Principles of Foundation Design for Engines and Compressors by W. K. Newcomb O ‘Ingersoll-Rand, Painted Post, N. ¥. Reprinted from the Transactions of the ASME for April, 1951, with permission of copyright owner oem ( Simegersoii-Remed co [ . 11 BROADWAY NEW YORK, N ¥, 10004(Ceeprinted from the Tronsacions of ha ASMER for April 1961, wih permission of Copyright Oumar) Principles of Foundation Design for mgines and br WK. NEWCONB? ‘This paper discusses the elaatic character of the ground ‘and chows that foundations for reciprocating machinescan, ‘be treated as epring-supported masses, the machine ‘and foundation representing the mass, and the ground ‘the spring. Such elastic systems have natural periods of sibeation; if the frequency of unbalanced inertia or other (tuciting force is near the natural frequency of the founda Son, rewonance will occur producing excessive vibration. Examples of this phenomenonare given and texts on foun- dations having resonance are described. Also, the nature of inertia forces found in reciprocating machines is ex- plained. Examples of good and bad foundation design re shown and « rational method for foundation design is outlined. Ieracoucriox HIS peper is an attempt to rationalise foundation design, 1 rien by » Gre and baler of recroaing ‘ove not ia the foundation business, and it therefor tects the aberrations of a mechanical soqatet rather than » civil engineer ot expert on soi! mechanics. Foundations for reciprocating machines difler from foundations for buildings or similar structures since dynamic rather than atic lade are involved. With a static load, only, the bearing, ‘capacity of the eoil ned be considered, and there are various well Known rules to follow. With a dyoamic load, however, these rules do not apply, since the frequency of the forces and danger ‘of resonance with attendant excesive vibration must govern foundation design. ‘Biase Cxanacren ov Guo ‘Although ground chatacterietcs and subtol elasticity have been discuved or mentioned in engineering literature (1,2, 3, 4, ‘and others) from time to time, their relationship to the prin- ‘ple of foundation design haa not been emphasized adequately ‘or thoroughly understood. Ax a result, the elastic character of ‘the ground is not always recognized. ‘The following examples show this characteristic: ‘An oibetorage tank war observed to settle when filled and rise ‘0 its original position when emptied. The test was made with a ‘transit and repeated several times with the same results, There vets a detsite relation between lod and defection which i trpi- ‘eal of elastic materials, ‘The elastic character of the ground ia alm shown by a soil defection test at another sta, A loading platform with 2 aq ft bearing area was loaded gradually. The defections obtained * Machanical Enginear, Interell-Rand Company. Mem. ASME. | Numbers in parenthens refer tothe Bibliography atthe end ofthe peer. Contributed by the Oil and Gas Power Division and prevented st the Ol and Gas Power Conlerencs, Dultimore, Mdy June 12-10, {1000 of Tae Aarenrcan Sorter? oy Mxcaaican ENGUEEAR ‘Nore: Stasrnente and opisions wlvasted in onzers are to be ‘wovertood an itive expressions of ther auzors and aot hoe of {he Scciety.” Manuscript recevea at ASME Headguartern, March 30, 1980. "Paper No. 60—OGP-S. Compressors PAINTED POST, N. ¥. are plotted in Fig. 1, curve AB. With 1500 Ib per aq ft load, the deflection B, was 0.30 in, This load was maintained for 72 Ihr and no further deflection took place. ‘Then the losd was re- moved quickly, and the defection changed to 0.25 in, C. The Joad was applied quickly, and the delection went to 0.905 in., D, ‘which is very close to the original deflection, B. The line BC therefore representa the load-defection characteristics of this ound. ‘Because of this elastic nature of the ground the foundation ‘and aubeol form an elastic aystem consisting of a mass (foundse ton block tod machine) supported by a apring (subsoil). Tf sruch an elastic syntem ix excited by period foroes having a fre- ‘quency near the natural frequency? of the elastic nystem, reso- ‘ance will occur producing excemive vibration. Fonces Acrnva ox Fouxpanions his