ACOUSTIC PROPERTIES OF INDUSTRIAL MACHINERY. design techniques to minimize gear drive radiated noise

Transcription

1 ACOUSTIC PROPERTIES OF INDUSTRIAL MACHINERY design techniques to minimize gear drive radiated noise

2 This white paper is to inform you generally on the topic of noise as related to Philadelphia Gear Corporation industrial products. Realizing that you lack the time, background or inclination to read texts on acoustics, this outline will be kept as brief as possible. There are five topics that will be generally covered, which should assist in your better understanding of your industrial product sound generations: 1. Basic introduction to noise, its measurement, and characteristics of instrumentation. 2. The affects of environment and system characteristics on generated noise. 3. Noise tolerance of personnel and equipment with present regulations and controls. 4. What sound levels can be expected during spin test, load test, and field operation. 5. Method of lowering generated sound levels, design absorptions, shielding, etc. Just about everything a person does involves motion. This motion produces sound. When machines are built to improve man s abilities, the situation becomes worse, because machines are usually noisier than people. In fact, the situation had become so bad that on May 17, 1969, the occupational noise-exposure regulation of the U.S. Department of Labor went into effect (Walsh-Healey Act) to protect employees from hearing damage. Then the Occupational Safety and Health Act of 1970 (84-Stat. 1950) took the same noise regulations and extended them to cover over 55 million workers in any business engaged in interstate commerce.

3 SECTION I. NOISE WHAT IS IT? The word noise implies a loud, harsh, undesired, confused, or nonharmonious sound. Because we know most of our products generate harmonious sounds, we will refrain from using that word and talk in terms of generated sound levels. Sound is a disturbance that propagates through an elastic material at a speed characteristic of the medium. But, this is a very cumbersome definition, so we will talk of sound as incremental changes in air pressure (vibration of air molecules). The audible frequency range of sound is often taken as between 20Hz (Hz = Hertz, as in cycle per second) and 20,000 Hz (20KHz). However, the human ear has trouble hearing much below 200Hz or above 10KHz. Although our hearing frequency range may be limited, we can distinguish very low levels of pressure fluctuation to very high levels. The loudest sound pressure possible is ±1.0 atmosphere (±14.7 psi or ±1 bar), which may occur at the center of an explosion. The threshold of hearing, at about 1,000Hz, is normally around ± bars (.002 µ bars =.0002 µ atmospheres = 20 µ Newton/meters sq. =.0002 dynes/cm sq. 3 x 10-9 psi). This gives the ear a dynamic sensitivity of about 10-9 power. Because of this high sensitivity, a decibel (db) scale has to be used, and a sound pressure level (Lp) is defined as: Lp = 20 log 10 ( P ), db (re 20 µn/m 2 ) P P is the reference pressure taken as the threshold of hearing 20 µn/m 2 (International Standard). Example: The sound pressure near a punch press is measured as being.0025 psi. What is the sound pressure re 20 µn/m 2 in db (.0025 psi = N/m 2 )? Lp = 20 log 10 ( 17.22N/m2 ), db (re 20 µn/m 2 ) 20 µn/m 2 Lp = 20 log 10 (8.612 x 10 5 ), db (re 20 µn/m 2 ) Lp = db (re 20 µn/m 2 ) So we would commonly say the noise of the punch press is 119 db. If one does not like to use such formulas, then Figure No. 1 gives a table of common sound levels and some simple ratios in terms of decibels (db). Like our ears, we can use a pressure sensitive instrument called a sound level meter to measure the intensity of sound. Sound level meters commonly have a broadband frequency response (2Hz to 40KHz) and high dynamic range (10 db to 140 db). However, microphones used with these meters are often limited in range and sensitivity (a standard 1 microphone normally has a frequency range flat from 20Hz to 12 KHz and sensitivity of 40 db to 14dB). Also, like the human ear, a sound level meter can be fooled and indicate the wrong level and frequency. The sound level meter with the wrong microphone or in the Figure 1 - Typical A-Weighted Sound Levels presence of wind, electro-magnetic fields, etc., may also indicate false readings of sound pressure level. Therefore, it is very important that the person operating a sound level meter be qualified, knowing the characteristics of his instrument system and the electro-acoustical characteristics of the area in which he is measuring. It was mentioned that the frequency response of the human ear is not as good as a sound level meter. Therefore, various weighting networks (filters) have been established so that the objective meter measurement will come closer to indicating what the ear hears. Figure No. 2 (on page 3) shows the

4 Figure 2 - A, B and C Electrical Weighting Networks for the Sound-Level Meter. These numbers assume a flat, diffuse-field response for the sound-level meter and microphone attenuation of the A, B and C weighting scales of a sound level meter. The A scale is a filtering system that roughly matches the human ear s response at sound levels below 55 db. The B scale roughly matches the ear at levels between 55 db and 85 db, and the C scale is to match above 85 db. However, the A scale (sound pressure level measured in db A ) has received prominence due to its use by OSHA, for measuring levels up to 115 db. It is interesting to note the tremendous attenuation the A scale exhibits on low frequencies. At about 95 HZ, for example, there is about a 20 db attenuation (only 1/10 of the actual sound level is indicated on the meter). Therefore, machinery generating low frequency sounds are more likely to pass a db A specification, and be less annoying to the ear. Another filtering system often used in the measurement of sound is the octave and 1/3-octave bands. These are discrete filters which only register a limited range of frequencies. The octave and 1/3-octave bands are used more for analytical work and are usually specified by their center frequencies. (See Figure No. 3) The 63 Hz octave band to the 8,000 Hz octave band are most commonly used in industry specifications. A narrow band filter (spectrum analyzer) is similar to octave band filters; however, each band is greatly reduced in width to allow better resolution of component frequencies in a noise spectrum. It may be becoming increasingly apparent that the field of acoustics is a complicated science; however, it can be kept in simple understandable terms if the preceding basics are fully comprehended. In the appendix is a glossary of common terminology used in acoustics. A more complete list is available from the American National Standards Institute (ANSI) as Standard S American Mentioned Standard Acoustical Terminology. 2

