You are on page 1of 14
Machine foundations provide a robust platform for machinery to operate in a smooth manner with imal maintenance requirements. An unexpected vibration can be detrimental to the machine components, ground settlement, structural integrity of foundation, and other machines or working personnel adjacent to it. If vibrations become excessive and uncontrollable, a machine can be forced to shut down or a catastrophic failure can result from fatigue failures of machine components. To avoid this, the serviceability limit state design takes into account of vibration control. Machine vendors usually specity different levels of foundation performance and serviceability vibration limits depending on the machine types for engineers to design the machine foundation, However, in design office, dynamic design is often considered as a secondary check or sometime even omitted due to the lack of time and budget allocated to the project. This paper will present various design approaches from a rule of thumb static design to a comprehensive dynamic analysis and discuss their limitations. A case study of large machine foundation with stringent vibration limits will be presented to demonstrate the necessity and complexity of a full dynamic analysis. This study aims to deliver the message that a static force does not always give the full picture, Project managers and engineers can achieve a better insight into dynamic analysis for those machine foundations that are sensitive to vibration, and then be upfront with the clients about the design time and budget required for the dynamic analysis and evaluation. 1, INTRODUCTION Machine foundations provides a robust platform for machinery to operate efficiently and reliably. Vendors continuously improve the productivity and efficiency of their machine by increasing the machine size and operating speed. On the other hand, the requirements of vibration control can become ‘more stringent due to higher machinery specifications. Consequently, modern machine foundations are required to resist larger dynamic forces at higher operating speeds while controlling vibrations. Unexpected vibrations can be detrimental to the machine components, foundation settlement, structural SEG ane DRE ER ES OS SENATE Ey a ines NC ee RP Rae Oe AY De CNS Si RRO nS of machine foundation can be underestimated. A case study of high-speed machinery supported on shallow foundation will be presented to demonstrate the importance of considering the higher-order dynamic responses. 2, DYNAMIC EFFECT DUE TO HARMONIC FORCE The dynamic response of a single degree of freedom system subjected to a harmonic force can be represented in Figure |. A dynamic amplification factor (DAF) is given by: max. of dynamic response DAF = lax. of static response Fig. | - Dynamic Amplification Factor of SOF System nr, Dac snine cursed G T Undamped 7 [Y 8 STINT Th 4 iv LF LA asa ol ZA 00% 2 po os 0 05 10 15 20 25 3 Frequency ratio(w,/e,) frequency of dynamic force natural frequency of system (Vk/m) system stiffness TEE 3.1 Category I- Static Analysis with Rule of Thumb Machine vendor sometime is confident that the induced unbalance force of machine is insignificant to cause any vibration issue based on the empirical results (e.g. mobile generators or small pump stations). The design can be simplified to rule-of-thumb approach by providing sufficient mass into the foundation system to control the vibration. ACI 351 recommends the weight of foundation system at least three times the weight of a rotating machine and at least five times the weight of a reciprocating machine. The considered machine weight includes both moving and stationary machine parts. No rigorous dynamic analysis is needed. 3.2 Category 2— Natural Frequency Analysis Vendors may require the machine foundation system to be designed such that no system resonance is oceurred during the normal operation. Natural frequency analysis can be used to demonstrate sufficient frequency separation between the system resonance and the excitation frequency. A sufficient frequency separation is achieved when any primary natural resonance of the foundation system is less than 0.5 or greater than 1.5 times the excitation frequency. Unfortunately, in many cases, the frequency separation cannot be achieved and a vigorous dynamic analysis (Category 3) would then be recommended, 3.3 Category 3 — Forced Vibration Analysis Vendor or the client may identify that the machine is crucial and sensitive to vibration. In this case, a forced vibration response