not always possible to obtain perfect balance in recipro- cating machines and, a « result, unbalaneed periodic foroes may ‘exit. These forces are caused by aconleation and deceleration of the piston or other reciprocating para. Also, centrifupal forees and torque reactions are sometimes factors, Fig, 2 shows ‘the inertia forces in a typical single-cylinder engine or compres- sor resulting from thin acceleration and deceleration. At top dead éeater (in a vertical machine), the acceleration and resulte fng inertia force are maximum in the direction away from the crank, At bottom center they are maximum in the opposite direction ‘This inertia force F, which acta along the axis of piston of ‘erombead ax thown in Fig. (0) can be expreaved mathematically ‘at a Fourier series aa follows = 00000284 WRN (con 8 + A 00028 + B oe 49 + Coon 0+... sell 1 Satara feaueney, the frequeney at which an lati nyatera tonde to vitrate afer bring dsplaced from the equiibsum position and ay ‘Ralerance (6) and other taxta on engine dynumica, ~ anem as tet me ad Wro.2. femerss om Accaamartow Fonces 14 Cua aio Comma ‘pre Roo Mzcassisw where Fm inertia foree, To (@ — position of crank in degrees after top or outer dead center W = weight of reciprocating parts, Ib N = speed, pm L = length of connecting rod, in. B= crank radius, in. B,C — constante determined by L/Rratio. When L/R = 4, 4 = 0.25402, B = —0,00409, C = 0.00007 Equation [1]is abo written F = 0.000028 WRN? K.. K ia called the inertia force (acceleration) factor and represents the terms im parentheses in Equation [1]. Valuesf K for difer- ‘eat L/R ratios are found ia most mechanical-engineering hand- books. : "Tho inertia force F is represented graphically in Fig. 20), where K ia plotted against crank angle. Note at outer dead center (9 — 0) that X is maximum having a value of 1.250 (eben L/B = 4), Ax the crank rotates, K decreases to 2270, ten becomes negative, reaching « maximum negative value of 0.750 at bottom dead center (@ = 180°). When K ia positive, ‘he inert force F acts away from the crankshaft; when nege- ‘ive it acta toward the crankshaft. ‘Examination of Equation [1] or (2] shows that the inertia force F varies directly with the reciprocating weight, directly with the stroke (or crank radius) and with the square of the peed. ‘TRANSACTIONS OF THE ASME ‘The primary inertia force F” and wecondary inet fore P*) become ‘FY = 0,0000284 RW? cos 8. 5) ‘The primary and secondary forces having different frequencies have diferent edects on vibration and must be dealt with in- dividualy, Since they vary siousoidally, the maximum values fare used for foundation and vibration ealculations. The maxima fSecur when cos 0 = 1 and cot 26 = 1, and the primary and sec- ‘ondary forces then become . Prisaary, Frasn = 0.0000284 WRN?.. i} Secondary, Frese cu} Fam ‘In single-crank machines both primary and secondary inertia forees are unbalanced, With two cranks at 180 deg the primary foree of one crank is opposed to the primary foree of the other ‘rank and completely balancos it ifthe reciprocating weights of ‘the two cylinders are the same. The resultant primary force vwill then be zero. However, the secondary forces actin the same direction and add. With two cranks at 90 deg, the secondaries faré balanced and primaries partly balssced. With 3 or more equally spaced eranks, both primary and secondary forces are alanced. However, while the forces may be balanced, couples ‘can be produced by the forces acting aloag the axes of the dif- mame "ED INERTIA FORCES AND COUPLES FOR yea DHFSEEN? CHANE ARRANGEsCENTS craw senensnes rec crea RO Se ss = = Eel [-|- une cruncens @ we for leon. - orem cures = = - - ts Ee - Ea cel en) ™ “CHLNCERS AT 9" Saar - ~ L™ "The frst term of the part of the equation in parentheses, that in fos 6” Equation, {1], representa the “primary” inertia force ‘which bas one complete eycle per revalution of the erankshaft a3 shown in Fig 2(c). The second term, “A cos 2," represents the “eoondary” inertia force and is much smaller; it has two com- plete cycles per revolution, Fig. 2(q). The next term isthe fourth (Order with four cycles per revolution, but itis very small and is ‘unully