While I think we were a bit "over zealous" in our testing procedures, I would say that it was more common than not. I base that on my experiences in other companies and dealing with component suppliers, etc.

I honestly can't remember the nature of many of the surprises, but there was usually a lot of "there is no way that should have happened".

Interesting side note - there was a stem / seatpost company that we would NOT spec....because their products always failed VERY quickly in our fatigue testing. I won't name the company, but it was a brand that is widely recognized to be "bullet proof" and very durable. But we couldn't use them....

---

Chicago Cubs - 2016 WORLD SERIES Champions!!!!

"If ever the time should come, when vain and aspiring men shall possess the highest seats in government, our country will stand in need of its experienced patriots to prevent its ruin." - Samuel Adams

i think that you view product testing - fatigue testing, most notably - as valuable and necessary. no disagreement between us there. i think the gap between us is that you think the question of whether a bicycle product will fail before it's supposed to is knowable. i do not.

let me qualify. is it possible to know whether a fork is going to fail? or a crankset? a frame? a chain stay? stem? handlebar? no. yes, theoretically. yes, eventually. but not today.

what you have today in this industry is that razor thin margin between light weight and fatigue life. if you feel you need to make a 300g road fork, but it cannot fail, no, you can't simply engineer that with enough precision to anticipate all the stresses on that fork. to those who think they can, i submit it's only because you're starting with a design that stands on the shoulders, and benefits from, field testing, fatigue testing, ending in hundreds or thousands of fork failures over the past 25 years.

had an engineer the job today of designing a 300g bicycle fork that would stand up to 100,000 cycles (or however many cycles the industry is asking for these days) in a typical fork testing machine (as prescribed by ASTM or CPSC or whomever), without the aid of knowing how it is forks have evolved and how they're build today, not a chance in hell he could do it from behind his computer.

---

Dan Empfield
aka Slowman

The other big variable that isn't being discussed is manufacturing/QC issues. You can build and test prototypes from supplier first samples all you want but it only takes one batch of die cast parts with porosity issues to lead to a rapid unanticipated failure.

The more I think about it the more I come to the conclusion that using a die casting process for a stem is borderline negligent. I can't believe that many engineers would chose that material and process for a critical component.

Hi, because of the second recall of the TriRig Sigma Stem iám checking possibilitis for a new Stem/Bar solution for my Shiv comp in black satin.

The big problem ist that i had to cut the steerertube realy low for the Sigma.

Can anybody tell me the stack of the original stem of the S-Works models?

The next big problem is that no new forks are available in europe, not even with a different paint, so if my steeretube is to low for any other stems i have to wait for the new Sigma, and that´s no solution for me because i don´t wana use this steme anymore.

Iám realy angry about this because it is the second time i can´t use my racebike in the midseason!!

See if Dengfu will sell the FM087 Flat Stem Separately. Very similar steerer clamp requirements.

---

---

IF the load case is properly measured/defined and the properties of the materials are well understood...then yes, in a statistical sense it is possible to determine the likelihood of failure of a particular component on a bike. Engineering like this is done on a daily basis, and has been done for a VERY long time.


And I would submit that is a false dichotomy (i.e. "fool's choice") based on a misunderstanding of the driving performance factors of a bicycle ;-)


Once again...IF the engineer or organization properly defines the use/load case, then YES it's possible to properly engineer a solution using well established principles and calculations. If not, then yes, one is confined to make/test/break/redesign.

That's where we seem to be hung up. Your opinion appears to me to be that the only way to get the knowledge of how a fork is used is to make and break a shit-ton of forks and empirically come up with "what works". I'm merely pointing out that this isn't so. It's also possible to FIRST use instrumentation to investigate/define the expected load case and then engineer a proper design. This is done every day in a myriad of industries, and in some cases where multiple prototypes and/or destructive testing isn't possible.

