Could the laser tracker AT401 replace digital levelling and “Ecartometry” for the smoothing and realignment of the LHC ?
D. Missiaen, CERN, Geneva, Switzerland,
M. Duquenne, Institut National des Sciences Appliquées(INSA), Strasbourg, France.
Abstract
The laser tracker AT401 appeared on the market a couple of years ago with a very accurate distance meter, as a matter of fact the one of the Mekometer ME5000, and angular encoders almost as accurate as the ones of the best total stations. For the smoothing and realignment of the LHC components, the Survey team at CERN normally uses digital levelling measurements to determine the vertical position and offsets to a stretched wire measurements, also called ecartometry, for the horizontal one. During the last winter technical stop, a measurement of an LHC sector (3km) was realised using these three technologies in order to compare the AT401 capabilities with respect to the others. The paper will present the methodology applied, the data processing, the results obtained and the conclusions drawn for the future LHC realignment campaign which will take place during the long shut-down of 20 months starting at the end of 2012.
Introduction
In order to evaluate the capabilities of the laser tracker AT401 for the determination of the vertical and horizontal position of the LHC components, a measurement campaign has been realised during the last winter technical stop in one sector in the LHC. The sector 78, the most unstable of the LHC, has been surveyed using digital levelling measurements, offset to a streched wire measurements and AT401. For the first time at CERN, a pure 3D method has been compared to a 2D + 1 method in the frame of the smoothing of the components of a huge particle accelerator.
Cern standard survey techniques and methodology
Vertical measurement
The vertical position of the 220 cold magnets of the sector 78 was measured using the digital level DNA03 using the Cholesky double run method. The closure between the outward and return runs being almost 4 mm, which is above the manufacturer specifications, the middle part of the sector was re-measured with an optical level Na2. The deviations of the components with respect to their nominal position appear in Figure 1. It can be seen that the curve of the outward and return are similar, except in the middle part where the sector 78 “hole” is almost visible for the outward (red) and not at all on the return curve (green). The combination of both measurements as well as the Na2 ones show a flat curve (blue), the comparison with the theoretical values closing at 1 mm and the “hole” is not visible.
Figure 1 : vertical deviation with respect to theoretical
After having fixed the calculation on a point at each extremity of the sector and processed the data using the PLANE software [1][2] with a window of 63 points and a re-alignment threshold of 0.2 mm, the deviations with respect to the smooth curve appeared onFigure 2.
Figure 2 : vertical deviation with respect to a smooth curve
Horizontal measurement
The horizontal position of the components was realised using distance measurements with respect to a stretched wire of 120 m long [1][2]. No distance measurements were taken as the most critical deviation for the particles is the one perpendicular to the beam direction. It has to be mentioned that this technique is very efficient for the determination of the relative position of components but is not so accurate for a more absolute one. The calculation is fixed at each extremity of the sector and some radial constraints have been added in the adjustment to avoid a huge absolute deviation, which is not the reality, between the measured and the theoretical position. The deviation to the theoretical position is shown onFigure 3. Two magnets have a deviation in the range of 1 mm and have clearly to be realigned.
Figure 3 : horizontal deviation with respect to theoretical
These data were also processed with PLANE [1][2] using a window of 63 points as well a threshold for re-alignment of 0.25 mm. The deviations with respect to the smooth curve are presented onFigure 4.All the magnets whose deviation is above 0.25 mm are going to be realigned.
Figure 4 : horizontal deviation with respect to a smooth curve
The AT401 main parameters
The AT401 is an instrument between a total station and a laser tracker which measures horizontal and vertical angles and distances. It is equipped with the most accurate encoders, like the ones of the TDRA6000/TS30 category and with an absolute distance-meter (ADM) equivalent to the one of the Mekometer ME5000. The Powerlock technology ensures the detection of the reflector thanks to a camera located above the telescope and locks the laser beam onto it. The accuracy at 2 (MPE) given by the manufacturer of the instrument is :
-For distances : +- 10 microns at 80 m
-For angular encoders : 3.8 dmgr
preliminary studies
Some simulations were done using the LGC software [3] to determine what was the best sequence of measurementsto realise with the AT401 taking into account the following parameters :
- A free position of the instrument in the middle of the passage area at 1.5 m from the beam line
- the technical limitations of the AT401 which are a minimum distance of 2 m and a maximum of 160 m
- the technical specifications of the instrument
- the redundancies required in order to propagate accurately the measurements.
Six cases were studied :
- Case 1 : one station in front of every quadrupole measuring one half-cell (53 m) on each side
- Case 2 : one station in front of every two quadrupoles measuring one a half half-cell (79 m) on each side
- Case 3 : one station in front of every quadrupole measuring one a half half-cell (79 m) on each side
- Case 4 : one station in front of every two quadrupoles measuring two half-cells (106 m) on each side
- Case 5 : one station in front of every quadrupole measuring two half-cells (106 m) on each side
- Case 6 : one station in front of every three quadrupole measuring two half-cells (106 m) on each side
The cases 2, 3 and 6 were rapidly eliminated because the quadrupoles, which are the most critical elements of a particle accelerator, were measured with less redundancies than the dipoles for the first two cases and for the last one the overlap of measurements was too small. The 100 simulations for the cases 1, 4 and 5 gave the results shown in Table 1 for the determination of the positions of the components.
