Académique Documents
Professionnel Documents
Culture Documents
The Importance Of A
abound for soil-foundation
Sound Foundation > Methods
modeling for rotating, reciprocating
machinery
By THOMAS A. CHIRATHADAM
AND BENJAMIN A. WHITE
he seamless performance of rotating and reciprocating machinery relies on the effective design and
installation of their foundations. The foundation design involves careful consideration of foundation block size,
block constructability, use of oil barrier, equipment chock
mounts, anchor bolt design, use of rebars for crack management, and soil-structure interaction [1].
In addition to the foundation block size, block embedment, soil parameters, and dynamic forces also influence
the overall system response [2]. This article introduces several methods for soil-foundation modeling, presents a few
commonly used models, and illustrates the application using an example analysis of a pump foundation.
The foundation blocks are sized sufficiently large for dissipating the unbalanced forces, moments and torque generated by the machinery. However, the center of gravity of
the foundation must be kept as low as possible to avoid
large rocking motion and the foundation bearing pressure
must not exceed the soil-bearing capacity. Hence, the foundation block design involves trade-offs between the foundation size/mass and surface area.
As a general rule, the foundations are sized such that the
block is at least five times heavier than the machinery. The
transmissibility of the dynamic forces from the machinery
to the ground depends on the soil properties such as the
shear modulus and Poisson ratio. The foundation blocks
resting on soft soil may resonate in various modes as an
isolated system, instead of dissipating the forces. Installation of piles can provide additional stiffness and damping to
the foundation due to the soil-pile interaction.
The prior art offers several analytical methods for modeling the machinery foundation impedance, of which the elastic half-space method has been quite extensively used in
the industry. The soil is an infinite inhomogeneous medium,
and most modeling methods approximate it conveniently to
either a series of uniformly distributed springs or a set of
translational or rotational springs. See Gazetas [3] for a detailed review and description of several approaches avail-
TECH brief
ance functions of foundations embedded in stratified soil.
The procedure involves developing dynamic stiffness matrices for each cone layer and assembling them together to
form the complete dynamic stiffness matrix of the entire soil.
Further, imposition of suitable kinematic constraints gives
the impedance functions. Pradhan, et al. [9], compares the
predictions from a cone frustum model for a two-layered
system against measurements and finds good correlation.
Gazetas and Roesset [10] demonstrate that the properties of the soil layers do not affect systems mounted on
heavy foundations. Hence, approximate solutions from
elastic half-space method can predict the system responses with fair amount of accuracy. Flexible skids and light
foundations are sensitive to the number of layers of soil and
the presence of bedrock. A finite element analysis modeling
the soil as uniformly distributed spring stiffness beneath the
skid and foundation can provide more accurate solutions.
However, no ready reference comparing predictions to field
measurements of resulting equipment dynamic responses
are available.
The shape of the foundation (circular, rectangular, etc),
and soil properties (homogeneous, layered, or above a rock
layer), embedment (foundation on surface, embedded), and
foundation block rigidity (rigid/flexible) largely affect the soilfoundation interaction. Elastic half-space theory has been
widely used to model dynamic foundations, due mainly
to the simplicity of the lumped parameter approximation
employed [3]. Also see Richart, et al. [4], for detailed lumped
parameter modeling of foundations using half-space methods.
The half-space method assumes that the foundations
are circular in shape. In order to model foundations of any
other geometry, an approximate equivalent radius must be
used. For instance, the equivalent radius of a rectangular
foundation with width 2B and length 2L for translational
modes is
The simplest method of modeling a soil-foundation system is by assuming that the rigid foundation block rests on
a homogeneous soil half-space as illustrated in Figure 1.
Table 1 lists the static stiffnesses for the rectangular rigid
foundations, along with typical correction factors. Note that
several publications provide varying correction factors, and
hence only a range is provided in Table 1. For more information, see Reference [3].
ii) Soil stratum over rigid base
Figure 2 shows a schematic representation of a foundation block resting on a soil layer above bedrock. Table 2
lists the stiffnesses for the foundation block. Note that the
expressions for the foundation stiffnesses are valid only
for certain ratios of soil layer depth and foundation equivalent radius.
(1)
n Figure 1. Rigid
foundation on homogenous elastic
half-space.
