We went to Germany to test the most popular bicycle chains. Here’s what we found:

Picture your bicycle chain spinning around the gears and back through the pulley of your rear derailleur. With every revolution, each chain link is slowly elongating. How does this happen? As a chain link turns, the cylindrical pins holding the each link together rub against the inside of each link’s collar bushings. Over time, this rubbing wears away at each pin, and the internal diameter of the collar bushings expands, growing the link’s length.

This process is exacerbated not only by the number of miles the chain is ridden, but also by how much grit and water gets in between those two parts. The sharpness of the angle the chain runs at between the front and rear sprockets, plus a lack of lubrication, can also speed a chain’s wear. That’s how, over thousands of revolutions, a metal chain can stretch and wear out.

Over my three decades of writing for VeloNews, I have endeavored to quantify chain wear through a laboratory test. Accurate quantification of it has evaded me, and I resorted to looking around among chain manufacturers, who have in-house chain-durability testing facilities.

In that vein, I spent a few days at the Wippermann/Connex chain factory in Hagen, Germany, last year, where I participated in a chain breakage test. I also observed the running of the company’s continuous chain-durability tests.

I worked with Wipperman to conduct long-term durability tests on the most popular 11-speed chains on the market. The tests work by running each chain on a machine for days on end until the chain stretches to a specific length. Engineers record the time it took for the chain to stretch, or wear, to that specific length. The company has been running this same durability test on many types of bicycle chains over the last 20 years on two, identical machines.

RELATED: Measuring chain wear accurately

This test took 80 days to complete its run on 13 different chains. Due to the length of this test, I was unable to observe the entire thing in person. But I did observe the methods, and I can vouch that the protocol is sound. Access to the facility is limited, and I believe the engineer operating the test is a straight shooter.

The test

Our test included 13 11-speed chains, listed alphabetically: Campagnolo Record, Connex 11sX, 11sB, and 11s0, FSA N11, KMC X11SL Gold, X11-93 and X11SL DLC, Shimano HG 901, HG 701 and HG 601, and SRAM PC-XX1 and PC-1110. We did not test any 12-speed chains, nor did we test chains from the new Shimano or SRAM gravel gruppos.

Each chain was tested on the test machine, which runs inside an enclosed cage and has a motor that drives a solid steel, 52-tooth chainring at 90 rpm. The loop of test chain is wrapped around that chainring and around a 17-tooth solid steel sprocket located 400 millimeters away (representing a bike in a 52 X 17 gear with 400mm chainstays). A 60-kilogram (132-pound) load cantilevered out on a rod attached to the pivoting plate onto which the 17-tooth sprocket is mounted pulls tension on the chain. Given the 10 m/s2 acceleration of gravity, this represents a 600-Newton load on the chain, which simulates a 70kg (154-pound) rider on an 8kg (17.5-pound) bike riding an eight percent climb at 13kph (8mph) and 90rpm.

To represent chain angles when shifting across a 40mm-wide cassette, the 17-tooth sprocket can be manually shifted over 20mm laterally in either direction. A tube through the top of the cage enclosure allows periodic addition of water, sand, and oil directly onto the chain while it is running.

Each test chain is joined with Connex master links to an equal length of the (new) reference chain, which is a Connex 11sX chain. The entire double-length chain is run on the machine. Connex technicians have tested the 11sX chain model multiple times on this same apparatus. Thus, should the measurements on the Connex half of the loop of chain be inconsistent with past results, the technician would stop the test and investigate what was awry.

A mark is scratched into the end of the test chain with a hardened steel scribe. Since the chain is removed daily and measured, this mark is made to orient the chain in the same direction on the test machine and on the measurement bed.

The chain initially runs noticeably roughly on the test apparatus, causing the 60kg mass to bounce. After a prescribed five-hour run-in period with only the factory lubrication on the chain, and no side angle applied to the chain, it runs smoothly, as the 400+ component parts of the chain settle in and polish each other, and the extra assembly grease rubs off. FrictionFacts tests have showed that two hours of break-in reduces chain friction by 1-2 Watts, illustrating the folly of racing on a brand-new chain.

After this run-in period, the chain is removed and cleaned with a solvent. Technicians make an initial measurement, and then the standard test protocol begins. The chain is run at 90rpm on the test machine. Each day, a precise amount of oil, water, and sand is gradually applied via the metering tube onto the running chain. This mixture creates an emulsion representative of what occurs on wet roads. The chain runs straight for a day, then with the rear cog offset 20mm to the right the next day, straight again for a day, and offset 20mm to the left the following day; this pattern repeats until the test of that chain ends.

The length measurement, made daily, is done on a long micrometer bed. The chain is pulled taut on the measuring bed with a 20kg mass running over a double pulley. This creates an effective mass of 10kg, or a 100N load, which is approximately one percent of the average 10,000N force required to break the chain. With the chain under tension, precise measurement on a Vernier scale is made while looking through a magnifying glass.

The average chain length is 108 links on a road bike with 50/34 chainrings and an 11-29 cassette. At 12.7mm per link, this equals 1371.6, or, removing the master link, 107 links equaling 1358.9mm. The chain-wear test keeps running, day after day, until the length has grown by one percent, or 13.6mm; the test stops when the 107 links measure 1,372.5mm.

The steel chainring and rear sprocket are machined in-house in by Connex, and new ones are used for each chain tested. Applying greater force to the chain would result in a faster test, but that would also produce fewer data points to place on a chain-wear curve. To ensure the integrity of the test, the door to the test room is always locked, and only the technician running the test has the key. An auto shut-off mechanism is triggered by an electric eye if the chain jumps off the machine or breaks.