neglected. There are also 6th, Sth, and higher orders, but they, mre likewise very small and.can’be neglected. For foundation design we need daly consider the primary and second~ ary forces, and the simplifed approximate form of the inertia- fores equation is generally used, Here the inertia force F ia expresoed a3 x comers 323 (w 6+ Feu 28) oofNEWCOMB—PRINCIPLES OF FOUNDATION DESIGN FOR ENGINES AND COMPRESSORS ferent cylinder Table 1 shows unbslané foro and couplen ‘obtained sith some common crank arringemente, ‘Resoxaxee mv Fooxoanions, ‘Since the foundation isan elastic system, these periodis forces (and to 2 lemer extent the couples) tend to induce vibration Uf the frequency of the exciting foree is near the natural fre ‘queney of the foundation, resonance will obcur and may cau ‘tzoemsive vibration. Fig 3 shows a gne-cngine-driven compressor foundation havitg resonance withia the operating speed range. While adequate unerete yardage was used (four times that generally necessary for 2 machine of this type), it was not placed effectively. The due to unbalanced seoondary forces, the natural frequency i 72, the vibrations per minuta occurring at 236 rpm, the resonant, peed. ‘An engine or compressor foundation can be represented graphi- cally as spring-supportad mass with a dashpot to provide damp- ing as shown in Fig. 5. When the nias, spring charactritics, and damping are known, also the magnitude and frequency of the fxciting force, the amplitude of the resulting vibration can be determined, Jo ene] T eee “ hi St le 2 HIN ee ? PACE nemo Les | & Fra. 3 Fourmanon Haro Resowance mt Oreaarura Srrxo = PT | — ‘ tert “ar if a - = 2 1. tit Tt i ~ : Teac noon i ie. Viaurionssomines Conras 2 Foz Me 6 tom tcl memnnc care for the cate een | : repeated in Fig 5. Here “amplitude ratio” x plotted acninat Leer] “tur te! tr eitowat taping faa” Nota ht Fro. 4 Viszanox Tear Saowmra Resowancs mr 4 Fouxpaow ‘tround was soft, and:the result was excessive vibration, Vibra- ‘Hon measurements showed tat the foufdstion rocked back and forth about an axis slightly below the base as marked in Fig. 3. ‘This movement was caused by horizontal wecondary forces in the ‘compremor. ‘Tbe horizontal movernent at top of the foundation is plotted in Fig. 4. Note that at 354 rpm the amplitude of vibration it AAs in. As the soed decree be amps inca ‘4 maximum of 0.016 in. at 236 cpm, although the inertia, {ic (abich vain a tw aquae of the tpeed) i 30 por eat {eas than at 354 rpm. Below 236 rpm the amplitude drops off sharply. Ia this particular machine the unbalanced force was 10,800 Ib (seonndary) at 354 rpm, and 4500 Ib st 236 rp. ‘The vibration peaking in thia zanser is typical of an elastic aymem. As there were bro complete vibrations per revolution tos C= eta + Be w/a ‘Amplitude ratios © 1 when y= actus! amplitude 1 = free amplitude, the amplitude with which the foundation ‘ase would vibrate if it were oot restrained ¢ = damping factor, viscous damping sssumed ‘w= frequency of exciting force a, = natural frequency of foundation ree amplituie, r = 35,200 F/Wa 0h 1 Damping ete reutin ip amplitude of vibration duet ihe dary ‘Vncoue damping is tesumed reel tae seneny Wh terntration, ¢ = 1 when e * elerence (i) aad other tazta on vibeation, O‘TRANSACTIONS OF THE ASME PF = maximum valun of primary or secondary inertia foros, ‘+ whichever is being coowidered, Ib W @ combined weight of foundation and machine, Tb (@ = Seequency of inertia foree F, in eycles per min (opm) ‘An emmination of Fig. 6 shows the amplitude ratio ia sero, atten the frequency ratio is sero, which means there is 20 Vibration when the machine is standing tl. Then, aa the speed of te wechine and frequency and magnitude of the exciting orcemmimereased, the arplitade increases, reaching a maximum, at ormamr the frequency ratio of 1 (for the smaller damping fao- {ors mien ¢ = 0.25 oF les). At thisspeed the periodic xciting foreedeiwesonance with tho natural frequency of the foundation. [At bagearepeeds the amplitude falls off and approaches the free ‘ample. Let us see how these resonance curves ean be ap- pled te Elerent types of foundations ‘Themame two ceneral types of foundations; (a) the