28 - 02 Report

TU Delft
Passive Suction
under Mud Mats
Model – Testing - Validating
Christiaan den Hertog

3-6-2017
Administrative Data
Author:
Christiaan den Hertog, Bsc

1509969
[email protected]
Committee
Prof. Dr. Ir. C van Rhee

[email protected]
Faculty of Mechanical Materials and Maritime Engineering
Dr. Ir. S.A. Miedema

[email protected]
Dr. Ir. A.M. Talmon

[email protected]
Dr. Ir. M. van Damme

[email protected]
Faculty of Civil Engineering
Abstract
The subject of this master thesis is related to the passive suction under mud mats. Mud mats are
used in the offshore industry to prevent structures from sinking in the soil after installation. If for
some reason a structure has to be lifted, on the one side due to an installation error, on the other
side due to removal the force needed to lift the structure sometimes exceeds the total submerged
weight. This report contains a Literature survey conducted regarding the above subjects. This report
also shows the design of a test setup and procedure for lifting a plate from a sand bed. In the end the
report follows the modelling of an analytical model which is validated using the test data
The focus of this report lies with finding out what basic parameters dominate this force and what the
influence of these parameters are during lifting is sand. Experiments will be conducted to test these
parameters. In the end these parameters will be used to develop a simple analytical model to predict
the order of the breakout force. The model will then be validated with the experiments. The main
research question is:
How are the breakout force and breakout time influenced by the permeability of the sand and the
lifting force? Can an analytical model, that uses these parameters, predicts the lifting force within a
certain margin?
Literature states that no breakout forces are to be expected due to the large permeability of sand but
from cutting theories it is known that under pressures exist around the blade tip, especially for the
smaller grain sizes. A test setup was built to study the lifting process and measure the pressure under
the plate and his displacement for a given load. The tests were performed in two different sands
(Silverbond and Geba Weiss) for a range of different loads with two different plates: A 2-Dimensional
setup and a 3-Dimensional setup. An analytical model was created to predict the lifting force for a
given permeability and upward velocity of the plate. In the end the model was validated with the test
data.
The test data showed that for the 2-Dimensional case the pressure profile was of a rectangular
shape. A factor 10 in breakout time between the sands can be observed. The 2-Dimensional model
gives a good estimation of the lifting load in Silverbond sand when using velocities from the
beginning of the lifting process. For the Geba sand, after adjusting the length of the flow paths, the
model also gives a good fit.
The experiments with the 3-Dimensional plate showed that the pressure profile under the plate is of
a rectangular shape with steep slopes towards the edge of the plate. Nothing can be concluded
about a relation between the breakout time and the permeability for the same load between the two
different sands due to an inertia dominated process.
The 3-Dimensional model predict the lifting force accurately for the Silverbond sand using the
velocities from the beginning of the lifting process. For the Geba sand the permeability had to be
scaled to give a good fit. This is because of model assumptions and using flow paths. Adjusting the
tune factor did not give satisfactory results.
Contents
Nomenclature............................................................................................................................................
List of Figures.............................................................................................................................................
List of Tables ..............................................................................................................................................
1 Introduction ..................................................................................................................................... 1
1.1 Problem Description ................................................................................................................ 1
1.2 Purpose of Research ................................................................................................................ 2
2 Theory.............................................................................................................................................. 4
2.1 Earlier Research ....................................................................................................................... 4
2.2 Parallel Resistor method ......................................................................................................... 5
2.3 Conclusions.............................................................................................................................. 8
3 Testing ............................................................................................................................................. 9
3.1 Introduction ............................................................................................................................. 9
3.2 Test setup ................................................................................................................................ 9
3.3 Test Parameters .................................................................................................................... 12
3.4 Test Procedure ...................................................................................................................... 14
4 General Test Results ...................................................................................................................... 16
4.1 Introduction ........................................................................................................................... 16
4.2 Net loads................................................................................................................................ 16
4.3 Measurement Analysis .......................................................................................................... 17
4.4 Conclusions............................................................................................................................ 21
5 2 Dimensional Results ................................................................................................................... 22
5.1 Expectations .......................................................................................................................... 22
5.2 Time of breakout ................................................................................................................... 22
5.3 Mean Displacement............................................................................................................... 24
5.4 Mean Pressures ..................................................................................................................... 26
5.5 Comparing the Two Sands ..................................................................................................... 28
5.6 Conclusions............................................................................................................................ 29
6 3 Dimensional Results ................................................................................................................... 30
6.1 Expectations .......................................................................................................................... 30
6.2 Mean time of breakout ......................................................................................................... 30
6.3 Mean Displacement............................................................................................................... 32
6.4 Mean Pressure....................................................................................................................... 34
6.5 Comparing the Two Sands. .................................................................................................... 37
6.6 Consolidation......................................................................................................................... 38
6.7 Conclusion ............................................................................................................................. 40
7 Analytical Model ............................................................................................................................ 41
7.1 2D Analytical Model .............................................................................................................. 41
7.2 3D Analytical Model .............................................................................................................. 43
7.3 Expectations .......................................................................................................................... 47
8 Validation ...................................................................................................................................... 50
8.1 Introduction ........................................................................................................................... 50
8.2 2-Dimensional Comparisons.................................................................................................. 50
8.3 3-Dimensional Comparisons .................................................................................................. 54
8.4 Scaling .................................................................................................................................... 58
8.5 Conclusions............................................................................................................................ 60
9 Conclusions and Recommendations ............................................................................................. 62
9.1 2-Dimensional Conclusions ................................................................................................... 62
9.2 3-Dimensional Conclusions ................................................................................................... 62
Bibliography........................................................................................................................................... 64
Appendix A Sensor Data ................................................................................................................... 66
Appendix B Permeability Tests ......................................................................................................... 67
Appendix C Sieve Analysis. ............................................................................................................... 69
Appendix D Layer Thickness.............................................................................................................. 70
Appendix E 2-Dimensional Test Results ........................................................................................... 71
Appendix F 3-Dimensional Test Results ........................................................................................... 85
Nomenclature
𝑞 𝑆𝑝𝑒𝑐𝑖𝑓𝑖𝑐 𝐹𝑙𝑜𝑤 𝑚/𝑠
∆𝑝 𝑃𝑟𝑒𝑠𝑠𝑢𝑟𝑒 𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑃𝑎
𝑘 𝑃𝑒𝑟𝑚𝑒𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑚/𝑠
𝑘𝑚𝑎𝑥 𝑀𝑎𝑥𝑖𝑚𝑢𝑚 𝑝𝑒𝑟𝑚𝑒𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑚/𝑠
𝑘𝑚𝑖𝑛 𝑀𝑖𝑛𝑖𝑚𝑢𝑚 𝑃𝑒𝑟𝑚𝑒𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑚/𝑠
∆𝑠 𝐹𝑙𝑜𝑤 𝑝𝑎𝑡ℎ 𝑙𝑒𝑛𝑔𝑡ℎ 𝑚
𝑑ℎ 𝐻𝑦𝑑𝑟𝑎𝑢𝑙𝑖𝑐 ℎ𝑒𝑎𝑑 𝑚
𝑑𝑙 𝐿𝑒𝑛𝑔𝑡ℎ 𝑚
𝜌𝑤 𝐷𝑒𝑛𝑠𝑖𝑡𝑦 𝑊𝑎𝑡𝑒𝑟 𝑘𝑔/𝑚3
𝑔 𝐺𝑟𝑎𝑣𝑖𝑡𝑎𝑡𝑖𝑜𝑛𝑎𝑙 𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑚/𝑠 2
𝜖 𝐷𝑖𝑙𝑎𝑡𝑎𝑡𝑖𝑜𝑛 −
𝑣𝑐 𝐶𝑢𝑡𝑡𝑖𝑛𝑔 𝑉𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑚/𝑠
𝛽 𝑆ℎ𝑒𝑎𝑟 𝐴𝑛𝑔𝑙𝑒 𝑅𝑎𝑑𝑖𝑎𝑛𝑠
𝑅𝑥 𝑅𝑒𝑠𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑜𝑓 𝑓𝑙𝑜𝑤 𝑙𝑖𝑛𝑒 𝑥 𝑠
𝑅𝑡 𝑇𝑜𝑡𝑎𝑙 𝑟𝑒𝑠𝑖𝑠𝑡𝑎𝑛𝑐𝑒 𝑜𝑓 𝑎𝑙𝑙 𝑓𝑙𝑜𝑤 𝑙𝑖𝑛𝑒𝑠 𝑠
𝑝 𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝑝𝑟𝑒𝑠𝑠𝑢𝑟𝑒 𝑜𝑣𝑒𝑟 𝑠ℎ𝑒𝑎𝑟 𝑧𝑜𝑛𝑒 𝑃𝑎
𝑑𝑛 𝑆𝑡𝑒𝑝 𝑠𝑖𝑧𝑒 𝑜𝑣𝑒𝑟 𝑠ℎ𝑒𝑎𝑟 𝑧𝑜𝑛𝑒 −
𝑝𝑖 𝐼𝑛𝑡𝑒𝑔𝑟𝑎𝑡𝑒𝑑 𝑝𝑟𝑒𝑠𝑠𝑢𝑟𝑒 𝑜𝑣𝑒𝑟 𝑠ℎ𝑒𝑎𝑟 𝑧𝑜𝑛𝑒 𝑃𝑎
𝐷50 50% 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑠𝑖𝑧𝑒 𝑑𝑖𝑠𝑡𝑟𝑖𝑏𝑢𝑡𝑖𝑜𝑛 −
𝐷15 15% 𝑝𝑎𝑟𝑡𝑖𝑐𝑙𝑒 𝑠𝑖𝑧𝑒 𝑑𝑖𝑠𝑡𝑟𝑖𝑏𝑢𝑡𝑖𝑜𝑛 −
𝑑𝑍 𝐻𝑒𝑖𝑔𝑡ℎ 𝑖𝑛 𝑣𝑒𝑟𝑡𝑖𝑐𝑎𝑙 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑢𝑛𝑑𝑒𝑟 𝑝𝑙𝑎𝑡𝑒 𝑚
𝑑𝑡 𝑡𝑖𝑚𝑒 𝑠
𝑑𝐿 𝑆𝑡𝑒𝑝 𝑠𝑖𝑧𝑒 𝑝𝑖𝑒 𝑠𝑙𝑖𝑐𝑒 −
𝑑𝑅1 𝑊𝑖𝑑𝑡ℎ 𝑜𝑓 𝑃𝑖𝑒 𝑠𝑙𝑖𝑐𝑒 𝑜𝑛 𝑡ℎ𝑒 𝑝𝑙𝑎𝑡𝑒 𝑚
𝑑𝑅2 𝑊𝑖𝑑𝑡ℎ 𝑜𝑓 𝑃𝑖𝑒 𝑠𝑙𝑖𝑐𝑒 𝑜𝑢𝑡𝑠𝑖𝑑𝑒 𝑡ℎ𝑒 𝑝𝑙𝑎𝑡𝑒 𝑚
𝑟 𝑟𝑎𝑑𝑖𝑢𝑠 𝑚
𝑄 𝑉𝑜𝑙𝑢𝑚𝑒𝑡𝑟𝑖𝑐 𝑓𝑙𝑜𝑤 𝑟𝑎𝑡𝑒 𝑚3 /𝑠
𝑣𝑝 𝑈𝑝𝑤𝑎𝑟𝑑 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 𝑝𝑙𝑎𝑡𝑒 𝑚/𝑠
𝑆 𝑆𝑢𝑟𝑓𝑎𝑐𝑒 𝑏𝑒𝑛𝑒𝑎𝑡ℎ 𝑝𝑙𝑎𝑡𝑒 𝑎𝑡 𝑑𝑅1 𝑚
𝑎 𝐴𝑟𝑐𝑙𝑒𝑛𝑔𝑡ℎ 𝑓𝑙𝑜𝑤 𝑙𝑖𝑛𝑒 𝑚
𝑛 𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑓𝑙𝑜𝑤 𝑙𝑖𝑛𝑒 𝑠𝑡𝑒𝑝𝑠 −
𝑛𝑠𝑙𝑖𝑐𝑒𝑠 𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑠𝑙𝑖𝑐𝑒𝑠 −
𝑑𝑃𝑠𝑙𝑖𝑐𝑒 𝑃𝑟𝑒𝑠𝑠𝑢𝑟𝑒 𝑑𝑟𝑜𝑝 𝑜𝑣𝑒𝑟 𝑠𝑙𝑖𝑐𝑒 𝑃𝑎
𝑑𝑝 𝐷𝑖𝑠𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡 𝑑𝑢𝑟𝑖𝑛𝑔 𝑏𝑟𝑒𝑎𝑘𝑜𝑢𝑡 𝑚
𝐸𝑣 𝑣𝑜𝑙𝑢𝑚𝑒 𝑐𝑟𝑒𝑎𝑡𝑒𝑑 𝑢𝑛𝑑𝑒𝑟 𝑝𝑙𝑎𝑡𝑒 𝑏𝑦 𝑙𝑖𝑓𝑡𝑖𝑛𝑔 𝑚3
𝑆𝑝 𝑠𝑢𝑟𝑓𝑎𝑐𝑒 𝑎𝑟𝑒𝑎 𝑜𝑓 𝑡ℎ𝑒 𝑝𝑙𝑎𝑡𝑒 𝑚2
𝑁𝑚𝑎𝑥 𝑀𝑎𝑥𝑖𝑚𝑢𝑚 𝑝𝑜𝑟𝑜𝑠𝑖𝑡𝑦 𝑜𝑓 𝑡ℎ𝑒 𝑠𝑎𝑛𝑑 −
𝑁𝑚𝑖𝑛 𝑀𝑖𝑛𝑖𝑚𝑢𝑚 𝑝𝑜𝑟𝑜𝑠𝑖𝑡𝑦 𝑜𝑓 𝑡ℎ𝑒 𝑠𝑎𝑛𝑑 −
𝐼𝑣 𝐼𝑛𝑖𝑡𝑖𝑎𝑙 𝑣𝑜𝑙𝑢𝑚𝑒 𝑖𝑛𝑓𝑙𝑢𝑒𝑛𝑐𝑒𝑑 𝑧𝑜𝑛𝑒 𝑚3
𝐿 𝐿𝑎𝑦𝑒𝑟 𝑡ℎ𝑖𝑐𝑘𝑛𝑒𝑠𝑠 𝑜𝑓 𝑖𝑛𝑓𝑙𝑢𝑒𝑛𝑐𝑒𝑑 𝑧𝑜𝑛𝑒 𝑚
List of Figures
Figure 1.1 – Model of the Under pressures versus the lifting loads ....................................................... 1
Figure 2.1 – The Equipotential lines and the flowlines for a cutting case............................................... 6
Figure 2.2 – The Water flow towards the cutting zone. .......................................................................... 7
Figure 3.1 – Diagram of the test setup .................................................................................................. 10
Figure 3.2 – Layout of the location of the pressure sensors on the 2-Dimensional plate .................... 11
Figure 3.3 – Layout of the location of the pressure sensors on the 3-Dimensional plate ................... 11
Figure 3.4 – Photos of the test setup and equipment........................................................................... 12
Figure 4.1 – An example of the unfiltered velocity. .............................................................................. 18
Figure 4.2 – The load and the displacement versus the time for a 3-Dimensional test in Silverbond
sand, (3D-S-70N-1). ............................................................................................................................... 18
Figure 4.3 – The pressure and the velocity versus the time for a 3-Dimensional test in Silverbond
sand, (3D-S-70N-1). Note the filtered velocity ..................................................................................... 19
Figure 4.4 – A failed 2-Dimensional test, as can be seen the breakout time is very short, (2D-S-150N-
3)............................................................................................................................................................ 20
Figure 4.5 – A failed 2-Dimensional test, the pressure sensors registered a lot of dynamic results, (2D-
S-150N-3). .............................................................................................................................................. 20
Figure 5.1 – The time of breakout for the Silverbond sand, the spread of the different tests is visible.
Note that it looks like the trend is having an Asymptotic value, beware that it is only a trend line. ... 23
Figure 5.2 – The time of breakout for the Geba sand. Here the trend is going to zero, again it is only a
trend line and it is not suspected to cross the Net load axis. ............................................................... 24
Figure 5.3 - The mean displacement of for the tests performed with the Silverbond sand. ............... 25
Figure 5.4 - The mean displacement of for the tests performed with the Geba sand. ........................ 25
Figure 5.5 – The mean plate pressure versus the measured pressure for the Silverbond sand.
Especially for the 124 Newton load the difference is big...................................................................... 26
Figure 5.6 The mean plate pressure versus the measured pressure for the Geba sand. Here the
differences occur with almost every test. ............................................................................................. 27
Figure 5.7 – The Measured Pressure versus the mean plate Pressure. ................................................ 27
Figure 5.8 – The results of DP4 versus the mean plate pressure, DP4 carries the information about
the pressure profile under the plate. .................................................................................................... 28
Figure 5.9 – The differences in breakout for the different sands ......................................................... 29
Figure 6.1 – The mean time of breakout for the Silverbond sand, notice the descending trend. ........ 31
Figure 6.2 – The mean time of breakout for the Geba sand, notice that the difference between the 5
Newton load and the 13N load is large. ................................................................................................ 32
Figure 6.3 - The mean displacement of for the tests performed with the Silverbond sand. ............... 33
Figure 6.4 - The mean displacement of for the tests performed with the Geba sand. ........................ 33
Figure 6.5 – The mean plate pressure versus the measured pressure from DP1 for the Silverbond
sand. Notice again the difference in measured pressure and the mean plate pressure. ..................... 34
Figure 6.6 – For all the test the mean pressure of DP1 is higher than the other 2 sensors. ................ 35
Figure 6.7 – The mean plate pressure versus the measured pressure for the Geba sand. .................. 35
Figure 6.8 – For all the test the mean pressure of DP1 is higher than the other 2 sensors. ................ 36
Figure 6.9 – Comparison between the mean plate pressures and the measured pressures. .............. 36
Figure 6.10 – Time of breakout for the 2 sands. Notice the huge difference. ...................................... 37
Figure 6.11 – Underwater pre-loading .................................................................................................. 38
Figure 6.12 – Consolidation test 1 ......................................................................................................... 39
Figure 6.13 – Consolidation test 2 ......................................................................................................... 40
Figure 7.1 – Flow lines underneath the plate........................................................................................ 41
Figure 7.2 – The plate seen from above, now the pie slice model becomes clear. .............................. 44
Figure 7.3 - Flow paths. ........................................................................................................................ 45
Figure 7.4 – The tuning factor of the flow paths. .................................................................................. 46
Figure 7.5 – The limitation of the model. .............................................................................................. 47
Figure 7.6 – Linear pressure development under the plate. ................................................................. 47
Figure 7.7 – The force development under half the plate. ................................................................... 48
Figure 7.8 – The pressure development under the total plate ............................................................. 48
Figure 7.9 – Corrected pressure development under the plate............................................................ 49
Figure 8.1 – 2D - Model comparison for the Silverbond sand, the trend is visible but the model over
estimates. .............................................................................................................................................. 50
Figure 8.2 – First predictions of the model, a large over estimation. ................................................... 51
Figure 8.3 – The point where the velocity is derived from the test data. ............................................. 52
Figure 8.4 – The model versus the test data with the new velocities for the Silverbond sand. ........... 52
Figure 8.5 – The model versus the test date with the new velocities for the Geba sand. .................... 53
Figure 8.6 – The model versus the test data using a tune factor for the flow path length. ................. 54
Figure 8.7 – The measured data set out versus the model data for the Silverbond sand with the mean
velocities over the complete process. ................................................................................................... 55
Figure 8.8 - The model versus the test data with the velocities from the beginning of the process for
the Silverbond sand. Note that the permeability stays the same. ........................................................ 56
Figure 8.9 – The predictions of the model for the Geba sand using the mean velocities over the
complete lifting process. ....................................................................................................................... 57
Figure 8.10 – - The model versus the test data with the velocities from the beginning of the process
for the Geba sand .................................................................................................................................. 57
List of Tables
Table 3.1 – Load Parameters for the Silverbond sand. ......................................................................... 13

