just limited to how fast something is as a common misconception might suggest, it is also how efficient it is, how durable it is, how long it can maintain high levels of performance, what the peak performance levels are and many other factors. For example, it would be useful to know how long it takes for a mountain bike to climb a certain hill, but it would also be useful to know how much more effort it took compared to a previous attempt. If a transmission part is tested for durability, it might be tested under incremental load to determine at which point the part starts to deteriorate. An effective test will take into account all factors that are relevant to the product and allow for margins of error due to the limitation of manufacturing consistency in any given product. This, in effect, will provide information valuable for both product development and for the consumer when making a product choice. 2.2 Safety Testing Imagine a job where you get first pick and hands-on experience with the latest and greatest in biking technology, and the satisfaction that comes with knowing that your testing results could directly influence the way we ride in the future. That’s what it’s like to be a bike tester. They do the tough job of sifting through all the good, the bad, and the ugly and determine its fate in the bike industry. Testing doesn’t stop short to only bikes, it goes for all bike components and accessories. Any item that is to be sold specifically for biking needs to undergo testing to ensure it is safe and reliable (Brady, 2004). The public’s safety is of huge concern and if a product fails here, it likely won’t even make it to the market. Official safety standards are regulated by the Consumer Product Safety Commission. Tests are performed for helmet standards, mechanical safety for bikes, and flammability resistances for all bike accessories (CPSC, 2009). Some companies choose to go above and beyond and perform their own safety tests, usually bringing in a third-party company specifically specialized in that type of testing. Any found data is always better for the manufacturer since all results can be used toward making a safer end product for the user. Safety testing methods may vary depending upon the product, but often involve impact, fatigue, and static loading tests. These are commonly carried out in test labs and specialized testing equipment may be used to simulate real-life situations. 3. Testing Equipment and Tools These are the basic equipment used for testing bicycles. The advances in material strengths of bicycle parts have increased tremendously during the past two decades. The increase in material strengths, combined with the need for increased safety, has forced an acute interest in understanding the exact loading conditions a bicycle will experience throughout its lifespan. As discussed earlier, the primary loading condition a bicycle experiences is the input of the rider. To understand the loads created by the rider, a system to measure the input must be established. Utilizing the fact that power is defined as force times distance divided by time (P=FD/T), we can see that the truer measurement of a rider’s input is taken as torque and angular velocity of the wheels. A bicycle can be simulated as a two-dimensional system. To measure the torque about the rear axle, it is necessary to measure the force on the rear hub and the distance from the hub contributing to movement. A conventional way to measure force is to use a proving ground with strain gauges. The method to measure torque, however, is not conventionally simple. An effective way to measure torque on wheels is to use a dynamometer. A dynamometer is an apparatus which absorbs power from an engine or, in this case, a bicycle and measures the rate at which work is being done. There are many varying types and complexities of dynamometer systems, and the appropriate one should be chosen based on the type of testing to be done. Because of the ability to control test parameters and simulate real-world conditions, dynamometers are quite useful for product development testing. Often times, private companies or large bicycle manufacturers will contract testing services to independent engineers who operate dynamometer labs. Most modern bicycles are made with the assumption that stiction forces from static equilibrium positions are negligible. Because of this, it is an incorrect assumption to say that the force and torque of a rider will always result in strictly linear movement. Even so, appropriate measurements of force and torque by a rider should still be calculated as it is a valuable parameter for comparison in the effects of varying test conditions. Types of force and torque measurement on bicycle parts can vary from using load cells on actual bicycle components to more indirect methods by measuring cable tension and pedal forces. Step one to measure force, however, will first require an understanding of what areas a bicycle will allocate the most force from a rider. This is an extensive topic, and a separate report on the force vector analysis of a bicycle will be available in the future. Depending on the test, the use of force and torque measurement can be simple or quite complex. It usually can be determined by the areas and components being tested and the necessary degree of data accuracy. In any case, the more advanced the test, the more a data acquisition system will be needed. Data acquisition systems vary by cost and complexity and have the potential to be the most valuable of all testing tools for any test requiring multiple data parameters. A simple example could be a comparison test in component strength between two different designs. In this test, it may only be necessary to measure force/torque on a single component to compare as a load is cycled in varying directions. A more complex example could be an all-out product test of an entire suspension system. Data parameters in such a test could be quite extensive, ranging from rider inputs of various terrains to component loading, deflection, and actual stress in comparison with FEA data. If customer satisfaction is guaranteed by a product meeting specific design criteria, the end results of tests logged in data acquisition can serve as proof that indeed a given product meets customer expectations. 