5 FREQUENCY OCTAVE ONE-THIRD OCTAVE LOWER UPPER LOWER UPPER BAND BAND BAND BAND BAND LIMIT CENTER LIMIT LIMIT CENTER LIMIT ,000 1, ,000 1, ,122 1,250 1, ,413 1,600 1, ,420 2,000 2,840 1,778 2,000 2, ,239 2,500 2, ,818 3,150 3, ,840 4,000 5,680 3,548 4,000 4, ,467 5,000 5, ,623 6,300 7, ,680 8,000 11,360 7,079 8,000 8, ,913 10,000 11, ,220 12,500 14, ,360 16,000 22,720 14,130 16,000 17, ,780 20,000 22,390 Figure 3 - Center and Approximate Cutoff Frequencies for Standard Set of Contiguous-octave and One-third-octave Bands Covering the Audio Frequency Range TO ADD LEVELS: Enter the chart with the NUMERICAL DIFFERENCE BETWEEN TWO LEVELS being added. Follow the line corresponding to this value to its intersection with the curved line, then left to read the NUMERICAL DIFFERENCE BETWEEN TOTAL AND LARGER LEVEL. Add this value to the larger level to determine the total. Example: Combine 75 db and 80 db. The difference is 5 db. The 5-dB line intersects the curved line at 1.2 db on the vertical scale. Thus the total value is or 81.2 db. TO SUBTRACT LEVELS: Enter the chart with the NUMERICAL DIFFERENCE BETWEEN TOTAL AND LARGER LEVELS if this value is less than 3 db. Enter the chart with the NUMERICAL DIFFERENCE BETWEEN TOTAL AND SMALLER LEVELS if this value is between 3 and 14 db. Follow the line corresponding to this value to its intersection with the curved line, then either left or down to read the NUMERICAL DIFFERENCE BETWEEN TOTAL AND LARGER (SMALLER) LEV- ELS. Subtract this value from the total level to determine the unknown level. Example; Subtract 81 db from 90 db. The difference is 9 db. The 9-dB vertical line intersects the curved line at 0.6 db on the vertical scale. Thus the unknown level is or 89.4 db. Figure 4 - Chart For Combining Levels of Uncorrelated Noise Signals There is one other characteristic of sound which should be covered in this section; that is, how to add the sound level generated from two separate sources. Any school student will tell you that ( = 89) is an invalid equation. However, if we state that in the same environment 82 db + 88 db =89 db, we would be correct. Figure No. 4 shows a chart which can be used to assist in adding and subtracting sound pressure levels stated in db units of measure. Consider a theoretical example: There are four gear motors generating equal amounts of sound energy (power). Together they produce a level of 94 db A. To cut the noise by 3 db to 91 db A two of the gear motors would have to be turned off. To be below the OSHA limit of 90 db A, a third gear motor would have to be shut down, resulting in a level for one gear motor of about 88 db A. In a real situation, one might not be 3

6 able shut down equipment on a production line, so improved quality and enclosures might have to be used. The important point of this example is understanding the magnitude of the sound energy reduction required to achieve a 3 db reduction in sound pressure level. THE EFFECTS OF ENVIRONMENT ON SOUND Anyone who has held a conversation in a completely empty room and compared it with a similar conversation in a room filled with heavy cushioned furniture, carpet and draperies, has a subjective feeling on the effect of sound in various environments. ANSI Standard S (R-2001) gives four definitions for environments in determining the sound power (energy) of sound sources. These four environments are described as a free field, free field above a reflecting plane, a diffused field, and a semi-reverberant field. However, to describe what is meant by these fields, it is necessary to introduce the concept of sound power level. The sound power of a source is the total sound energy radiated by the source per unit of time. The level of sound power (Lw), like sound pressure, is expressed in terms of decibels. Lw = 10 log 10 W db re watts W 0 W 0 is the reference power level of 1 x watts. (Some older specifications still may use watts.) It is extremely important to recognize the difference between sound power level (Lw) and sound pressure level (Lp). Sound power is the energy at the source. The sound pressure is the fluctuation of air pressure at the point of measurement. The sound power of a source is independent of the environment. Whereas, the sound pressure measured near a source is highly dependent on environment. The paradox is that we must measure sound pressure to calculate sound power (sound power cannot be measured directly). A free field environment around a sound source can be thought of as being like a fan running on top of a tall flag pole in the middle of 1,000 flat acres of open land. In a free field, all the sound is radiated away from the source and the pressure waves lose their strength as they dissipate into the surrounding air (6 db drop for double the distance). The further away from the source, the lower the measured sound pressure level. The opposite of a free field is a reverberation chamber (diffused field) with a source placed in the center of the empty room with hard skewed walls, floor, and ceiling. Almost all of the sound from the source will be reflected around the room and back at the source so that the sound pressure level measurements at any point in the room are about the same. It is quite obvious that practical industrial equipment environments are neither free field or diffused field, but rather some form of semi-reverberant field. In a semi-reverberant field depending on room size, reflecting objects, absorption and other acoustical conditions, the reduction in db from 5 feet away to 10 feet away from the source could be less than 1 db or as great as 6 db. The point of this discussion, as mentioned, is that a db sound pressure measurement is highly influenced by its acoustical environment. Another extremely important consideration when evaluating generated sound pressure level of machinery is that sound can be structure borne for considerable distances without significant attenuation. Structural steel beams may provide a path for structure borne sound (vibrations) to travel significant distances and then radiate airborne sound pressure levels at nearly the same level as at the source. How often is a screw driver used to transmit structure borne noise from the gearcase to the ear? Furthermore, structure borne sounds may excite natural resonances of other equipment and structures, and thus create a sound pressure level louder than the source under investigation. If the sound levels of a gas turbine driven-gear-compressor system are being measured to determine the gear noise, one could ask the following questions: 1. What is the major noise source; turbine, gear unit, compressor, piping or structure? 2, How much of the noise is traveling through the support structures and radiating at some point other than its source? 3. Is the gear unit mesh frequency exciting a natural resonance in the sheet metal cover of the turbine, or the piping, etc.? 4. Is a blade pass frequency exciting a natural resonance of the bull gear web or the gear housing? 5. What are the sound levels at different loads or speeds? This list could be continued at great length; however, one can see that there are many different influences when trying to determine the sound level of a gear unit in the middle of a power transmission system. The exact same gear unit may generate completely different sound levels in two different systems. Therefore, if an accurate sound power level can be established for a source, it is the only method for rating the source independent of environment. If a db A, octave band, narrow band, etc. sound pressure level is used as a criterion, it must be remembered that the environment and mounting will influence the results. 4

