Analysis of Soil Machine Interactions (Part 1): Processing of TBM-Machine-Data and Extraction of excavation-specific Data

Transcription

1 EURO:TUN rd International Conference on Computational Methods in Tunnelling and Subsurface Engineering Ruhr University Bochum, April 2013 Analysis of Soil Machine Interactions (Part 1): Processing of TBM-Machine-Data and Extraction of excavation-specific Data Jan Düllmann 1, Fritz Hollmann 2, Markus Thewes 2 und Michael Alber 1 1 Engineering Geology / Rock Engineering, Ruhr-University Bochum, Germany 2 Institute for Tunnelling and Construction Management, Ruhr-University Bochum, Germany Abstract Every drive with a shield machine provides the user with a large amount of machine data, which may be used to analyse the interaction between the ground and the TBM. A special focus is on the data, which is connected to the movement of the cutting wheel and the shield. Variations of these data are often interpreted as subsoil changes or as a result of adverse effects on the cutting wheel (e.g. clogging or tool wear). However, these data also depend significantly on technical and human factors, which are not connected to the excavation process. Data analyses without extensive revision of these effects therefore may cause misinterpretations. In this first part of the paper, the various influences, particularly from face support, on the machine raw data are specified. As a result a methodology will be presented to extract relevant parts from the raw-data. The importance of the data revision is shown in practical examples, which could have led to false interpretations when analysing the soil-machineinteraction with unrevised machine raw-data. For future projects, the extraction of the excavation-specific data components will allow analyses with improved relevance. 1

2 Jan Düllmann, Fritz Hollmann, Markus Thewes, Michael Alber 1 INTRODUCTION Data collected during shield tunneling are frequently used for drawing conclusions about the interaction between machine and the geologic media. Changes in recorded machine raw-data are often attributed to changes in ground conditions. However, the machine data are influenced by numerous factors such as the state of the cutting wheel, the steering of shield or errors / tolerances in measurements. This first part of the paper deals exclusively with quality and processing of the machine raw-data. The status of the documentation of and at the site as well as the effects of certain ground conditions on machine behavior/data is not discussed. Possible sources for errors in raw-data as well as remedial measures are given. Thus, it is possible to define excavation-specific components of machine data which are assumed to be independent of technical or human influence. In this paper the analysis is limited on thrust force on cutting wheel, penetration and total thrust force. It has to be considered, that such processing of raw-data may be necessary for other parameters as well. The analysis of the excavation-specific components itself will be presented in part 2 [1] of the paper. 2 CURRENT PRACTICE OF MACHINE DATA ANALYSES Some raw-data are significant for evaluating the interaction between machine and ground conditions and refer to either the shield progress or face [2]. Sudden changes in the aforementioned data may reflect a change in ground conditions as the data directly or indirectly depend on those conditions. However, those data must be separated in active and passive parameters: An active parameter is given by the operator and is thus a pre-determined value. Active parameters include cutting wheel rpm as well as pressure of thrust cylinders which pushes the TBM forward. Additional active parameters are face support pressure and slurry flow rate with the use of mix-shields or the rotation speed of the screw conveyer with EPB shields. Passive parameters (thrust force on cutting wheel, torque on cutting wheel, advance speed and rate of penetration) are the result of active parameters. The determination of values for active parameters aims for a specific advance speed / penetration under given site conditions. The cutting wheel torque and thrust are, however, not target rather than limiting values for active parameters. 2

