Accuracy of a torque wrench (and other torque-related questions?)

Factual Questions
1

I did a search, and found a number of threads regarding torque, but not this…

I have a torque wrench. I know how to use it. (Heh)

However, I have no idea if the torque put on the bolt is actually the torque specified.

Is there a way to test a torque wrench, to make sure it is working properly, without buying a second torque wrench?

Beyond this, how accurate must a torque wrench be to make sure work on, say an automobile engine, will not actually damage what it is you are trying to torque?

For example, say I have a spec that says a bolt needs to be torqued at 20 ft/lbs. Does that permit a ± 10% error margin to allow for torque wrench calibration errors?

As I understand torque (in layman’s terms), it is basically the measure of the tightness of a bolt/nut that holds one object to another. With my 20 ft/lb example, if you torque it too much (over 20 ft/lbs), you could strip/break the bolt, damage the gasket between the objects, etc. Not enough torque, and the bolt could work itself loose, cause a gasket to leak, etc. Is this a fair, general description?

In reading about torque wrenches, there seems to be a number of options… The best are extremely expensive, and unless you are making your living as an auto-mechanic, working on aircraft, or some NASA project, there are inexpensive ones that a DIY’er can buy and use to make most general repairs. However, these cheaper ones are not as accurate, nor do they claim to be. But people seem to use them routinely without major catastrophic results (or maybe they die as a result of their torque errors, and no one knows WHY their engine blew up).

Even the best torque wrenches need to be re-calibrated to ensure accuracy… And I doubt very seriously that most torque wrenches out there in use have been calibrated/tested on a regular basis to ensure that the torque required is properly delivered by each torque wrench. So there must be some wiggle room in the torque specifications to account for this…

Am I right? Or must torque specifications be met within ± 1% to ensure things function as designed?

And an added question. When I was a kid, I would work on cars with my friends, and none of us ever used a torque wrench. Things were just done by “feel”. I now know that this was an incorrect approach, but to be honest, I can’t remember anything that we did that failed because of an incorrectly torqued bolt, outside of the occasional stripped nut/bolt. Were we just lucky? Is ignorance truly bliss?

2

In a previous life, I had a garage full of motorcycles. Because a lot of the engine work was aluminum, I bought two torque wrenches. One for the soft touch 20/ft-lbs, and another for the heavy 100/ft-lbs.

I was fortunate to be working at a factory that had a calibration lab. Being frugal, I bought the torque wrenches at the China store. I took the torque wrenches to the lab for testing. And they were within specs. I was surprised.

I think the main thing is to know the material you are working with. Aluminum/ steel and fine verses coarse thread. Stripped threads create bigger problems. Best to avoid that.

I hope this helps.

3

Yeah, the calibration lab where I work calibrates torque wrenches, plus when any torque wrench is signed out for a job it has to be checked for accuracy (there is a tool for that) in the tool crib before being issued, but I’m not sure what tolerance is acceptable.

I doubt automotive shops check anything after the wrench is purchased.

4

A torque wrench is measuring force so it would be possible to rig up something like a weight scale with a sliding weight or add the appropriate weight at the correct location. Here’s an illustration.

5

I think torque bolts will give you the most accurate measurement; I think they may be prohibitively expensive though.

6

For what it’s worth, my dad worked as an auto mechanic from the time that he was discharged from the Army after WWII until the mid-1970s. This would have been on iron blocks and engines, not aluminum, but he never had a torque wrench and had a good reputation around town.

7

I used to sell Snap On tools. We had a service where mechanics could send in their torque wrenches and have them calibrated. Usually cost less than $100. Most guys didn’t do it but the ones who did engine rebuilds or were more professional sent them in yearly.

One thing to note is their accuracy range. I don’t have the specs but rule of thumb is don’t count on it in the top or bottom 10% of its range.

8

A common reason for warped brake discs on cars is improper torquing.

A couple of years ago I ruined a $200 carbon fiber seat post on my bicycle because I didn’t have a torque wrench that went as low as 5 NM. Now I do.

9

Very simple to hook a fishing scale up to the handle on your wrench and see if it matches the setting when pulled. Just keep in mind 12" pounds = 1 foot pound and find out your mechanical advantage by measuring the exact length on the handle where you have your scale pulling from.
I have checked a couple of snap on’s and they were within 1#.