Table 3.2 – Load Parameters for the Geba Sand ................................................................................... 14
Table 3.3 – Test matrix for the 2-Dimensional test in the Silverbond sand. ......................................... 14
Table 3.4 – Test matrix for the 2-Dimensional test in the Geba sand................................................... 15
Table 3.5 – Test matrix for the 3-Dimensional test in the Silverbond sand. ......................................... 15
Table 3.6 – Test matrix for the 3-Dimensional test in the Geba sand................................................... 15
Table 4.1 – The net load values for the Silverbond sand in the 2D situation........................................ 16
Table 4.2 – The net load values for the Geba sand in the 2D situation. ............................................... 17
Table 4.3 – The net load values for the Silverbond sand in the 3D situation........................................ 17
Table 4.4 – The net load values for the Geba sand in the 3D situation. ............................................... 17
Table 8.1 – The mean layer thickness per test setup and sand type. In the right column, the layer
thickness is defined as a percentage of the 2-D plate width and the 3-D plate diameter.................... 59
1 Introduction
1.1 Problem Description

As a big part of the extracted oil comes from offshore areas, large structures are needed to create a
horizontal, safe, working space above the water level for oil drilling. One of the structures is the so-
called jacket which consist of several hollow pipes welded together. A commonly used method for
attaching big structures to the sea floor is the use of long piles which are driven into the soil. In the
case of a jacket the piles are driven through pile sleeves which provide guidance.
When jackets are installed on the sea bottom several problems can occur. Apart from installation
errors such as for instance lifting errors, one of these problems is related to the situation below the
water surface. Most offshore areas are formed of soft, unconsolidated soil on which it is difficult to
perform the pile driving operations. The jacket has the tendency to sink in the soil on the side where
piles are driven into the bottom.
To prevent the jacket from falling over mud mats are used. These mats support and distribute the
weight of offshore structures on the soft soil. In this way, the structure is able to stand on the sea
floor.
When the structure is finally standing on the sea floor the weight of the structure pushes the water
out of the soil which creates a more consolidated soil state at the interface between the mud mat
and the soil. Lifting of the structure after it has been lowered now means that not only the weight
but also the (passive) suction of the soil has to be overcome. According to (Liam Finn, 1972): If the
object is embedded and an attempt is made to lift it adhesion will develop under the base of the
object. The difference between the force now required to lift the object and the submerged weight is
the breakout force (𝐹𝐿𝑖𝑓𝑡 ), see Figure 1.1. With breakout defined as the moment the plate comes
loose from the bottom.
Figure 1.1 – Model of the Under pressures versus the lifting loads
The passive suction occurs as water has to flow into the pores under the mat during lifting and
encounters a certain resistance by doing so.
1
Passive suction can occur in the following situations:
- Decommissioning of the structure: When for example a jacket is removed by lifting it

from the bottom a certain force is needed to overcome the suction of the soil on the
mud mat.
- Installation of the structure: During an installation error when for example the
orientation is wrong or the structure is not levelled correctly.
This thesis is focused on the modelling of the under pressures under a mud mat using an analytical
approach. The model is supported by using a test setup to generate data to validate the model.
1.2 Purpose of Research

This research is aimed on to the prediction of the force needed to overcome the passive suction
between soil and the mud mat using an analytical approach and to get insight in the physical
parameters behind this process. During the literature survey a lot of information regarding the topic
was found but most of it is focused on other subjects of the breakout phenomena. As a dredging
student, an analytical model used for sand cutting was found and the question arose if this model
could be modified to predict the breakout force and time. Afterwards this model could be validated
with test data.
During the research the following objectives are encountered:
- Literature study: Previous research regarding this subject has been done. However,
on different subjects and structures.
- Modelling: the formulation of an analytical model. Use a proven model from
dredging cutting theory and adapt this to model the under pressures under mud
mats.
- Testing: Using a self-developed test setup to get insight in the governing variables
and procedures of lifting a plate from a sand bed.
- Validation: Use the test data for validating the analytical model. Compare the test
data with the analytical model.
The research itself is dependent on time and budget therefore the research itself has boundaries:
- Only breakout phenomena in sand will be investigated. Breakout phenomena in clay have
been extensively researched. Sand however not, early researchers (Liu, 1969) even stated
that lifting in sand does not lead to a breakout force. But, from cutting theory (Miedema,
2014) it is known that small grained sand can lead to under pressures. Therefore, it is
interesting to investigate the suction problem for sand.
- The test structures are simplified; this is done to give easy insight in the governing
parameters.
- Testing will be done with two different sands, Silverbond of 50 µ and Geba Weis of 125 µm
- Only vertical upright testing will be done, lifting under an angle is a totally different subject
involving other parameters.
The main research question to be answered is as follows:
How are the breakout force and breakout time influenced by the permeability of the sand and the
lifting force? Can an analytical model, that uses these parameters, predicts the lifting force within a
certain margin?
2
To help answering the main research question the following sub questions have been composed:
- Can the Parallel Resistor model from cutting theory (Miedema, 2014) be adapted to use in
Breakout Phenomena?
- How does the permeability scale over different sorts of sand using the same breakout load?
- Does pre-loading the plate result in a difference in breakout time in Sand? (Liu, 1969) states
that for clay the longer the settlement time the longer the breakout time. For clay soils the
situation is different, pore water pressures do not dissipate that easy as in sand. To help
investigating the following sub question can be proposed:
o Can pre-loading the structure change the permeability of the soil when placed onto
the sand bottom?
- How does the breakout time scale when increasing the load, and thus the weight of the
structure?
- What is the displacement a plate needs before breakout?
3
2 Theory
2.1 Earlier Research