---

http://bikeblather.blogspot.com/

"Your opinion appears to me to be that the only way to get the knowledge of how a fork is used is to make and break a shit-ton of forks and empirically come up with "what works"."

i don't disagree with you at all, theoretically. i'm just saying that in practice, in this industry, there's a gap between how engineers think a bicycle is ridden and how a bicycle actually is ridden. i am certain we will someday be able to predict weather patterns. there is an engineering solution to cloud prediction. we just don't have that solution yet. we're not there yet.

the difference between you and me is that you think engineers have figured bikes out. i know for an absolute fact that they haven't. do i think this means we dump engineering? no. i don't believe we dump engineering anymore than i believe we dump climatology. i'm saying there's a gap between engineering and real world cycling, and that gap is spanned by testing.

this isn't a binary choice between trial-and-error and engineering. you begin with engineering. you end with testing. when you test, and you are surprised and disappointed with your result because it does not comport with your expectations based on engineering, the result of the test generally informs the design changes you make (rather than going back and re-engineering the entire thing again).

as i have observed the bicycle business this is the typical chain of events. it's not that the trial-and-error guys overrule the engineers. it's that the engineers themselves end up with a final product based on the template i describe above. if i am mischaracterizing how it is that products in our industry come into being, i'm happy to hear from engineers and product managers that i'm behind that times.

this isn't to say that engineering always or usually falls short. i'm certain that in many or most cases you engineer a product and that product makes it right to market. but if it's something that might break, and might cause damage to the rider, that product is never brought to market through engineering alone. that product is ALWAYS tested, and if there is an early failure the nature of the failure informs the changes in product design, and it is at this point that the value of engineering lessens and trial-and-error now becomes the dominant design input.

---

Dan Empfield
aka Slowman

What he said.....

---

Chicago Cubs - 2016 WORLD SERIES Champions!!!!

"If ever the time should come, when vain and aspiring men shall possess the highest seats in government, our country will stand in need of its experienced patriots to prevent its ruin." - Samuel Adams

Engineering is a process...much like the scientific method. You take a best guess, you build it, you test it, and you discover why your best guess was wrong....so you loop back to the beginning with better assumptions than you started with and try it again. Computer aided design tools help shorten the amount of iterations this cycle takes before you get something you deem 'acceptable' but they do not eliminate this loop.

Interesting side note - there was a stem / seatpost company that we would NOT spec....because their products always failed VERY quickly in our fatigue testing. I won't name the company, but it was a brand that is widely recognized to be "bullet proof" and very durable. But we couldn't use them....

Oh come on! Will you tell me if I guess right? Are they still in business?

>IF the load case is properly measured/defined and the properties of the materials are well understood...then yes, in a statistical sense it is possible to determine the likelihood of failure of a particular component on a bike. Engineering like this is done on a
>daily basis, and has been done for a VERY long time.

Right. But how do you do that? It's often an iterative process. In my experience we design something based on a load case, lab test it. Then we give it to a Marine. Marine breaks it in 5 minutes. So then you ask the Marine exactly what he was doing when it broke, and your mind is blown, because WTF, why would anyone do that? But then adjust the load case, and do a design iteration.

As Slowman says, sometimes you can nail it the first time, and it's all good. The holeshot. That's often the case for relatively mature, understood things. But often if you're doing something innovative - something with little-to-no existing real-world time - the likelihood of the holeshot decreases rapidly.

Computer stuff helps. It's evolutionary - maybe revolutionary - but, again,not the be-all, end-all.

Hi everybody, this is Nick from TriRig. I used to post here as JudgeNick.

I want to thank Dan for letting me back onto the forum. I've been gone for a couple years, but glad to be back. I thought it would be nice for Slowtwitch forum members to be able to ask me questions here, since there is quite a bit of regular activity regarding my products. I'm happy to answer any questions you might have about the Sigma, how it was developed, what's going on right now, etc.

---

--
TriRig.com

Nice, glad you're back in the door. Kudos for the way you guys have handled the issue.

---

-Alex

I am personally curious as to the root cause of the failure and any significant contributing factors.

Why did you chose a die casting process and ADC12 as a material? I also assume the castings are made in china or SE asia?

Kudos on doing the right thing and getting out in front of the issue. Always great to see.