3.1 Dynamometers and Power Meters These devices are used to measure the power output of a cyclist. The most direct method of measuring this is by using a dynamometer in which the cyclist is required to pedal a modified exercise bike with the rear wheel removed and a mechanical resistance applied to the pedal cranks. When using this method, it is possible to obtain very accurate and repeatable measurements as the data are not influenced by changing environmental conditions. The main limitation with using a dynamometer is that this method of testing is not conducted on the cyclist’s own bike and the relationships between pedaling torque, cadence and power output, and the influence of varying body position on the pedals cannot be accurately reproduced. This can be problematic when testing is to be conducted on saddle designs. Power meters and ergometers built specifically for use with the cyclist’s own bike have been developed to eliminate these issues. Power meters can provide accurate data, but their cost and the range of values available means it can take longer to find the best meter and/or equipment configuration for a particular study. 3.2 Force and Torque Sensors Direct offering to the force torque sensor is the most recent addition to stationary legal and illegal drugs cost has made it the most cost effective. A standard off the shelf trainer for testing power output can be used and the sensor simply mounts in line between the frame and the rear wheel. At present, this is by far the most common way to measure forces acting on the rear wheel. Its simplicity and cost are attractive, however, with the rapid progression of smart trainers, the above technology may soon become competitive. This system is currently under development in the United States. It measures the force applied to the pedals by the rider. Force sensors are mounted in the pedal spindles and data is relayed by wire to a small box on the frame. When force is applied to one pedal, the box sends a signal to a motor in the rear unit, guaranteeing the pedal travels at the same speed around the whole pedal stroke. This is a highly sophisticated and expensive system, and the development team is working on ways to improve it as the data is often inaccurate. The ability to reproduce results and obtain accurate data makes it an appealing concept for future testing. 3.3 Data Acquisition Systems Materials testing is the most basic form of testing and is often used in research and development of new bicycle components. The goal is to determine the physical properties of a material, commonly a metal or a composite. These tests are often used to compare different materials or grades of materials. The data is used to determine which material is best for a given application. Property tests can be destructive or non-destructive. An example of a destructive test would be tension testing of a batch of spokes to determine their strength. The best way to acquire data for this test would be to perform it in a test lab using a materials testing machine. A data acquisition system could be used to record the load and elongation of the spokes. Static or dynamic load tests on bicycle frames are an example of a non-destructive materials test. The data acquisition system would be used to record the forces and displacements at various points on the frame. This data can then be compared to finite element analysis to determine if the frame is capable of withstanding impacts and loads that are encountered during normal use. Acquisition of accurate data that can be used to draw unbiased conclusions about how a bicycle or a component performs is a critical part of bicycle testing. Modern instruments are capable of collecting an enormous quantity of data and storage and organization of the data is an important consideration when selecting what types of tests to run and what equipment to use. Some common types of controlled tests that are used in bicycle testing are materials tests, fatigue testing, impact testing, and testing of new designs or competitive products. A data acquisition system consists of sensors that collect data, a computer to control the test conditions and record the data, and software to manipulate and analyze the data. Data acquisition systems are used in all types of controlled testing on bicycles and bicycle components. 4. Testing Procedures and Protocols Finally, any methodology of product testing must contain procedures and protocols. This methodology explains the specific steps for test preparation, test execution, data collection, data analysis, and report generation. These procedures must be made explicit and in sufficient detail so that generalizability and repeatability of the test is ensured. Controlled scientific inquiry is characterized by methodical and well-thought-out testing with clear goals and procedures. These goals and procedures need to be written down. The following three subsections describe the specific procedures and protocols for testing bicycle products. These procedures and protocols are specifically for testing within the laboratory or field of the Mechanical Engineering Department at UC Santa Barbara. They may be modified in the future to accommodate different tests or tests conducted in different locations. 