3 Analysis of Soil Machine Interactions (Part 1) Immediate observation of instant values is mandatory during operations. This, however, cannot be achieved with the use of standard spreadsheets making specialized tunneling information systems necessary [2]. The data of the active parameters are already displayed in real-time in the operator s cabin [3]. Typically, it is the operator who has continuous access to and supervises these parameters. Additionally, some data are automatically compared to limits which are specified before shield drive [4]. During operations those data are of particular interest which are important for safe tunneling with little settlements (i.e. face support pressure, delivery, densities and pressure for backfill grouting). Of similar importance may be a sudden increase in cutting wheel torque and thrust force while the penetration remains the same or is decreasing. This observation may hint at the necessity of either a cutting wheel inspection with respect to tool wear and clogging or a documentation of changed ground conditions. 3 MEASUREMENT AND STORAGE OF DATA Today s TBMs integrated automatic data logging are based on complicated measurement systems. Hundreds of sensors are placed on various locations along the tunnelling system and continuously transmit analogue electrical signals (typically Voltage in mv). The signals are processed by AD converters into physical units such as pressure. Each sensor features a specific range of application (e.g. pressure sensors with a range of bars) as specified by the manufacturer. Outside the specified range the converted values may be erroneous. With significant overloading, even for a very brief time, above the sensor s design range the device may be damaged or destroyed. In any case, electrical zero or the linearity of the sensor may experience an offset leading to a need for a recalibration of the senor (which in many cases is never done). Thus, the use of under-dimensioned sensors may lead to erroneous measurements which cannot be compensated later. Over-dimensioned sensors negatively affect the accuracy of a measurement as a sensor features error margins as a function of its dimension (e.g % FS (of the full scale)). In addition, the resolution of the measurements is coarser by using only a small part of the allowable range. The calibration (the correct reflection of the actual physical status) of the sensors is typically done ahead of tunnelling operations. Control and re-calibration during operations are difficult, because of the large number of sensors. This may lead to unnoticed zero-shifts and changes in linearity of the sensors. Comparable sensors in 3

4 Jan Düllmann, Fritz Hollmann, Markus Thewes, Michael Alber laboratory practice need to be checked and re-calibrated on a regular basis because zero shift and changes in linearity occur even without exceeding the sensor s limit (manufactures specify long- term accuracy of sensors in the range of 0.1% FS per year). Compared to the controlled laboratory conditions, the sensors on a construction site are exposed to very adverse conditions. They include large variations in temperature and humidity, vibrations and damages by force as well an extensive cable length. The sensors endure these conditions over a period of several months which may have negatively affects the sensor s accuracy. The reliability of the recorded machine data suffers and re-calibration of the sensors is necessary. Sensors adapted to those conditions are necessary to install. 4 EXCAVATION-SPECIFIC DATA COMPONENTS Ground conditions as well as the current state of the cutting wheel influence the machine raw-data and their change by trend. The actual effect of these two factors may only be assessed by additional investigations. We introduce the excavationspecific component, combining both (ground conditions and state of the cutting wheel) as the respective effects on the raw-data in many cases cannot be clearly distinguished. Additional influences on raw-data stem from operators skills, face support pressure and friction forces. These effects are independent of excavation processes. The raw-data in the database therefore consist of excavation-specific and excavation-independent data components as specified in table 1. Table 1: Excavation-specific and excavation-independent components of machine raw-data Machine raw-data Excavation-specific data components Excavation-independent data components Influence from resistance to excavation Influence from state of cutting wheel Technical components Human components - Influence from ground - Design of cutting wheel - Influence from face - Operator s skills conditions (e.g. soil (diameter, grade of support (adjustment of active density) aperture, type of tools) - Influence from friction parameters) - Actual state of tool forces wear and clogging 4

5 Analysis of Soil Machine Interactions (Part 1) It is obvious that machine raw-data has to be partially revised to allow for meaningful interpretations of interactions between TBM and ground conditions. Therefore the excavation-specific data components (see table 1) have to be identified. 4.1 Thrust force on cutting wheel For shield tunneling without face support, the parameter thrust force on the cutting wheel (F Th ) represents the sum of friction forces (F C + F B ) and contact force of cutting wheel at the tunnel face (F T ). This parameter (F Th ) is also used for analyzing ground conditions when applying face support. The raw-data (F Th ) are calculated from the pressure and the cross-section of the rams. Three groups of cylinders, each equipped with 2 or more individual cylinders, allow with many TBMs controlled displacement and tilting of the cutting wheel. They also transmit the forces from the cutting wheel and cutting wheel drive to the shield. With face support, the pressurized chamber is separated from the area with atmospheric condition by the submerged wall. For all shield types with face support, the face support pressure acts not only on the face itself but also on the area of the cutting wheel drive. This force acts opposite to the drive direction and has to be compensated for by the cutting wheel displacement cylinders. Figure 1: Schematic presentation of forces acting on cutting wheel of a hydro-shield machine: F T -contact force of cutting wheel at tunnel face; F S - resulting force of face support pressure acting on area of cutting wheel drive; F B - internal friction forces of main bearing; F C - internal friction force of hydraulic cylinders; F Th - thrust force on cutting wheel raw-data. 5

