From the Rolls-Royce experimental archive: a quarter of a million communications from Rolls-Royce, 1906 to 1960's. Documents from the Sir Henry Royce Memorial Foundation (SHRMF).
Analysis of valve-spring surge and stress using an electric telemeter to inform spring design.
| Identifier | ExFiles\Box 56\2\ Scan090 | |
| Date | 1st May 1929 guessed | |
| [Page 10] 10 ELECTRIC TELEMETER AND VALVE-SPRING SURGE cam was carefully checked and found to impart almost a pure sine-wave motion to the valve. Inspection of this oscillogram shows no resonant points; however, some very feeble waves are present in the lift curve. It is hard to account for the presence of the waves, as their frequency of about 700 cycles per second is not high enough to be caused by the C-spring or the oscillograph elements. It is probable that their source is in some roughness or chatter-marks on the cam or in the valve train, or is in a slightly loose fit of the push-rod or valve-stem. In any event the amplitude of these vibrations is so low that they have no effect on the results. The prime feature of Fig. 20 is its freedom from resonance points. DYNAMIC STRESSES It is common practice to calculate the stresses present in valve-spring operation by the conventional formula for stress, which can be expressed S = 8PD/πd³ (2) where S is the fiber stress, in pounds per square inch; P is the load, in pounds; D the pitch diameter of the spring, in inches; and d is the wire diameter, in inches. Such a formula can be used to good advantage in computing the stresses present in the spring and in determining the stress range through which the spring operates, provided the motion of the spring is non-oscillatory with respect to its free frequency. In other words, the use of this formula will give results when, and only when, a load of P, is maintained continuously while the valve is closed and a load of P₁ is attained, exactly and always, when the valve is wide open. However, in dynamic operation, the oscillatory nature of the vibration of the spring causes the load to fluctuate greatly and with extreme rapidity. For this reason, the conventional stress-formula does not give an indication of the conditions under which the spring operates, except at very low engine-speeds. In a general way, the values of the actual stresses and stress-ranges vary with the magnitude of the surge, as indicated in the telemeter oscillograms. Since these actual stresses and stress-ranges are factors which are varying continuously whenever the motion of the spring is influenced by its own free frequency, it is impossible to assign definite concrete values to the stress conditions. However, to make clear the high orders of the stresses encountered and the extreme values for the stress ranges under surge conditions, it is well to present some figures which are believed to be of sufficient accuracy to give a picture of the true conditions. For this purpose attention is called to Fig. 8, which shows a spring to be vibrating badly at a speed of 1650 r.p.m., excited by a strong tenth harmonic. SURGE STRESSES EVALUATED If particular attention is paid to the lift curve nearest resonance at 1650 r.p.m., a peak swing can be noted very close to the maximum-lift position. The stress at the point would be approximately 60,000 lb. per sq. in. if there were no surge. However, the peak swing of the wave motion produces an additional stress at this point, so that the actual stress at the crest is probably about 146,000 lb. per sq. in. One half-wave later, a trough in the wave motion occurs. At this minimum point, the actual stress is probably 49,000, instead of about 60,000, lb. per sq. in. The true stress-range is therefore 97,000 lb. per sq. in., and it must be noted that this range has been traversed during a half-cycle of the free wave-motion of the spring, or at a rate of 33,000 cycles per minute. That the stress conditions vary from wave to wave can be seen upon further inspection of Fig. 8. It will be noted that the minimum point of the wave motion for the same lift-curve occurs just before the next lift. Here it is apparent that the crest of the wave motion causes a large addition to the valve-closed stress. The normal valve-closed stress is 21,500 lb. per sq. in., but the stress at the crest of the wave seems to be about 68,000 lb. per sq. in. A half-cycle later, at the trough of the wave, the stress would be zero; hence, the stress range for this half-cycle is 64,500 lb. per sq. in. At another point, intermediate between the two described above but on the base-circle part of the valve-lift curve where the calculated stress has a constant value of 21,500 lb. per sq. in., inspection discloses the stress at the crest of the wave to be 68,000 lb. per sq. in., while the stress at the trough of the wave is 28,500 lb. per sq. in. less than the normal valve-closed stress. This seems to cause a stress reversal of 7000 lb. per sq. in. and a stress range of 75,000 lb. per sq. in. for the half-cycle. Conditions of zero stress and reversed stress seem to indicate that the spring is in a free position in the first case and is acting as an extension spring in the second case, which is impossible