analysis such as a harmonic analysis or time-history dynamic analysis is required to evaluate the foundation dynamic response. If the soil radiation damping and the dynamic soil-structure interaction are to be considered in the machine foundation design, a frequency domain dynamic analysis together with impedance functions would then be recommended, This paper will focus Gunter Klein aad Lvetich Bien (2005) Disseat tyyecn! Delors Iequences Of Imectine Kuncetson system (Figure 2). When the machine such as turbo-generator operates at 30Hz and is supported on a shallow rock foundation, the foundation system may be difficult to avoid resonance. The dynamic shear modulus of foundation material is an important parameter to determine the machine dynamic behaviours. The client and project managers sometime underestimate the importance of in-situ geotechnical data such as dynamic shear modulus which may require additional site access or expenditure to obtain, Fig. 2 - Natural Frequency of Support System {Source from Gunter Klein and Dietrich Klein (2003)}, (Gependent on the area of the foundation) —— piles —~ springs elastic area isolation The dynamic shear modulus is commonly obtained by measuring shear wave velocity in the field (e.g, eross-hole method, down-hole method, up-hole method and seismic reflection). The relationship between dynamic shear modulus and measured-in-field shear wave velocity are as follows: G= ou)? where impulsive force. Discussions in this paper will be limited to the reciprocating and rotating machines. Salesmen always describe their machines as perfectly balanced and with no significant dynamic force. The vendor often does not explicitly provide the unbalanced force for the client or designers. In these cases, the unbalanced force can be estimated by ISO 1940/1 and ACI 351 when the rotating mass and balance quality grade are given by the manufacturer: Fy = My€mWo”Sp where Fy dynamic force (zeto to peak) (KN) m, rotating mass (ke) €m mass eccentricity based on the balance quality grade in ISO 1940/1 (mm) © circular operating frequency of the machine (rad/s) 5; service factor allow for increased unbalance during the machine service life generally at least equal to 2 or stated by the manufacturer 5 DYNAMIC SOIL-STRUCTURE INTERACTION The design of machine foundation is complex dynamic soil-structure interaction problem, Academia has been investigated the soil-structure interaction problem since early 1920s. However, very few design standards or guidelines have addressed the dynamic soil-structure interaction. 5.1 Review of Dynamic Soil-Structure Interaction Research activities on the effect of soil-structure interaction in the seismic application had been considerably increased in the 60s and 70s due to the extensive developments of Nuclear Power Plants (Roesset 2013). In the early development, linear and frequency independent springs were used to simulate the stiffiness of the foundation. By 70s, the effect of soil-structure interaction was generally BOs Wer Say Sa SS SS ee, Nae ee eee Se oS eee damping. The impedance stiffness approach has been used to represent the dynamic soil-foundation behaviours. Practicing engineers often are not familiar with impedance functions involved in dynamic soil-structure interaction. Indeed, many design standards and regulations are only qualitatively to describe the general effects and significance of dynamic soil-structure interaction such as in EN 1998- 5:2004. Only few design references such as ACI 351 and Canadian Foundation Engineering Manual provide guidelines to evaluate the dynamic behaviours of machine foundation by the impedance stiffness approach. To help understanding of the basic concept of dynamic impedance stiffness, the equations of motion in time domain and frequency domain are discussed, The typical equation of motion in the time domain (t) is Mil(t) + CHC) + Ke u() = PCC) where M_ mass constant C damping constant Kee static spring constant P(t) dynamic action Frequency domain representation for the harmonic force P(t) = P(w)el* @ Frequency domain representation for displacement, velocity and acceleration uC = uwye"* @ At the low frequency, the real part of dynamic stiffness (Kp — @*M) is a positive value, the displacement response is in phase with the driven dynamic force. In contrast, at the high frequency, the real part of dynamic stiffness becomes a negative value which represents the displacement response is 180° out of phase to the dynamic force. At the resonance frequency (« = y/Kee/M) , the real part