4.1 Test Planning and Design Depending on the test type, designs can range from a simple comparative test or control trials to more complex repeated measures or factorial designs. Controlled comparative testing involves simultaneous testing of different products under the same conditions. This may involve testing components or using specific product features and is often used to compare existing and new versions of a product. An example of this would be testing different types of tyre using the same tyre pattern for all, in order to gain an understanding of how the rubber compound affects performance. This design would be followed by a single repeated test (e.g. a set downhill course) to compare the tested tyres’ performance to old tyres. Control trials involve testing a new product against an industry standard or similar existing product. This will involve specific tests of the product against set criteria and is often used in review style testing to determine if a new product is worth purchasing. The most complex of these designs is the factorial design, which involves testing all possible conditions of 2 or more variables. This would be rare in product testing and usually reserved for testing the product’s effect on a separate system or human physiological response. Product testing within the consumer market can be classified as qualitative, providing an understanding of product function and performance, or quantitative, producing numerical data which can be compared to other products and provide a level of product quality or performance. Most clinical or bench tests can be considered qualitative; the performance of the test and the product are both measured, but comparison to other products is not possible. Detailed product testing requires a carefully designed test plan to provide valuable results. The first concern the researcher must have is the intention of the test. If the researcher is seeking to test the performance of a certain part or the reliability of a single mudguard, then they will only need to design a small and simple test. However, if the intention is to compare products in order to write a review, then a larger, more complex test will need to be designed. This will involve setting the terms of the test by defining what is going to be tested and how the results will be recorded. For performance testing, this may mean setting a test course with a specific distance and identifying how product performance will be measured (e.g. time taken to complete the course or physiological measures such as heart rate or oxygen consumption). For reliability testing, this may involve identifying the environmental conditions the test aims to simulate and how test product durability or failure will be defined and recorded. Once the test plan is clear and agreed upon by all involved parties, the next step is to design the test. 4.2 Test Execution and Data Collection The goal of testing is the thorough collection of data that shows product performance under many different conditions. To achieve this, the variables must be identified, appropriate levels of these variables selected, and a plan formed on how data will be collected. Variable identification and specification is a very important part of testing. Often, all of the things that can affect the product output or dependent variable are not identified. These unidentified independent variables can confound the test and ruin any chance of determining whether changes in the dependent variable were due to the treatment variable. Specification of variables must be done in a way that allows others to replicate the test. This is usually done in a written plan or procedure. Test data is recorded information. This can be measurements, observations, and in some cases, verbal statements or written records. Many tests in the social sciences or marketing require the use of verbal or written information. In our case, data will be measurements on many different factors. Data can be affected by the method of data collection. For this reason, the data collection method must be determined in order to avoid inconsistency. Data collection will be taken to a higher level in a later portion of this article. 4.3 Data Analysis and Interpretation Finally, to summarize the evidence for the research hypothesis and the associated inferences, it is important to distinguish between statistical significance and the practical relevance of the results. Although testing the null hypothesis can provide an estimate of the probability of a sampling error, it does not provide an estimate of the probability of the null hypothesis being true. Confidence intervals provide an estimate of the effect of an independent variable at different levels of probability and also show the direction and the range of possible values for the research hypothesis. If the confidence interval includes the mean difference specified in the research hypothesis, then this can be interpreted as evidence in support of the research hypothesis. In the first instance, the research hypothesis may assume that there is a significant effect or a significant relationship for a specific comparison. This is known as a one-tailed hypothesis. It is often best to initially test this with a more conservative approach using a two-tailed hypothesis test, whereby the critical p-value for significance is divided between the two tails of the distribution, giving a more stringent assessment of statistical significance. If the effect of the relationship is in the hypothesized direction, using a two-tailed test can provide insight into the power of the test in confirming the research hypothesis by obtaining a p-value that is closer to the critical value. This usually results in a staggered approach to testing the research hypothesis. Two-tailed tests are indicated by a difference from the null hypothesis with the symbol = on the alternative hypothesis, whereas one-tailed tests will use the symbols ‘>’ or ‘<‘. In certain research designs, data analysis can only begin when data collection has been completed. However, it is often best to begin the data analysis process while still collecting data. This helps the researcher to know what to look for in the data and also to assess whether the experiment is proving successful. Data analysis and interpretation are ongoing throughout the research process. The goal of the analysis of the measured variables is to examine the hypothesis of the research. This usually involves testing the null hypothesis of no relationship or no difference between groups. 