Earlier research concerning the so-called break out forces has been done. The first to do some testing
is (Muga, 1968) with the NCEL, Naval Civil Engineering Laboratory. Muga conducted theoretical and
experimental studies in order to derive an empirical formula to estimate the force required to lift an
object of a certain size from the sea bottom. His formula is dependent on the contact area of the
structure, a time allowed for the breakout and a supporting pressure of the soil (recognized as the
soil bearing capacity by Therzaghi).
(Vesic, 1969) studied the factors affecting the magnitude of the breakout force, he gave some good
recommendations for further research.
(Liu, 1969) also in a NCEL study stated that the breakout force is more dependent on the embedded
depth of the object. He was able to specify the breakout force within a certain time frame but could
not accurately predict the time of breakout. He also stated that the magnitude of the breakout force
is dominated by the soil cohesion, area of structure time of breakout and the permeability of the soil.
(Lee, 1973) continued with research for the NCEL, he divided the breakout of partially embedded
objects from cohesive seafloors into immediate and long term breakout problems. The conclusion
was the breakout will eventually occur under any net uplift force, however much time may be
involved. No existing theoretical models were applicable; the estimations of the breakout force gave
a range of plus or minus one hundred percent.
(Liam Finn, 1972) used in the basis the same formula as Muga and added some factors for the
submerged weight of the base and soil. Also, the added the bearing capacity factor which is a
function of the shape of the base.
(Rapoport V, Young A.G, 1983) stated that the empirical relations found before are easy to use but
limited to a certain soil type. they derived analytical equations for one dimensional cases. Also, their
approach assumed general soil failure and is therefore an upper limit of the breakout force. (Foda,
1982) used a boundary layer formulation, he stated that when an object is lifted from the seafloor a
gap will arise between the structure and bottom. Using mathematics, he describes all the governing
effects for gap flow and eventually gives a prediction of the breakout force based on this flow.
(Das, 1991) further proposed modifications for the formulas regarding the breakout force using the
contact bearing pressure, a relation between the object area and the object weight and he stated
that the time an object rests on the sea bottom effects the breakout force. (Al-Shamrani, 1995) did a
finite element analysis on the break out phenomena. His analyses were in agreement with the
empirical formulas stated by earlier research. However, limited field information is available for
confirmation of his model.
(Mei, C.C., Yeung, R.W., Liu, K.F., 1985) investigated based on the assumptions from (Foda, 1982) the
breakout phenomena and showed that breakout also occurs without the assumptions of an elastic
soil skeleton.
4
(Craig, 1987) and (Chen, 2012) both did a series of centrifugal test to investigate the breakout
phenomena. (Craig, 1987) did this to investigate the installation of a jacket. (Chen, 2012) to
investigate the uplift capacity of the mud mats in slightly over consolidated clay.
For tilted lifting (Huang, H.M., Lin, M.Y., Huang, L.H., 2010) investigated the flow field under the
structure induced by the tilted lifting and the lift force using an analytical approach.
What can be observed from the literature is that few to no research in to breakout phenomena in
sand have been carried out. Also, no modelling is done into this subject. From cutting theory it is
known that cutting in sand develops under pressures around the blade tip. An analytical model exists
where these under pressures are calculated. Therefore, it is interesting to investigate the breakout
phenomena in sand.
To model the lifting process in sand the first idea was to use Plaxis as a program. After a couple of
weeks using Plaxis it was decided that Plaxis was too complex to give easy insight in the governing
parameters.
For modeling the lifting process an Analytical model will be used and modified. (Miedema, 2014)
published a book about cutting theories used in Sand, clay and Rock. For this report the analytical
sand pore pressure calculation model will be used that describes water flow in the sand bed but does
not take the stresses in the grain skeleton of the soil in to account. The pressure calculation is useful
for prediction of the lifting force.
2.2 Parallel Resistor method

As an analytical model an adaption of the Parallel Resistor model for water under pressures in the
sand on and round a blade during saturated sand cutting will be used. This model will be adapted for
the under pressures under the mud mat.
Miedema uses his model as a method to use the basics of the sand cutting theory in a very practical
and pragmatic way. The model is primarily used to give some fast and easy insight in the pressure
development around the blade. From the theory and comparisons with FEM calculations it is known
that the model has an accuracy of 10% (Zhao, 2001). The model itself is calibrated with measurement
data. For such a model, given the accuracy of the input parameters, this is a good result. The problem
with FEM calculations is that they are time consuming to make and therefore expensive. Also, they
are most of the time case dependent and therefore non-universal in use.
5
Figure 2.1 – The Equipotential lines and the flowlines for a cutting case
The model of Miedema is based on the flow lines in the sand bed. Miedema reasons that following
the flow lines through the bed to the shear zone the water flow will encounter a certain resistance.
This resistance is proportional to the distance the water has to travel and dependent on the
permeability of the soil, see Figure 2.1.
The longer the flow line, the bigger the resistance and the smaller the permeability the more difficult
it is for the water to flow through the soil.
Dilatation is described as an increase in volume due to shear deformations. During cutting dilatation
takes place with as result an increase in pore under pressures (Miedema, 2014).
Now the law of Darcy describes a relation between the specific flow 𝑞 (m/s) and the pressure
difference ∆𝑝 (𝑃𝑎):
∆𝑝
𝑞 =𝑘∗𝑖 =𝑘∗ (2.1)
𝜌𝑤 ∗ 𝑔 ∗ ∆𝑠
Where:
∆𝑠 = Flow length
𝑘 = Permeability
𝑖 = 𝑑ℎ/𝑑𝑙
𝜌𝑤 = Density water
𝑔 = gravitational constant
The total specific flow for the cutting case originates from the flow caused by the dilatation, in the
cutting case this is defined as follows:
𝑞 = 𝜖 ∗ 𝑣𝑐 ∗ sin(𝛽) (2.2)
6
Where:
𝜖 = Dilatation
𝑣𝑐 = Cutting Velocity
𝛽 = Shear Angle
Therefore, one can state:
∆𝑝
𝜖 ∗ 𝑣𝑐 ∗ sin(𝛽) = 𝑘 ∗ (2.3)
𝜌𝑤 ∗ 𝑔 ∗ ∆𝑠
Figure 2.2 – The Water flow towards the cutting zone.
Now the length of the four flow lines can be calculated, see Figure 2.2. Using the corresponding
permeability in combination with shifting some variables the following relation is found:
∆𝑝
𝜌𝑤 ∗ 𝑔 ∗ 𝜖 ∗ 𝑣𝑐 ∗ sin(𝛽) = (2.4)
∆𝑠
(𝑘)
For the different lengths and permeability’s:
∆𝑝 ∆𝑝 ∆𝑝 ∆𝑝
𝜌𝑤 ∗ 𝑔 ∗ 𝜖 ∗ 𝑣𝑐 ∗ sin(𝛽) = + + + (2.5)
∆𝑠 ∆𝑠 ∆𝑠 ∆𝑠
( 1) ( 2 ) ( 3) ( 4)
𝑘𝑚𝑎𝑥 𝑘𝑚𝑎𝑥 𝑘𝑖 𝑘𝑖
As the length of a flow line divided by the permeability can be seen as a resistance the following
applies:
∆𝑝 ∆𝑝 ∆𝑝 ∆𝑝
𝜌𝑤 ∗ 𝑔 ∗ 𝜖 ∗ 𝑣𝑐 ∗ sin(𝛽) = + + + (2.6)
(𝑅1 ) (𝑅2 ) (𝑅3 ) (𝑅4 )
7
Following the rule of parallel resistors and using the resistance values found above the following
equation is found:
1 1 1 1 1
= + + + (2.7)
𝑅𝑡 𝑅1 𝑅2 𝑅3 𝑅4
Combining equation 2.6 and 2.7 results in an equation describing the pressure difference in the shear
zone on a certain point:
∆𝑝 = 𝜌𝑤 ∗ 𝑔 ∗ 𝜖 ∗ 𝑣𝑐 ∗ sin(𝛽) ∗ 𝑅𝑡 (2.8)
Integrating over the shear zone gives the average pressure over the shear zone, With 𝑑𝑛 the step size
over the shear zone.
𝑑𝑛
1
𝑝= ∑ ∆𝑝𝑖 (2.9)
𝑑𝑛
𝑖=0
2.3 Conclusions
Early publications showed that already in the sixties a lot of attention was put in the research on
breakout phenomena. A lot of this research was aimed at testing in the field using for example ships
and submarines. Also, a lot of publications were dependent on the location of the test site. As
specific soil conditions are present at these sites it is not evident to draw general conclusions. A lot
of the earlier research is indirectly or directly based on the research done in the sixties primarily by
Muga. In the eighties Foda looked on the breakout phenomena from another angle proposing new
formulas to predict the breakout force. (Das, 1991) performed some experiments in soft clay with
cylindrical objects.
It can be concluded that no real tests on scale are performed in fine grained sand. As clay behaves
different than sand it is not evident to use the same formulas found in the early days for soils based
on sand. To get some understanding of the basic physical parameters an analytical model is created
to predict the breakout force in sand. The model used is based on the parallel resistor method used
in (Miedema, 2014). This model has some similarities with the lifting case, for example, in cutting
cases under pressures are present under and around the cutting blade. This gave the inspiration to
adapt this model for the lifting case. To validate this model laboratory test will be performed.
8
3 Testing
3.1 Introduction
In many scientific researches testing with a certain setup is done. In this case testing is done to give
the analytical model some fundamental backup. Also in comparison with field testing the small-scale
laboratory setup has some advantages:
- More parameters are easier to control.

- Good visibility on the setup.
- Costs are low compared to field tests.
- Tests are easier to reproduce, faster and under same conditions.
The test results will be used to validate the model outcome. As mentioned before the testing will
only be done using fine grained sand as soil.
At first it was important to check whether some results can be derived from the test setup at all.
Therefore, a preliminary test setup was build according to some expectations.
Expectations:
- Suction can be measured using differential pressure meters.

- With a transparent acrylic plate the water flow might be observed during the lifting
process.
- After lifting a hole in the sand bed can be observed, under pressures stick the sand to
the plate during lifting.
After careful testing, some improvements were made to the test setup, in the early stages a strong
nylon rope was used for lifting but this proved to be too elastic generating unrealistic results and too
much fluctuation in the load. Also, it proved to be difficult to level the sand bed correctly resulting
most of the time in immediate breakout of the plate. The visibility in the reservoir was unfortunately
very bad. Especially the fine-grained sand carried some dust with it.
Suction under the plate was eventually registered by the Differential Pressure sensors after making
sure the right sequence of venting was used. At first it was questionable to use the DP sensors as it
was suspected the pressure was small. Testing showed that the sensors could be used. After every
test a gap in the sand bed was present so it seemed that the plate was taking a portion of sand with
it during the lifting process.
After fine-tuning the test and learning from the preliminary testing the final test setup was in
operation.
3.2 Test setup

The tests are performed in the Dredging Laboratory of the technical university in Delft.
The test setup is built around an existing acrylic reservoir with the following dimensions: 80x80x80
cm. Around the reservoir a rigid frame is placed which holds the pulleys. A simple diagram of the test
setup is shown in Figure 3.1
9
Figure 3.1 – Diagram of the test setup
In the reservoir, an acrylic plate is placed with a thickness of 30mm. This is done to make sure the
deformations due to bending can only occur with a very high, unrealistic, loading. It was expected
that only small displacements are needed to lead to breakout of the plate, therefore only minimal
deformations due to bending can be allowed. Testing will be done with 2 types of acrylic plate, a
rectangular and a round version. The rectangular plate is sealed off against two sides of the reservoir
with thin rubber gaskets with low friction, this is done to create a pure 2-Dimensional case. The
round plate is modeled to use for the 3-Dimensional case.
In the plates, several 3 mm holes are drilled to make connections for the Differential Pressure
sensors, DP sensors. These are connected via rigid poly flow hoses; the holes are divided over the 2-
Dimensional (Rectangular) and 3-Dimensional (Round) plate according to Figure 3.2 and Figure 3.3
respectively. The plate will rest on the sand bed, the sand bed is always of the same thickness,
namely 700 millimeters.
10
Figure 3.2 – Layout of the location of the pressure sensors on the 2-Dimensional plate
Figure 3.3 – Layout of the location of the pressure sensors on the 3-Dimensional plate
The plates are lifted via a flexible steel rope which runs over two frictionless pulleys. At the other end
of the cable a bucket is placed to hold weights needed to create the specified load.
11
Figure 3.4 – Photos of the test setup and equipment
The load is transferred to the plate by a rigid stainless steel lifting frame which is connected via four
chains to the plate, see Figure 3.4
Between the lifting frame and the steel cable a load cell is placed to measure the load exerted by the
bucket on the plate. During the lifting the displacement is measured via a displacement sensor placed
onto the plate, the sensor is calibrated for every test.
As mentioned before there are connections in the plate for hoses leading to DP sensors. The sensors
measure the under pressure beneath the plate and are therefore negatively calibrated.
3.3 Test Parameters