6 Jan Düllmann, Fritz Hollmann, Markus Thewes, Michael Alber In this case the excavation specific data component is the cutting wheel contact force at the tunnel face (F T ), and will be called contact force CW in the following. In order to evaluate the contact force CW the force of the slurry on the area of the cutting wheel drive (F S ) has to be considered. Even at low support pressures, evaluations show a significant influence of face support force on raw-data. The resulting face support force increases nonlinear with the diameter of machine, as it is presented in figure 2. Figure 2: resulting force from face support pressure at different pressures, depending on cross section diameter (size of effective area) Calculations of face support forces have been carried out for three projects and the findings are presented in figure 3. The column charts show average values for whole project distance. Raw-data of thrust force on cutting wheel represent 100 % and calculated forces from face support pressure are given as a percentage of these rawdata (friction forces are not considered). Figure 3: face support force portions on total thrust force raw-data (friction force are not considered) for different hydro-shield projects 6

7 Analysis of Soil Machine Interactions (Part 1) It was found, that in any case of face support, independent of machine type, geology or used pressures, the influence of face support force on raw-data is very significant. Similar results were found by Festa et al. [5]. Additionally the results given in figure 3 show a trend to lower portions of face support force with increasing excavation resistance of the ground. Additional forces to be taken into account are various frictional forces. These forces include internal friction of guidance within the main bearing of the cutting wheel drive and friction of the individual thrust cylinders for cutting wheel displacement. It is neither possible nor useful to distinguish between single frictional forces. However, they amount to significant magnitudes as may be seen when closely analyzing instantaneous data of cutting wheel displacement without face contact. Several bar of oil pressure are necessary to move any pistons (F C ) as well as the cutting wheel drive (F B ). This pressure is termed oil pressure for idle. The idle pressure has to be converted to constant frictional force and this force has to be included in the analysis of raw-data. The (theoretically) constant friction force has to be added to the raw-data because parts of contact force CW and face support forces are compensated by those friction forces. As discussed above the computation of the contact force CW may be summarized in the following formula (1): F T = F Th - F S + F C + F B (1) F T = contact force of cutting wheel at the tunnel face [kn] F Th = thrust force on cutting wheel raw-data [kn] F S = resulting force of face support [kn] F C = internal friction force of hydraulic cylinders [kn] F B = internal friction forces of guidance within main bearing [kn] By processing the data of different projects no consistent values for friction forces, but variations of more than 100 % (within each project) were found. A clear reason for this remains ambiguous, so that for further estimations values had to be assumed. 7

8 Jan Düllmann, Fritz Hollmann, Markus Thewes, Michael Alber 4.2 Penetration The analyses of the excavation process using the penetration raw-data it is not meaningful without regarding other parameters, because penetration is directly influenced by the active parameters (see section 2). A change of penetration magnitude during excavation hints first of all only a change of active parameters by the operator. To receive an excavation-specific component in this case, not the technical but the human influence has to be identified. For that reason, scaling of penetration to the contact force CW is suggested and the obtained parameter is called specific penetration [6]. Pen spec = Pen raw / F T (2) Pen spec = specific penetration [mm/u/kn] Pen raw = penetration raw-data [mm/u] F T = contact force of cutting wheel at the tunnel face [kn] The relationship between specific penetration and rock mass conditions (UCS, distance of discontinuities) is well known for TBM projects in hard rock conditions, and can be used even for advance rate estimations [7]. Therefore the use of specific penetration values for analyzing shield driven tunnel projects is not a reinvention. Processing of penetration raw-data is not a splitting of different components but a combination of an active and a passive parameter. Related to evaluations of three different projects the findings show that it may be much easier to use specific penetration as one combined parameter, as to analyze two or more parameters separately (e.g. penetration, contact force CW and slurry density). 8