7 TOLERANCES AND SPECIFICATION OF SOUND LEVELS There are many different types of specifications for limiting sound levels. However, it may be best to classify two general areas of concern: 1. The level that is normal or satisfactory for the type of equipment or application. 2. The level that is acceptable or safe for people who must hear without damage in the area of the operating equipment. Both of these areas of concern are difficult for accurate evaluation because of the lack of information on sound power levels of all sources, characteristics of all environments, and because human sensitivity to noise varies. Therefore, general or normal guidelines have been established as acceptance criteria. Satisfactory equipment sound level specifications are normally stated in terms of acceptable octave band levels for eight principal bands. AGMA Standard 6025-D98 Sound for Enclosed Helical, Herringbone and Spiral Bevel Gear Drives is the current industry standard. This standard describes the instrumentation, measuring methods and test procedures necessary for the determination of a gear unit s sound pressure levels for acceptance testing. It must be recognized, therefore, that a specification level has to be pointed toward a specific environment whether it be a test stand or a field measurement. We normally want to use the test stand, at the point of manufacture, as the acceptable environment. It is reasoned that the semi-reverberant field of a test stand is similar to many installation environments and therefore, within limits, the sound pressure levels will be similar. Rules and Regulations Occupational noise exposure. O.S.H.A. AND WALSH-HEALEY PUBLIC CONTRACTS ACT (a) Protection against the effects of noise exposure shall be provided when the sound levels exceed those shown in Table 1 of this section when measured on the A scale of a standard sound level meter at slow response. When noise levels are determined by octave band analysis, the equivalent A - weighted sound level may be determined as follows: (see graph) (b) When employees are subjected to sound exceeding those listed in Table 1 of this sections, feasible administrative or engineering controls shall be utilized. If such controls fail to reduce sound levels within the levels of the table, person protective equipment shall be provided and used to reduce sound levels within the levels of the table (c) If the variations in noise level involve maxima at intervals of 1 second or less, it is to be considered continuous. (d) In all cases where the sound levels exceed the values shown herein, a continuing, effective hearing conservation program shall be administered. Exposure to impulsive or impact noise should not exceed 140 db peak sound pressure level Equivalent sound level contours. Octave band sound pressure levels may be converted to the equivalent A-weighted sound level by plotting them on this graph and noting the A-weighted sound level corresponding to the point of highest penetration into the sound level contours. This equivalent A-weighted sound level, which may differ from the actual A-weighted sound level of the noise, is used to determine exposure limits from Table 1. 1 When the daily noise exposure is composed of two or more periods of noise exposure of different levels, their combined effect should be considered, rather than the individual effect of each. If the sum of the following fractions: C1/T1 + C2/T2 Cn/Tn exceeds unity, then the mixed exposure should be considered to exceed the limit value. Cn indicates the total time of exposure at a specified noise level, and Tn indicates the total time of exposure permitted at that level. Figure 5 - O.S.H.A and Walsh-Healey Public Contracts Act 5

