Sei sulla pagina 1di 9

Conference for Landmanagement, Geoinformation,

Building Industry, Environment


"I N T E R G E O E A S T"
1-2 March 2007, Sofia

INVESTIGATION OF MAXIMAL ACCURACY


OF CIVIL ENGINEERING STRUCTURES LEVELLING
BY USING LASER LEVELS
Gligorije Perović, Stojančo Vučkov, Darko Anđić
Serbia, FYR Macedonia, Montenegro

Key words: Laser level, civil engineering structures, levelling, accuracy.

SUMMARY

The geometric levelling accuracy by means of laser (digital) levelling instruments ranges from 0.2
mm to 1.0 mm per kilometer (Torge, Geodesy; W. de Gruyter, Berlin – N. York, 2001), the
accuracy of 0.2 mm to 0.5 mm per kilometer being achieved in favourable measurement conditions.

However, the construction and auscultation of some civil engineering structures requires a high or
the highest accuracy of levelling, which has to be better than 0.5 mm per kilometer, while
measurements have to be realized in very hard, i.e. unfavourable conditions. Therefore authors
investigated variance components, i.e. possibilities of levelling by using laser leveling instruments,
which was done during the auscultation of a dam and the renewal of a fortress. Thus, a very high
accuracy of levelling was achieved - of the order of 0.2 mm - 0.3 mm per kilometer.

INTRODUCTION
In the building, reconstruction or auscultation of many civil engineering structures precise geodetic
measurements, including also a precise levelling, are necessary. However, all these measurements
are performed in unfavourable conditions. Especially, the precise levelling of civil engineering
structures is realised under unfavourable atmospheric conditions, such as high and low air
temperatures, strong wind, high concentrations of dust and humidity in the air, bad visibility, etc.
Besides, the terrain upon which the levelling takes place, is also unfavourable: steep and covered by
sand - gravel - stones, soft or rugged, with large altitude differences, and the measurements are
disturbed by activities of various machines in the vicinity of the level which produce the terrain
vibration, etc. As a consequence, major measuring errors appear, as well as a high error
accumulation caused by a significantly enlarged number of stations per kilometer of levelling
section compared to the case in national levelling networks where the measurements are performed
under favourable conditions. All this leads to an accuracy in geometric levelling of civil engineering
structures lower than that in the national levelling networks.
The measuring errors for the classical levelling have been studied in details (e.g. GUGK 1974,
Pelzer and Niemeier 1984), whereas for the levelling with digital levels this is not the case.

For all the reasons given above and also for the purpose of achieving a high accuracy of geometric
levelling the present authors have studied the accuracy of the measuring method with digital levels
at the following civil engineering structures: Perović and Vučkov – at GEOPOLYGON ''NOVO
SELO'' in Macedonia in 1996 and 1997 (RGU Makedonije 1996, 1997, Vučkov 2001), and Perović
and Anđić in 2006 – at GEOPOLYGON ''TVRĐAVA NIKŠIĆ'' and at two levelling bases in
Podgorica (Montenegro). Geopolygon ''NOVO SELO'' was formed in 1988 for the purpose of
auscultation of dam ''NOVOSELKA'', ''TVRĐAVA NIKŠIĆ'' is under reconstruction, whereas the
levelling bases in Podgorica have been specially formed in order to study the levelling errors.

DESCRIPTION OF THE PRECISE-LEVELLING METHOD

The precise-levelling method of civil engineering structures can be briefly described in the
following way (also Perović 2007):

1) Levels 1) Digital, precise;


2) Line of sight setting 2) ≤ 0.1′′ , (≤ 0.5 ⋅ 10−6 ) ;
3) Atmospherical conditions 3) Without precipitation;
4) Temperature adaptation of level 4) More than 60 minutes;
5) Tape of levelling rod 5) Invar, bar codes;
6) Rod supports 6) Foot plates for stable terrain;
Pegs for soft terrain;
7) Level protection against insolation 7) Parasol;
8) Line of sight length 8) 2 m ≤ d ≤ 50 m ;
9) Measurement reliability at the station 9) Two measurements with different heights
above the ground of instrument;
10) Measurement reliability in the section 10) Measurement forward and backward;
11) Weakening of error influence for 11) Setting the compensator before reading.
compensator

STUDY OF MAIN VARIANCE COMPONENTS

In the course of establishing the measuring method the distributions and their parameters for each
error, which is a component of the total error of measurement results, are studied. Then the main
variance components, describing to a sufficient accuracy the total error, are separated. The entire
procedure cannot be presented here, so only a part of it concerning the main variance components
will be presented.