To get some insight in the physical processes it is important to define the test parameters correctly.
Different parameters are to be tested.
3.3.1 Sand types

The pressure difference predicted by the parallel resistor model is dependent on the permeability of
sand. Therefore, to check the validity of the model different types of sand will be used, however the
permeability cannot be to large otherwise no sufficient under pressure can build up. The two sands
used are the Silverbond D6 and Geba 0.06-0.25 due to availability in the laboratory.
These two sands are tested in the laboratory of civil engineering. An earlier sieve analysis has been
done by Ir. Rik Bisschop. Test data showed that the D50 of Silverbond and Geba is 51 µm and 125 µm
respectively.
Both sand have endured a falling head test to check the maximum value of the permeability.
Silverbond got a value of 5e-6 m/s. Geba, having a larger grain size, had a value of 7e-5 m/s. This is
the smallest permeability for this type of sand as it was make sure the sand was packed as tight as
possible during the test. The values of the porosity are tested at the civil engineering laboratory. The
values for the Geba sand tested as:
12
𝑁𝑚𝑎𝑥 = 0.47
𝑁𝑚𝑖𝑛 = 0.40
And for the Silverbond sand:
𝑁𝑚𝑎𝑥 = 0.54
𝑁𝑚𝑖𝑛 = 0.39
3.3.2 Consolidation
Another factor that defines permeability in soil is the consolidation. One can reason that if a weight is
placed on the sea bottom for a certain amount of time the water will flow out of the soil bed
underneath the mat. How much water and how fast is dependent on the weight above the mat,
duration, compressibility of the soil and the permeability of the soil. This variable is not of great
interest for structures who have been placed on the sea bottom a long time ago as the soil
underneath the mats has reached its maximum consolidation given its weight. The results will be of
interest for modelling the installation process of such a structure. This variable will be tested by
placing a big weight onto the 3-Dimensional plate and let the plate rest for a certain amount of time.
Afterwards it can be checked if different weights placed onto the plate will result in different
breakout times.
3.3.3 Lifting loads

The lifting loads are an important variable, actually they are of influence on two other variables
namely the velocity and the breakout time. Lifting with a bigger load means that the upward velocity
of the plate will be larger which means that the expected under pressures increase. Lifting with a
larger velocity means that the breakout time is shorter. The question is how these variables relate
with the use of different sands. Lifting with the same load on Geba should give an overall faster
process than lifting on Silverbond.
The lifting loads are chosen according to the dry weight of the lifting frame added by the submerged
weight of the plates. As mentioned before breakout can only occur if there is a net upward force,
therefore the minimum load needs to be larger than the combined weight.
Silverbond sand Load 1 Load 2 Load 3 Load 4
2D 130N 150N 170N
3D 65N 70N 90N 110N
Table 3.1 – Load Parameters for the Silverbond sand.
13
Geba sand Load 1 Load 2 Load 3
2D 90N 110N 130N
3D 50N 60N 70N
Table 3.2 – Load Parameters for the Geba Sand
Table 3.1 and Table 3.2 give the total load imposed on the complete structure. These loads can be
verified by the use of the load cell. When comparing the under pressures with the loads it is
important to note that the loads should be corrected for the weight of the frame and the underwater
weight of the plate. When correcting for this weight the net pulling load under the plate is known.
3.4 Test Procedure

The sand is placed into the reservoir with water and mixed thoroughly to exclude possible trapped air
bubbles. The mixture is vibrated to speed up the settling en left resting for a week. This creates a
solid sand bed onto which the plates can be placed. This has to be done for the two sand types
ensuring equal sand – water distribution.
Before placing a plate onto the surface the sand is smoothened with a steel ruler en left for setting.
The plate is hanging in the water so that the DP-sensors can be vented. The plate is than placed on
the smooth sand surface and vibrated around the plate to ensure no air is trapped under the plate
and to create a homogeneous sand bed. After settling for an hour (water is clear again) the test can
begin. Weights are carefully placed into the bucket in such a way no dynamic peak load is exerted
onto the frame.
Every test is run three or more times with the same weights to get some good average values. The
testing find place according to the test matrices shown in tables below. The red colored tests shown
in the tables below are the test that are discarded due to reasons specified in chapter 4.
2-D Tests Silverbond

Testname
Load (N) Test 1 Test 2 Test 3
130N 2D-S-130N -1 2D-S-130N -2 2D-S-130N -3
150N 2D-S-150N -1 2D-S-150N -2 2D-S-150N -3
170N 2D-S-170N -1 2D-S-170N -2 2D-S-170N -3
Table 3.3 – Test matrix for the 2-Dimensional test in the Silverbond sand.
14
2-D Tests Geba
Testname
Load (N) Test 1 Test 2 Test 3
90N 2D-G-90N -1 2D-G-90N -2 2D-G-90N -3
110N 2D-G-110N -1 2D-G-110N -2 2D-G-110N -3
130N 2D-G-130N -1 2D-G-130N -2 2D-G-130N -3
Table 3.4 – Test matrix for the 2-Dimensional test in the Geba sand.

Testname
Load (N) Test 1 Test 2 Test 3 Test 4
65N 3D-S-65N -1 3D-S-65N -2 3D-S-65N -3
70N 3D-S-70N -1 3D-S-70N -2 3D-S-70N -3 3D-S-70N -4
90N 3D-S-90N -1 3D-S-90N -2 3D-S-90N -3
110N 3D-S-110N -1 3D-S-110N -2 3D-S-110N -3 3D-S-110N -4
Table 3.5 – Test matrix for the 3-Dimensional test in the Silverbond sand.
3-D Tests Geba

Testname
Load (N) Test 1 Test 2 Test 3 Test 4
50N 3D-G-50N -1 3D-G-50N -2 3D-G-50N -3 3D-G-50N -4
60N 3D-G-60N -1 3D-G-60N -2 3D-G-60N -3
70N 3D-G-70N -1 3D-G-70N -2 3D-G-70N -3
Table 3.6 – Test matrix for the 3-Dimensional test in the Geba sand.
15
4 General Test Results
4.1 Introduction
The data from the test setup was recorded with Labview and further processed by Matlab. The data
was recorded with a sample frequency of 100hz. The results are then corrected for offset values, this
means that the test setup will run idle for a minute every test to record the idle values of the sensors.
The values are then later subtracted from the complete data set.
As there were in fact two different tests performed, namely the 2-dimensional and the 3-dimensional
test, these will be treated separately.
In advance of testing a test plan was written to check how many tests were to be performed.
Unfortunately, not every test came out as a reliable result. Some test results are therefore discarded.
This is done after carefully examining the results. Most of the time unrealistic short breakout times
led to removing of the test results, this is probably due to the fact the plate did not made good
contact with the sand bed, air was trapped beneath it or the sand itself had air in it. In other cases
there was a clear problem with air in the polyflow hoses to the DP sensors so no correct pressure
registration took place. The specific test results are reported in Chapter 5 and 6
4.2 Net loads

For analyzing the measurements, it is important to reason how the load is interpreted. The load
imposed onto the lifting frame is not equal to the load under the plate, this is because when lifting
the structure first the dry and submerged part of the frame have to be overcome, after that the
under pressure is created. For analyzing the net load is of more importance, therefore in the tables
below the load is converted to a net load. The factor used is, as mentioned, dependent on the dry
weight and the submerged part (e.g. the plate) of the structure. For the 2-Dimensional situation the
difference between the load and the net load is 52 Newtons., for the 3-Dimensional the difference is
calculated as 45 Newtons

Load (N) Net load (N)
130N 78N
150N 98N
170N 118N
Table 4.1 – The net load values for the Silverbond sand in the 2D situation.
16
2-D Tests Geba
90N 38N
110N 58N
130N 78N
Table 4.2 – The net load values for the Geba sand in the 2D situation.

65N 20N
70N 25N
90N 45N
110N 65N
Table 4.3 – The net load values for the Silverbond sand in the 3D situation.
3-D Tests Geba

50N 5N
60N 15N
70N 25N
Table 4.4 – The net load values for the Geba sand in the 3D situation.
4.3 Measurement Analysis

The test data was recorded with Labview at a sample rate of 100hz per second. This generates a lot
of raw data which need to be filtered, see Figure 4.1 . Using Matlab to filter the signals with a moving
average filter the resulting signals are used to generate plots like those in Figure 4.2 and Figure 4.3.
Figure 4.2 shows the Load and Displacement versus the Time for the 3-Dimensional plate. What can
be seen is that the load is located around the 26 Newton as the Displacement increases steadily. The
Load has a short interruption in the signal at around 20 seconds. This is because all the weight used
in the test setup are placed into the bucket carefully one at a time to make sure no dynamics are
disturbing the data. The Peak load is derived from the values at roughly 26 seconds, the maximum
displacement is the displacement taken just before the breakout.
In Figure 4.3 the pressure and velocity versus the time is plotted. The velocity is derived from the
displacement. After filtering the trend is visible. The velocity is used to determine the time of
breakout. Looking closer at the velocity at around 90 seconds it can be seen that the slope of the
velocity signal increases rapidly, this means that the plate including the lift frame are accelerating.
17
Figure 4.1 – An example of the unfiltered velocity.
Figure 4.2 – The load and the displacement versus the time for a 3-Dimensional test in Silverbond sand, (3D-S-70N-1).
From these data plots the peak values are used for further analysis in Chapter 5 and 6. The reason for
using the peak values can be understood if one looks at the signal for the pressure. At around 15
18
seconds the Load in Figure 4.2 starts to increase like a ramp function. The pressure signals in Figure
4.3 response different. They increase more like a first order system. The reason for this is to be
answered later on. At the point the velocity was increasing rapidly also the pressure signals begin to
act different. Large jumps in the signals can be spotted. This is due to the dynamics affects taking
place when the plate begins his journey to the surface with a sudden increase in lift velocity. The
peak values to be used for the pressure signals are in this case taken right before breakout.
Figure 4.3 – The pressure and the velocity versus the time for a 3-Dimensional test in Silverbond sand, (3D-S-70N-1). Note
the filtered velocity
Unfortunately, not all test performed were usable. Sometimes imposing the load on the plate and
lifting frame initiated immediate breakout as by other test the breakout time was normal but little or
no pressure registration happened. These situations can be linked to trapped air underneath the
plate or the sand bed proved not to be perfectly smooth. In Figure 4.4 and Figure 4.5 the results of a
2-Dimensional test that went wrong are plotted. What can be seen is that the load was increasing.
During the increase the velocity also started to increase rapidly leading to a breakout time of under 5
seconds. These kind of test results are excluded from further analysis.
19
Figure 4.4 – A failed 2-Dimensional test, as can be seen the breakout time is very short, (2D-S-150N-3)
Figure 4.5 – A failed 2-Dimensional test, the pressure sensors registered a lot of dynamic results, (2D-S-150N-3).
20
4.4 Conclusions
What can be concluded from analyzing the test data is that pressure registration is difficult. A lot of
tests are dealing with delays in pressure registration. This is probably due to the pressure mounting
points on the plate. Some damping can occur if these holes are getting constipated with sand. A
method for avoiding this kind of results in future testing would be the use of more pressure sensors
linked to each other so that the mean value of a couple of mounting points is used. Also, equipping
the mounting points on the plate with little filter stones could be a solution. Another reason for bad
pressure registration can be air in the hoses. Before every test the sensors are getting vented for a
couple of seconds. However, air inside the sand bed could influence the pressure measurement.
21
5 2 Dimensional Results
The first tests to be performed were done with the 2-Dimensional plate. The plate has connections
for up to five DP sensors. Four of them were used and distributed over the plate according to Figure
3.3. The sand types used to build up the sand bed were the Silverbond and Geba sand mentioned
before. The magnitude of the different loads that were applied on the plate are according to Table
3.1. This chapter will show the results of the 2-Dimensional test and show some of the relations
found.
5.1 Expectations
As with every laboratory test there are some, theoretical, expectations. As for the 2 dimensional
plate these were as following:
- Breakout time is directly related to the permeability of the sand. The permeability of the
sand is linked to the D15 quadratic (Den Adel, 1989), this holds that if the grain size doubles
the permeability becomes 4 times as big. As the permeability increases roughly with a factor
8 it was estimated that the breakout time also decreases roughly a factor 8.
- According to the model outcome (See chapter 7) the pressure development for the 2-
Dimensional case is expected to be linear given any input over the width of the plate. By all
means the under pressure measured on a point closer to the plate edge is expected to be
lower than a point on the center line of the plate, furthest away from the plate edge. In this
case DP1, DP2 and DP3 should give similar results as DP4 should give a lower output. As the
2-dimensional model is linear and DP4 is located on a quarter of the plate the under pressure
is expected to be half of the under pressure measured by the other sensors.
- The average time of breakout is supposed to decrease as the load increases. One can expect
that doubling the load should give half the breakout time as the model itself is linear.
5.2 Time of breakout