8 Trying to comply with specifications that are pointed toward hearing conservation (OSHA, state and local ordinances) presents a different situation. To meet OSHA requirements, (Figure 5) one must know the position of the people exposed to the generated sound and the duration of the exposures. Presently, if an operator of a rolling mill has to stand near a gear unit generating 98 db A at 5 feet for only two hours out of every day, the area meets the environmental requirements of OSHA. However, the same installation would not meet OSHA requirements if the operator had to stay in the area of 98 db A for the three hours or more each day. It must be emphasized that a manufacturer cannot supply equipment to meet an OSHA db A rating if the position of the operator or exposure time is unknown. It therefore becomes more realistic for the user to specify permissible sound levels in terms of octave band measurements at a specified distance, or overall db A level at a specific location and environment. Another highly important consideration is that most environments contain more than one noise source. Therefore, we must sum their amplitude, as stated earlier. This means when supplying individual components in a system, each must be significantly below 90 db A if the whole system is to be below 90 db A at a specific position. Noise not only may impair hearing it has been known to increase blood pressure, heart rate, gastrointestinal tract responses and cause bad psychological effects. The loudness of a sound depends on its frequency, and there are different scales used to determine relative loudness. Noise does not have to be loud to be annoying. The sound of a dripping faucet, chalk squeaking on a blackboard, or a high pitch gear mesh frequency are examples. Impact noises about 45 db A will generally awaken people asleep at night, no matter how innocent it may look on a sound level meter. Generally broadband steady-state noises (hum) are less annoying than an intermittent noise (beats), pure tone (whine) or impulsive noise (impact). Harmful steady-state noise may be overlooked, whereas quiet beats may be subject to scrutiny. Therefore, even if a gear unit meets a specified sound level it can be annoying and become a subject of complaint. THE EXPECTED SOUND LEVEL GEARS MAY GENERATE Philadelphia Gear has had years of experience measuring sound, both on the test stand, and in field installations. This experience has given us ability to know the approximate levels that may be expected on qualification spin or load tests, and in field service. For a particular application, knowing the gear unit size, type, operating speeds, load, and system, the expected sound level may be given within a range of a few db. This sound level can be obtained from test results of identical or comparable units and/or empirical data extrapolated from similar equipment. The levels will not include driving or driven equipment noise and system influences. Due to variances from system to system and environmental differences these values generally cannot be guaranteed. However, the values will be found to be accurate within the limits imposed by differences in environments, measurement techniques, and interpretation. If the frequency components of the overall sound generated by gear units are reviewed in general there will be many similarities. The most common frequencies will be the rotational speeds, their multiples, tooth mesh frequency, windage, bearings, and natural resonances. Figure 6 reveals some of these common sources of airborne and structure borne sounds generated in gears. It is interesting to note that the majority of bearing noises, natural resonances and critical speeds of moderate and high speed industrial gear units are in the 500 to 8,000 Hz octave bands where the ear is most sensitive. This in itself is not a problem. However, when noise problems do exist, and levels must be lowered, the task becomes a little more difficult. METHODS OF LOWERING GENERATED SOUND LEVEL Upon reviewing the expected sound level, or hearing a noisy gear unit, the first question may be how could the generated levels be improved or lowered? There are three general approaches to lowering generated sound levels: Enclose or isolate the source so that the operators are shielded from the noise Absorb sound that may be normally reflected Eliminate or lower the sound energy produced at the source by improved quality or design Generally absorbing or isolation of sound from sources is the most economical means of lowering sound levels (see Figure 7 & 7A). Because some generated sounds are machine characteristics, no amount of redesign or improved quality will eliminate these sounds. However, from an operation standpoint, when possible the most desirable method of lowering sound levels is by design or quality improvements. Improved sound or vibration levels achieved from smoother operation will also tend toward lower dynamic loads and improve the durability life of the gears, bearings and couplings. (continued, pg. 8) 6

9 Instruments that provide the operator with not only the amplitude of the vibration or noise, but, also the predominant frequencies can be a tremendous aid in determining problem cause. These causes normally present themselves as follows: 1. Imbalance: Presents itself at a frequency equal to once per shaft revolution and it will increase in amplitude as speed is increased. 2. Misalignments: Will present themselves also at once or sometimes twice and three times per shaft revolution. However, the amplitude will remain fairly constant with speed changes 3. Friction: This is difficult to pinpoint by vibration and noise frequency of occurrence, which may be very high when continuous sliding occurs. It may also be random, high-amplitude, shock-type pulses, as in hydrodynamic bearing rubbing. It may be irregular and often violent. 4. Looseness: This may cause unbalance, misalignment and friction rubbing at moderate and high speeds. At low speeds, it may display itself as an irregular rattle. Often it shows up at twice shaft rotational speed. 5. Gear Casing Distortion: This is often an indirect cause of vibration and noise, which also leads to unbalance, misalignment, or friction. It will tend to change in amplitude with load or operating temperatures, when speed is held constant. 6. Critical Speeds: These occur through any given speed range and are points at which a rotating system vibrates torsionally or laterally at a particular frequency. Rotors characteristically show violent increase in amplitude at particular critical speeds but are fairly stable above and below these speeds. A critical speed may change frequency with load and temperature. 7. Resonances: these also display themselves as frequencies at which system members vibrate. The distinction from critical speeds is that resonances occur in other than rotating members, and can affect alignment. Resonances occur at fixed frequencies and change in amplitude with load and temperature. 8. Tooth Profile Wear: This will show up at tooth mesh frequency (i.e. rotating speed times number of teeth) and multiples of this mesh frequency. 9. Bearing Instability: Improperly designed rollling element bearings will cause high-frequency vibration at several times rotational speed, also friction vibration will occur. Lightly loaded hydrodynamic bearings will tend to whirl at.43 to.47 times the rotational speed. This so-called half-frequency whirl will on-set rotor instability with speed or temperature changes and may continue until the rotor is completely stopped. 10. System Pulses: These may occur in many types of systems, such as the vane-pass frequency of a pump or compressor (rotational speed times the number of vanes), and the beating of reciprocating engines which cause frequencies at one-half and one-quarter rotational speed at various amplitudes. 11. Windage: From couplings and other rotating parts (generally broadband noise), but can be at a bolt pass frequency or fan blade pass frequency. Note: All of these types of vibration and noise frequencies can be generated in a gear drive. Major frequencies can interact and cause frequency modulation and phase shifts. Any combination, sum, difference and multiple (harmonics) of the prime frequencies can occur if the forcing magnitude and system degrees of freedom are such that they will cause and allow the generated vibration to become predominant. Generally, only the prime frequencies will present themselves as problem modes. However, sometimes very elusive frequencies appear, such as a periodic cutting machine error appearing on one of the gears. Figure 6 - Common sources of airborne and structure borne sounds generated in gears. 7