resiliently smoustel Soundation supported on eprings, rubber, cork or felt te, eal () the conventional concrete foundation poured irediyam the ground, with or without pling. While this paper ‘aay Bela with the second type it may be of interest point ‘out mmadiicronces between them so as to emphasise the char- ‘eteditim of the latter, ‘To be eGective, the resilient founda Sousmast have a natural frequency well below the frequency of snyemiting force so thatthe frequency ratio is 2or 3" On the ‘there, i in desirable that the conventional foundation have t frequmey ratio of 0.5 oF less to avoid excessive vibration, Damping factors of sole, vary widely, an average value (1) beigg RAK However, thé foundation in Fig. 3 had a damping. factarorm 0.082, which is unusually low and was revult of the rocliggaetion displacing little of the ground. The vibration Remmpamests of this foundation are plotted in Fig. 6, curve A, showing that the actual performance agrees with vibration the: ‘ory. Bwrtunately, it ie not necemary to know the damping fae- he soil because in a good foundation the frequency ratio jar Yess, Note that the damping hss little effect on the ‘amgtale in this part of the resonance curve, and if neglected (email be small sad oa the safe side, ‘Thus, to predict the etlemmnce of a foundation it is necessity to know only the “tem emplitude,” ‘natural frequency,” and the “magnitude” and ingueny” of the “exciting forces.” 2 Beem Arrrcrovo Naromat, FREQUENcr oF A FOUNDATION Yathawey 2 foundation can have six degrees of freedom. That a, amg vibrate in six diferent ways as shown in Fig. 7,0r 8 ‘combimtion of them. However, in the sctual foundation the awemmmt is generally either horizontal or vertical, depending on ‘wildhtlrce predominates or ia near resonance. The horizontal re ‘Vansamiom or Foonnanions, Six Deaaszs ov Fazsvow t ‘natural frequency ia usually the same as the vertical natural fre! quency, The natural frequency in rocking is influenced by the dimensions of the foundation, For shallow foundations the rocking natural frequency will be very nearly the same an the vertical natural frequency. For high foundations (also deep foundations) it will be much lower. ‘The approximate natural frequency of a foundation can be determined from the stat defection ss described Iater in Equation [10] or expresoed in terms of soibbearing preasure aa abown in Fig. 8° Here the gues pasaee 8 BUH EE I ‘4 Fra, 8 Naroms Funquanor or Fovxsuniow ox Two Treas oF Son atural frequencies for two different soils are given. Nota ‘that the softer ground has the lower natural frequency. The raph also shows how tho natural frequency is acted by soil losding; the lower the soil bearing pressure, the bigber the natural frequency. This fact provides « meana for controlling the natural frequency. If the natural frequency of proposed foundation is too low, it ean ba raised, within limits, by reduoing the soil beating preseure, This improvement can be accome lished by spreading the foundation or placing the foundation Dock on & mat. Fig. 9 is an example ofyraising the natural frequency of foundation by extending the base. This foundation is the eame fone shown in Fig. 3, but a reinforced mat, 2 {¢ thick, was added to tie to another foundation on one side, This change raised the resonant pes (u/a = 1) from 236 rpm, curve A, to 346 rpm, sand substantially reduced the amplitude as sown in eurve B. ‘The new natural frequency ia 692 vibrations por min and is a big, {improvement over the original foundation.* Curve A (eame as Fig. ), shows amplitude of the original foundation for compari- * Fig. 8, while empirial is based on published data on soils, The upper ends ofthe eurves, where the soil bearing prenture i igh, ap Drosch the nagural frequescing calculated from static deecton (I). "Te lower ends of the carver here the tol bearing preaure ia Lom approach the natural frequeacy of the round determined by vibes: tee teats, such as (2). "Thus ie takes lato account the fact that the tqound acts like a spring haviog appreriablo mess, Halerence (0) ‘Brot additional information on natural frequency of foundations. ice euros nth raat inicanan mabe ton