5.2.1 Introduction
The time of breakout is defined as the time the plate needs for breakout over the different tests
under the same conditions. Expected was that for the Silverbond sand the breakout time is
noticeable larger than for the Geba sand. This, as mentioned before, has to do with the difference in
permeability for the both sands which leads to a higher resistance for water to flow along the
flowlines and thus a larger breakout time.
5.2.2 Silverbond 51µm Sand

The average breakout time for the Silverbond sand is plotted in Figure 5.1. What can be seen is that
the breakout time for the net load of roughly 80 newtons has a range around 120 seconds. For the
100 and 125N net load the breakout times are in the range of 90 and 80 seconds respectively. The
trendline is added to help visualize the descending trend. Notice that the Load values plotted are the
Net load values under the plate. This means that the submerged weight is subtracted.
22
Figure 5.1 – The time of breakout for the Silverbond sand, the spread of the different tests is visible. Note that it looks like
the trend is having an Asymptotic value, beware that it is only a trend line.
5.2.3 Geba 125µm Sand

The average breakout time for the Geba sand is plotted in Figure 5.2. Again, a descending trend is
visible
23
Figure 5.2 – The time of breakout for the Geba sand. Here the trend is going to zero, again it is only a trend line and it is not
suspected to cross the Net load axis.
5.2.4 Conclusion
From analyzing the breakout time of the 2-Dimensional plate it can be concluded that the breakout
time does not scale linearly with the force. For the Silverbond sand an increase of around 25N
decreases the breakout time by almost a factor 2, further increase of the same load gives a marginal
decrease in breakout time. For the Geba sand however the breakout time follows a more linear
slope, see Figure 5.2
5.3 Mean Displacement

5.3.1 Introduction
The mean displacement is defined as the displacement the plate travels before breakout. This
displacement can easily be extracted from the test data as the displacement follows a relatively
horizontal line before breakout, after breakout the plate accelerates fast so an increasing slope can
be spotted in the line of displacement. Again, the displacement is an average value over the different
tests under same conditions.

In Figure 5.3 the mean displacement for the different loads is plotted. As can be observed the mean
displacement does have a quite constant trend. They are all located around the 2 millimeters of
24
displacement needed before breakout. Notice that the loads are the imposed loads on the structure,
not the net loads
Figure 5.3 - The mean displacement of for the tests performed with the Silverbond sand.

For the Geba sand the mean displacement has some more variation, the displacement for the 90
Newton load is twice the displacement than that of the 130 Newton load.
Figure 5.4 - The mean displacement of for the tests performed with the Geba sand.
5.3.4 Conclusions
The mean displacements of both the Silverbond and the Geba sand are all of the same order. The
difference between the points is small, only the Geba sand had some deviation. What can be
concluded is that this displacement is needed before breakout can happen, the displacement is
25
related to a certain amount of water needed before breakout can occur. The differences in the mean
displacement are small between the two sand, therefore nothing can be concluded about this.
5.4 Mean Pressures

5.4.1 Introduction
The mean under pressures are the pressures derived from results measured by DP1, DP2 and DP3.
This means that the mean under pressures are the peak under pressures found in the center line of
the plate, see Chapter 4. The under pressures are corrected for offset values as mentioned before.
Together with the measured under pressures the expected under pressures are shown. These are the
peak under pressured derived from the imposed load by dividing the load by the surface of the plate.
5.4.2 Results
A good verification of the readings of the DP sensors is to compare the measured pressure with the
load. After all the pressure underneath the plate should be equal to the load. In other words, there
should be an equilibrium if the setup is static. The mean plate pressure is derived by dividing the load
over the surface of the plate, this means that the mean plate pressure is a mean pressure over the
plate. For the three different loads the pressure is averaged and compared to the mean plate
pressure. The result for the Silverbond sand is shown in Figure 5.5, the Geba sand results are shown
in Figure 5.6
Figure 5.5 – The mean plate pressure versus the measured pressure for the Silverbond sand. Especially for the 124 Newton
load the difference is big.
26
Figure 5.6 The mean plate pressure versus the measured pressure for the Geba sand. Here the differences occur with almost
every test.
As can be seen there is a difference in measured pressure and the expected plate pressure. To get
some better understanding of what is happening all data points are plotted and the measured
pressure is plotted against the mean plate load. The plot in Figure 5.7 shows the mean plate pressure
versus the measured pressure. However, the measured points do not represent the mean pressure
but show the local pressure. The trend shows that the measured pressure is for most points higher
than mean plate pressure which is to be expected as the flow lines are longer.
Figure 5.7 – The Measured Pressure versus the mean plate Pressure.
27
Figure 5.8 – The results of DP4 versus the mean plate pressure, DP4 carries the information about the pressure profile under
the plate.
A reason for this is that one only measures at one point and the real pressure profile underneath the
plate is not homogeneous as with the mean plate pressure. At the edge of the plate a strong
pressure gradient is present, as the pressure at the edge is zero. Therefore, the pressure in the
middle is obviously higher than the mean pressure.
The first three DP sensors are all located in one line. DP sensor 4 is located more to the edge of the
plate and therefore has some more information. In Figure 5.8 the results of DP4 are plotted against
the mean plate pressure. What can be seen is that they mostly fall within the 10% margin. This
means that they are roughly equal. Thus, at the location of DP4 the pressure is almost the same as in
the center of the plate. This says that the pressure profile is nonlinear and has steeps slopes towards
the edge of the plate. More cannot be concluded as more sensors are needed.
5.4.3 Conclusion
Taking a closer look at both sands the only thing of notice is that the absolute pressure of the
Silverbond sand is higher than the pressure measured of the Geba sand. This makes sense as the
Silverbond sand creates a higher under pressure than the Geba sand. This is off course the result of a
different load range.
5.5 Comparing the Two Sands

To see if the difference in permeability between the Silverbond and Geba sands is visible a
comparison between the two-different sand with the same imposed load has to be made. One test of
28
both sands is done with the same load, this load was measured by the load cell and due to friction in
the pulleys has some spread. This spread can be observed in Figure 5.1 and Figure 5.2 respectively.
From this spread the mean values had been calculated. What can be observed in Figure 5.9 is that
even if the load during the test in the Geba sand is lower the breakout time is roughly a factor 8
smaller in the Geba sand than the test in the Silverbond sand. Comparing this result with the two
measured permeability values from paragraph 3.3.1 which gave roughly a factor 10 the result seems
plausible.
Figure 5.9 – The differences in breakout for the different sands
5.6 Conclusions
The 2-Dimensional test setup was not easy to work with, despite that fact the results do not
disappoint. The pressure development can partly be reconstructed and for the time of breakout a
nice trend between the 2-different sand can be spotted according to the literature (Lee, 1973).
29
6 3 Dimensional Results
The tests to be performed with the 3-Dimensional plate are in basis the same as with the 2-
Dimensional plate. The biggest difference is that the plate now is not sealed off against 2 sides
making it a 3-Dimensional problem. Also, there is less room for DP sensor connections. The plate has
connections for up to three DP sensors. These are distributed across the plate according to Figure
3.3. The sand types used to build up the sand bed were the Silverbond and Geba sand mentioned
before. The magnitude of the different loads that were applied on the plate are according to Table
3.2. For the Silverbond sand four different tests have been performed. For the Geba sand again 3
tests. This Chapter will show the results of the 3-Dimensional tests.
6.1 Expectations
Again, for the 3-dimensional model some expectations raised before the testing started.
- Breakout time is directly related to the permeability of the sand. The permeability of the
sand is linked to the D15 quadratic (Den Adel, 1989); this holds that if the grain size doubles
the permeability becomes 4 times as big. As the permeability increases roughly with a factor
four it was estimated that the breakout time also decreases roughly a factor four
- The model outcome of the 3-Dimensional model has, as mentioned before, a nonlinear
behavior. Due to the assumptions, the pressure development is evened out in the middle. As
with the 2 Dimensional plate the pressure towards the edge of the plate is expected to be
lower than the pressure in the center of the plate, the question is of the DP2 and DP3 sensor
location is far enough from the center of the plate to measure a decrease in pressure.
- The average time of breakout is supposed to decrease as the load increases.
6.2 Mean time of breakout

For the mean time of breakout, the principle is the same as with the 2-Dimensional plate. Al the tests
are plotted against the load. With the help of a dashed line a trend can be visualized.

The Silverbond sand has been test, as mentioned, with four different loads. What can be seen is that
the time of breakout drop hard as the load increases, see Figure 6.1. Notice again that the loads
plotted are the net loads under the plate.
30
Figure 6.1 – The mean time of breakout for the Silverbond sand, notice the descending trend.

As for the Geba sand the time of breakout was hard to determine for the bigger loads. The sand is a
lot more permeable. In Figure 6.2 it can be seen that at the 2 highest net loads the breakout is almost
immediate after applying the load. This fast breakout is directly linked to the large permeability of
the sand.
31
Figure 6.2 – The mean time of breakout for the Geba sand, notice that the difference between the 5 Newton load and the
13N load is large.
6.2.3 Conclusions
What can be concluded over the breakout time for the Silverbond sand is that as the load increases
the breakout time decreases following a nice slope. Enough tests were performed to see the trend
which is plotted in Figure 6.1. For the Geba sand the results are more difficult to read. What probably
happened is that the scale of the applied weights was to large leading to tests results at the
boundaries of the breakout time. When a load of 50 Newton is applied, a net load of 5 newton in the
graph, the mean breakout time is just above the 100 seconds. Increasing the net load with 10
Newtons the breakout time is already under the 10 seconds. This means that for the larger load the
breakout time is more dominated by the inertia of the setup than the under pressures under the
plate. A more refined load scale during the testing had been of better use. However, if one plots the
breakout time the same behavior as with the Silverbond sand can be spotted, see Figure 6.2
6.3 Mean Displacement

The mean displacement of the 3-Dimensional plate is defined in the same way as the mean
displacement for the 2-Dimensional plate.
32
For the Silverbond sand the mean displacements are shown in Figure 6.3. The displacements before
breakout are all in the order of 2-2.5 millimeters, around 400 particle diameters. A descending trend
can be observed however nothing can be concluded about this trend.
Figure 6.3 - The mean displacement of for the tests performed with the Silverbond sand.

For the Geba sand the mean displacements are plotted in Figure 6.4. The mean displacements are all
in the same order of around 2 millimeters before breakout.
Figure 6.4 - The mean displacement of for the tests performed with the Geba sand.
6.3.3 Conclusions
For the Silverbond sand nothing can be concluded regarding the descending trend of the mean
displacements. The differences in displacement between the different loads are so small that the
displacement can be regarded as equal over the test. Again, no differences can be seen between
33
both grain sizes. What can be said about the mean displacement is that is does not scale with the
particle diameter.
6.4 Mean Pressure

6.4.1 Introduction
The mean pressure of the plate is the mean pressure in the center of that plate and derived from the
result of DP1. The difference with the 2-Dimensional plate is that now only one sensor can be located
in the center. DP2 and DP3 are located on the same distance at two sides of the plate. Therefore, the
result of these 2 sensors is averaged in the plots below. Again, the mean plate pressure is calculated
according to the imposed load.
6.4.2 Results
For the Silverbond sand the mean pressure of DP1 versus the mean plate pressure is shown in Figure
6.5. What can be observed is that the measured central plate pressure is higher than the mean plate
pressure.
Figure 6.5 – The mean plate pressure versus the measured pressure from DP1 for the Silverbond sand. Notice again the
difference in measured pressure and the mean plate pressure.
Comparing DP1 with the mean result of DP2 and DP3 gives results plotted in Figure 6.6. What can be
seen is that the result of DP1 in the center is higher than the mean values of DP2 and DP3
34
Figure 6.6 – For all the test the mean pressure of DP1 is higher than the other 2 sensors.
For the Geba sand the results are shown in Figure 6.7
Figure 6.7 – The mean plate pressure versus the measured pressure for the Geba sand.
Again, comparing DP1 with the values of DP2 and DP3 for the Geba sand gives Figure 6.8.
Comparing the Pressures with the mean plate pressures results in Figure 6.9
What can be seen is that DP1 is still higher than DP2 and DP3, though it is not much. This is probably
because the permeability of the Geba sand is higher compared to the Silverbond sand. Therefore, the
pressure gradient is smaller.
35
Figure 6.8 – For all the test the mean pressure of DP1 is higher than the other 2 sensors.
Figure 6.9 – Comparison between the mean plate pressures and the measured pressures.
36
6.4.3 Conclusions
What can be seen is that for the Silverbond sand the results of DP1 are quite a bit higher than the
results of the DP2 and DP3 sensors. This makes sense and is a good indication of the pressure profile.
To calculate the pressure profile exact more sensors are probably needed. For the Geba sand the
results are less different indicating a smoothened pressure profile in the center of the plate. As the
packing is different with the Geba sand, leading to differences in the permeability this is possible as
the pressure gradient is smaller.
6.5 Comparing the Two Sands.