First off, I want to express a big, heartfelt THANK YOU for all the support you have all shown me in this thread, and throughout this forum over the years. I sincerely appreciate it, and it's really great to hear especially during this replacement process. Now, as to some of these questions:

Details about Failures:
For business reasons, I won't get into too much detail regarding the stem failure. As mentioned, we discovered that under certain conditions, the cast parts could fatigue and fail. Rather than attempt to "screen" which stems might pose a problem for which riders, we are simply replacing ALL of them.

Why Die Casting?
We originally chose to cast the stems because, after the initial cost of the die cast molds (which is considerable), we'd have the ability to make a lot of stems quickly, reliably, consistently, and with extremely high quality. Our past experience with cast parts (the Omega) has been an overwhelming success, further motivating the decision to cast the stems as well.

Why ADC12?
We originally chose ADC12 because it's one of the strongest aluminum casting alloys around. If you google the data sheet for ADC12, you'll find it's very similar to T6-6061, one of the most widely-used metals in all of cycling. That's the material used in the Omega brakes, and those have performed reliably for thousands of cyclists all over the globe, at every level of competition. Further, in order to maximize the properties of the metal and eliminate any microporosities/air deposits, we use the highest casting pressure our casting presses can take. That adds a little bit of time and cost, but helps make sure the ADC12 performs to its highest potential.

Why switch to T6-7075?
Our testing revealed that switching to machined stems versus cast stems would eliminate the fatigue and fracture problems the old design experienced. We could have simply moved to T6-6061, and that would have been sufficient to eliminate the problem. However, I wanted to go above and beyond what was simply necessary, and shoot for a higher standard of excellence. So we decided to use T6-7075, virtually the strongest and stiffest aluminum alloy there is. If you have a look at the data sheets for 6061 and 7075, respectively, you'll find 7075 is in a class all its own. In some categories, it boasts metrics at double or even triple the values for 6061. Using T6-7075 costs us quite a bit more, because the alloy is so hard that it requires longer machining time and more frequent bit replacements (and CNC bits are not cheap). But we really want to inspire confidence in our customers, and create the highest-quality products we can.

I know that this whole situation can be very inconvenient, and for that I personally apologize. But right now, my priority is keeping everyone safe, and making things right.

Thanks!

---

--
TriRig.com

While I don't dispute the tool wear issue, a lot of machinists will say that machining 7075 is faster exactly because it is harder -- that is, it is less 'gummy' and the chips break more easily, allowing for deeper cuts and/or faster travel. You may want to talk with a shop that specializes in machining aerospace alloys (such as 7075 and 2024) because I'm sure they have processes that are optimized for them.

Thanks. I apologize if I said something misleading about 7075. In general, you will find that it is ALWAYS more expensive to machine with 7075 versus 6061, simply due to material cost, if not machine time, bit wear, and other factors. Maybe it's "faster" in the opinion of some, but it's (almost?) universally more expensive. As far as TriRig is concerned, the difference in cost between 6061 and 7075 from our factory was not prohibitive. It was definitely more expensive, but easily worth it from my perspective. Personally, I've NEVER seen anyone quote a 6061 CNC job cheaper than a 7075 job.

---

--
TriRig.com

Sure, I can believe that. I'm just saying that's not due to cycle time, as 7075 is more machinable due to it being harder and having nicer chip properties. But, hey, don't take my word for it -- even the Aluminum Association rates 7075 as being more machinable than 6061.

Jamaican wrote:

See if Dengfu will sell the FM087 Flat Stem Separately. Very similar steerer clamp requirements.

dengfubikes wrote:

• Stem for FM087 = $60
• Shipping to USA = $28 by EMS
• Stem length = 75mm
• Steerer tube length requirements: min 24mm, max 34mm

My steerer tube height = 22.5mm
Dust cover removed = 26mm


However, it appears that the FM087 drops down in front of the frame. I'm awaiting a response from Dengfu regarding the distance between the steerer tube and front of frame.


My NP3:

---

/Howie Nordström

No...I'm saying (and provided links to examples) that some engineers have actually instrumented up a bike and determined the typical loadings as an INPUT into the design process. You stated that this is a difficult thing to determine. It's not...if you know how to using mechanical instrumentation. Especially for the case of the loading on a stem, as we are discussing here.