10 When sound levels are being discussed, we must be mindful of the following considerations: 1. What specification(s) are applicable and what required levels must be achieved? 3. What improvement and costs are involved when a design change or improved quality is specified? 4. What costs are involved if enclosures, isolations, or absorbing materials are used? 2. Does the standard or normal unit of the type being considered meet the requirements? Figure 7 - Examples to illustrate the possible noise reduction effects of some noise control measures. Reprint Handbook of Noise Measurement 7th edition 1972 General Radio Corp. 8

11 Figure 7 A - Additional examples to illustrate the possible noise reduction effects of some noise control measures. 9

12 Figure 8 The 20 pressure angle is the best compromise between strength and quietness. Other tooth forms may be used where one factor or the other is more important. The 14-1/2 tooth gives less noise; the 25, greater strength. Figure 9 - (See illustrations to the left.) Comparison of involute gears, a, and full-recess gears, b. In full-recess gear sets, the teeth of one of the gears have all-addendm profiles, while those of the mating gear are all dedendum. TYPE OF GEARING For quiet operation, parallel-shaft gears, such as spur, helical, and double helical are preferred to right angle and crossed-axis gearing. The primary reason is that spur and helical gears make it possible to maintain tight tolerances and to operate with minimum friction. Helical gear types have the added advantage of maintaining more than one tooth in contact during operation (helical overlap). Because of this, it is possible to get as much as 12 dba reduction in noise by using them instead of spur gears. SECTION II. DESIGNING NOISE OUT OF GEARS The best time to solve a gear-noise problem is at the design stage. Retro-fixing is usually expensive and unsatisfactory. Many factors play a role in ensuring a quiet gear mesh. Adjusting one or more of them before the hardware is manufactured adds little to the cost, but may make a significant difference in noise level. However, to complicate matters, there is always the need for a balance between low noise and satisfactory performance. The most important design factors that influence the noise level of a set of gears are: Type of gearing Tooth profile Pitch Pressure angle Recess action Profile modification Overlap ratio Backlash Tooth loading Quality level Surface finish Gear runout and unbalance Gear ratio Resonance Lubricant viscosity Type of bearings Material Housing With double helical gears, there is the problem of correctly manufacturing the two helices with exactly the same phase and accuracy; in other words, without apex runout for weave. Any slight deviations in manufacturing, coupled with the axial mass inertia of the gear, will prevent equal load sharing and smooth operation; this will contribute to vibration and noise. Thus, for smooth, noise-free operation, the optimum type of gear throughout all speed ranges is the single helical gear. Any thrust or overturning loads that single helical gears may induce are easily offset by modern design techniques. TOOTH PROFILE Tooth profiles other than the common involute type have been developed over the years. None of them has been widely accepted, mainly because they lack sufficient advantages and manufacturing history to justify deviating from the involute form. Gears with circular-arc tooth profiles, for examples, can retain more lubricant between mating teeth, thus tending to reduce noise and wear. However, the circular-arc form is still largely experimental, and has not proved to be inherently quieter then involute gears. Additionally, manufacturers are generally tooled for involute types, which is one of the main reasons why other tooth forms have not been adopted. 10

13 Figure 10 Tooth modification is important if a mesh is to be quiet. Shown here are tip and root relief, a, and barreled tooth flank, b. In each case, the solid lines represent the modified form Figure 11 Action of teeth with contact ratio of 1.5. In a, two teeth are in contact, at A and B. However, as the mesh turns, b, the leading pair of teeth disengages at C and only one set is in contact until the next set engages. Such action can set up oscillations that can add to noise. Other forms have been designed specifically to carry much higher loads than equivalent involute types. These may be just as noisy, if not noisier, than conventional involute types. Extensive experience with involute gears has led to profile modification for quieter operation. This experience is not available for other tooth forms. PITCH For quiet, smooth operation, the finest possible pitch should be selected for the given load conditions. The finer the pitch, the higher the number of teeth in contact. This increases the amount of tooth overlap, both transverse and helical. The higher tooth overlap produces a smoother transfer of load, reducing dynamic oscillation of the gear mesh. This will result in a quieter gear train even tough the finer pitch will produce a higher mesh frequency. Of course, the finer the pitch, the smaller the tooth size and the lower the strength rating of the gear set. To compensate for this, there are several choices: enlarge the pitch diameters, make the gears wider, or use the higher hardness material and precision-grind. One drawback to the use of finer-pitch ground-tooth gears is that more time is needed to manufacture the teeth, which adds to the cost of the gearing. PRESSURE ANGLE Where a low noise level is a key requirement, the lowest possible pressure angle should be chosen. With the ever-increasing need for higher-capacity gears, many designers have tended to the 25 pressure angle in place of the more common 14 1/4 and 20 pressure angles, Fig. 8. The 25 pressure angle gears do indeed provide greater load capacity; but such gears tend to be noisier, because the contact angle and transverse overlap ratio are lower. In general, a 20 -pressure angle is a good compromise between quiet operation and high load capacity. RECESS ACTION In cases where a gear drive will be transmitting load in only one direction of rotation, recess-action gears can provide a further reduction in noise. In gears of the full-recess type, the teeth of one of the gears in mesh have all-addendum profiles, while the mating gear has all-dedendum teeth, Figure 9. In these gears, contact occurs during the recess portion of the tooth s engagement and disengagement cycle. Contact during the approach action, and some of the resulting detrimental scraping effects, are avoided. The reason for the undesirable effects of approach-action, contact can be understood by looking at a standard involute gear pair, as shown in Fig. 9a. Initial contact occurs at A when the lower gear is rotating counterclockwise. The approach action takes place from A to B, and recess action from B to C, when contact ceases. During contact, the action is a combination of sliding and rolling. Approach-action friction tends to increase scuffing, wear, and noise. There is a change in the direction of sliding at the rolling point of the pitch line that tends to break down the oil film, causing increased friction. The recess phase of the meshing, on the other hand, is a sliding-out action. Friction is lower and in a direction to help rotating, Fig. 9b. Thus, the effect of recess action on the teeth is less severe than the approach action. It is possible to design a semi-recess action system that combines the benefits of full recess action with those of standard action. Such gears are more common than the full-recess types. 11