When comparing the Silverbond and the Geba sands for differences in the mean breakout time as
with the 2-Dimensional test, the breakout time for the same loads has to be observed. The mean
values of the spread in Figure 6.1 and Figure 6.2 have been calculated. Plotting this breakout time
for the same loads gives Figure 6.10. What can be observed is that the mean breakout time for the
Silverbond sand is 75 seconds, the mean breakout time for the Geba sand is around 5 seconds. This
gives roughly a factor 15.
Figure 6.10 – Time of breakout for the 2 sands. Notice the huge difference.
As for comparing the breakout times for the same load and different sands a remark has to be made.
Forces slightly larger than the submerged weight plus lifting frame may cause an upward
displacement but not enough for breakout. As the force increases the breakout time decreases to
almost zero (Liu, 1969). What probably is happening is that the load imposed on the plate is too large
for the 3-Dimensional setup in the Geba sand. As can be seen the breakout time of the 70 Newton
load is almost the same as the 90 Newton load. An even bigger load will probably give the same time
of breakout. The breakout is not dominated anymore by the parameters of the soil but more by the
inertia of the setup. For future testing, more research can be done in the area between the 70
Newton and the 90 Newton load.
37
6.6 Consolidation
In earlier research a lot of testing was done regarding the effect if pre-loading a structure has effect
on the breakout time. In fact, pre-loading the structure will lead to the consolidation of the soil. The
pre-load will load the soil for a certain amount of time, after removal the breakout time is increased
with respect to a soil which was not pre-loaded.
For testing the effect of consolidation, the 3-dimensional plate is placed onto a smooth sand bed and
loaded for 5 hours with a 20-kg weight using specially designed frame. The setup can be seen in
Figure 6.11. Note that this is not the correct Silverbond sand but the Geba instead. This is done due
to visibility issued with the Silverbond sand.
Figure 6.11 – Underwater pre-loading
Every test has been performed twice, the test with pre-load and the test without pre-load. This gives
a total of 4 tests.
The results for the pre-load test are plotted in Figure 6.12 and Figure 6.13 respectively. What can be
observed is that the breakout time for both test is practically the same. Actually, the difference is
smaller than the mutual difference of normal tests. What can be concluded is that the difference in
breakout time is so small that Pre-loading in sand does not make a difference in breakout time. For
clay this difference is noticeable. Pre-loading clay gives more plastic deformations. If these
deformations will be present in sand it will be in the first couple of minutes, pre-loading for a longer
time does not make any difference.
A recommendation for further investigation could be to test on a loosened sand bed instead of a
vibrated bed. By vibrating the sand all the stresses are dissipated, looser sand reacts stronger on
consolidation. Problem is that testing in loose sand are not easy to reproduce.
38
Figure 6.12 – Consolidation test 1
39
Figure 6.13 – Consolidation test 2
6.7 Conclusion
Concluding from the results the 3-Dimensional test can be seen as a success. The test setup
functioned well with only the pressure registration as a difficult part. The pressure development can
be simulated but more sensors are needed regarding this subject. As for the mean time of breakout
between the two sand smaller load steps should be tested for the 3-dimensional setup in the Geba
sand. These same loads should be tested in the Silverbond sand to expose the trend. What can be
concluded is that the time of breakout follow a nonlinear trend as the load increases.
40
7 Analytical Model
As mentioned before in chapter 3 one can see the Parallel Resistor method is intended for use on 2D
cases of sand cutting. Lifting structures, or plates, from the sea bottom is a different case. The model
has to be adapted to this case to give a good result. This chapter will explain how the model is
adapted for both the 2-Dimensional and the 3-Dimensional case.
7.1 2D Analytical Model

First some insight in the situation. The 2D case is, as earlier explained, modelled as a plate on a sand
bed. The situation is explained according to Figure 7.1
Figure 7.1 – Flow lines underneath the plate
As can be seen the situation is similar to the Miedema setup only no cutting takes place. In this case
the force exerted on the lifting frame will try to pull out the plate and by doing so creating an under
pressure under the plate. For this under pressure to exist water needs to flow to the area under the
plate surface. Assuming circular flow lines, one can make a similar approach as Miedema did for the
cutting case. The difference is that now only flow lines from one side are used making the 2D case
relatively simple. This is done because if multiple flow lines were used the crossing of the flow lines
would have to be analyzed. Mathematically it can be done but the question is if it makes any
difference for the breakout force predicted by the model as the influence of this effect can be
neglected as the model will predict a force in the same order as the test data. If one want to know
the force precisely a finite element model should be used. Analogue to the earlier equations the Law
of Darcy is used:
∆𝑝
𝑞 =𝑘∗𝑖 =𝑘∗ (7.1)
𝜌𝑤 ∗ 𝑔 ∗ ∆𝑠
41
In this case the change in volume is caused by dilatation and therefore water has to flow in the
increased pore volume. Therefore, the total specific flow rate comes from the velocity at which the
plate is lifted. So, the specific flow rate is defined as:
𝑑𝑧
𝑞= (7.2)
𝑑𝑡
Where:
𝑑𝑧 = Height in vertical direction

𝑑𝑡 = time
By combining equation 7.1 and 7.2 the following relation is found:
𝑑𝑧 ∆𝑝
𝑞= = 𝑘∗ (7.3)
𝑑𝑡 𝜌𝑤 ∗ 𝑔 ∗ ∆𝑠
So, the equation for the pressure difference is found as:
𝑞 ∗ 𝜌𝑤 ∗ 𝑔 ∗ ∆𝑠
∆𝑝 = (7.4)
𝑘
The difference with the parallel resistor method is that the pressure difference is calculated only for
half the plate, as the plate is symmetrical it is easy to calculate the force for the complete plate.
Assuming the flowlines are again halve circular they are calculated as following:
∆𝑠 = 𝜋 ∗ 𝑟 (7.5)
With the resistance defined as:
∆𝑠
𝑅 = (7.6)
𝑘𝑖
Substituting equation 7.6 in equation 7.4 gives:
𝑑𝑧 ∆𝑝
𝜌𝑤 ∗ 𝑔 ∗ = (7.7)
𝑑𝑡 (𝑅)
42
Gives the final equation for the pressure difference in the 2D case:
𝑑𝑧
∆𝑝 = 𝜌𝑤 ∗ 𝑔 ∗ ∗𝑅 (7.8)
𝑑𝑡
Integrating over half the plate gives the average pressure:
𝑛
1
𝑝 = ∑ ∆𝑝𝑖 (7.9)
𝑛
𝑖=0
The average pressure times the surface of the plate now the gives the force needed to lift the plate
given a certain speed.
7.2 3D Analytical Model

The 3D model is based on the same approach. Now a round plate is lifted from the sand bed. This
situation cannot be modelled in a 2D model. The problem lies with the form of the plate.
Cutting the plate in half gives the same situation, see Figure 7.1:
However, seen from above it is obvious that using flow lines gives a problem with the total flow.
Given a fixed step size, the surface at the edge of the plate is bigger than the surface at the center of
the plate. In other words, water from a location outside the place with a bigger surface flows to a
location on the plate with a smaller surface. In fact, this is a geometrical problem. To accommodate
for this effect the plate is divided into pie slices. Now the pressure drop over the pie slice can be
calculated using flow paths, the slice is divided using a fixed step size which determines the number
of flow paths. The paths itself are divided into steps, see Figure 7.2
43
Figure 7.2 – The plate seen from above, now the pie slice model becomes clear.
The surfaces are radius dependent and can be calculated according to:
𝑑𝐿 ∗ 𝑑𝑅 (7.10)
Where:
𝑑𝐿 = Number of flow paths pie slice

𝑑𝑅 = width
The width is defined as:
𝜃
𝑑𝑅 = 2 ∗ 𝑟 ∗ sin ( ) (7.11)
2
Where:
𝑟 = radius
Dividing the flow path in a certain amount of steps the surface at each step can be calculated, see
Figure 7.3.
44
Figure 7.3 - Flow paths.
To calculate the total pressure drop over a flow path the pressure drop over 1 step is calculated and
integrated afterwards. The volumetric flow rate right below the plate, at dR1 follows from the
upward velocity of the plate which is known as input.
𝑄 = 𝑉𝑝 ∗ 𝑆 (7.12)
Where:
𝑉𝑝 = Upward velocity of the plate

𝑆 = Surface located beneath the plate at dR1
Now the specific flow rate is known as function of the radius
𝑄
𝑞(𝑟) = (7.12)
𝑆(𝑟)
The pressure drop per step analogue to Darcy:
𝑞(𝑟) ∗ 𝜌𝑤 ∗ 𝑔 ∗ 𝑑𝑠
𝑑𝑝(𝑟) = (7.13)
𝑘
With:
𝑎
𝑑𝑠 = (7.14)
𝑛
Where:
𝑎 = arc length
𝑛 = number of steps of the flow line
45
Integrating the pressure over the flow path gives the total pressure drop per flow path. Integrating
over the pie slices gives the pressure drop per pie slice. Knowing the surface of a pie slice the force
per slice and thus for the plate is known.
𝐹 = 𝑛𝑠𝑙𝑖𝑐𝑒𝑠 ∗ 𝑑𝑝𝑠𝑙𝑖𝑐𝑒 (7.15)
Where:
𝑛𝑠𝑙𝑖𝑐𝑒𝑠 = Number of slices

𝑑𝑝𝑠𝑙𝑖𝑐𝑒 = Pressure drop per slice
In the above scenario, it is assumed that the flow path distance on the plate is the same as outside
the plate, in other words the distance is mirrored around the edge of the plate, see the blue path in
Figure 7.4. This might not be the case; therefore, the flow path distance can be adjusted using a
mirror factor. However, the surface of the flow path also changes, not only the length. The step size
of the flow path becomes a variable. This makes sense because the surface outside the plate must be
completely covered by the flow paths.
Figure 7.4 – The tuning factor of the flow paths.
It is assumed the during the lifting the plate will extract sand from the bottom, the maximum amount
of sand is considered as an upper limit for the analytical model, Finn and Byrne (1972) state that very
high suction can develop under embedded object during pullout. When such suctions do develop, a
general failure can occur. This general failure is set as the upper limit of the model. Initially this is set
as the volume of half a sphere times the density of the soil. See Figure 7.5
46
Figure 7.5 – The limitation of the model.
7.3 Expectations
7.3.1 2-Dimensional model
Both the 2-Dimensional and the 3-Dimensional model can simulate the pressure development under
the plates.
The 2-Dimensional Analytical model is directly based on the cutting model of (Miedema, 2014). The
plate is simplified in the test setup to force the situation to act like a 2-Dimensional model by closing
of the 2 sides of the plate. The outcome of the pressure development of the 2-Dimensional analytical
model therefore is linear, see Figure 7.6
Figure 7.6 – Linear pressure development under the plate.
7.3.2 3-Dimensional Model

The 3-Dimensional model uses the pie slice method to accommodate for the effects occurring around
the 3-Dimensional round plate. The pressure drop, which eventually gives an estimate for the lifting
force, is strongly dependent on the length of the flow path and the surfaced of the flow path. At the
47
center of the plate the surface of the flow path is very small. This flow path gets its water supply
from the largest area outside the circle. In other word, the 1 in the equation drops as it is divided by
the surface of the flow line which increase as we move alongside the flow path away from the center.
The Ds in the equation has its maximum for the flow line going to the center. The increase of the
length of the flow line cannot match the decrease in surface area of the flow path and therefor in the
pressure development under the plate according to the 3-Dimensional analytical model an optimum
is present. This optimum, also visible in the force development, lies at about 40 percent from the
plate edge, see Figure 7.7
Figure 7.7 – The force development under half the plate.
Combining the development for the total plate gives the results shown in Figure 7.8
Figure 7.8 – The pressure development under the total plate
48
The question arises if this is realistic. Based on the measurements this phenomenon is not possible.
What is happening?
At first suppose the plate is infinite long pulled at with a certain velocity assuming a stationary
situation. Water has the tendency to flow form areas with high pressure to areas with a lower
pressure. Flow of water alongside the bottom of the plate is not incorporated in the model.
The second point is that during the calculation there was assumed water in incompressible. In fact,
water is for a small part compressible. In real the real situation the pressure will due to these 2 points
be evened out. The pressure development will look more like that in Figure 7.9
Figure 7.9 – Corrected pressure development under the plate.
49
8 Validation
8.1 Introduction
In this chapter the comparison between the model and the measurement data will be investigated,
both for the 2-Dimensional model and the 3-Dimensional model. The models itself are explained in
Chapter 7.
To validate the model, the force calculated by the model must compared with the test data. If, for
example, the test data has a standard deviation from the model the model can be corrected for this.
At the beginning of the comparison a certain permeability must be chosen, the velocity is derived
from the test data. The point where the velocity is chosen is not obvious. The first approach is to take
the mean velocity over the total breakout process.
As mentioned before in paragraph 3.3.1 the permeability of the sand was tested and found at 5e-6
m/s for the Silverbond sand and 7e-5 m/s for the Geba sand. This are the first input values for the
permeability as they will result in the maximum force, no smaller permeability is possible in the sand
bed
8.2 2-Dimensional Comparisons

