Dynamic design of the foundation of

reciprocating machines for offshore

installations case study


by:
Dr.-Ing. Jan Steinhausen
KTTER Consulting Engineers KG
Rheine, Germany


5
th
Conference of the EFRC
March 21
th
/ 23
th
, 2007, Prague





Abstract:
The installation of large and heavy reciprocating machines on offshore constructions demands specific
requirements in the design of the foundation with respect to vibrations. Because the use of those piston
engines implies high dynamic loads at the substructure of offshore systems, special measures are
required. This paper shows how problems can be avoided by applying a detailed vibration engineering
design. An offshore project in the German North Sea involving the installation of three heavy
reciprocating mud pumps is taken as an example. It is demonstrated how vibration engineering aspects
can be considered within a project at the very beginning. The developed installation concept and its
specific technical implementation at the construction are part of this paper.


Page 2
1 Introduction
For the foundation of heavy reciprocating machines
at offshore platforms one has to keep an particular
eye on the dynamic loads which will be induced
into the substructure below the installation.
Normally, block foundations made of concrete
cannot be used because of their heavy weight and
shortage of space on site. This paper presents the
vibration engineering design for the example of the
foundation of three reciprocating mud pumps on an
offshore platform.

On Mittelplate (see figure 1), the artificial
platform for drilling and mining of petroleum in the
German wadden sea, a new drilling rig for
increasing the exploration capacity has been
planned since 2003. The drilling rig of type T150
has been built upon the existing skidding beams (of
the former drilling rig) since August 2005 and has
been set into operation in December 2005.










Figure 1: The drilling and mining platform
Mittelplate in the state with the old drilling rig.

The studies in the run-up to the project resulted in
minimizing the weight of the drilling rig which
moves on the skidding beams, see figure 3. For that
reason, the idea was born to relocate the mud tanks,
the pump installation, the silos for auxiliary
flushing and parts of the electrical installation
(SCR-unit) above the area of the Fangedamm.
The Fangedamm is placed in the west of the
platform. Its task is to absorb the water wave
energy during heavy weather and to protect the
Mittelplate. For the relocation of these units the
Fangedamm should be covered with a concrete
slab founded with piles into the wadden sea. Figure
2 presents the planned installation with the pump
house in the middle of the slab above the
Fangedamm. In order to minimize the
transmission of vibrations due to the operation of
the three mud pumps (power 2,200 hp each) to
other units and especially to the sensitive SCR-unit,
the vibration engineering aspects have been
considered from the very beginning of the
construction.
2 Feasibility analysis
For the basic design in the run-up to the project in
2003 first computations should investigate the basic
vibration behaviour of the pump installation.
2.1 Modelling
For the setup of the computational model (Finite-
element-method) the new platform construction
(slab) above the Fangedamm has been considered
as a plane construction consisting of prefabricated
elements of reinforced concrete. The inter-
connections between the elements are considered as
monolithic (rigid). The concrete slab is founded
with steel piles into the ground of the wadden sea.
Figure 3 shows the design of the slab and the pile
foundation with the positions of the piles below the
slab.
The basic concept of the vibration insulation of the
pump house provides a decoupling of the platform
areas in the north and in the south. This decoupling
is obtained by slits in the slab, see figure 9. Hence,
the area of the pump house foundation can be
treated separately. Therefore, the installation of the
reciprocating machines has been modelled for eight
rows of the piles between the axes C and F, see
figure 4.



























Figure 2: Application areas of the new platform
above the Fangedamm, left: plan, right: side
view.

Page 3
























Figure 3: Pile foundation design for the new
platform above the Fangedamm, top: section in
north-south direction, middle: top view of slab
(grey) with pile grid, bottom: section (in east-west
direction) of the slab and pile foundation.