5. Challenges and Limitations of Bike Product Testing The major issue with any type of testing is making sure that the testing is relevant, goes smoothly, and is completed in a reasonable time frame. Product testing, particularly for bicycles, can be greatly affected by external factors and environmental conditions with the potential to greatly skew data. An example of this would be testing a set of brakes in wet and dry conditions, although finding a location where there is a consistent and reliable slope that can run a variety of weather conditions is in itself a challenge. The most difficult area to overcome, however, is replicating real-world scenarios, with the controlled environment of a laboratory or test track being far removed from the trails and tracks riders use their products on. Attempting to mimic testing under these circumstances can often result in not only inaccurate findings but damage to the test equipment and unusable data. The third problem is cost and time constraints, with the manufacturer wanting to get a new product to market as quickly as possible, but skipping any sort of testing phase can be very detrimental to the product’s lifetime and the company’s reputation. Testing to destruction can be extremely time-consuming and costly, and depending on the product and its intended use, may not be a viable option. Testing can be a difficult area for any company and fraught with no shortage of potential problems, but employing a well-thought-out plan and the correct methodology can yield extremely valuable data and greatly improve the quality of a product. This, in turn, can result in increased customer satisfaction and product confidence, which is the end goal for any company. 5.1 External Factors and Environmental Conditions Another significant factor is the wide range of user differences in cycling disciplines. Mountain biking, for example, can place higher stresses on equipment than road cycling due to the impacts and vibrations from rough terrain. Rider skill also has an impact, in particular for crash testing of safety equipment. “Humidity, heat, and UV light can all have significant effects on a range of different materials used in cycle components,” says Dr. Konstantinos Poulios. “Sometimes the way that they affect the safety of a product can be critical. For example, carbon fiber composites used in bicycle frames may suffer a reduction in impact strength and toughness with long-term exposure to UV light. Metals can suffer surface or subsurface corrosion that can severely affect their mechanical properties.” Cycling is a sport which is hugely influenced by its environment, and as a result, products on the market are designed with a wide spectrum of uses in mind. It is crucial for bike and component manufacturers to be able to test the reliability and performance of their products, as well as for consumers to gain an accurate representation of what a product will offer them. However well a product performs in a lab or artificial test environment, the answer to how accurate a representation this is of the product’s real-life performance is always “not very”. This is due to the wide range of factors that affect a product’s performance in real-world conditions. Some of the main environmental factors are weather and riding conditions, and the long-term effects of these such as wear and corrosion. 5.2 Replicating Real-World Scenarios Whilst the control and repeatability of lab-based tests is desirable, it is often necessary to take testing out into the real world to get an accurate understanding of a product’s performance and durability. This is particularly important for load bearing components and in the crowd of manufacturers claiming their product is ‘stronger, lighter and faster’, real world testing is the only way to determine which is best. Unfortunately, the complexity and unpredictability of the real world creates huge challenges for test repeatability and establishing any meaningful degree of control. A law of physics states that the simple act of observing a system has an effect on said system. This is certainly true in testing, as the presence of test engineers and equipment in an area open to the public is likely to attract attention and possible theft of the equipment. Even taking photographs of the equipment for reference can have unwelcome consequences. During recent real world fork testing, the photographer only narrowly escaped a mob of irate park users who were indignant at what they perceived as an underhand council decision to authorize mountain bike activity in their local beauty spot! 5.3 Cost and Time Constraints Bike product testing is a time-consuming and complicated affair. With so many variables and options, it is tough to know if you have tested the product fairly and equally against the competition. This is the first problem in deciding how to test products in a manner that is fair and just; are we comparing apples with apples. Time and cost constraints are the biggest restriction to this type of work. Often manufacturers and designers fail to understand the level of time required to test a product to the best of its ability. This is a relatively new concept in the bike industry: test a product to its limit and break it; then you know how strong you need to make it to last a reasonable amount of time. Usually, the product will need to be tested to destruction, then it needs to be tested in variations of conditions relative to how it will be used, and lastly it needs to be continuously tested for a period of time to compare against other similar products on the market. Often the cheapest way to test is to take a product and test it as others would in their everyday lives. This is usually done with a multitude of devices and machinery that has different purposes, i.e. abrasion testing, impact testing, loading fixtures, and corrosion testing to name a few. The second best way to do this is actually using the product in the places and conditions it was designed for. An example of this is suspension or frame testing done on dirt roads or rough terrain. The product then needs to be compared with others in an identical situation. This is a big ask for many companies and often this type of work is limited to high-performance products only because it is very costly. 