2.2.1HWD deflection measurements vs deep anchor information
A wide-scale HWD survey has been performed on the LCPC’s fatigue carrousel between April 2008 and July 2009 on pavements dedicated to a multiwheel effect study [Homsi et al., 2010] consisting in studying influence of axle configuration on different road pavement structures.
Several studies have been carried out using HWD, as - a deflection study on the several structures of the annulus,
- a specific study of the water table influence (not presented in this work) ,
- HWD measurements over embedded gauge profiles. This work will be detailed in part 3. It has been performed on the S2 structure described in appendix A, which is the thicker
pavement of the annulus and thus the closer to airport structures. - a comparison between deflections given by HWD and a deep anchor.
Let us interest here to the latter experiment. It has been performed July 2009. The deep anchor is composed of a LVDT (Linear Variable Differential Transformer sensor) measuring the surface displacement with regard to a deep reference (here a 3 m deep concrete blocking). Measurements have consisted in (manually, due to sought precision) placing, for geophones 2 to 9 successively, the HWD so that the considered geophone metal tip is placed over the deep anchor head, and performing a HWD test in this configuration, when recording in parallel the LVDT response.
Fig. 1-33 displays the comparison between HWD and deep anchor time-related deflections measurements for the second geophone. Note that time was manually adjusted as HWD and external recording systems are independent. Applied peak force was about 75 kN. Tests have been performed in the early morning to minimize temperature variations and gradient. This gradient is all the same 7 °C. Temperature ranges from 24°C (surface) to 31°C (see appendix 1.2). A slight non return to zero at the end of signal is observed. As this phenomenon is not observed in Fig. 1-3, at low temperatures, this may reflect some viscous behaviour of asphalt materials.
Correlation between the two signals is of same quality for the 3 falls and the 8 considered geophones, so that it can be assumed that deep anchors measurements are reliable for HWD dynamic assessment. Geophone 1 has not been tested because positioning was too
complicated.
Fig. 1-33 Correlation between HWD and deep anchor surface deflection measurement; geophone 2
This experiment opens interesting new prospects. Whereas HWD record is limited to a 120 ms time frame, the external instrumentation presents the advantage of allowing deflection measurement on time frames as large as requested. It allows:
- following the response of the pavement under static setting of the HWD foot, - observing successive occurring rebounds,
Fig. 1-34 displays results for all geophones. Force and geophone 1 response are provided by HWD. Time origin has been fitted using HWD geophone 1 response. Fig. 1-35 shows the good repeatability of applied load for all tests.
Fig. 1-34 Long-time time-related deflections provided by a deep anchor
Fig. 1-35 Repeatability of applied force between tests
Partial conclusion
The deep anchor of LCPC’s fatigue carrousel has allowed checking deflections values provided by HWD. Thus it appears that thistype of external instrumentation is an interesting tool to assess HWD deflections reliability. Two deep anchors have been set on the STAC’s Bonneuil test facility in this purpose.
Second interest of deep anchors is allowing measurements on extended time frames. It will be taken advantage of these findings in the following (see § 3.2).
2.2.2Comparison withexternal instrumentation data
The study presented here is the second part of the accelerometer-based external
instrumentation experiment. The main purpose is here to check the value of time delay observed in Fig. 1-3. Direct measurements of force sensor and central geophone raw signals are used, as it is practically too complicated to measure central deflection with an
accelerometer. Besides acceleration measurements are used to assess the applied force and deflections. These values are compared with data from HWD data measuring chain.
Once more a spider has been used. Unfortunately, because of limited memory of the
apparatus, when desiring recording at least principal impact and first rebounds with a 3 200 Hz acquisition rate, only two channels can simultaneously been followed. Nevertheless repeatability of results is good enough so that tests can be multiplied.
• Preliminary studies
- influence of external parallel connection
Before testing, it was necessary to check that parallel connection on the spider has no influence on the HWD results. This study has been performed by repeating tests with and without external connection. It appears that the latter has no effect. Results are not presented here.
- integration of geophone raw signals
Fig. 1-36 presents results of integration of the geophones raw signals, compared with signals provided by HWD measuring chain, for geophones 1, 2, 5 and 9. No transfer function has been applied as the latter is unknown. A “k” gain has been applied, which seems to be common to all geophones.
A very good match is found, when only considering the increasing part of signal. It demonstrates that geophones raw signals can be used in the frame of this study.
• Time delay study
Two sets of data are compared. On the one hand the force and central geophone signals given by HWD, and on the other hand the same signals acquired by spider. When the time origin is manually fitted to superpose both force signals (Fig. 1-37; note that a gain has been applied to electrical force signal), a 3,5 ms gap is observed between the peak values of central geophone (Fig. 1-38).
Fig. 1-37 Time fitting performed on force signals
Results of the two test surveys (winter and summer) performed on the S1 structure have been
used to study evolution of time delay with the different parameters (buffers type, plate, fall height, temperature).