2015 EDITION
TECH brief
2015 EDITION
n Figure 6. This graph shows the maximum velocity in a pumpfoundation model for various frequencies. Increasing running
speed generates proportionately larger unbalance forces.
Pump foundation analysis example
An example foundation analysis is presented here for
a skid-mounted pump, API: BSAFP(1CW-FL). A 5500 hp
(4101 kW) motor, operating at 1500 rpm, is
driving the 20,000 lb. (9072 kg) heavy centrifugal, single stage, HSB pump via a hightorque power transmission coupling. The
pump and motor are securely fastened to
a base plate, which then is anchored to the
foundation block with 16 bolts.
The block foundation, which is nine times
heavier than the equipment mounted on it,
is embedded in about 4 ft. (1.2 m) of soil.
The finite element (FE) model of the pumpmotor-baseplate assembly and the foundation block is built in ANSYS, as depicted in
Figure 5. The soil in which the foundation is
laid has properties listed in Table 5. During
operation, dynamic imbalance loads from
the pump and motor shafts are transmitted
to the base plate. Additionally, static loads
from the suction and discharge piping also
act on the pump. The anchor bolt stresses
are due to a combination of the static and
dynamic loads on the equipment attached to
the base plate.
The analysis focuses on finding the system
overall response and the loads acting on the
anchor bolts during normal operation, with the
pump and motor rotating unbalanced forces
acting on the system. The rotating unbalanced
forces arise due to the rotor center of mass not
being coincident with the axis of rotation. The
mass imbalance in the rotor can be represented in terms of an unbalance mass mu at a
radial distance of e from the rotor centerline.
Per API 610, the maximum allowable unbalance for a machine of weight W operating at
N rpms is 4W/N oz-in. In the current analysis,
imbalance forces corresponding to twice the
allowable unbalance mass are applied at the
WWW.CTSSNET.NET CTSS
TECH brief
rotor midspan. The centrifugal reaction (unbalance) force
Fu is
Fu = 2.Mu .e. 2 = 2.(4W/N). 2
where is the rotor speed in rad/s. Note that the factor 2 doubles the maximum allowable reaction force per API 610. The
vertical and horizontal loads attain their
maximum value only once per cycle, and
are 90 out of phase with each other. In
order to estimate the maximum reaction
loads, harmonic loads of equal magnitude are applied along the horizontal and
vertical directions.
In the current analysis, since the depth
to bed rock is unknown, another model
[15] that is similar to model (iv) is used and
the formulae are listed in Table 6. Note
that, even if model (iv) is used assuming a
very large H (depth to bedrock) value, the
resulting foundation stiffnesses would differ only less than ~10%. Table 7 presents
the resulting stiffness coefficients. Soil
provides some damping, however, in the
current analysis only a very small damping ratio of 2.5% is used in order to determine the maximum response amplitudes.
Figure 6 shows the resulting peak vibration amplitudes versus frequency.
The maximum pump-motor response
amplitude of 0.06 ips (0 peak) is within acceptable range. The anchor bolt
stresses were also evaluated and found
to be acceptable. The current foundation
block design is acceptable for the overall dynamic performance. Note that, the
foundation block is nine times heavier
than the total weight of equipment installed on it. For foundation blocks with
weight less than four times the machinery weight, the possibility of the block
natural frequencies being present in the
operating range is high. It is thus recommended to always keep the foundation
block weight at least five times that of the
combined total equipment weight.
Conclusions and recommendations
Reciprocating and ron Table 4. Static tating machinery foundastiffnesses for tions must be adequaterigid foundation ly designed to handle
embedded
in large dynamic operating
homogeneous loads. A sound geotechsoil stratum over nical study, characterizing the soil profile and
bedrock [3]
properties, is paramount
for proper design and
installation of a process
equipment foundation.
The soil modulus, num-
2015 EDITION
WWW.CTSSNET.NET CTSS
TECH brief
References
[1] Smalley, A. J.; and Pantermuehl, P.J., 2006, Systems
Mounting Guidelines for Separable Reciprocating Compressors in Pipeline Service, Report to Gas Machinery Research Council, Dallas.
[2] Bhatia, K.G., 2006, Machine Foundation Design A
State of the Art, J. Structural Eng., 33(1), pp. 69-80.
2015 EDITION
WWW.CTSSNET.NET CTSS