Torque Testing

At a guess, the motor is only supplying torque when there is a current flowing through it. At the point when the current has been shut off and the motor begins to slow the only torque that the motor can supply is as a result of its own inertia, i.e. as its self-contained 'flywheel' slows to zero.

Could you supply a sketch ofr the test stand cinematics? And give a detailed description of the test? Some times words are subject to subjective interpretations so that a drawing will give more objective base for analysis.

Just wanted to dismantle the inertia theory that was posted yesterday. (But thanks for the inspiration anyway!)

The torque necessary to turn the LSD is (depending on the internal configuration) between 50 and 70 N-m. When the motor starts slowing down, the readings given by DAQ are in the 10-15 N-m range, which is in theory not enough to even turn the diff.

Must be something else........maybe ghosts?

Have you records for current and torque versus time ? This could be of help. And of course the sketch i already mentioned.

You may have a problem with hysteresis. The torque sensors we use are in house designed cam actuated force gages. The cam action allows adjustable sensitivity and acts to limit the hysteresis inherent in strain gage sourced measurements.

Heat is another factor which may affect the output of your torque sensor. Calibration should occur at operating temperatures if any of the components in your measurement system change values with heat. At one point during a frustrating trouble shooting session I used a hair dryer to find the heat sensitive component.

Incidentally i develop sensors based among other on strain gauges. It is a well known principle : every mechanical interface in the transmission of efforts to the sensor leads to a more important hysteresis.

Many times this is due to 2 factors:

- the position where the vector is introduced in the sensor does change in friction of the vector magnitude which also leads to a nonlinearity.

- there are minute glidings in the interface which lead to friction and an augmented hysteresis.

Your considered a simplified model of the strain gauge transducer and avoided a temperature compensation -which is a very simple procedure- as well for the bridge unbalance as for the sensor sensitivity. A good compensation from the start would have made the use of the hair dryer unnecessary.

This is often the case when engineers develop for "economical" reasons sensors which at the end do cost more than bought sensors but such costs are "hidden" so that for the management there is a "tremendous" economy done.

You understand that i am a bit surprised that you introduce cams (a mechanical interface with sliding possibilities) to reduce hysteresis. I would appreciate to know more since i am very open to any new thing i can learn.

"A good compensation from the start would have made the use of the hair dryer unnecessary". Unfortunately I didn't design the system. I just inherited it and had to make it meet specifications.

The cams are roller and ramp designs with low friction values. The load cells are capable of measuring 1.5 million pounds of force. Solid rocket motors shake and vibrate. Cams/rollers are used to diminish those effects and set ratios for more or less sensitivity.

I see you are involved in the test of boosters and deal with quite high loads. If I assume the mentioned force as the "nominal" then assuming further that such tests are done with a rig on 3 sensors for stability you deal with total forces of 2E4kN.

From an other point of view such projects are not economical since usually state financed so that economy is not the first concern.

In the field you are active uncertainty of up to 5% makes no differences so that that cam+roller systems are not disturbing the measurement in a major way.

Never the less my comment is valid.

Of course working in your field it is some times difficult to buy.

I wonder which way you chose for the calibration: pyramids of lower capacity sensors, hydraulic intensifiers or direct comparison with a reference sensor in a hydraulic press? Which uncertainty have you obtained? To be good for the 5 % errors at the end you should have at calibration results better 1%. Have you obtained it?

Years ago i developed a system to weight and lift off-shore modules for own weights of up to 4.000 metric tons. But since weighing was the main goal i aimed at less 0.2% maximal total error. We got on 8 sensors repetitively errors less 1/1000 of measured weight and were able to demonstrate that weight is an invariant (what was strongly doubted before by the weighing engineers).

I am asking the questions about calibration just because i am curious.

Nick Name, calibration is performed using dead weight, local gravity measurement and direct comparison with a reference sensor. Many issues constrain the development of calibration data with less than 5% error as you pointed out however we have reduced the error budget to 2%.

Dealing with massive loads, vibration, heat, cold, wind and other influencing factors make for an interesting day at the test site. Lab developed measurment techniques are often far removed from the realities of measuring the various parameters associated with the "controlled explosion" of a million pounds of gunpowder, powdered aluminum , rubber cement and the oxymoron known as "stabilized nitroglycerin".

Very interesting.

It could be quite a work to calibrate with so big dead loads. I know that in the UK they have this type of calibrating rig. Such a calibration is very time consuming. In order to avoid handling of enormous loads the PTB(Germany) developed a hydraulic intensifier system using a much smaller dead load and amplifying force via hydraulic cylinders (seal and friction free of course).

It is very interesting to analyse the uncertainty of such systems. I do not understand why if you use dead weight loading (with the care to measure the local gravity ) you also use a reference sensor. Is that not a non needed redundancy? The 5% error I thought for the jet force value. To be sure of this the uncertainty for the results, the sensor it self should be better 1.5% assuming a normal distribution of errors and dispersions. The budget of 2% is with k=2? Does it consider only calibration or is it the estimated measuring error?