Temperature has only a limited influence, what is a first argument confirming that such an important time delay is not attributable to viscous effects.
The most important parameter is the fall height.
Respective mean time delays for the 400, 300, 200 and 100 mm fall heights are
3,39/3,35/3,41/3,59 ms for the winter experiment (mean pavement temperature: 6°C) and 3,56/3,61/3,67/3,90 ms for the summer experiment (mean pavement temperature: 30°C) No answer about the observed bias has been found from the manufacturer. It may be due to a buffering in the HWD acquisition chain which would start the deflections measurements only when a threshold signal is observed on the force value. This issue is to be deepened.
• Force vs acceleration of the falling mass
Fig. 1-39 compares applied force recorded by the HWD measuring chain and the one deduced from acceleration measurement for a 100 mm fall height. Both signals match relatively well. Nevertheless rapid oscillations are observed on the mass acceleration. They are of constant period so that they can traduce the natural frequency of a piece of the HWD foot.
• Deflection signals; geophones vs accelerometers
The relationship between HWD measured deflections and accelerometers is studied here. As a limited range of accelerations is required, 5 G accelerometers are used to provide better precisions. They are made interdependent with geophones thanks to a metal plate fixed on the geophones metal boxes (Fig. 1-40).
Fig. 1-40 Fixing of accelerometers on the geophones boxes
Fig. 1-41 and 1-42 display the comparison between deflections provided by HWD and a double integration on accelerometer signals, respectively for geophones 5 and 9. Note that acceleration and velocity signals are not at scale, but have been adjusted for readability. Matching is excellent for both geophones.
Fig. 1-42 Comparison between HWD and accelerometer deduced time-related deflection on geophone 9
• Possibility of measuring wider deflections basins, during longer times
A feasibility study to measure time-related deflections on a longer time and at higher distances is proposed here. Let us here interest to geophone signals, recorded by the spider. The protocol involves a long external heighten beam on which geophone boxes are fixed (Fig. 1-43) for high distances. First, it has been demonstrated that the external beam choice has no influence on the results. In that goal raw geophones data measured at 90 cm and 210 cm respectively mounted on the HWD and on this external beam have been compared. Conclusion is geophone measurements using external beam are valuable.
Then 6 distances have been chosen: geophones 1, 2, 5 and 9 in standard configuration, and geophones at 3 m and at 6 m on the external beam.
In order to have significant signals at 6 m from load centre, the H0 = 400 mm is retained.
As previously explained, only two signals can simultaneously be recorded. The chosen
protocol consists thus in performing 5 successive tests, the central geophone being common to all tests, and performing a time fitting on it.
Fig. 1-44 shows the electrical signals of central geophone after time fitting for the different tests. Curves match very well, what demonstrates the repeatability of the test. Elapsed time between principal impact and first rebound is this time about 450 ms.
Fig. 1-44 Electrical signals related to the different tests (central geophone)
Integrated time-related signals corresponding to the 6 geophones are displayed in Fig. 1-45 and Fig. 1-46.
Fig. 1-45 Surface displacements at several distances from load centre; integration from geophone electrical signals (1/2)
Fig. 1-46 Surface displacements at several distances from load centre; integration from geophone electrical signals (2/2)
These curves are of valuable help when interesting to waves propagation or to deflection basins. Nevertheless, it was initially expected that geophones raw data could be used, but in reality decreasing part of signals is unfortunately incorrect, so that they can not be used for dynamical analysis. It is expected that the experiment is performed again with accelerometers.
Partial conclusion
This paragraph has shown that the observed time delay between occurrence of force and central deflection signals is at least partially attributable to a measurement bias and not viscous phenomena. It will be emphasized in part 3 that this bias can have significant influence on dynamical backcalculation results, so that the issue needs to be deepened. It has also been shown that fixing an accelerometer on the falling mass is not the appropriate method to evaluate force applied to the pavement due to mechanical vibrations in the HWD foot. A dedicated calibration bank is in preparation on the STAC’s test site.
Feasibility of measuring wide deflection basins on long times has been demonstrated. The last mentioned experiment is to be performed again with accelerometers. STAC has acquired rapid acquisition system unit allowing 10 kHz acquisition rates on 24 measuring tracks.
2.2.3Crossed tests with other F/HWD
Crossed tests with the FWD of road ministry have been performed in the frame of the LCPC fatigue carrousel experiment. 22 test points have been tested in July 2009 (pavement
temperature in the 24-31°C range; see appendix 1.2). Same load level and same loading plate have been used. Results are very good in terms of maximal deflections. Results are not presented in this document.
Nevertheless time delays between force and central deflection signals are different, what confirms the sensed aforementioned bias. Fig. 1-47 summarizes results.
Mean gap between the two apparatus values is about 2 ms i.e. about 50 % of the HWD value.
Fig. 1-47 Differences between STAC’s HWD and French road ministry’s FWD occurrence of the central deflection