While writing a response to a question from an engineer regarding a government standard I helped write, I came across an interesting paper I had not seen before. Often when a thread comes up about torque wrenches I will say you there is so little known about the joint you are working on that at best they give you a (false) sense of security that you have a good joint. Most joints on a bike are slip critical, and as such it is more important to have high friction at the interface. Home mechanics are concerned about crushing their composite bits. So what is one to do? To me, the answer is obvious.
The paper at the link below is quite interesting. It is a study of motorcycle suspension parts, and bolt preload is monitored via a strain gage with the following variables
1) lubricated or unlubricated
2) cast or forged parts
3) spray painted or anodized parts
4) first installation versus the sixth installation
My experience would tell me, for a fixed input torque
1) lubrication causes preload increases
2) never looked into this, but would expect no difference if parts are equally polished after manufacturing
3) painted would tend to lower friction thereby increasing preload
4) repeated installations without cleaning surfaces between could swing either way. For lubricated, you could get migration of the lubricant to other places causing preload to increase. For dry, assuming no contamination of body oils, wear particle build up will cause more friction and reduce the preload
The variables they found to have the greatest effect on preload were (ANOVA ranked)
Surface finish - painted results in higher preload
Lubrication - lubed increases preload
Combination of lube and surface finish
Number of tightening - preload dropped after multiple install cycles
Combination of lube state and installation cycle
Far and away, the 2 biggest factors were lube state and surface finish.
The test data shows that for a constant applied torque, the bolt preload ranged from 2165 N up to 21648 N. Taking out all the painted surface finish data cuts the range to 2165 N to 15153 N, still quite a spread. From there, the data is clearly clumped, with the unlubricated tests providing the lowest 12 preload (and 6th cycle generally the lowest of those) and the lubricated being the 12 highest (and with reinstalls generally occupying the highest preload slots).
Unlubricated - 5075 N mean, 1546 N standard deviation
Lubricated - 10571 N mean, 2221 N standard deviation
In our standard we require 90/95 probability/confidence levels, which for the test data would lead to preload ranges of
Unlubricated 945 to 9205 N
Lubricated 4640 to 16502 N
It is not that I am against torque wrenches, it is that home mechanics often don't have the knowledge nor design detail to know how to correctly torque the joint. That torque callout on the part doesn't come with information regarding if to lube, and if so, where to lube, Aftermarket parts have no idea about the surface finish of the mated parts, is applied torque the total torque or torque above running torque? There is simply too little information provided to the end user, so there is no telling if the joint is safe or unsafe. You are flying blind. The torque wrench isn t the villain, but rather the lack of information.
So for those slip critical interfaces with carbon bits in the joint, what would I do? Friction paste at the shear plane, no lube, and a "dirty" thread surface to increase nut factor thereby reducing clamp force. Torque, test for slip via expected service load application, and increase torque if necessary. But creep up on it, Yes unlubricated comes with higher variability, but the overall benefit of a higher nut factor gives me more assurance.
Anyway, interesting paper on realistic maintenance operations.
https://www.researchgate.net/...minium_bolted_joints (or PDF at https://www.researchgate.net/...um-bolted-joints.pdf)
The paper at the link below is quite interesting. It is a study of motorcycle suspension parts, and bolt preload is monitored via a strain gage with the following variables
1) lubricated or unlubricated
2) cast or forged parts
3) spray painted or anodized parts
4) first installation versus the sixth installation
My experience would tell me, for a fixed input torque
1) lubrication causes preload increases
2) never looked into this, but would expect no difference if parts are equally polished after manufacturing
3) painted would tend to lower friction thereby increasing preload
4) repeated installations without cleaning surfaces between could swing either way. For lubricated, you could get migration of the lubricant to other places causing preload to increase. For dry, assuming no contamination of body oils, wear particle build up will cause more friction and reduce the preload
The variables they found to have the greatest effect on preload were (ANOVA ranked)
Surface finish - painted results in higher preload
Lubrication - lubed increases preload
Combination of lube and surface finish
Number of tightening - preload dropped after multiple install cycles
Combination of lube state and installation cycle
Far and away, the 2 biggest factors were lube state and surface finish.
The test data shows that for a constant applied torque, the bolt preload ranged from 2165 N up to 21648 N. Taking out all the painted surface finish data cuts the range to 2165 N to 15153 N, still quite a spread. From there, the data is clearly clumped, with the unlubricated tests providing the lowest 12 preload (and 6th cycle generally the lowest of those) and the lubricated being the 12 highest (and with reinstalls generally occupying the highest preload slots).
Unlubricated - 5075 N mean, 1546 N standard deviation
Lubricated - 10571 N mean, 2221 N standard deviation
In our standard we require 90/95 probability/confidence levels, which for the test data would lead to preload ranges of
Unlubricated 945 to 9205 N
Lubricated 4640 to 16502 N
It is not that I am against torque wrenches, it is that home mechanics often don't have the knowledge nor design detail to know how to correctly torque the joint. That torque callout on the part doesn't come with information regarding if to lube, and if so, where to lube, Aftermarket parts have no idea about the surface finish of the mated parts, is applied torque the total torque or torque above running torque? There is simply too little information provided to the end user, so there is no telling if the joint is safe or unsafe. You are flying blind. The torque wrench isn t the villain, but rather the lack of information.
So for those slip critical interfaces with carbon bits in the joint, what would I do? Friction paste at the shear plane, no lube, and a "dirty" thread surface to increase nut factor thereby reducing clamp force. Torque, test for slip via expected service load application, and increase torque if necessary. But creep up on it, Yes unlubricated comes with higher variability, but the overall benefit of a higher nut factor gives me more assurance.
Anyway, interesting paper on realistic maintenance operations.
https://www.researchgate.net/...minium_bolted_joints (or PDF at https://www.researchgate.net/...um-bolted-joints.pdf)