For the 2-Dimensional model an input permeability is first chosen at 5e-6 m/s. combined with the
different velocities from the tests the model gives a prediction for the force. As this (Net) force was
also measured it can be compared with the model data. For the Silverbond sand his is done in Figure
8.1. As can be seen the model over estimates the force roughly by a factor 10.
Figure 8.1 – 2D - Model comparison for the Silverbond sand, the trend is visible but the model over estimates.
50
For the Geba sand the permeability used was 7e-5 m/s. Using the velocities from the test data the
model outcomes versus the test data can be seen in Figure 8.2. Again, the model over estimates.
Figure 8.2 – First predictions of the model, a large over estimation.
Scaling back the permeability with a factor 10 is not realistic. The question arises if the right velocity
was chosen. If the velocity is the mean velocity over the process it is assumed that the model
assumptions are valid during the complete process, but are they?
An important observation of the tests is that after the test almost every time a large gap can be
spotted in the sand bed underneath the plate. During the lifting process the sand bed starts to
expand with the result that the permeability of the sand increases. The water flows with less
resistance trough the sand bed.
As an input parameter of the model a constant permeability is taken. If this permeability increases
during the process the model will give an over estimation. So maybe the model is not valid during the
complete process. To validate this the velocity is analyzed again, now only at the beginning of the
process. This makes sense as the sand bed is thoroughly vibrated after every test to ensure maximum
packing and thus, in the beginning of the lifting process, the assumptions of the model should be
valid.
The velocity follows from the point where the load has reached its maximum, see Figure 8.3.
51
Figure 8.3 – The point where the velocity is derived from the test data.
Using the new velocities, the model predictions seem to give a much better fit to the measured data.
Figure 8.4 gives the model predictions for the Silverbond sand. What can be seen is that, using the
measured permeability from the laboratory, the model predictions are within the 10% margins. For
the Geba sand a factor three has to be used to fit the model predictions with the test data, this can
be seen in Figure 8.5
Figure 8.4 – The model versus the test data with the new velocities for the Silverbond sand.
52
Figure 8.5 – The model versus the test date with the new velocities for the Geba sand.
Still, using a factor three for scaling the permeability is quite large. In the beginning, circular flow
paths were assumed and they seem to work for predicting the force using the Silverbond sand. As
the Geba sand has a larger porosity it is possible that the water travels a larger distance in the sand
bed. To investigate this the tune factor was built into the model. Using a tuning factor of 2 and the
initial measured permeability of the Geba sand gives Figure 8.6. What can be observed is that using
longer flow path lengths in the model the fit, using the measured permeability and the new
velocities, is quite good.
53
Figure 8.6 – The model versus the test data using a tune factor for the flow path length.
8.3 3-Dimensional Comparisons

For the 3-Dimensional model the permeability is initially chosen at the value determined by the soil
testing at civil engineering. This value is the maximum value possible and therefore determines the
maximum force. The velocities were initially the mean velocities over the complete lifting process.
Setting out the model data for the given permeability and versus the test data for the Silverbond
sand gives Figure 8.7.
54
Figure 8.7 – The measured data set out versus the model data for the Silverbond sand with the mean velocities over the
complete process.
What can be observed is that the model over estimates the data, just as with the 2-Dimensional
model.
Adjusting the velocities in the same way as with the 2-Dimensional model gives Figure 8.8. What can
be observed is that the predictions of the model fit the measurement data within a 10% margin. The
same as with the 2-Dimensional model.
55
Figure 8.8 - The model versus the test data with the velocities from the beginning of the process for the Silverbond sand.
Note that the permeability stays the same.
For the Geba sand the first predictions can be seen in Figure 8.9. As can be seen the model over
estimates the force roughly a factor 10. Using the velocities from the beginning of the lifting process
the permeability needs a scale down of a factor 4 to give a good fit, see Figure 8.10
56
Figure 8.9 – The predictions of the model for the Geba sand using the mean velocities over the complete lifting process.
Figure 8.10 – - The model versus the test data with the velocities from the beginning of the process for the Geba sand
57
8.4 Scaling
Now the model is validated with the test result the question arises how the results can be scaled to a
real-life situation. From the test results the mean displacement of the plate needed for breakout is
known. This mean displacement combined with the load which results in under pressures determine
the amount of water that is needed during the lifting process.
All tests results showed that the plate was lifted roughly 2 millimeters before breakout. This
displacement can be translated into a volume created under the plate which lead to dilatation of the
sand. This volume increase has to be filled with water. This volume increase is due to the
displacement of the plate.
To scale the results properly it is important to check the thickness of the sand layer under the plate
which is influenced by dilatation. In the real-life situation, the sand influenced will probably have a
non-rectangular shape but for the sake of simplicity a rectangular shape will be assumed.
The dilatation is defined as
𝑁𝑚𝑎𝑥 − 𝑁𝑚𝑎𝑥
𝜖= (8.1)
1 − 𝑁𝑚𝑎𝑥
The displacement combined with the plate area determines the added volume due to the lifting
process to the influenced zone.
𝐸𝑣 = 𝑑𝑝 ∗ 𝑆𝑝 (8.2)
Where:
𝐸𝑣 = Volume created under plate by lifting
𝑑𝑝 = Displacement of the plate during breakout
𝑆𝑝 = Surface area of the plate
Now the extra volume is known the volume of the influenced zone can be calculated by dividing the
added volume by the dilatation.
𝐸𝑣
𝐼𝑣 = (8.3)
𝜖
Where:
𝐼𝑣 = Initial volume of the influenced zone before the lifting process.
The equivalent layer thickness of the influenced zone is then found by dividing this volume by the
surface area of the plate.
𝐼𝑣
𝐿= (8.4)
𝑆𝑝
58
This layer thickness can be calculated for every test with the minimum and maximum porosity as
input. These are tested at the civil engineering laboratory. The values for the Geba sand tested as:
𝑁𝑚𝑎𝑥 = 0.47
𝑁𝑚𝑖𝑛 = 0.40
And for the Silverbond sand:
𝑁𝑚𝑎𝑥 = 0.54
𝑁𝑚𝑖𝑛 = 0.39
The layer thicknesses are than calculated for every test and can be found in Appendix D. For Both the
2-Dimensional and 3-Dimensional test the mean of the layer thickness is calculated per sand type,
see Table 8.1. The layer thickness is then defined as a percentage of the plate width in case of the 2-
Dimensional setup and the diameter in case of the 3-Dimensional setup.
Test Mean Layer Thickness (mm) Percentage of width or

diameter
2-D Silverbond 13.9 6,9%
2-D Geba 6.8 3.4%
3-D Silverbond 17.9 9%
3-D Geba 6.7 3.4%
Table 8.1 – The mean layer thickness per test setup and sand type. In the right column, the layer thickness is defined as a
percentage of the 2-D plate width and the 3-D plate diameter
The influenced zone can now be scaled to a real-life situation and used to determine the size of the
real-life mud mat.
No relationship, between the mean layer thickness and the width or the dimeter of the plate, other
than a linear relationship can be concluded until more tests are conducted.
To do so the right scaling factor has to be found. At first the definition of a scaling factor:
A scaling factor of a quantity is the ratio between the value of the quantity in the prototype and the
model:
𝑥𝑝
𝑛𝑥 = (8.5)
𝑥𝑚
The time of breakout is dependent on the volume under the plate created by lifting the plate and the
flow rate. Therefore:
𝑛𝐸𝑣
𝑛𝑇 = (8.6)
𝑛𝑄
With the volume defined as length to the power three:
59
𝑛𝐸𝑣 = 𝑛𝐿3 (8.7)
The flow rate is defined as:
𝑛𝑄 = 𝑛𝑎𝑟𝑒𝑎 ∗ 𝑛𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 = 𝑛𝐿2 ∗ 𝑛𝑣𝑝 (8.8)
The upward velocity of the plate is given by Darcy:
𝑛𝑣𝑝 = 𝑛𝑘 ∗ 𝑛𝑖 (8.9)
Where 𝑖 is defined as:
∆ℎ 𝑑𝑝
𝑖= = (8.10)
𝐿 𝜌𝑔𝐿
The pressure p is linked to the imposed load per area:
𝐵
𝑝= (8.11)
𝐿2
Where B is the imposed load.
So, with 𝜌 and 𝑔 constant in the model and prototype environment, the scale factor is defined as:
𝑛𝑝
𝑛𝑖 = (8.12)
𝑛𝐿
Substituting the above in equation 8.6 gives the final scale factor:
𝑛𝐿 𝑛𝐿2
𝑛𝑇 = 𝑛𝑝 = 𝑛 ∗ 𝑛 (8.13)
𝑛𝑘 ∗ ( 𝑛 ) 𝑘 𝑝
𝐿
So, when scaling the time of breakout, the time is dependent on a length scale, pressure and
permeability if the soil conditions are not constant. For instance, when the length scale becomes 10
times larger than the breakout time becomes 100 times larger when keeping pressure constant.
It is not certain if this scale rule is valid until full scale test have been conducted.
8.5 Conclusions
For the 2-Dimensional model the predictions compared with the measured data lie within the 10
percent margins for the Silverbond sand. Using the model for predictions in the Geba sand the tune
factor needs to be adjusted to give a good fit.
Apparently when using the model in Geba sand in a 2-Dimensional situation the flow line length
needs to be adjusted. The question is of this factor used to fit the model on to the test data is
realistic. The model is built around a number of assumptions and one of the assumptions was that
the flow lines are circular. What probably is happening is that the flow from the free surface of the
sand next to the plate to the section covered by the plate is more of an ellipse form, in other words
more horizontal flow. In this way, the travelled distance of the water in the sand is lower which
results in a lower prediction of the force of the 2-Dimensional model with the Geba sand. This is
something that adjusting the tune factor proves.
60
What can be observed is that the model can give reasonable predictions given the velocity is derived
from the beginning of the lifting process. The model is valid with a constant permeability.
For the predictions of the 3-Dimensional model the same trend can be spotted. What can be seen
with the Silverbond sand is that the fit of the model over the test data is reasonable accurate. For the
Geba sand roughly a factor 4 is needed. Using a tune factor in the 3-Dimensional model did not give
satisfactory results. A tune factor of 400 was needed to fit the model over the test data with
reasonable accuracy. The problem here is that the model uses flow paths instead of flow lines. This
was done because of the circular shape of the 3-Dimensional plate as explained in paragraph 7.2.
Adjusting the tune factor does not only increase the length of the flow path but also increases the
area of the flow path over the length. Adjusting the flow path length does therefore not have such a
great impact as with the 2-Dimensional model. This is something that needs further investigation.
61
9 Conclusions and Recommendations
This chapter gives an overview of all the conclusions and recommendations made in this report. First
the 2-Dimensional conclusions and recommendations are answered, later the 3-Dimensional case.
9.1 2-Dimensional Conclusions

The preliminary experiments already showed that under pressures were present during lifting
operations in fine grained sand. The 2-Dimensional experiments show that the breakout time is
related to the permeability of the sand. A factor 10 in breakout time given the same load between
the Silverbond and Geba sand is the result. For the breakout time it can be seen that doubling the
load does not result in dividing the breakout time in halve. The Geba sand seems to have this trend
but especially the Silverbond sand differs a lot.
The pressure profile under the 2-Dimensional plate was assumed linear, the testing showed
otherwise. A more rectangular pressure profile is present.
The mean displacement before breakout for the two different sands is about the order of 2
millimeters during the testing with different loads. It can be concluded that the grain size of the sand
does not influence the mean displacement before breakout.
The 2-dimensional model seems to give reasonable predictions when using velocities from the
beginning of the lifting process. The Geba sand needs adjusting of the tune factor to give a good fit.
Further research concerning the 2-Dimensional case would be useful. The model could use some
attention. It would be useful if the model could make predictions over the complete lifting process. A
variable permeability should be taken into account as the porosity of the sand changes due to the
lifting process. If this is done correctly a time prediction could be built in to the model.
To say something about the flow lines in the sand one should consider using pressure sensors in the
sand bed. Also, more pressure points are needed to say more about the pressure profile underneath
the plate.
9.2 3-Dimensional Conclusions