9 Analysis of Soil Machine Interactions (Part 1) 4.3 Total Thrust Force Total thrust force comprises many different components and forces within the cutterhead and shield of a TBM [3], [8]. Different forces are displayed in figure 4 and have to form an equilibrium. Processing of total thrust force aims for the calculation of shield friction force as the excavation-specific component. Hints for a theoretical calculation are given in Herzog [9]. An additional resistance to the cutting edge of the shield is neglected due to the larger diameter of the excavation than the diameter of the shield [3]. Figure 4: Schematic presentation of forces acting on shield and cutting wheel of a hydro-shield machine: F Th total - total thrust force raw-data; F S - resulting force of face support pressure acting on TBM cross section; F T -contact force of cutting wheel at tunnel face; F Sh - shield friction force, resulting from earth pressure and bedding stresses; F C - internal friction force of thrust cylinders; F B - pulling force of backup-equipment. As shown in figure 4 it is obvious, that for processing of F Th total many different components have to be considered. First of all, resulting force of face support pressure (F S ) and friction forces (F C ) are crucial. In this case, face support force acts on the total cross sectional area of TBM. Due to that fact, resulting forces (F S ) have a large influence on total thrust force raw-data. The friction force of the thrust cylinder have to be taken into account, too. Moreover, the pulling force for the backup equipment (F P ), the contact force CW (F T ) and the shield friction force (F Sh ) have to be subtracted from raw-data. The magnitude of the contact force CW was discussed in section 4.1. There are other forces such as friction between brush seal and segments which should be included, their magnitude is however considered negligible. 9

10 Jan Düllmann, Fritz Hollmann, Markus Thewes, Michael Alber Theoretically, the shield friction force may be calculated as follows: F Sh = F Th total - F S - F C - F T - F P (3) F Sh = shield friction force [kn] F Th total = total thrust force raw-data [kn] F S = resulting force of face support pressure [kn] F C = internal friction force of thrust cylinders [kn] F T = contact force of cutting wheel at the tunnel face [kn] F P = pulling force of backup-equipment [kn] Calculations have been carried out for one project and the findings are presented in figure 5. The regularity of support force portion on raw-data is shown in figure 5, left side. The black line represents calculated face support force percentage on raw-data. Raw-data of total thrust force is represented by 100%. Figure 5: left side: portion of resulting face support force (black line) on total thrust force rawdata (100%); right side: calculated average values for each component The deviation between face support force and raw-data comprises contact force CW, pulling force of backup-equipment, frictional forces of thrust cylinders and shield friction force. In this case the calculated average amount of these different components represents only approx. 24 % of the raw-data, which can be seen in figure 5, right side. 10

11 Analysis of Soil Machine Interactions (Part 1) 5 CONCLUSION AND SUGGESTIONS In this paper (part 1 of 2), options and limits of processing and analyzing of machine data were demonstrated. Thrust forces on cutting wheel, penetration and total thrust force were regarded. It was shown that the relevance of automatically generated rawdata without further processing is very limited for interpretations of soil conditions. Interpretation of automatically generated raw-data with respect to ground conditions may be particularly misleading. Raw-data may be split into different components. Excavation-specific components (influences of soil conditions and cutting wheel state) in many cases show only little portions on raw-data, whereas excavation-independent components like face support force or friction force show major influence. Thus, it is recommended to separate excavation-specific components as accurately as possible. Although theoretical approaches are given, subsequent evaluation of feasible values is difficult, because of very alternating conditions in the course of shield driven tunnel projects. Measuring errors may occur as a result of strong vibrations, temperature and moisture deviations in addition to large cable length, influencing complicated measurement setups. Frictional forces may change in course of construction as a result of wear effects and entry of fines (mud, dust, clogging). Thus, the significance even of processed data is influenced negatively and subsequent estimations or even corrections in most cases are hardly to realize. Despite this persistent inaccuracy, processing of raw-data is very useful, because soil conditions will show larger influence on excavation-specific components as on rawdata. Variations of machine data may be shortly detected as a result of technical or excavation-specific reasons on the one hand. On the other hand, little variations of excavation-specific components will not lead to any variation of raw-data. Continuous observation of the excavation-specific components may form the chance to recognize critical influences (e.g. clogging) much earlier. For further projects the following suggestions are offered in the following table 2: 11