14 TOOTH MODIFICATION Often overlooked, proper tooth modification is very important if the gear is to be quiet and long-lived. Dynamic studies have shown that most high-powered gears, as well as many lightly loaded ones, require modification of the tooth profile. This ensures smooth sliding of the teeth into and out of contact without knocking, and compensates for misalignment, errors in manufacture, and deflections under load. Tip and root relief are important. The tips of the pinion teeth are thinned slightly, beginning at a point halfway up the addendum, Fig. 10A. This modification prevents the tooth tip from striking one of the opposite teeth before it begins to pick up the load, or striking the trailing flank of the preceding teeth when rolling out of mesh. The root of the pinion is also thinned to allow for the tips of the gear teeth. In some cases, even the hydrodynamic effect of the lubricating oil can cause the teeth to emit a rapping sound during operation, even though there is no physical contact at or near the tips. Figure 12 Effect of tooth loading factor, K, on dynamic load. As load decreases, so does noise. Crowning of the tooth flank is another extremely important tooth modification. Load deflections and slight errors in alignment of the gear housing in the bearing bores can cause hard contact on the flank ends of the tooth. It is, therefore, desirable to crown the gear (form a barrel shape, fig. 10b) to center the contact area. As the load increases, a smooth spreading of the contact occurs until the entire flank is loaded. OVERlAP RATIO Designing can reduce gear noise at the mesh so that the total overlap ratio is an integral number of teeth. Tests have shown that if the ratio is exactly 2.0 so that exactly two teeth are always in contact the smoothest transfer of load is obtained. Overlap ratio expresses the average number of teeth in contact; in spur gears, it is obtained by dividing the length of the line of action by the normal pitch. With helical gears, the total overlap is the combination of the transverse overlap and the helical overlap. As an example, assume that a gear pair has an overlap ratio of 1.5. At first, there will be two pairs of teeth in contact, fig. 11a, until a point is reached, fig. 11b, where the leading pair disengages at C. After disengagement, the trailing pair will be the only pair in contact until another pair engages. Thus, at some point, two teeth are in contact sharing the load, and at another point, one tooth will assume the entire load. This sets up an oscillation about the deflection of the tooth that, although small, adds to sound generation. With a higher overlap ratio, more teeth are in contact to share the load, providing increased capacity. An overlap ratio of 3.0 is often best for maximum load capacity. However, a value of 2.0 is generally best for reducing noise and dynamic loads. Figure 13 Effect of quality level on total dynamic noise. Although more costly, higher quality levels do offer a significant reduction in noise. BACKLASH Adequate tooth clearance (backlash) must be provided for thermal and centrifugal expansion. A tight backlash is usually required only with reversing drives, where lost motion and impacts during reversing may be a problem. The operating temperature of a steel gear or gear rim might be 100 F or more above ambient conditions; also, at high speeds, centrifugal forces can become considerable. Expansion results, and the too-tight gear set begins to howl. It is relatively simple to calculate fairly accurately the total anticipated expansion, and the resulting amount of backlash needed. 12