I can imagine how complex the problem you are confronted with can be since the boosters are quite long and the side wind can introduce a lot of disturbances. With respect to temperature may i say that the usual approach of static compensation is not always satisfactory since it is linear and the transducer bridges are in the temperature field not linear. I used an active compensation (such huge system can afford the cost for it) via measuring actual temperature and compensating according to a temperature previous calibration. The problem remains for the sensitivity changes since they can be determined only under load at different temperatures (effect on young modulus) but the dead weight is the best loading for it.

Thanks for your explanations I appreciate. If you will ever need an other opinion please do not hesitate it will be a pleasure.

Nick Name, your analysis regarding the difficulties of obtaining accuracies below 5% with such a massive structure is correct.

FYI the loading crane is capable of lifting 4 million pounds with a boom height of 450 feet. Someone before my time tried (or studied)the German system you mentioned. However for some reason the idea was abandoned. Perhaps it was the friction of the seals ("sticktion") in the hydraulic cylinders which made the system unsuited for our purposes.

The reference sensor is actually part of a system which is used as one of the backup Data Aquisition System in case of an "incident". Remember, at the temperatures and pressures inside a solid rocket motor, we're only a few "points" (degrees and or psig) away from a rather violent decomposition of the fuel/oxidizer mixture. During the test and development phase of any rocket motor that possibility becomes a probability.

In any case, the methods now employed to obtain thrust measurements are satisfactory to current engineering management. As you mentioned in your previous post, costs are often the ultimate regulator of technological innovation and utilization.

Thank you for your offer of consultation.

The two cylinders building the force amplifier are -as i mentioned- seal and friction free. I presume that the design problem was a brake for your (may i say) colleague.

There is a problem related to the expansion of the cylinders under pressure and the supply for leaks compensation. The design of a dead weight system is a lot easier being only a problem of mechanical structural design. It is more a strain calculation as a system design.This may be the reason. Due to your not too frequent tests the time spend for calibration is a secondary problem.

A last question: the sensors are designed for traction or compression? The reason is that I am accustomed to methods used for trust measurements in main engines (liquid fuel + oxidizer and own weight small with respect to trust) but not at all familiar with solid boosters which have an own not neglect able weight.

Again thanks for the very interesting comments.

Nickname

PS If you will ever want a more specific comment send your mail for a direct contact.

Was last July (2006) a very rainy month in your country?

Nick Name, The sensors are designed for compression. Since solid rocket motors must produce more thrust more than their weight, at ignition we see an increase of compressive force (thrust) on the load cells which is less violent that of a liquid fueled engine.

Liquid fueled engines can be "tuned" so their vibrations and harmonics are less destructive. The resonance and harmonics of solid motors vary thoughout the burn and are not precisely predictable. A motor 30 meters in length may "grow" as much as 25 centimeters during the burn which also adds to its changing fundamental resonance and harmonics.

Mount the motor on a hinged plate with a load cell under the free end (or a small cylinder with a gauge and transducer). Calibrate the gauge output by putting a level bar on the motor shaft and put a test mass on the bar and calculate how much torque per gauge readout you have with the test mass. Then throw that stupid thing you are using away and use the torque and rpm.

Of course a torque can be measured either in the direct line or a a reaction torque as you suggest. There but a difference in the dynamic behaviour of the system. The reaction torque measurement is ok only if the required band pass is rather low since the combination of the motor mass and the hinge+ transducer has an own frequency and can filter high harmonics. If the test is a low frequency or "static" test then your suggestion is ok. The use of a cylinder with pressure gage depends on the type of cylinder and internal friction which can introduce quite important errors again the possibility to use such a device depends on the uncertainty which can be accepted.

The general problem is that measuring requires several definitions : uncertainty, band pass and so on before a measuring system can be accepted as "good" for the project.

If the torque or force are to be determined with all their dynamics the hinge system can be too un precise.

Would it be possible for you to be reading the inertial forces of the circulating fluids in the differential? Maybe replacing the contained gear lube with a spray system would help eliminate some of that.

RichH

Nope. Diff is running dry for this specific series of tests. I also tried recalibrating the sensor sensitivity -- unsuccessful: The offset value shown in the last readings of each series I can make bigger or smaller, yet not disappear.

Unfortunately, "the force is with me, all the time!"

Anybody know where I could find a semi-used brushless rotary transducer of 200-250 N-m capacity at a reasonable price? I already have offers from companies all over the world for new units, but those puppies all run above 2000 Euros each.

You mention "recalibrating" what do you understand by it ? Full recalibration against a reference sensor or the zero compensation with a potentiometer? It is a pity that you do not give the informations i asked for since they could allow a full explanation of the "happenings". It could be possible that you (or some body else) overloaded the sensor and that could lead to a wrong result.

Without data to analyse it is very un-probable to give you a near to truth solution.