For a first design of the pile foundation steel pipes
with a diameter of 500 mm, a thickness of 10 mm
and a length of 12.5 m have been considered. The
outer piles (protection piles, diameter 700 mm)
should protect the inner piles the load piles
against the wave impact on the Fangedamm.
They do not have any contact to the load piles.
Therefore, they are not involved in the
considerations here.
For a first evaluation of the foundation the (elastic)
bedding of the piles has been substituted by a fixed
length of 8.5 m from the top to the foot of the pile,
see figure 3. There are three piles planned in a row
in east-west direction. Each pile foot is considered
as totally fixed.

The mud pumps are triplex-type reciprocating
pumps, type 7.5x14 / 2200 HP, manufactured by
WIRTH. The weight mass of 46 tons of each
machine has been distributed (lumped mass)
according the fixation positions of the skid onto the
slab of the FEM-model. The speed of the pumps is
within the range from 0 rpm to 110 rpm. The
maximum dynamic load reactions arise at the 1
st

order due to the free mass moments of the
machines (1
st
order: 0 1.83 Hz).












Figure 4: Positions of the mud pumps and
alignment of the piles, first design.

2.2 Modal analysis
A modal analysis has been carried out for a primary
assessment of the dynamic behaviour of the first
construction design. The results indicate the three
lowest natural frequencies are within the 1
st
order
frequency range (0 1.83 Hz), thus in the range of
the largest reaction load. Figure 5 presents the
natural mode shapes up to about 20 Hz. The 1
st
and
2
nd
mode shape indicate each primarily a translation
of the concrete slab, the 3
rd
indicates a torsional
movement (with vertical axis). The 4
th
mode shape
indicates a deformation of the concrete slab itself.















Figure 5: Natural mode shapes of the first design
of the pump installation, undeformed: unfilled grid,
deformed: filled grid.

2.3 Response analysis at operating
conditions
For a first assessment of the forced vibration level
during operation of the pumps the exciting mass
moments within the 1
st
order frequency range have
been obtained from the data specified by the
manufacturer. The exciting moments are applied on
the model with respect to the location of the crank
shaft of each pump, see figure 6. The oscillating
moments represent a harmonic excitation with
direction normal to the slab. The direction of the
rotating moments rotates with the rotating
frequency in the y-z plane. All moments are acting

Page 4
synchronously (no phase shift). Thus, the worst
case is considered, when all pumps are in
operation. The critical damping ratio is set to 0.025
for the complete frequency range. The simulated
results show that the vibration velocity on the
concrete slab achieves in case of resonance up to
70 mm/s (target is: 3 mm/s rms) in horizontal
direction.













Figure 6: Applied moments for first response
analysis.

In order to avoid exceeding resonance vibrations
within the frequency range of the 1
st
order, the
target has been to shift the 1
st
(lowest) natural
frequency of the complete construction above 3 Hz.
Therefore, the slab of the pump house should be
connected with a steel-framework to the massive
drilling cellar (reinforced concrete), see figure 7.
First calculations for this 2
nd
design indicated the
lowest natural frequency far above the critical
exciting range of the 1
st
order of the pump speed.
Thus, exceeding resonance vibrations are
eliminated within this frequency range.















Figure 7: FEM-model of the 2
nd
design of
reciprocating pump foundation, connection of the
slab of the pump house to the massive drilling
cellar by framework construction.

However, the natural frequencies of the
construction (2
nd
design) will be excited by the load
reactions of the 2
nd
, 3
rd
and other higher orders. For
a more accurate estimation of the expected
vibration level (during operation of the pumps) the
exciting reaction forces have been determined by
measurements on the test bench at the
manufacturer. For this, a single of the reciprocating
pump has been set up on especially prepared steel
profiles (beams) on which strain-gauges have been
applied. The forces transmitted to the ground have
been measured at selected locations of the skid. The
forces have been determined in x-, y- and z-
direction. The maximum forces have been detected
at 110 rpm with maximum load (pressure). For
example, figure 8 presents the time-frequency
spectrum of the force in y-direction at a selected
position on the skid during run up (0 110 rpm),
steady state (110 rpm) and run down (110 0 rpm).
Local resonances of the test bench setup can be
seen here, too. The relevant frequency range goes
up to about 30 Hz. From the measured forces (at
different locations) the exciting forces and
moments of the reciprocating machine for the
further calculations have been derived.
The reaction forces and moments have been applied
to the FEM-model of the 2
nd
design. The results of
the response analysis show that even by the higher
order forces (moments) exceeding vibrations of the
slab can be excited (resonance cases). To take this
into account, the resonance vibrations should be
decreased by special damper elements and should
be integrated in the framework design, see
figure 11. The selected dampers are acting as
viscous dampers [3].