of dynamic stiffness becomes zero, and only imaginary part (J«xC) is remained. Interestingly, the dynamic response at resonance is only controlled by damping, and the displacement response is 90° out of phase to the dynamic force. In dynamic design of machine foundation (Wolf and Deeks 2004), a dimensionless frequency (a) is commonly used instead of frequency (w). The dynamic stiffness is rewritten in terms of spring coefficient k(aq) and damping coefficient c(ag) as a function of dimensionless frequency (ag) Ka(@o) = Keel (k(ao) + idoc(ag)] ae ay = 2% % where To characteristic length of the foundation Vs. shear wave velocity of the first soil layer Manual computation of the impedance response functions and dynamic responses can be tedious. Computer programs such as Matlab and DYNAG can be used to determine the impedance stiffness and damping coefficients, DYNAG has been developed in University of Western Ontario, Canada by El Naggar and adopted in Canadian Foundation Manual for the machine foundation design. 5.3 Limitations of Impedance Stiffness Approach The impedance stiffness approach, which assumes the interface of soil-foundation to be rigid, is only ALTOURTSTOR WHE COS OF TIO INES HPSEC TNSCIUNS CRN DES COMQSRE CYTES SOUL-SEOTS REERCLON. problem, While this case can be analysed by assuming the interface of soil-foundation to be rigid with no flexural deformation of foundation, the effects of flexural deformation can be detrimental to the vibration control. Some advanced finite element analysis (FEA) packages, such as SAP2000, are capable of modelling a flexible machine foundation with the dynamic soil stiffness and damping coefficients in the form of impedance functions.. 6.1 Problem Definition Multiple high speed machines supported on a single block foundation is not uncommon for gas compression station. As the case study, two high speed centrifugal compressors driven by an electric motor (E-motor) were supported on a monolithic foundation shown in Figure 3 Fig. 3 - General Arrangement of Gas Compression Unit Emote This gas compression station was identified as the crucial facility along the production line. The E~ 6.2 Machine Foundation Resonated with Shallow Rock DYNA with the consideration of dynamic soil-structure interaction was used to evaluate the natural frequency of the foundation system. Although the machine foundation was supported on relatively good shallow sandstone, the vibration response still could not meet the stringent vibration limit given by the vendor due to the fact that the machine foundation system could not avoid the resonances at the machine operational speed. In the early development of the design solution, a concrete block foundation supported directly on the shallow rock was considered, The natural frequency of this type of foundation, however, appeared to be fairly closed to the machine operational frequency. Figure 4 illustrates the primary foundation natural resonances in relative to the machine operational frequencies. Primary natural resonances of the machine foundation were horizontal (Y), vertical (Z) and rocking (RX) modes. The E-motor operated at low frequency (24Hz to 32Hz) whereas the compressors operated at high frequency (134Hz. to 208Hz). Horizontal natural frequency (Y) appeared to be at the machine operational frequency in this case. The maximum velocity response thus exceeded the vibration limit of 0.7Smm/s shown in Figure 5. ‘A number of attempts were used to provide sufficient frequency separation between the foundation natural frequencies and the machine operational frequency, however very little success was achieved. By enlarging or reducing the foundation footprint, and by increasing or decreasing the mass of foundation, the horizontal and vertical natural frequencies shifted together in the same manner. Since the horizontal and vertical natural frequencies were fairly closed to each other, it always one of them fell within the machine operational frequency. As a result, it was concluded that the block foundation, in this case, was not a practical solution to avoid resonance, Fig. 4 - DYNA6 - Primary Natural Frequency (Block Foundation) In terms of adjusting individual natural frequency and providing frequency separation, the pile foundation was a better solution. The concept of pile foundation (Figure 6) is similar to table-top foundation, which consists of large diameter of piles with sleeves to isolate the machine foundation from the surrounding rock. The sleeved pile (Figure 7) reduces the horizontal stiffness (reduces horizontal natural frequency). The large diameter of pile increases the vertical stiffness (increase vertical and rocking natural frequencies). Fig. 6 - Pile Foundation isolated from Surrounding Rock Fig. 7 - Pile Sleeve Detail s \ meg Fig. 9 - DYNAG - Dynamic Response (Pile Foundation) It is noted that DYNAG did not indicate any vibration response at high operational frequencies. The flexure behaviour of foundation was not considered in DYNAG, hence, a finite element model coupling with the frequency dependent pile stiffiness and damping coefficient was established to ensure that any vibrations due to flexure of the foundation were acceptable. 