Wind Tunnel Balance Calibration, Willard Smith, 1972
NASA-AMES RFS:213-2
Moffett Field, California
September 26, 1972
MEMORANDUM for Assistant Chief, Experimental Investigations Branch
From: Willard G. Smith, Electro-Systems Engineering Branch
Subject: Wind Tunnel Balance Calibration Correlation
On several occasions project engineers have complained that balance sensitivities obtained by check loading in the tunnel differ from values obtained in the calibration lab by as much as 1% of full scale. There are a number of possible reasons for discrepancy, such as improper hook-up of readout instrumentation, improper calibration technique, or actual differences in balance sensitivity within the repeatability range of the balance.
I recently made some tests to see if simulated balance sensitivities, established in the calibration lab could be duplicated in the wind tunnels. A Bytrex,variable bridge calibrator was used to electrically simulate a perfect balance. A standard sensitivity was selected, representative of Task balance outputs, and the ratio of load output to calibrate reading was obtained for calibrate resistors #10, #11, and #12. The data from the calibration lab were then compared with data taken in the tunnels (8X7, 9X7, 11-foot, 12-foot and 6X6), and all were identical within the resolution of the readout. These tests verified the integrety of the complete electrical system including lead wires calibrate resistors.
The calibration technique is straightforward and could hardly alter the primary sensitivities although the interactions are strongly affected by variations in load and fixture alignment, etc. Accuracy of the calibration weights would not be a source of significant error. Our brass calibration weights were checked after ten years of service, and those few which were not within ±0.03% were taken out of service. In 1970, the Unitary 50# lead weights were found to be uniformly low by about 0.25%, apparently due to abrasion over a period of ten years, and were corrected to ±0.02%.
It must be concluded then that the sensitivity measured in the wind tunnel is correct within the repeatability of the balance. Over the years a great many tests have been made to establish the repeatability and accuracy of wind tunnel balances. Unfortunately, the data are not entirely systematic and we obviously do not understand all of the factors that influence balance performance (at fractional percent of full scale levels). It is apparent that the fit and location of the balance pin and the fit of the calibration body influence the balance performance
A number of tests have been run in recent years in an attempt to verify performance of the calibration mechines and to look at balance performance. Some typical examples are presented below. These results are mostly point-forpoint comparison of the raw data and so do not reflect the benefit of curve fitting.
Description of Test Maximum Discrepancy in % Full Scale Sensitivity Interaction
1) 1″ Mk V Task balance recalibrated after 8 months time 0.3% 0.5%
2) 1″ Mk V Task balance machine loaded, same balance pin, 0.5 0.8
roll angle in adaptor of 0, 90, 180
3) 1″ Mk V Task balance comparison of machine and hand-load 0.9 0.8
(different calibration bodies)
4) 1″ Mk IV Task balance comparison of machine and hand load 0.5 0.5
using machine balance adaptor in both cases
5) Same as above using hand load calibration body 0.6 0.4
6) 2″ Mk I Task balance comparison of machine and hand load 0.6 1.0
7) 1.5″ Mk XII hand loading in roll only with roll angles of 0, 90, 180, 0.3 3.7
270 and each with a different balance pin location
8) 1″ Mk XII Task balance machine load axial sensitivity at beginning 0.21 –
and end of calibration
9) 1.5″ Mk XII Task balance machine load axial sensitivity at beginning 0.12 –
and end of calibration
10) 4″ Mk II Task balance machine load of side-force with a plus 0.17 –
full scale preload and a negative full scale preload
11) 3/4″ Solid balance comparison of machine and hand load 0.6 0.7
12) 3/4″ Solid balance machine loaded at roll angles of 0′ and 90′ 0.4 0.7
As previously stated, these data are worst case, single-point readings and include random as well as systematic errors. Some idea of the repeatability of balances under identical physical conditions is shown in Cases 1, 8, and 9. Case 10 illustrates the effect of previous load on repeatability. The computed sensitivity changed 0.17% although each set of data had a maximum deviation of only 0.06%. Comparison of Cases 3 with 11 and 2 with 12 shows that solid balances offer only slight improvement over Task balances.
It has long been known that Task balances are sensitive to internal strains generated by attachment of fittings, especially the balance mounting pin. Cases 5 and 6 show the effect of changing the calibration body and Case 3 adds the variable of different balance pin location. Fortunately, the same balance pin location can be used in the wind tunnel test and in the calibration. Case 7 shows an unusually bad example of balance pin influence. The large discrepancies may include a cross-product effect since a tare weight of 1/6 the normal force capacity was present when loading roll, but this is not believed to be a significant effect. Distortion of the outer case of the balance is more likely the cause. The outer case is not concentric, so pin stresses in the thick and thin walls will be different. Also, the pin acts as a fulcrum for relative motion of the calibration body about the balance due to rolling moment. The distortion of the outer case is different than for each of the four pin positions, If the calibration body had fit more snugly on the balance in Case 7, the distortions would have been limited and the data might have looked better.
In summary, a great deal of work has been done in an attempt to evaluate the performance of wind tunnel balances and calibration techniques using both hand loading and machine loading rigs. The data presented above is a small but typical example of the comparative tests we use. Because of the many factors which appear to affect balance performance, it is very difficult to positively isolate and identify individual factors. Much of the data is inconclusive, but some general patterns are evident.
In Task balances axial force is the most accurate component followed by normal force. Rolling moment and side force are the least accurate, and the performance of all components is degraded by the presence of large rolling moments. All balances, solid as well as Task balances, are sensitive to the fit of the calibration body and model and to the fit and location of balance pin. For maximum balance accuracy, the model must be a light push fit on the balance and the balance pin must be in the same location used during calibration. Each project engineer should verify that model pin hole is consistant with the limited pin locations available in the calibration rig. This verification is important and should be done during the pre-test conference.
It is difficult to assign an accuracy to wind tunnel balances because of the many factors involved. The accuracy of normal balances of 1.5 inch diameter or larger is ±1/2% of balance capacity if the fit is good and the correct pin is used. If the wrong pin location is used, accuracy may be degraded to ±1.0%. Accuracy of the smaller balances is no better than ±3/4%. There is considerable evidence that the accuracy of balances is degraded to ±1% in the presence of large rolling moment loads.
In order to achieve the maximum accuracy from a balance, the strain field throughout the balance, generated by a load, must be the same in the wind tunnel as during calibration. This means either the attachment in the two cases must be identical or the strain path to the measuring elements must be extended to attenuate any irregularities in strain input. I think that if a balance were designed with its diameter and length increased about 50% over our present balances of the same capacity, its,accuracy might be improved to ±1/4%. This means a trade-off of 1/4% increase in accuracy at a cost of 1/3 reduction in Reynolds number. The performance of the two-plane solid balances being built for Lowell Keel will be watched to see if they offer better accuracy than Task balances.
Willard G. Smith
Research Scientist
RKH
WGSmith:ea 9-26-72/5459
cc: A. Giovannetti, 213-1