under the conditions. What really is indicated is that, because of the wave motion of the spring, its coils successively assume positions corresponding to that of all the coils when the spring is free or when it is acting as an extension spring. SPRING FREQUENCY DETERMINES STRESS CYCLE From the foregoing it can be established definitely that, under highly oscillatory motion such as is encountered at resonant speeds, the spring operates under very severe stress-conditions. At high speeds the stress conditions are very bad, even without resonance. As a rule, it is only at low engine-speeds that the spring can be said to passing through the stress range calculated by the stress formula and that the rate of the stress cycle can be said to be the camshaft speed. At the higher engine-speeds, if the motion of the spring is oscillatory, the stress ranges always will be much higher than the calculated range and the stress-cycle rate will be double the spring frequency—many times the camshaft speed. In the design of valve-springs it has been found that high-frequency springs are preferable to low-frequency springs. The prime reason for this is that, for the same engine-speeds, the resonant points for the former are due to the higher harmonics. These generally have considerably smaller amplitudes than the lower harmonics, which cause the resonant points in the same camshaft-speed range for the low-frequency springs. The lower harmonic amplitudes cause the surge amplitude to be less for the high-frequency spring, and the freedom from surge seems to increase the fatigue life of the high-frequency spring. Invariably, increasing the frequency of a valve-spring necessitates making it stiffer, which in turn gives rise to an increase in the stress range as calculated by the usual formula. In reality the actual stress conditions in the spring that is relatively free from resonant points approach the calculated values more nearly, whereas the conventional stress calculations give absolutely no clue to the true stress conditions in the surging spring. A [Page 11] ELECTRIC TELEMETER AND VALVE-SPRING SURGE 11 spring breaks because of the stresses set up in it; no spring breaks when it has a lower actual stress than a similar spring which stands up in service. CONCLUSIONS Through the investigations thus far conducted by means of the electric telemeter, we are able to form several conclusions as to the cause of surge, the effect of surge on the behavior of the springs, and the influence of surge on the design of the springs. Where the cause of surge is concerned, the telemeter has proved to be a valuable instrument, as it furnishes a means for checking the theory that surge is a function of the amplitude of the harmonics of the complete valve-lift curve. This is important because it enables us to work toward the elimination of surge through cam design as well as spring design. The telemeter enables us to demonstrate that, where surge is encountered, the stress conditions existing in the spring are very severe. In a spring which has resonance points, the calculated stress conditions are only valid at low engine-speeds. At high speeds, the stresses are much greater than the calculated values, and the rate of the stress cycle is many times greater than at low speeds. The telemeter enables us to draw the following conclusions in regard to the design of valve-springs: (1) Low-frequency springs are to be avoided because, at engine speeds within the driving range, the springs tend to come into resonance when excited by the lower order of harmonics of the complete valve-lift curve; and the lower harmonics have very appreciable amplitudes, as a rule. (2) High-frequency springs are to be desired because, within the same driving range, the resonance points are due to the influence of the higher harmonics, which usually have much less amplitude than the lower harmonics. (3) Springs having a continuously variable pitch are valuable in minimizing the effects of surge. It is important that a few of the active coils should close up solid when the valve is open. The effect of the change in the number of active coils is to alter the frequency of the spring continuously throughout the lift of the valve. Thus, resonant conditions are prevented by interference in the wave motion, although the surge amplitude at high speeds may have certain points of minima and maxima. (4) High-frequency variable-pitch springs are preferable to low-frequency variable-pitch springs. (5) As a rule, if more than one to three coils of a variable-pitch spring are closed as the valve lifts, the total number of active coils must be rather large. This will cause the spring to have a low frequency when it is on the base-line portion of the valve-lift curve. The lower harmonics, which act throughout the whole lift-curve, will cause the spring to vibrate badly during this portion of its cycle, even though severe resonant points are eliminated. In conclusion, the authors wish to express their appreciation to J.{Mr Johnson W.M.} L. Whiteman* and J.{Mr Johnson W.M.} L. Sjolander* for their assistance throughout the investigation. * Cleveland Wire Spring Co., Cleveland. | ||