12 Jan Düllmann, Fritz Hollmann, Markus Thewes, Michael Alber Table 2: Suggestions for machine data interpretations of future projects before start of excavation calibration and check of all sensors measuring excavation relevant parameters (linearity and zero shift for AD conversion) * determination** of internal friction forces within the guidance of main bearing of cutting wheel and hydraulic cylinders by moving the cutting wheel forward and backward determination** of internal friction forces within the thrust cylinders by moving each singe cylinder group forward and backward (determination of idle pressures) visualization of excavation-specific components in operator s cabin as an additional tool for excavation survey after start of excavation Regular interval checks of all sensors measuring excavation relevant parameters and recalibration if needed (linearity and zero shift for AD conversion) Regular interval checks** of determined values for internal friction forces within the guidance of main bearing of cutting wheel and hydraulic cylinders by moving the cutting wheel forward and backward Regular interval checks ** of determined values for internal friction forces within the thrust cylinders by moving each singe cylinder group forward and backward (idle pressures) * adequate dimension of sensors has to be considered in course of TBM development ** logging frequency should be raised for determination and check of internal friction forces 12

13 Analysis of Soil Machine Interactions (Part 1) It is strongly recommended to use excavation-specific data components for all kinds of analysis, focusing on the interaction of machine data and ground conditions. After identification of technical and human influences it has to be regarded, that excavationspecific components are influenced by ground conditions and state of cutting wheel as well. To compare different projects, cutting wheel design (diameter, degree of aperture, type of tools) has to be taken into account. Influences of tool wear and clogging always have to be regarded within every project. For an exact determination of those influences, a continuous documentation at project site is necessary. ACKNOWLEDGEMENTS The presented research is part of project A1 within the Collaborative Research Center SFB-837 (Interaction modeling in mechanized tunneling), funded by the German Research Foundation DFG. Additional funding was provided by Herrenknecht AG, Germany. REFERENCES [1] Hollmann, F., Düllmann, J., Thewes, M, Alber, M. (2013): Analysis of Soil Machine Interactions (Part 2): Influences on the excavation-specific Data on TBM-Machine Data. EuroTun2013 [2] Maidl, U., Nellessen, P., Zukünftige Anforderungen an die Datenaufnahme und auswertung bei Schildvortrieben. Bauingenieur 78, [3] Maidl, B., Herrenknecht, M., Maidl, U., Wehrmeyer, G., Mechanized Shield Tunneling. Ernst &Sohn, Berlin. [4] Stahl, F., Babendererde, L., Site Supervision and Quality Assurance of a Project with several TBM Drives. Tunnel 3/2009, 2-9. [5] Festa, D., Broere, W., Bosch, J.W., An investigation into the forces acting on a TBM during driving Mining the TBM logged data. Tunneling and Underground Space Technology 32,

14 Jan Düllmann, Fritz Hollmann, Markus Thewes, Michael Alber [6] Rutschmann, W., Mechanischer Tunnelvortrieb im Festgestein, VDI- Verlag, Düsseldorf. [7] Alber, M., Prediction of penetration and utilization for hardrock TBMs. Proc. ISRM Int. Symp. Eurock 96. Rotterdam: Balkema, [8] Girmscheid, G., Baubetrieb und Bauverfahren im Tunnelbau, Ernst & Sohn, Berlin. [9] Herzog, M., Die Pressenkräfte bei Schildvortrieb und Rohrvorpressung im Lockergestein, Baumaschine und Bautechnik 6/85,