15 TOOTH LOADING As shown in Figure 12 the higher the tooth loading factor, K, the lower the total dynamic load and, consequently, the lower the amplitude of vibration and sound. Factor K is proportional to the tangential driving force, W, on the gears: K = W, ( m G + 1 ) Fd m G where f = face width, d = pinion pitch diameter, and m G = speed ratio Quieter operation results from higher loadings because there are always some inaccuracies in gears, regardless of the quality level to which they are manufactured. With higher loadings, tooth deflections are greater and errors have less effect on uniform transmission of power through the gear set. QUALITY LEVEL One of the easiest ways to reduce noise is to specify higher AGMA quality levels. As shown in Fig. 13, at higher quality numbers, the total dynamic load is reduced. Generally, AGMA Quality 12 or better is needed for smooth operation. Of course, higher quality levels cost more. For example, going from Quality Number 8 to 12 means an increase in cost of 25% to 50%, mostly because of the need for special, high-accuracy finishing and grinding equipment and techniques. SURFACE FINISH Surface finish should be as fine as possible. Several methods - shaving, for example can be used to obtain a good finish. However, where a superior finish is needed, grinding is usually chosen. Machining can cause periodic undulations on the gear teeth; these should be kept to less than 50 to 100 µ in. Such undulations usually result from cyclic run outs and inaccuracies in the master wheel of hobbing machines used to machine the gears. They can cause such odd phenomena as, say, a mesh noise frequency indicating 132 teeth when the bull gear actually has 215 teeth. Upon investigation, the 132 frequency will turn out to be the number of teeth in the master wheel on the hobbing machine. GEAR RUNOUT AND IMBALANCE There is little sense in having, say, Quality Number 14 gears, with tooth tolerances held to within a few ten thousandths of an inch, and then specifying two or three thousandths tolerances on the alignment and parallelism of the bores. All alignments and runouts should be in keeping with the quality level used. For optimum quiet operation, the unbalance in the rotors should be limited to less than U = 3W N where U = rotor unbalance, oz-in.; W = weight of rotor, lb; N = speed, rpm GEAR RATIO A non-integral gear ratio should be selected to prevent a tooth on the pinion from contacting periodically the same teeth on the mating gear. In most applications, an exact gear ratio is not necessary; by juggling the ratios slightly, it is possible to obtain an odd tooth, called the hunting tooth. As an example, a gear pair with 30 teeth on the pinion and 120 teeth on the gear has a speed reduction ratio of 4:1. A particular tooth on the pinion will contact teeth 1, 31, 61, 91, and continue to repeat the cycle, contacting the same four teeth. If a 30/121 pair is used instead, the reduction ratio becomes 4.03: 1, and the odd tooth on the 121-tooth gear will not contact the same tooth on the pinion every four cycles. For example, a tooth on the pinion will contact teeth 1, 31, 61, 91, 121, 30 60, 90, etc. and will not repeat the cycle until all teeth on the gear have been contacted. RESONANCE Any resonating member is a source of vibration and sound. There are two types of resonances to be concerned about. One is caused by the rotating parts and is associated with critical speeds; the second involves the support cases, foundation, and structures. The critical speeds (resonances) of the rotating parts should be at least 20% from the operating speeds, from their multiples (harmonics), and from the mesh frequencies of the gear teeth. The farther the operating speed is from the critical speeds, the less chance there will be of detrimental effects. Resonances of gear cases and other supporting members should also be 20% from operating speeds, multiples, and tooth-mesh frequencies. This may be difficult to obtain at times, so that a more practical minimum may be 10 to 15%, depending on how accurately the resonant frequencies can be calculated. LUBRICANT VISCOSITY The higher the viscosity of the gear lubricant, the greater the damping action at the mesh, and the quieter the operation. However, higher viscosities will result in some loss in operating efficiency and power. Most designers of high-speed gears choose a light grade of turbine oil (ISO 32 or ISO 46) because it is the same oil used in the turbine or compressor. But from the standpoint of the gear box itself, oils of ISO 68 or ISO 100 are better suited, provided that the power loss is not a problem. Besides quieting the gears, the oils with a higher ISO will cut down gear tooth scoring and wear. 13

16 TYPE OF BEARINGS Both hydrodynamic (sleeve type) bearings and rolling element (anti-friction type) bearings can be used to support gear shafting. Hydrodynamic bearings, which operate on an oil film, provide for quieter operation. Hydrodynamic bearings typically have higher friction losses and require more lubrication oil than rollling element bearings. For standard applications, rolling element bearings are usually less expensive than hydrodynamic bearings. The particular application often dictates the choices of bearing types. MATERIAL The type of material chosen for the gear can have a significant effect on noise. Certain metals, such as the manganese-copper alloys, can dampen vibrations considerably. However, such alloys lose their damping characteristics at about 150 F. Unfortunately, many enclosed gears operate above this temperature in industrial applications. Non-metallics, which also have good damping characteristics, have relatively low loadcarrying capabilities, and are more difficult to manufacture accurately. In general, alloy steels are still the best choice. A surface hardened precision tooth ground gear-set today is the preferred tooth treatment method to maximize gear-set ratings. Superior ground tooth accuracy and fine tooth surface finishes made possible by current state-of-the-art tooth grinding machinery also provide quieter operating characteristics. HOUSING Proper housing design can go a long way toward blocking the transmission of sound to the environment. Generally, a cast-iron housing will be deader or quieter than a weldedsteel housing. However, welded steel has many advantages. It is stronger and more easily designed to meet custom requirements. To improve the damping characteristics of a welded steel housing, thin, flat plates should be avoided, because they can be excited by rotating components and mesh frequencies. Additional mass can be added, as well as stiffeners and reinforcing ribs, to break up place-section resonances. If further noise damping is required, a layer of synthetic material, such as felt, synthetic rubber, or polymer can be sandwiched between two steel plates to form a constrained layer of damping material. To cut down the transmission of noise through the floor, the housing should be mounted on a vibration-isolating material placed directly under the gear box. A relatively stiff laminated rubber or other type of material should be used to absorb the higher frequencies. ACKNOWLEDGMENT: Philadelphia Gear Corporation would like to acknowledge William A. Bradley, current V.P., Technical Division of AGMA, whose work on this subject while employed as PGC s Chief Test Engineer provided important source materials for this paper. 14