Figure 8: Time-frequency spectrum, example of
force in y-direction (at a selected position) during
variation of the pump speed: 0 - 110 - 0 rpm.

2.4 Result of the feasibility analysis
The basic computations for the mud pump
foundation reveal that the critical exciting
frequency can be shifted above the critical exciting
frequency range (1
st
order) by stiffening the
structure with a steel-framework connected to the
drilling cellar. The damping of the expected higher
frequency vibrations (structural resonances) will be
provided by special dampers. All in all, from the
vibration point of view the installation of the
reciprocating pumps on the slab above the
Fangedamm is feasible.
1
X
Y
Z

APR 5 2004
19: 14: 33
RWEDEA_V100
ELEMENTS
REAL NUM
RWE_DEA032_F_AS: MP12_DMS2
0
50
100
150
200
250
300
350
s
0 5 10 15 20 25 30 35 40 45 50
Hz
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
kN

Page 5

3 Detailed layout Structural
dynamic calculations
The feasibility study was followed by the detailed
design of the framework and the damping elements.
After completion of the work in the underground at
the construction side, the following changes have
been found compared to the existing (fundamentals
of the calculation) model:

1. Strongly varying lengths of the piles and thus
of the penetration depths into the supporting
(ground) layer.
2. A change in bedding functions
1)
depending on
the penetration depth of the piles.
3. A change in position of several rows of pillars
made the pillar distance no longer equidistant
in north-south direction.

3.1 Measurements to obtain the
properties of the construction
To verify the model for the bedding function of the
pile foundation, dynamic measurements of the
construction properties have been carried out after
completion of the bare concrete platform for the
pumps (i.e. without the pumps and other
constructions on the platform). Following the
measurements, the calculation model has been
adapted.
Preceding calculations with the original
assumptions for the bedding functions have showed
that the lower three mode shapes of the concrete
platform (placed on piles) can be expected in the
frequency range between 2.5 and 3 Hz. Because of
this a special unbalance exciter was used during the
measurements that excited with an adequate force
at frequencies as low as 1 Hz. The filled slits for
vibration decoupling at the north and south edge of
the concrete platform for the pumps (see figure 9)
were emptied. No contact bridges existed with
adjoining platforms in the north and the south. The
slab for the pumps could thus move freely in
horizontal plane.
The FEM-model has been adapted to the measured
natural frequencies and damping. Apart from the
individual lengths of the piles and their penetration
depths, the framework under the concrete platform
has also been considered in this adaptation. The
adaptation of the bedding function has been carried
out on the following parameters:


1)
bedding function =change of the modulus of
elasticity of the soil-layer with depth

1. dynamic bedding modulus, K
S, dyn

2. depth of (the top of) the supporting ground
layer, h
BA

3. damping in the bedding function, D
S

For the location of the natural frequencies primarily
the first two parameters are of importance.











Figure 9: Location of the measurement points S1 to
S4 and the locations of the unbalance exciter A1
and A2 (top view of the concrete platform).

3.2 FEM-model of the foundation
For the final model of the construction above the
Fangedamm in the vicinity of the pumps eight
rows of piles in north-south direction have been
taken into account (see figure 10). The concrete
plate has been assumed to be monolithic, i.e. with
rigid interconnections. The connection of the steel
pile heads to the concrete slab has also been
assumed to be rigid. The masses of the three
Triplex reciprocating pumps have been distributed
as four masses per pump (so in total 12 masses) at a
height equal to the centre of gravity of the pumps,
i.e. at 1.25 m above the centre of gravity plane of
the concrete platform (thickness 30 cm).