10

Snap-on was one of the higher end brands I was thinking about. For the serious, professional mechanic, they want great, reliable tools. For that, they pay a premium price. But I am surprised to learn that most guys who have a snap-on torque wrench wouldn’t have it re-calibrated annually.

Do you know how out of spec. an average snap-on torque wrench would be once it was sent back for its yearly recalibration?
Is using a torque wrench over-blown in importance? Because I personally have never used a torque wrench to guarantee my wheels are bolted onto my car at the proper “Maximum allowable torquage” (Marisa Tomei voice, as Mona Lisa Vito), and I have never (yet, anyway… Thank god!) had a wheel fly off of my car as I was driving down the street. Am I just lucky? Abnormally strong? Have an amazing ability to torque something by hand to meet the required specs without actually, you know, measuring them, or is this just a fluke?

11

For most jobs that an owner would do on the average modern car, going by feel will be fine. If you’re putting a cylinder head back on, that’s another story. I’ve never seen a shop (ranging from cheap oil or tire change places to European luxury dealers) use a torque wrench on an oil pan plug or lug nuts/bolts. Most seem to just run the impact gun until things stop turning - an owner doing it by feel is likely to be closer to factory spec.

Regarding torque wrenches, the beam type can be more accurate (and less expensive) than the click type. But the advantage of the click type is that you can be looking directly at the work and still hear the click, while with the beam type you have to switch from the work to the indicator and back. Plus, with a beam wrench you have to compensate for parallax issues when reading the indicator. However, the largest source of inaccuracy is probably inexperienced users who use extensions, operate the tool at an angle to the fastener, etc. It is also important to “unwind” the click-type back to zero (or its minimum setting) before putting it away. If you have an impact tool and are only concerned about a few torque values you can just get a torque stick, which looks like a driver extension but is made of spring steel.

For more than you ever wanted to know about this stuff, I suggest reading through the text of Fastener Black Book. Don’t buy it just for that, though - most of its considerable size is composed of tables, etc.

Here’s a humorous parody of a torque wrench ad.

12

Onced used a torque calibration unit over the weekend from our satellite facility to recal a torque wrench I owned.

Couple of things I noticed.

  1. The torque wrench settings were slightly non-linear.

  2. Never could get the top torque or bottom torque to align to the scales of the calibrated instrument, so I just set it for the middle and let it go at that.

Talking to the mechanical engineer, he said the key was ensuring that it was more important to ensure the torque settings between the different screws were close to one another rather than the absolute cal. Hence, the allowable range in torque settings on some spec sheets.

FWIW, I’ve had the thing for 25 years, and had checkout the calibration of couple of times; didn’t change more than 5%. That being said, this a clicking type and so I always back off on the spring when done.

(ETA: A beam type is always more accurate and now I see I’ve been beaten by TK. Also, I seem to recall when my brother bought these little plastic sticks to ensure that the torque wrench was in spec when cranking down on the crankshaft. They would squish to a particular size, and from that you determine if the torque was Goldilocks).

13

You need to go to an actual tire shop. They use torque wrenches on aluminum alloy wheels where I go after the initial take up. Costco, Volvo dealer, Discount Tire.

Back (way back) in my Nuke weapons days; everything had a torque, a pattern, and sometimes torque application in multiple steps. Using a non-calibrated or out-of-calibration torque wrench or screwdriver would get a team de-certified in a heartbeat.

14

Yes. The idea of a fastener is to stretch to apply a load securing the interface. Torque is a measure of how much load is on the fastener, which translates into friction of the fastener interface to prevent back out as well as anchoring of the surfaces to each other. Too much torque damages something, too little and it does not hold solidly. This means a leak or the interface slips or the fastener backs out.

When I was working on space hardware, torques were speced for all applications. When determining the application torque for a bolt, it was set around 65% of the yield, but we were always* given a range rather than a number. Standard was 60 - 75% IIRC.

We were required to use calibrated torque wrenches, and get them checked annually. Failing a cal test meant going back and reviewing all the documentation that used that wrench and determine what would be the effect of the out of tolerance.