Slowman wrote:

the difference between you and me is that you think engineers have figured bikes out. i know for an absolute fact that they haven't.

Perhaps the ones you are familiar with haven't. But, I've already pointed you to examples to ones that DO figure out the loads and inputs.

Slowman wrote:

this isn't a binary choice between trial-and-error and engineering. you begin with engineering. you end with testing. when you test, and you are surprised and disappointed with your result because it does not comport with your expectations based on engineering, the result of the test generally informs the design changes you make (rather than going back and re-engineering the entire thing again).

as i have observed the bicycle business this is the typical chain of events. it's not that the trial-and-error guys overrule the engineers. it's that the engineers themselves end up with a final product based on the template i describe above. if i am mischaracterizing how it is that products in our industry come into being, i'm happy to hear from engineers and product managers that i'm behind that times.

You're making my case for me ;-) I'm sure what you have observed has been the dominant "process" in the bike industry for a long time. I've already stipulated that this was the case until relatively recently. However, nowhere above in those 2 paragraphs do you mention 2 things critical to the engineering process. First, you need to actually take measurements to determine the load case. This is the thing you claim is extremely difficult. I've already given you examples of how that is just not so. Go read the Cervelo links again. Second, AFTER the testing, you need to "close the loop" on the initial modeling to make sure that: a.) the test apparatus/protocol faithfully replicates the actual loading (again, see the Cervelo link...they found that the "industry tests" were lacking in that regard), and b.) if the results don't agree with the modeling, you need to determine what was missing or incorrect in the model so that the models can improve for future analysis. To not do so is to just condemn yourself to the crude "open loop" engineering process you described above.



Resorting to just "trial and error" without understanding is merely tinkering, not engineering.

---

http://bikeblather.blogspot.com/

To be clear, I've never said that modeling alone is sufficient.

In the case you mention above, how was the original load case defined? Did you take a similar device that was instrumented and let the Marines "play" with it?

---

http://bikeblather.blogspot.com/

well...not quite true.i built a stem specifically for the shiv TT with it's 1" steerer while also allows for use of 31.8 bars and features fully internal wiring that will enter the shiv TT frame without ever being exposed to air (from bar to stem to frame, with a slight frame mod) and have been racing it since 2012. machined 7075.

:)

it seems to me that you're trying to bend this into a battle of two approaches: engineering versus trial-and-error. that's not how this industry works and that's certainly not how i've characterized it. i know most of the more senior engineers at most bike companies in the U.S., and i know how they work, and how they work is how i've described it. using your reasoning there is no need for wind tunnels, or fatigue and high impact testing machines. maybe that's what you believe, i don't know. but until all the loads and forces applied to all parts of a bike - by the wind, by the road, by the rider - are understood, the industry is going to keep doing things the way i describe (and i don't want to ride anything made by anybody who thinks his engineering knowledge alone is enough to make safe products). everybody tests. not only do they test, in many cases it's enshrined into law that they test. and were it not a legal requirement, they would still test, because the engineers in this industry (fortunately) are not so arrogant as to think their engineering knowledge supplants the need to test.

---

Dan Empfield
aka Slowman

"To be clear, I've never said that modeling alone is sufficient."

i think that is certainly your implication. that is the take-away of many or most, if not all, of us who read your posts earlier in the thread. you write as if all the forces subjected to a bike and all its constituent parts are kind of like a species whose genome has been entirely mapped. the implication is that testing is therefore superfluous. it's pro forma. it's ministerial. nothing to learn here.

---

Dan Empfield
aka Slowman

What? I am most certainly not and have never even implied it's either/or. That's something you've been putting forth, not me. In fact you yourself above in post#52 said that "i think that you view product testing - fatigue testing, most notably - as valuable and necessary. no disagreement between us there."

I haven't said anything in between then and now which stated or implied anything different. So why do you now think I hold the opinion that testing is completely unnecessary? I get the feeling you haven't been thoroughly reading my replies.

Let me sum up. You said "mechanical and structural elements of bikes are so complex they defy mathematics." I would like to see if others with structural engineering backgrounds agree or disagree with your claim. I certainly do not, especially in regards to the loads on a stem.

---

http://bikeblather.blogspot.com/