Table 1 : accuracy of determination of the points
Case / Radial(mm at 1s) / Longitudinal (mm at 1 s) / Vertical
(mm at 1 s)
1 / 0.43 / 0.05 / 0.02
4 / 0.27 / 0.03 / 0.06
5 / 0.07 / 0.02 / 0.03
It was therefore decided to realise the measurements using the 5th case, i.e. taking more measurements than strictly needed but allowing a much more extensive data processing. The sequence can be seen on Figure 5.
Figure 5: measurements sequence with the AT401
3D measurements with the AT401
The measurements
The measurements shown took place during 12 days to cover 2.5 km of the sector 78. More than 100 positions of the instruments and 6500 observations were realised by a team of two persons. The data acquisition was done with a CERN home-made software. The initialisation of the instrument was done at each station at a distance of 10 m. The H and V angles were measured with the two faces of the instrument, the distances only with face 1. 50 points were measured per station, each station lasting more than 1h.
Heating problem
The recommendation of the manufacturer is that the AT401 has to be warmed up at least 3 hours before starting the measurements. Therefore it was decided to let the instrument warm up all the night but for security reasons it was locked inside a container. The problem was that the instrument was overheated and it was reflected in the poor quality of the closure of the round of Horizontal angle measurements during the following morning. It was thereforedecided to switch on the AT401 only 3 hours before the measurements and to keep it in a ventilated place. The closures were much better as shown on Figure 6. For the vertical angles, this phenomenon was not so visible as for the Horizontal ones.
Figure 7 : closure of round of H angles
Most of the measurements with a bad closure were redone at the end of the campaign and all of them were part of the data post-processing.
Data processing with LGC
The observations were processed in 3D using the LGC software [3] with a fixed point at one extremity of the sector and an orientation point at the other end, taking into account the CERN geoid model. One calculation was done with all the observations and another one with only the observations taken at a distance of less than 53m from the instrument.
The residuals at 1of the calculations are shown in Table 2.
Table 2 :residuals (1of the LGC calculation
Observa-tions / < 53 m / allr.m.s / average / r.m.s / average
H. Angle (dmgon) / 0.6 / 0 / 1.5 / 0
V. Angle (dmgon) / 2.8 / 1.6 / 5.2 / 3.8
Distances (mm) / 0.11 / 0.00 / 0.05 / 0.00
The residuals are consistent with the specifications for the distances and the horizontal angles measurements but not for the vertical angles.
It was then decided to post-process the horizontal angles and distances measurements with LGC and to make some deeper investigations for the vertical angles. (see next chapter)
The radial (perpendicular to the beam) position of the components measured with AT401 (red) and with “ecartometry”(blue), recalculated without any radial constraints to be consistent with the AT401 data processing, is shown onFigure 8.
Figure 8 : horizontal position of sector 78 with AT401 and “ecartometry”
It can be seen that the global shape calculated with the AT401 measurements doesn’t deviate from the theoretical one by more than 4 mm, the equivalent deviation being bigger than 12 mm for the “ecartometry” measurements. This is due to the fact that the AT401 measurements are at least equivalent to a wire of 200m which is the double length of what is done in the “ecartometry” process.
Vertical angles inconsistencies
Following the results shown inTable 2, some investigations were done to understand the origin of the problem with the vertical angles.
For each position of the AT401, a comparison was done between the altitude of the 50 points along the 200 m measured by the AT401 and the one measured by direct levelling.Figure 9shows that at 106m from AT401, there is a significant difference up to 2.5 mm between both determinations.The shape is exactly the same for all the stations, it is not symmetrical and not linear w.r.t the distance of the points to the position of the AT401.
Figure 9 : comparison of altitude between direct levelling and AT401 for each position of the AT401
Moreover, a measurement was done in a test area at CERN (called TT1), where 20 monuments along 120 m were measured with the optical level N3 and with the AT401 located in two positions, one in the middle of the zone (on the left) and the other at one extremity (on the right). The same deviation as in Figure 8between the altitudes determined with direct levelling and trigonometric levelling were observed. There is a possibility that it is coming from some refraction error, as this error is not taken into account LGC but the problem is still under investigation. At the end it was decided to modelise the error using a parabola curve and to correct the vertical angles before processing them with LGC. Two different calculations were done with the AT401 data, one with all the observations and another one with the observation taken at less than 53 m from the AT401.
Table 3: residuals of the LGC calculation after correction
Observa-tions / < 53 m / allr.m.s / average / r.m.s / average
V. Angle (dmgon) / 1.7 / -0.6 / 2.6 / -1.2
The residuals of the calculations using the corrected measurements appear in Table 3. The values are now within the specifications of the manufacturer.