6.3 Finite Element Model ith Frequency Dependent Pile Stiffness ‘A mass foundation with a total number of 12 piles was simulated by using a finite element package (Strand7) together with frequency dependent pile springs from DYNAG (Figure 10). The concrete foundation was modelled using brick elements. Compressor skids were modelled using steel beams. Masses of compressors, motor and associated components were modelled as translational node masses. The frequency dependent pile springs were modelled at the bottom of conerete foundation. Frequency 6.4 Unexpected Vibration due to Flexural Deformation ye In general, the primary natural frequencies by Strand7 agreed with DYNAG results (Table 2). No primary resonances fell within the machine operational frequency. However, flexural deformation of foundation appeared to be triggered by the unbalanced force of E-motor shown in Figure 11 Table 2— Natural Frequency Dz th Mass Participation Ratio Mode Freq(Hiz) DX DY 1 42 “7 169 39.1 413 453 469 503 36.0 99.8 ou 973 19 12 os Ls 648 03 12 296 Ba RX 38 13 243 22.7 28 196 12 RY os 06 359 47 nz 24 38 RZ 32 441 31.0 46 26 06 $8.0 10 o7 1 Force vibration analysis was undertaken separately to evaluate the vibration response due to the E- motor and compressors. Figure 12 presents the dynamic response of the machine foundation subjected to the unbalance force due to E-motor or compressors. A clear frequency separation was achieved for the dynamic response triggered by the unbalanced forces of compressors. The sleeved piles under-tuned ‘the horizontal resonance; and the relatively large diameter piles over-tuned the vertical resonance. Fe we In contrast, the flexural natural frequency at 3911z (Mode 4) was triggered by the unbalanced force of E-motor, which was fairly closed to the machine operational frequency. Fortunately, with a further adjustment of pile sleeve length, the flexural resonance due to E-motor was shifted towards the higher frequency, and the total dynamic response satisfied the vendor vibration requirement shown in Figure 13, Fig. 12 - Dynamic Responses due to E-motor and Compressors motor compressors 6 : 3 Fa A Ba ee ° ° 20 «0 0 0 100 Freeney (ke) Fig. 13 - Dynamic Responses with Flexible Behaviours of Foundation GRRE ORY OE EE SEN CLS TOE WEEE PRR SIS. SENN GN GEER CHE I been presented. A sleeved pile foundation was adopted to avoid foundation resonances coincided with the machine operational frequency. The dynamic effect of flexible machine foundation has been found ‘that it can be detrimental to the vibration control. Consequently, a finite element model is recommended to consider the dynamic soil-foundation interaction for the vibration-sensitive machine foundation. REFERENCES, ACI 351.1R-99 ~ Grouting between Foundations and Bases for Support of Equipment and Machinery. Canadian Foundation Engineering Manual 4" Edition ~ Canadian Geotechnical Society 2006. Poulos, Harry G. "Behavior of laterally loaded piles: I-single piles." Journal of the Soil Mechanics and Foundations Division 97.5 (1971): TH1-731. Ginter Klein and Dietrich Klein (2003). Geotechnical Engineering Handbook, Procedures Vol.3 . Ernst & Sohn. Wolf John P. (1994) Foundation Vibration Analysis Using Simple Physical Models, PTR Prentice Hall. Wolf John P, and Decks Andrew J. (2004) Foundation Vibration Analysis: A Strength-of-materials Approach, Elsevier. Roesset Jose M. (2013) Soil Structure Interaction The Early Stages, Journal of Applied Science and Engineering, Vol. 16, No. 1, pp. 1-8 Veletsos, A. S., & Wei, Y. T. (1971). Lateral and rocking vibration of footings. Journal of Soil Mechanics & Foundations Div. Kausel, E.(1974) “Forced Vibration of Circular Foundations on Layered Media” Sc. D. Thesis, Massachusetts Institute of Technology. Novak, M. (1974). Dynamic stiffness and damping of piles. Canadian Geotechnical Journal, 11(4), 574-598,

You might also like