Figure 10 shows the optimized FEM-model for the
pump house including the framework connecting to
the drilling cellar and the trusses for the
implementation of the dampers. The framework
that is connected to the construction above the
drilling cellar has been adapted to the real north-
south orientation of the pile rows. From several
calculations, it has been found that the most
favourable position for damping, i.e. the dampers,
is under the overhang of the concrete platform.

3.3 Optimization of the framework
To reduce the resonances of the components above
the 1
st
order of the pump speed, dampers have been
implemented with almost viscous damping
properties. From parameter studies followed a
damping specification for the dampers of
d =500 kNs/m at 16 Hz.



Page 6
Within the scope of optimizing the framework
several configurations with different stiffness of the
framework and number of dampers have been
investigated. It has generally shown that for a
framework with a higher stiffness the resonance
amplitudes have been smaller compared to that of a
framework with less stiffness with the same amount
of dampers.

The implementation of 8 dampers has been
favourable for obtaining a targeted maximum
vibration velocity amplitude of 3 mm/s (rms) at the
concrete surface under operating conditions. The
calculated mode shapes of the optimized
installation are depicted in figure 13. The 1
st
mode
shape can be seen primarily as a movement of the
concrete platform in x-direction parallel to the
drilling cellar. The 2
nd
mode shape is a rocking and
translational mode of the concrete platform into the
drilling cellar. The 3
rd
mode shape is more a torsion
mode of the concrete platform. The higher mode
shapes are vibrations of the framework.

The results of the vibrational analysis at the
operating conditions have showed that with three
pumps operating in parallel the targeted vibration
velocity of 3 mm/s (rms) is reached. As an
example, figure 11 shows the time-frequency-
amplitude spectra of the calculated vibrational
velocity in y-direction at node 244 (location, see
figure 10), where the highest vibration level occurs.
The figure shows a resonance at 13 Hz that
occurred at a simulated sweep of pump rotation
speed up to 110 rpm.












Figure 10: FEM-model of the optimized and
realised framework with the 8 damping elements
under the concrete platform (slab).


















Figure 11: Time-frequency-amplitude spectrum of
the calculated vibrational velocity in y-direction at
node 244 (max. 3.1 mm/s rms) excitation in y-
direction, rotational sweep 90 110 rpm.

4 Verification measurements
After completion and start-up of the new drilling
rig, the vibrational situation during operation of the
pumps was checked by measurements. The
vibrations of the construction were measured at
several locations at the slab of the pump house and
at the neighbouring slabs (north: container, south:
tanks). The measurements have showed that the
structural vibrations (movements of the slabs) are
within the limit of 3 mm/s rms at all operating
conditions of the pumps. The slabs north and south
of the pump house showed a significantly lower
vibration level compared to the slab of the pump
house. An excitation of higher resonances at
neighbouring slabs could not be identified. Thus,
the slits (with filling) are sufficient as vibrational
decoupling. All in all, the function of the
framework with dampers proved to be very
satisfying.















Figure 12: Framework after completion.






0
5
10
15
20
25
30
35
40
45
s
0 10 20 30 40 50
Hz
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
mm/s

Page 7






















Figure 13: Undeformed model (top left side) and
natural mode shapes, minimum displacement: blue,
maximum displacement: red, natural frequencies
(from left to right and from top to bottom) 7.1 Hz
(top right), 13.0 Hz, 15.7 Hz, 17.8 Hz, 19.6 Hz.



5 References
[1] Riehwein, W.; Lesny, K.; Wiemann, J .:
Nachweise und Sicherheitskonzepte fr die
Grndung von Offshore-Windenergieanlagen
in der Deutschen Bucht, 2. Symposium
Offshore-Windenergie, Bau- und umwelt-
technische Aspekte, September 2002,
Hannover
[2] Schaumann, P.; Seidel, M.: Eigenschwing-
verhalten von Windenergieanlagen Berech-
nungen und Messungen, Vortrag zur
DEWEK 2000, Wilhelmshaven
[3] Gerb Schwingungsisolierungen, 11. Auflage,
Berlin, 2002