The vertical profiles of the components measured with the AT401 issued from the two calculation mentioned above as well as the profile measured with direct levelling (average of outward and return run) are show in Figure 10.
It can be seen that for both instruments, with a fixed point at one extremity, the deviation to the theoretical position at the other extremity is in the range of +1mm for the direct levelling and -0.5 mm for the AT401. These are surprisingly good results for the AT401 due to its principle of trigonometric levelling. But, the local position of the components is slightly different depending on the instrument used andthe AT401 measurements (red and green) show the well-know “hole” in the middle of sector 78 which was not visible this year with direct levelling.
Figure 10 : vertical position of sector 78 with AT401 and direct levelling
Data processing with PLANE
The AT401 data were also processed with PLANE as it was done for the standard “ecartometry” measurements using the same parameters. Two calculations were done, one with all the AT401 measurements and the second with the ones which distance to the AT401were lower than 53 m.
Figure 11 : deviation to a smooth curve in H
In the horizontal direction, Figure 11 shows the histogram of the difference of the deviations to the smooth curve issues from the measurements taken with the AT401 and the measurements taken with the stretched wire. It shows that both measurements give similar results”, 85% of the deviations being within 0.1 mm, with only two points being at more than 0.3 mm. The AT401 measurements taken at a maximum distance of 53 m (red bars) produces a comparison to the “ecartometry” which is a bit better than the longer ones. This means that the points taken at 106 m, which gives certainly a more robust global shape, don’t improve the relative position of adjacent magnets, which is the most important in the process of alignment of a particle accelerator.
In the vertical direction, the processing has been done with raw AT401 data and corrected data. Figure 12 and Figure 13 show the histograms of the difference of the deviations to the smooth curve issues from the measurements taken with the AT401 and the measurements taken with direct levelling. As for horizontal measurements, it shows that both measurements give similar results, 80% of the deviations being within 0.1 mm, with only one point being at 0.7 mm.
Figure 12 : deviation to a smooth curve in V for short distances (< 53m)
For the short distances measurements (Figure 12), there is no significant difference between the data with correction of modelisation and the ones without this correction. For the long distances measurements (Figure 13), the correction of modelisation generates a shift of 0.2 mm centring the histogram around 0 and narrows the histogram around it. In general, both figures don’t show clearly that long distances give better accuracy than shorter ones.
Figure 13 : deviation to a smooth curve in V (all measurements)
Benefits of the AT401
Longitudinal position of the cryostats
The measurements taken with the AT401 provided also with a very accurate longitudinalposition of the cryostats for the whole sector [4]. The calculation was done with only one fixed point at one extremity and the residuals for the distance measurements was much lower if all the measurements were considered (0.05 mm at 1 s) than if only the measurements up to 53 m were taken (0.11). This is shown in Table 2. Figure 14shows that the two calculations give different results and that the deviations are smaller with all the measurements included, the max longitudinal deviation being less than 2mm. Surprisingly also, a magnet in the middle of the sector (the dipole MB.30R7) has a longitudinal deviation at the entry and at the exit which is different by almost 2 mm. Even if this magnet is 15 m long, this difference cannot be explained by a thermal contraction. It seems also that the vertical “hole” of the sector 78 has an influence on the longitudinal position.
Figure 14 : longitudinal position
Radial deformation of the cryo-dipoles
During the measurements done with the AT401, for some cryo-dipoles, not only the points E and S at the extremities were taken but also the middle point M, which is not the case when “ecartometry” measurements are done [4].
Figure 15 : radial deformation
Figure 15 shows that for many dipoles there is a radial deformation of the cryostat, with a general tendency of the centre of the cryostat going towards the exterior of the LHC. This is visible because the radial deviations of the E and S points are rather different that the one of the M point. The more obvious case is the MB.C20L8 with a deformation of about 0.5 mm. This phenomenon was never observed in the past but it seems that the magnets inside the cryostats are not affected too much by this deformation, as it was never detected by the beam operation team.
Train studies
There is always the idea at CERN to realise the survey of the LHC with a train suspended to the existing monorail fixed to the tunnel vault. Some sensors are actually under studies, especially a portable HLS for the vertical determination and the photogrammetry technique with respect to a stretched wire for the horizontal one[5]. Another possible sensor could have been the AT401 as it is capable to measure the H and V position in a single shot. From the point of view of the accuracy, the measurements done during this campaign are very promising but many problems still have to be solved. The first one is the stability of the instrument. It seems difficult to have it stable enough to make accurate measurements without an external removable supporting system fixed to the floor [4]. The second one is the targeting of the points to be measured. It will not be possible financially to equip all the LHC fiducials with a corner cube CCR1.5. Moreover, the corner cubes need to be oriented towards the AT401 for each position of the instrument. In this case the only solution would be to have a small dedicated robot moving all along the train to realise this operation. A sketch of a possible set up is shown inFigure 16.