6. Advances in Bike Product Testing AI and machine learning Perhaps one of the most exciting prospects of the future in product testing technology is the use of artificial intelligence and machine learning. Artificial intelligence can be programmed to perform complex tasks without the need for guidance from a human, and machine learning is an AI application that provides systems the ability to automatically learn and improve from experience. Real-world data gained from physical testing or collected in harsh environments can be massively enhanced by the use of AI for data processing. An example of man and machine collaboration is data collection during downhill mountain bike testing. During a test session, suspension and frame data would be collected using onboard data logging sensors, but due to the chaotic nature of downhill riding, the sensors may be damaged during a crash from which the data was particularly useful. Simulation of such an impact may be too dangerous to attempt, but if footage of the crash exists, the data gained can be salvaged by an AI that can replicate the damage to the bike based on the video and identify the likely data changes. The days when a manufacturer designed a product, built a prototype, and broke/bent it until it failed are now largely gone. As powerful and sophisticated as computers have become, many traditional forms of physical testing can now be simulated. The powerful simulation of testing data is particularly relevant when dealing with high-end bikes as the particular material and construction methods used often make physical testing an expensive proposition. For example, analysis of a suspension fork can be simulated by making use of known loads and forces at the dropout transmitted through the stanchions. The stresses on the stanchions and resulting deflection can then be analyzed, giving a lot of useful feedback without the need to physically build the fork. Finite Element Analysis (FEA) is a method of structural analysis where a complex shape is divided up into a large number of small elements (blocks or triangles in 2D and tetrahedrons in 3D). Loads and material properties assigned to these elements will then define the way it deforms and the resultant stresses. FEA is commonly used in testing frame design and suspension components. Other areas of virtual testing are fatigue testing and impact testing, both essential in bike component testing and both possible to simulate using known real-world loads and data collected during real testing. Step forward to the future and perhaps we will see entire bikes being virtually tested, their strengths and weaknesses identified and modified before any physical product is actually built. With the progress currently being made in computer technology, there is no reason why this cannot happen. 6.1 Virtual Testing and Simulation Virtual testing involves the use of computer algorithms to attempt to simulate the effects of road or rider input on a bicycle and has been, and continues to be, a very active area of research in bicycle product testing. The idea is very appealing; a virtual test can be as simple as running a computer simulation to see how a new frame design affects ride quality, or as complex as running a fully parametrized optimization on a component to find the lightest weight design which meets all applicable strength and durability standards. The simplicity and low cost of most virtual tests compared to physically running tests on prototypes mean that a large number of design iterations can be evaluated, making it an excellent tool for exploring the design space for a new product. Finally, it can provide a means to quantify how a given design change affects some aspect of product performance, which is something that is often quite difficult to do with physical prototype testing. For example, it is fairly easy to test the strength of a prototype component, but testing how small design variations affect the strength can require building multiple prototypes which is both time consuming and costly. Virtual testing can offer a more efficient means to examine how design changes affect product performance. 6.2 Artificial Intelligence and Machine Learning in Testing Commercial aircraft manufacturers and similar high-tech industries are investing heavily in artificial intelligence and machine learning, and due to the high costs and risks of such projects, are seeking ways to accelerate its adoption in product development and manufacturing. Machine learning is a type of artificial intelligence that enables computers to learn and make decisions without being programmed to do so, by using data to find patterns or trends. Usually used to identify stock market trends, fraud detection, and in bioinformatics, machine learning has been used to now aid bicycle and component manufacturers to better understand and analyze the testing data that they accumulate. An example of machine learning is a program that may take fatigue testing data as its input, and classify the likelihood of a frame or component failing or exceeding certain parameters under certain conditions. This allows product developers to more easily interpret the data and make informed design decisions. A machine is said to have ‘learned’ when its performance in a certain task improves with experience, and this is seen as a powerful tool in an industry that has traditionally relied on the experience and intuition of individuals. Simulation data can also be used as a learning tool, and there are current projects investigating how a program might actually ‘learn to ride’ and control a bicycle, from the data gathered in bicycle or rider simulations. 