charactereica"althougS oiser iar (G) find onknearty. This illrence map by toe Teele oF shape of fouudation or magaitude of deGecton of the tol.NEWCOMB—PRINCIPLES OF FOUNDATION DESIGN FOR ENGINES AND COMPRESSORS het ent @ veuro= nein = = Se Tra. Rassno Naronas Pangvancr bx Exrannmne FoownaTon ‘The revised foundation was satisfactory for all epeeds up to 200 spm. Further improvement could be obtained by extend {ng the base on the other side. Curve C, Fig. 9, shows resulta ex: pected if the foundation had been placed originally on an ade- ‘quate mat, 20 a8 to provide a lower soil bearing presrure and ob- ‘iia a higher natural frequency. ‘Fig. 10 shows an example of resonance in « compressor founda- r meumeg 7 vouron rote EE Fre, 10 * ig abown in Fig. 11. ‘The effect of eoft and firm ground under identical foundations Fig. 11(0) shows the results of « vibration ‘tat where a foundation intended for frm ground was placed on oft ground. Here the natural frequency was too low, producing. Feaonaoe and excessive vibration at 434 rpm. a Fig. 110), the engine and foundstion were exact duplicates, but tbe ground was firm, A much higher natural frequency (beyond the range of ‘the tent) wat obtained and vibration was nil. ‘of low mi loadings for soft ground, auch a8 wet cannot be overemphasized. A common mistake in the design of engine ad compressor foundation isto follow the sil bearing pressures allowed by building codes such xs Table 2. ‘These sol bearing pressures are much Woo high for dynamic loud, Soft olay or wet sand, for instance, is considered suitable for 2000, 1b per aq ft under building footings, but an engine or compres ‘er foundation on this soil and with this bearing load wil have natural frequeney of 490 vibrations per min ((rom Fig. 8), which in suitable only for.s low-speed machine, TABLE 2 EXAMPLES OF ALLOWABLE SOUL BEARING PRES fied Fol RRNAW Loh be loads oniy & fraction of thee eoil bearing preseuree “cea eer eum oie ‘The fact that bigh or deep foundstions have a lower natural frequeney ia rocking than shallow foundations has already been mentioned. A comparison of typical shallow and high founda- tons shows thet the natural frequency of the later may be reduced to one half, unless epecial precautions are taken. When ‘necessary to install engines or compressors on high foundations ‘a mat ehould be wed to iower the sil londing and raise the natural frequency. The block on which the machine rests ean be relax ‘ively amall, the only requirement being that it is muticiently strong and rigid to provide adequate support and maintain ‘proper alignment of the machine and traamaje; the vibration forees to the mat. Pnawo "When poor soil is encountered, piling is often desirable. Tt is ‘beyond the scope of this paper to discus the detailed use of pile af this subject belongs to the foundstion specialist. How- fever, ples properly used will rise the natural frequency of the foundation, expecially the vertical aatural frequency if driven to firm ground, When substantial horizontal inertia forces sre encountered and piling is necemary, ube use of batter pilen is {important to keep the natural frequeney igh. A foundation supported on piles should always be checked for natural ire- ‘quency oo the assumption that the earth may abrink away leav- {ng the foundation supported on columas. This ean be done by determining the “static detection” caused by a horizontal force ‘equal to the total weight of the foundation and machine, ‘There {e's convenient relation between the static defection and the natural frequency aa follows oe 4 = natural frequency, vibrations yer min 3 = static defection, in. 110} 2wv 3a Fig 12 shows natural frequency plotted againat static deflec- Pie 12 Beamon Berwsze Grane Darccmow sx Nason ‘Demamnvo Fooxnarions ‘Manufacturers’ foundation drawings show foundations muite~ ble for Srm soils euch as well-cemented sand and gravel or hard lay. When tofter ground is encountered, additional precautions ust be taken, Many foundation probloms can be solved by ‘placing the foundation block on mat, the sizeof which is deter- Iined by the magnitude and frequency of the unbalanced forces fend the character of the ground. In