17 APPENDIX COMMON ACOUSTIC TERMINOLOGY These definitions include most of the common terms used in acoustics. Many of the definitions are selected from A.N.S.I. standard S (R1971). Others have been shortened and expressed technically to be as accurate as possible from references 1, 2, 3, 4, 5, 6, 7. A-SCALE: A filtering system that has characteristics that roughly match the response characteristics of the human ear at low sound levels (below 55dB SPL, but frequently used above this level). ABSORPTION COEFFICIENT: The absorption co-efficient of a material or sound-absorbing device is the ratio of the sound absorbed to the sound incident on the material or device. The sound absorbed by a material or device is usually taken a the sound energy incident on the surface minus the sound energy reflected ACOUSTICAL MATERIAL: Any material considered in terms of its acoustical properties. Commonly, a material designed to absorb sound. AIRBORNE NOISE: A condition when sound waves are being carried by the atmosphere. AMBIENT NOISE: Ambient noise is the characteristic (or background noise) of a given environment. AMPLITUDE DENSITY DISTRIBUTION (PROBABILITY DENSITY DISTRIBUTION/FREQUENCY DISTRIBUTION): A function giving the fraction of time that the pressure, voltage, or other variable dwells in a narrow range. AMPLITUDE DISTRIBUTION FUNCTION (DISTRIBUTION FUNCTION)/PROBABILITY FUNCTION/CUMULATIVE FREQUENCY FUNCTION): A function giving the fraction of time that the instantaneous pressure, voltage or other variable lies below a given level. ANALYZER: A combination of a filter system and a system for indicating the relative energy that is passed through the filter system. The filter is usually adjustable so that the signal applied to the filter can be measured in terms of the relative energy passed through the filter as a function of the adjustment of the filter response-vs-frequency characteristic. This measurement is usually interpreted as giving the distribution of energy of the applied signal as a function of frequency. ANECHOIC ROOM: An anechoic room provides a free-field acoustic testing environment like the out-of-doors. All of the sound emanating form a source is essentially absorbed at the walls of the anechoic room. Hence, there are no reflections and the spatial sound radiation pattern of a source may be determined. AUDIOMETER: An instrument for measuring hearing threshold level. AUTESPECTRUM (POWER SPECTRUM): A spectrum with the coefficients of the components expressed as the square of the magnitudes. B-SCALE: A filtering system that has characteristics that roughly match the response characteristics of the human ear at sound levels between 55 and 85 db. BAFFLE: A shielding structure or partition used to increase the effective length of the external transmission path between two points in an acoustic system as, for example, between the front and back of an electroacoustic transducer. C-SCALE: A filtering system that has characteristics which roughly match the response characteristics of the human ear at sound levels above 85 db. COHERENCE: A measure of the reliability of a transfer function estimate. It is zero when the transfer function has no statistical validity and unity when the estimate is not contaminated by interfering noise. CRITICAL FREQUENCY: The lowest frequency at which the wavelength of a bending wave traveling in a structure is the same as the wavelength in air at that frequency. Coupling between the air and the structure is very good at this point, and sound waves may move from the structure to the air and viceversa with ease. CRITICAL SPEED: A speed of a rotating system that corresponds to a resonance frequency of the system DAMPING: Dissipation of structure-borne noise (usually traveling bending waves) by conversion to some other form of energy, usually heat. For the most party, this is accomplished by using a material with a high internal energy-absorbing capacity. DEAD ROOM (see also ANECHOIC ROOM): A room that is characterized by an usually large amount of sound absorption. DECIBEL: One-tenth of a bel. Thus, the decibel is a unit of level when the base of the logarithm is the tenth root of ten, and the quantities concerned are proportional to power. Note 1: Examples of quantities that qualify are power (any form), sound pressure squared, particle velocity squared, sound intensity, sound energy density, voltage squared. Thus the decibel is a unit of sound-pressure-squared level; it is common practice, however, to shorten this to sound pressure level because ordinarily no ambiguity results from doing so. Note 2: The logarithm to the base the tenth root of 10 is the same as ten times the logarithm to the base 10: e.g., for a number X 2, log x 2 = 10 log 10 x 2 = 20 log 10 x. This last relationship is the one ordinarily used to simplify the language in definitions of sound pressure level, etc. DIFFUSE SOUND FIELD: A diffuse sound field is one in which the sound field at any given point is made up of sound waves of all angles of incidence. DIRECT FIELD: The sound in a region in which all or most of the sound arrives directly from the source without reflection DIRECTIVITY FACTOR: Of a transducer used for sound emission is the ratio of the sound pressure squared, at some fixed distance and specified direction, to the mean-square sound pressure at the same distance averaged over all directions from the transducer. The distance must be great enough so that the sound appears to diverge spherically from the effective acoustic center of the sources. Unless otherwise specified, the reference direction is understood to be that of maximum response. EARPHONE: An electroacoustic transducer intended to be closely coupled acoustically to the ear. Note: The term receiver should be avoided when there is risk of ambiguity. EFFECTIVE SOUND-PRESSURE (ROOT-MEAN-SQUARE SOUND PRESSURE): At a point is the root-mean-square value of the instantaneous sound pressures, over a time interval at the point under consideration. In the case of periodic sound pressures, the interval must be an integral number of periods or an interval long compared to a period. In the case of non-periodic sound pressures, the interval should be long enough to make the value obtained essentially independent of the small changes in the length of the interval. Note: The term effective sound pressure is frequently shortened to sound pressure. 15