7. Conclusion The recent trend has been toward more methodical testing. Instead of merely testing to see if a product fails, the goal is to produce a test that replicates the actual forces acting upon the product in its intended application. This way, the weak points in a design can be identified and the design can be improved. Material science is also a more recent consideration in testing. With advancements in materials such as carbon composites, testing to determine the material’s fatigue life in a given application is crucial. The information gathered from these types of tests played a critical role in the development of bicycle specific parts and away from direct transfers of motorcycle designs. In going through some of the current testing protocol of various companies in the bicycle industry, it is easy to see that a unified standard seems to have been borrowed from no specific industry or discipline. Testing ranges from very informal, such as first building the product and then using it in team or individual competition environments to evaluate it, to very formal, such as testing to motor vehicle or aerospace standards. At first glance, it might seem that more informal testing equates to cutting corners, but this is not always true. Information obtained from using a product in its intended environment is often irreplaceable. The very formal tests are often overkill, given the specific application of the product. The bicycle industry, in recent years, has seen extensive steps in design and development. No longer are the choices simple between steel or aluminum, hardtail or full suspension. Today’s mountain biker faces an array of products across multiple categories – products which have been innovatively designed and engineered. But how well these products perform, and how well they survive real-world use, is another matter. For the most part, the bicycle industry does not have standards, and, until a few years ago, had very little testing protocol that ensured that a given product would perform well and survive hard use. The consumer was most often the final test. A product’s early dismissal from a few product failures was common. The design team would then go back to the drawing board to rectify the failure. But without solid information on why the product failed and how to improve it, chances were high that the same failure would be repeated. This is where the emphasis on product testing can yield valuable information. 7.1 The Role of Product Testing in Bike Development The sheer variety of bikes available today is quite astonishing. In the 2003 edition of Bicycle Design: An Illustrated History, an attempt was made to catalog all the different designs of bikes that have been produced over the years. The editor, Garrett Glaser, conceded defeat, admitting that the book only covers designs up to around 1990 and as yet they had been unable to compile all the drawings to send to the publisher. The last successful count was 4200 different designs. Extrapolating trends from that data would indicate there are literally thousands of different types of bikes available today. Bikes are functional equipment that fulfill the needs of the owner to perform a given task. Therefore, for any given type of bike, creating the optimal design for its intended function is a complex process. The design process for bikes is based on an ever-increasing understanding of the complex interactions of forces that affect the performance of the bike. Using this knowledge, designers attempt to manipulate the magnitude and direction of these forces to create the desired dynamic of the bike. This manipulation is effectively a strategy to achieve a given force vector, e.g. a bottom bracket shell is widened to increase the stiffness of the frame at a given lateral load. The progression in understanding and the change of a force is effectively an evolution of the model of the system. So any change in design is an evolution of the model of the bike on a given terrain performing a specific task. Therefore, the rate of progression of design on any given bike type is dependent on the rate of change of the task and the understanding of the relevant forces and their interactions. If the task changes rapidly or there is a sudden advancement in relevant knowledge, designs will need to change frequently. This is the case in downhill mountain biking today. Any new design is effectively a prototype of the new model of the system created in an attempt to optimize the forces acting on the bike. However, it is not feasible to go straight to a new design based on current knowledge and then test it in the real environment, and a new design is likely to be a large financial risk. Therefore, the most effective method is to employ a simulation of the environment and its relevant tasks and then to test the current bike using the development model in that simulated environment. This is exactly the process undertaken in product testing. So the role of product testing is maximum when the rate of change of the relevant task and the understanding of the forces are high. With the increase of technology and knowledge relevant to bike designs today, the trend is towards an increasing rate of change of both task and model, therefore product testing is becoming ever more important. 7.2 Continuous Improvement in Testing Practices To achieve agreement on shared problems with a scientific approach to the task of testing, industry, safety, and standards organizations must work together. There are many areas of test method development where collaborative work may be possible. Simple test methods may be developed by in-house design teams or individual companies, but often the most time-effective method is to use expert external resources. Manufacturers and test houses working on the same tests have an interest in sharing resources, even where the test is applied to different products. Finally, there may be opportunities for joint development projects between industry and test experts such as those carried out in the public sector which benefit from government funding. In the UK, the safety of cycling is a research priority for the Department of Health whose interest is in reducing the number of cycling accidents. This has led to a greater use of cycle testing at test facilities run by TRL on behalf of the government.