Results

How did the chains perform? As it turns out, there was a wide range in the duration it took for each chain to reach 1-percent elongation. The Connex 11sX chain with stainless-steel inner links took 187 hours of continuous running, while KMC’s X11SL DLC hit that mark after 41 hours on the machine.

To add further context, I created a graph that charts each chain’s dollars-per-hour of use. I took the cost of each chain, in dollars, and divided it by the hours it took to hit 1-percent elongation. An expensive chain that lasts a long time might actually be cheaper to run than an inexpensive chain that lasts a considerably shorter amount of time.

The cost-per-hour ratio ranged from $0.28/hr. for the SRAM PC-1110 chain, to $2.93/hr. for the superlight hollow-pin/cutout-plate KMC X11SL DLC chain. The time spent actually changing the chain, while relatively unimportant for an individual bike, is worth considering for bike fleets, where replacing the chains on hundreds of bikes costs real money and makes long-running chains more valuable.

Now, this test was designed to wear chains out faster than they are likely to wear during actual use, as no rider only rides uphill without ever coasting downhill, or riding in a group. Also, the emulsion of sand, water, and oil applied to the chain is extremely abrasive. So, riders who clean and lubricate their chains are unlikely to experience similar wear times as the ones produced by this test. Even though it is generally recommended to change a chain when it hits the .75-percent growth barrier, rather than 1 percent, the cost-per-hour of riding will still be lower for each chain than this list shows.

Response From Manufacturers

We shared the results with representatives from each brand represented in the test, and gave them an opportunity to respond to our findings. Here’s what they had to say:

From Connex

Connex 11sX outwears all comers because it has stainless steel inner links and rollers. There are other materials and technology that contribute to 11sX longevity, but the stainless inner links and rollers are the primary reason. The 11sB also has stainless rollers and inner links, but it wears slightly less well than the 11sX because there’s a bling penalty – the treatment to make the pins gold and the chain black (similar to bluing) reduces longevity but it’s still miles ahead of all others except Campagnolo Record. For a chick-magnet like the 11sB chain, I think this is a small price to pay. The 11s0 is work-a-day rock-solid German steel.

All Connex chain pins are centerless ground to smooth the surface roughness from the standard 4-6 microns to 1.5 microns. The surface hardness is also very high. Smoother, harder pins wear less quickly.

From Campagnolo

What makes Campagnolo special is the fact that our chains are designed and produced entirely in Vicenza, both in terms of design and production. We do produce the molds and we do automatically assembly the part trough and computer controlled production line.

The inner link and the outer link are produced using a step mold that we design internally. The molding is done on our presses in the Vicenza plant.

The meshes, once printed, are heat treated (to give the necessary hardness and resistance) in our heat treatment department, after the heat treatment the meshes are treated superficially in our galvanic department to guarantee resistance to wear and tear (especially against atmospheric agents), after these steps, everything is sent to the assembly line and automatic control that allows the assembly and dimensional verification of the chains.

At the end of this line we have the greasing line which, through high temperature and ultrasound, allows the deposition of the lubricant in areas that are difficult to access.

The direct control of each production phase allows Campagnolo to have the process 100% under control. Improvements to the process have made it possible to improve the most critical area of ​​the chain: ex the area of ​​union between the internal and external link.

Attention to every detail of this area of ​​interface for joining the various parts has allowed us to continue to raise performance targets without reducing the useful life of the chain.

From SRAM

SRAM has been refining our chain wear test protocol for many years. We know from exhaustive lab testing and lab test validation through long-term field tests that this is extremely difficult to get right in the lab. A slight change in the exact amount, exact type, and exact application process of contaminant can all drastically change results in an unrealistic way. The test results from VeloNews do not match our own testing. Our own lab test show a dramatic difference in wear results of and between the PC-1110 and XX1 Hard Chrome chains, which is supported by feedback from the market.

From FSA

First off, I want to thank you for including us in your test and sharing the results with it. The test procedure seems like a very fair, unbiased, and accurate test. I will note the MAP for the Team Issue chain that I you tested is $28.99 instead of $37.

As any company we would prefer to be in the lead position, and see that we have improvements to make, but as a whole find the results quite positive. Based on our MAP of $28.99 for the Team Issue chain tested we also ended up with a great result in the cost vs. time comparison of the chains tested. As a brand, FSA has always been a product focused company that prides itself on the quality, dependability, and reliability of its products, and is always striving to improve upon those characteristics.

From Shimano

The testing protocol used by VeloNews may indicate differences between manufacturers, but it doesn’t create a result that replicates real world conditions. Most chains will far outlast the duration listed to 1 percent elongation.

There is more to choosing a chain than simply resistance to elongation. Using Shimano’s chain with Shimano chain rings and cassettes is the only way to get the best shifting quality. Both front and rear shifting are operations that rely on multiple components working together. There simply is no standard chain shape that can be easily subbed in. The asymmetry of Shimano chains is a fairly obvious example that third party chains aren’t going to function the same within Shimano’s shifting system.

Shimano has designed a drivetrain system that balances all the factors that are important to riders. Once you take the chain off the testing machine and put it on a bike in the real world, we would make the argument that our chains are the best in the industry.

Shimano’s product folks don’t believe the price per mile is accurate (for the companies you tested) and recommend you consider not including.