Typically when installing we would aim for the center of the range. You would get close. Then if there was an out of tolerance, if it wasn’t large, you could numerically verify it would still be in the acceptable range and you were good to go. Finding you could be out of range was more problematic. Hardware on the ground could be reprocessed, but stuff in space had to be either justified as acceptable or taken out of service and replaced. Yeah, a lot of justifying. Helped by things like fastener redundancy and design practices.

Anyway, you have leeway on your application. Part of design is picking the right size and number of fasteners to secure the item so that the 65% zone of the yield gives you proper grip. I’m sure auto designs have reasonable tolerance margin.

Trickier was usually tiny screws, because the urge to tighten until it feels snug. On tiny screws, that can be overtorqued and yield the screws.

Yes. Torque wrenches work by using a spring - either a torsion spring for click wrenches, or the flex of the beam on a beam torque wrench. Beam torque wrenches are simple mechanisms that basically get out of torque by the needle being bent or the beam being bent. Accurate and hard to get out of calibration, but require reading the scale during use and don’t have physical feedback when load reached.

Click wrenches have more moving parts. The two most likely ways to affect calibration is to bump it hard and knock the setting of the handle to the adjustment screw, and the torsion screw getting a set, i.e. change to it’s shape. That changes how much load it applies.

Springs are linear over the majority of their range, but near the ends of their range of travel, they get effects that make the spring rate less linear.

Found some interesting resources. Here is a torque wrench that a guy cut up to demonstrate how it works.

Also, link to someone who took data on a handful of wrenches.

Article on calibration and handling of torque wrenches.

*Some heritage designs did not have ranges, just single values. We had to play games on those installs. Proper design on anything new defined ranges.

15

ISTR that torque wrenches are generally regarded as being accurate to within ± 20%. The situation is even fuzzier because of the uncertain relationship between bolt torque (what you are measuring with your torque wrench) and bolt tension (what you are actually trying to achieve). There’s a lot of variability here because of thread conditions: dirt, corrosion, lubrication (or lack thereof), mechanical damage to the threads, speed of tightening, etc. The result is a wide variation between the torque you are applying to the wrench and the bolt tension you are actually creating. But the design of the joint allows for this, so unless things get really out of whack, you should be fine.

In situations where very precise/repeatable bolt tension is required, a simple torque spec is not good enough. Instead, you’ll see the service manual call for a “torque-turn” spec. In this situation you apply a light torque (maybe 20 N*m) with a torque wrench to take up all of the slack between the parts in the joint, and then you turn the bolt through a specified angle, like 1/4-turn or 1/2-turn, with a degree wheel attached to your wrench head so you can measure that angle accurately. The relationship between torque and tension may be quite variable, but this is not so for the relationship between bolt rotation and tension: you can measure 90 degrees of rotation to within 1 degree, so your accuracy there is close to 1%, and the geometry of the bolt (thread pitch, diameter, material properties, etc.) are very repeatable, so X rotation=Y tension gives you very consistent results. The large percentage of error in the initial snug-up with the torque wrench doesn’t matter a lot because 20% of a small torque = a small absolute error in tension next to the very large tension you create during the degree-turn. You’ll often see a torque-turn spec on head gaskets, con-rods, and crank main bearings, for example, where a reliable seal is required or where bolts are chosen so that a high percentage of their strength must be utilized to hold the joint together.

Some people are really bad at torque-to-feel; they will consistently undertighten, or regularly strip threads, despite their best effort. Using a torque wrench assures that even the village idiot can tighten a bolt properly. Where this really matters is in a professional setting, where stripping a fastener can result in extra/unpaid work to fix the customer’s machine, and undertightening a safety-critical fastener (like a lug nut) can result in a lawsuit if someone gets injured/killed.

16

If you’re talking about horizontal extensions or crowsfeet, yes, those will affect the reading and must be compensated for.

Vertical extensions will not change the reading.

17

Are you talking about Plastigauge (shown here before and after tightening the cap)? Plastigauge doesn’t tell you a thing about torque. It’s used to measure bearing journal clearance.

18

Handle extensions (cheater bars) are a definite No No.