other cases piling ean be ‘wed effectively. Tt ia good practice to refer the design of all special foundations to a foundation expert, one familiar with soil ynamice, ‘in designing the foundation for an engine or edmpressor, of ‘any machine having periodic forces, the procedure to follow sbould be: 1 Determine the magnitude and frequency of the unbel- ‘anced forces, This information is generally supplied. by the fanufacturer of the machine, The frequency of the unbalanced forces establiahes what the natural frequency of the foundation must be to avoid resonance and obtain a vibrationless instals- tion, ‘The natural frequency should be at least twice the fre- quency of any substantial unbalanced force. Expressing it an- ‘ther way, the frequency ratio should be less than 0.5. ‘2. Determine the character of the soil by borings, deflection teats, and, if possible, dyaamie testa (2,3, 6). ‘The maximum al lowable bearing preasure required to obtain the natural frequeney found in item 1 can be developed from these data or an approx imate value ean be read directly from Fig. 8. ‘3 Pick off the amplitude ratio from ig. 6. 4. Assuming some sllowsble amplitude of vibration auch as 0,002 in, determine the free amplitude 'S Keowing the magnitude and frequency of the unbal- anced force, item 1, and the free amplitude, item 4, the total ‘tum of the machine and foundation can be found, But the foundation-must be proportigned 0 a8 not to exoeed the maxi ‘mm allowable eo bearing pressure determined in item 2. ‘Taz Inet Fouxoanion Fig. 13 shows exaroples of good and bad foundation design ‘Avoid high or deep foundations. If tho latter type must be used bbe mure to provide » generous mat, Tuo ideal foundation is hallow (but suficiontly rigid to maintain aligoment) and is ‘spread out to obtain a low wil bearing preseure and place a farge TRANSACTIONS OF THE ASME surface in contast with the ground, ‘This constructign insures a high natural frequency which ia necessary for vibrationleas operation. Aexnowizpaucncrs ‘Tho author wishes to expres his appreciation to members of ‘the Ingersol-Rand Company staff for sssstanoo with the prepte ration of this paper. Ali, to Mr. F. E, Rasmers for data on aol ‘defection, and to Mr. W. P. Kinneman for suggestions on use of, piling. ‘BIBLIOGRAPHY 1 "Dosign of Diaal-Ensing Foundations,” by ICH. Larkin, ‘Trans, ABME, vo. Of, 142, pp. 341-300. 2 “Geophysical Study of Sol! Dynamica”" by R. K. Bernhard, ‘Trane, ATME, vol. 138, 1640, pp. 320-344. 12. “Dynamie Principles of Machine Foundations and Ground.’ tyJ. HA. Crockett and RE, R. Hammond, Proceedings of The Tewitution of Mochanical Engineers, London, England, vol. 160, 1049, pp. 612-523. 4 "Resonance of Machine Foundations and the Soil Cooficiate Wich Aiect Inv by G. 2. Tachebotariod and E.. Ward, Secoud ermal Contec oa Bo chat, vl 104, 5. 30 15 “Balancing of OU Eagloes.” by W. K, Wilson, J. B: Lippincott Company, Puiladaphia, Ps. 1929, pp. 7-10. ‘8 “Dynamic Testa by Seana of Ladueed Vibrationa” by RK. Barabard, Procedings ofthe ASTM, vol. 37, 1087, part 2, pp. 694 os. 7. ““Mechanieal Vibrations” by J. P. Dea Hartog, second edi- tio, McGraw-Hill Book Company, Inc, New York, N.Y. 1940, pid, p. Th ote. 8 “Theory of Elastic Engine Supports," by D. 8. Rowonsweig ‘Trans, ASME, vol. 61, 1990, pp. 31-28. 10 “Reduction of Ground Vibrations invo Structures," by J. HA. Crockett and RB. R Hammon, Iasttution of Orel Bagioesr, ‘Structral Paper 0, 18, London, Baglad, 1947 Appendix Ixerapieners Usto 10 Maasonz Vinnarion ‘The vibration testa described in thia paper were made with the ‘Type 761A vibration meter and Typo 762B vibration ansiyser rade by the General Radio Company, Cambridge, Maas. These Instrumente are inertia-operated crystal type and measure tbe root-mean-equare (rm) atmplitude which is plotted in the diter~ ‘nt ilustrations, The maximum amplitude from the mean position is 1.414 times tho rms amplitude, while the total over-all, isplacement (sometimes called double amplitude) ia 2.828 times the rms ampiituce. The test, abown in Fig. 10, wan mado with the foregoing vibrae tioa-measuring inatramenta and rooorded with a Brush Develop- ‘ant Company oscillograph.