The experiments with the 3-Dimensional setup showed that scaling the breakout time with the
permeability of the sand has it limits. During the experiments with the Geba sand a too large load
was chosen so that inertia dominated the breakout time. Nothing can be concluded about a relation
between the breakout time and the permeability for the same load between the two different sands.
What can be concluded is that the breakout time in Silverbond sand follows a nonlinear curve. A
same sort of relation can be seen with the experiments in the Geba sand but, as mentioned, the
experiments should be conducted with smaller increases in the imposed load.
The pressure sensors of the 3-dimensional plate gave a good indication of the pressure profile
beneath the plate. The experiments in the Silverbond sand gave a more or less rectangular shape
with steep slopes towards the edge of the plate. The experiments with the Geba sand showed a
smoother profile.
62
For the mean displacement, the conclusion is the same as with the 2-Dimensional plate. No relation
of the displacement is found with respect to the grain size of the sand.
Testing for consolidation effects showed that pre-loading the 3-Dimensional plate in sand does not
make a difference in breakout time. Pre-loading in the case of thoroughly vibrated sand therefore
does not make a noticeable difference in breakout time. Testing on loose sand is a recommendation
for future research.
The 3-Dimensional model predicts the lifting force reasonable accurate. For use with the Silverbond
sand the fit was within the 10% margins when using the velocities from the beginning of the lifting
process. For the Geba sand the permeability had to be scaled to give a good fit. This is because of
model assumptions and using flow paths. Adjusting the tune factor did not give satisfactory results.
Further research can be done in to modeling the prediction of the breakout time, this is not
incorporated in the model as is the case with the 2-Dimensional model.
63
Bibliography
Al-Shamrani, M. A. (1995). Finite element analysis of breakout force of objects embedded in sea
bottom. Riyadh, Saudi Arabia: College of Engineering, King Saud University.
Chen, R. G. (2012). Investigation of the vertical uplift capacity of deep water mudmats in clay.
Canadian Geotechnical Journal vol. 49, 853-865.
Craig, W. C. (1987). Extraction forces for offshore foundations under undrained loading. Journal of
Geotechnical Engineering vol. 116, 868-884.
Das, B. (1991). Bottom breakout of objects resting on soft clay sediments. International Journal of
Offshore and Polar Engineering vol. 1, 195-199.
Den Adel, H. (1989). Re-analyzing permeability measurements with the Forchheimer relation. .
Geotechnics report CO-272553/56.
Foda, M. (1982). On the extrication of large objects from the ocean bottom (the breakout
phenomenon). Journal of Fluid Mechanics vol. 117, 211-231.
Huang, H.M., Lin, M.Y., Huang, L.H. (2010). Lifting of a large object from a rigid porous seabed. (pp.
106-113). Taipei, China: Department of Civil Engineering, National Taiwan University.
Lee, H. (1973). Breakout of partially embedded objects from cohesive seafloor soils. Offshore
Technology Conference (pp. 789-802). Houston, Texas: U.S. Naval Civil Engineering
Laboratory.
Liam Finn, W. B. (1972). The evaluation of the breakout force for a submerged ocean platform. (pp.
863-868). Houston, Texas: Offshore Technology Conference.
Liu, C. L. (1969). Ocean sediment holding strength against breakout of embedded objects . Port
Hueneme, California: U.S. Naval Civil Engineering Laboratory.
Mei, C.C., Yeung, R.W., Liu, K.F. (1985). Lifting of a large object from a rigid porous seabed. Journal of
Fluid Mechanics vol. 152, 203-215.
Miedema, S. (2014). The Delft Sand, Clay & Rock Cutting Model. Delft.
Muga, B. (1968). Ocaen botton breakout forces. Port Hueneme, California: Naval Civil Engineering
Laboratory .
Rapoport V, Young A.G. (1983). Prediction for breakout resistance for object embedded into seafloor
soil. Houston.
Schriek, G.L.M. van der, Rhee, C. van. (2010). Physical scale modeling of dredging processes in sand
and gravel. 19th World Dredging Congress, 859-874.
Taylor, R. D. (1970). In place seafloor soil test equipment: a performance evaluation. Port Hueneme,
California: Naval Civil Engineering Laboratory.
64
Vesic, A. (1969). Breakout resistance of objects embedded in ocean bottom. Port Hueneme,
California: U.S. Naval Civil Engineering Laboratory.
Zhao, Y. (2001). The FEM calculation of pore water pressure in sand cutting process by SEPRAN. Delft.
65
Appendix A Sensor Data
For further research, it is important to note down the sensors used in the test setup. Four different
DP- Sensors were used in combination with a load cell. The manufactures and serial numbers can be
found in the tables below.
Sensor from test Manufacturer and Type Serial Number

DP1 Rosemount DP3 S22 7466512/0101

DP2 Rosemount DP3 S22 10110818

DP3 Rosemount DP3 10110817

DP4 Rosemount DP3 10110919

Loadcell Zemic H3G-C3-50kg TC7865

Displacement meter Messotron WT50K 31106
66
Appendix B Permeability Tests
The Permeability tests were conducted on the department of civil engineering of the TU delft. The
Permeability followed from a falling head test. The tests have been performed several times and the
Ks value followed from these tests. The first table gives the test for the Silverbond sand, the second
for the Geba sand.
KSAT
Software Version 1.2.0
Firmware Version 1.4
Last setting of the zero point 1-1-0001
Serial Number 0034
PARAMETER
Mode FallingHead
Sample name Test_50um_chris6_001
cross-sectional area of the burette [cm2] 4,536
Cross-sectional area of the sample [cm2] 50,18
Sample length [cm] 5,0
Plate thickness [cm] 1,0
Crown type SteelMeshCrown
Saturated plate conductivity [cm/d] 20000,000
Start of measurement 8-12-2016 11:16:47
Test duration 00:03:04
RESULT
Use auto offset adjustment True
Fitting Parameter a [cm] 6,15
Fitting Parameter b [s-1] -1,09E-03
Fitting Parameter c [cm] -,6
Fitting Parameter r2 [-] 1,0000
Ks Total [cm/d] 51
Ks Total [m/s] 5,94E-06
Ks Soil [cm/d] 43
Ks Soil [m/s] 4,95E-06
Ks Soil normalized at 25,0 Â°C [cm/d] [cm/d] 42

Ks Soil normalized at 25,0 Â°C [cm/d] [m/s] 4,94E-
06
KSAT
Software Version 1.2.0
Firmware Version 1.4
Last setting of the zero point 1-1-0001
Serial Number 0034
67
PARAMETER
Mode FallingHead
Sample name Gebat2_001
cross-sectional area of the burette [cm2] 4,536
Cross-sectional area of the sample [cm2] 50,18
Sample length [cm] 5,0
Plate thickness [cm] 1,0
Crown type FilterPlateCrown
Saturated plate conductivity [cm/d] 20000,000
Start of measurement 17-11-2016 11:54:34
Test duration 00:00:54
RESULT
Use auto offset adjustment True
Fitting Parameter a [cm] 5,18
Fitting Parameter b [s-1] -1,52E-02
Fitting Parameter c [cm]
Fitting Parameter r2 [-] 0,9996
Ks Total [cm/d] 750
Ks Total [m/s] 8,72E-05
Ks Soil [cm/d] 600
Ks Soil [m/s] 6,97E-05
Ks Soil normalized at 25,0 Â°C [cm/d] [cm/d] 595

Ks Soil normalized at 25,0 Â°C [cm/d] [m/s] 6,92E-
05
68
Appendix C Sieve Analysis.
The sieve analysis has been performed by Ir. Rik Bisschop during his Phd research. two of the sands
he used in his research were the Silverbond and the Geba sand. The sieve analysis is shown below:
D10 D15 D50 D60 D60/D10

Silverbond 0,092 0,098 0,125 0,133 1,45
Geba 0,017 0,021 0,051 0,057 3,35
69
Appendix D Layer Thickness
2-Dimensional test, Silverbond Layer Thickness (mm)
2D-S-130N -2 15.1
2D-S-130N -3 11.4
2D-S-150N -1 11.4
2D-S-150N -2 15.1
2D-S-170N -2 15.1
2D-S-170N -3 15.1
2-Dimensional test, Geba Layer Thickness (mm)

2D-G-90N -3 9.2
2D-G-110N -1 9.2
2D-G-110N -2 9.2
2D-G-110N -3 6.1
2D-G-130N -1 3
2D-G-130N -2 4.6
2D-G-130N -3 6.1
3-Dimensional test, Silverbond Layer Thickness (mm)

3D-S-65N -1 15.1
3D-S-65N -2 15.1
3D-S-65N -3 30.3
3D-S-70N -1 11.4
3D-S-70N -2 22.7
3D-S-70N -4 18.9
3D-S-90N -1 15.1
3D-S-90N -3 18.9
3D-S-110N -1 18.9
3D-S-110N -2 15.1
3D-S-110N -3 15.1
3-Dimensional test , Geba Layer Thickness (mm)

3D-G-50N -1 7.7
3D-G-50N -3 6.1
3D-G-60N -1 7.7
3D-G-60N -2 7.7
3D-G-60N -3 6.1
3D-G-70N -1 6.1
3D-G-70N -2 6.1
3D-G-70N -3 6.1
70
Appendix E 2-Dimensional Test Results
In this appendix, the 2-Dimensional test results are shown. From the many graph produced, there
was chosen to show the Load and the displacement versus the time in one graph and the Pressure
and the velocity versus the time in the other. All used data can be derived from these 2 graphs. On
the start of every page the test number is mentioned.
71
2D-S-130N -2
72
2D-S-130N -3
73
2D-S-150N -1
74
2D-S-150N -2
75
2D-S-170N -2
76
2D-S-170N -3
77
2D-G-90N -3
78
2D-G-110N -1
79
2D-G-110N -2
80
2D-G-110N -3
81
2D-G-130N -1
82
2D-G-130N -2
83
2D-G-130N -3
84
Appendix F 3-Dimensional Test Results
85
3D-S-65N -1
86
3D-S-65N -2
87
3D-S-65N -3
88
3D-S-70N -1
89
3D-S-70N -2
90
3D-S-70N -4
91
3D-S-90N -1
92
3D-S-90N -3
93
3D-S-110N -1
94
3D-S-110N -2
95
3D-S-110N -3
96
3D-G-50N -1
97
3D-G-50N -3
98
3D-G-60N -1
99
3D-G-60N -2
100
3D-G-60N -3
101
3D-G-70N -1
102
3D-G-70N -2
103
3D-G-70N -3
104

28 - 02 Report

Uploaded by

Copyright:

Available Formats

28 - 02 Report

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

28 - 02 Report

Uploaded by

Copyright:

Available Formats

TU Delft

Christiaan den Hertog

Christiaan den Hertog, Bsc

Prof. Dr. Ir. C van Rhee

Dr. Ir. S.A. Miedema

Dr. Ir. A.M. Talmon

Dr. Ir. M. van Damme

Table 3.1 – Load Parameters for the Silverbond sand. ......................................................................... 13

1.1 Problem Description

- Decommissioning of the structure: When for example a jacket is removed by lifting it

1.2 Purpose of Research

During the research the following objectives are encountered:

The main research question to be answered is as follows:

2.1 Earlier Research

2.2 Parallel Resistor method

Therefore, one can state:

Figure 2.2 – The Water flow towards the cutting zone.

For the different lengths and permeability’s:

- More parameters are easier to control.

- Suction can be measured using differential pressure meters.

3.2 Test setup

3.3 Test Parameters

3.3.1 Sand types

And for the Silverbond sand:

3.3.3 Lifting loads

Silverbond sand Load 1 Load 2 Load 3 Load 4

2D 130N 150N 170N

3D 65N 70N 90N 110N

Table 3.1 – Load Parameters for the Silverbond sand.

2D 90N 110N 130N

3D 50N 60N 70N

Table 3.2 – Load Parameters for the Geba Sand

3.4 Test Procedure

2-D Tests Silverbond

3-D Tests Silverbond

3-D Tests Geba

4.2 Net loads

2-D Tests Silverbond

3-D Tests Silverbond

3-D Tests Geba

4.3 Measurement Analysis

5.2 Time of breakout

5.2.2 Silverbond 51µm Sand

5.2.3 Geba 125µm Sand

5.3 Mean Displacement

5.3.2 Silverbond 51µm Sand

5.3.3 Geba 125µm Sand

5.4 Mean Pressures

5.5 Comparing the Two Sands

Figure 5.9 – The differences in breakout for the different sands

6.2 Mean time of breakout

6.2.1 Silverbond 51µm Sand

6.2.2 Geba 125µm Sand

6.3 Mean Displacement

6.3.2 Geba 125µm Sand

6.4 Mean Pressure

6.5 Comparing the Two Sands.

Figure 6.11 – Underwater pre-loading