As the main variance components for height differences are given the following ones (also Perović
2007):

1. Error of line of sight setting ε SET ;


2. Error of reference picture of rod graduation δ REF ;
3. Error of stability of instrument (STI) ε STI ;
4. Rod graduation error (RM) ε RM ;
5. Thermal extension of rod tape δt ;
6. Error of rod stability ε STL ;
7. Difference of vertical refraction at change points ∆r =δr ;
8. Air shimmering εtv ;
9. Scale factor error δE ;
10. Stability of bench marks ε STR ,

where " ε" denotes a random and " δ" a systematic error.

Thorough studies of accuracy the authors carry out on three quantities:


- on the differences of height differences at a station,
- on the differences of height differences forward and backward between bench marks, and
- from the adjustment of levelling networks.

The main variance components for height difference at a station


For the height difference at a station we have the following estimate:
1
∆h= (∆h′0 + ∆h′′0 ) (1)
2
where the superscripts (' ) and (" ) denote the first and second measurements.

Now the variance of height difference ∆h at a station, as the mean from two measurement series,
each with nč readings, will be:

σ2∆h =σ1,0
2
+ d 2σ2,0
2
, with (2)

σ1,0
2
= σ2RM ,ε + σ2STIL (a) 

1 1 2 , (3)
σ22,0 =σ2SET + σ∆2 r + σtv (b) 
2 nč 
where d is the line of sight length.
The standard errors, necessary in formulae (2) and (3), are not all known, so that the authors, on the
basis of their general knowledge from this field and of some thorough studies, estimate the
unknown standards and they are denoted here by superscript(A)1 reminding of the word authors:

σ SET =
0.1" (A) ; σ STIL 0.04 mm/st; 

γ= 10−6 / ° C; σ RM,ε = 0.003mm; 

σE = 2 × 10−6 (A) ; σ∆r = 0.21" ;  , with (4)

σ REF = 2 × 10−6 (A) ; σ STR : according to (5); 

σ∆t = 4° C (A) ; σtv = 0.2" . 

A)
According to the authors
=
σ STIL 0.02 mm/st − for rocks or pegs . (4')
The vertical stability of a bench mark depends on the type of point marking (founding) and on the
composition of the soil where a point is fixed and which is unknown. However, Perović (2007) has
estimated the standard of bench-mark stability for geopolygon “NOVO SELO” and obtained a
value of 0.2 mm, whereas for other cases of point marking the standards are assumed, i.e.:
=σ STR 0.1mm − bench mark or pillar in rock mass 

= σ STR 0.2 mm − pillars of geopolygon " Novo Selo " 
. (5)
0.2 mm ≤ σ STR ≤ 0.4 mm − pillars in hard ground , and 
0.5 mm ≤ σ STR ≤ 1mm − pillars in soft ground . 
The vertical self-displacement of bench marks belongs to a group of errors limiting the levelling
accuracy.

The main variance components for height difference between two bench marks

The height difference between two bench marks we estimate by means of the mean value from the
measurements forward and backward:

1 1 n n 
=
h (h′ + h′′=
)  ∑ ∆hi′ + ∑ ∆hi′′ , (6)
2 2 1 1 
where the superscripts (' ) and (" ) denote a measurement "forward", i.e. "backward".
For the variance of height difference (6) between two bench marks we obtain:

σ2h =σ21 + nσ22 + h 2σ32 , with (7)

σ12 =σ2STR (a) 



1 2 1 2 d2  1 2 1 1 1 2 
σ22 = σ RM , ε + σ STIL + 2  σ SET + σ2∆r + ⋅ σtv  (b)  . (8)
2 2 ρ 2 4 2 nč  

σ32 = (σ2REF + γ 2σ∆2 t + σ2E ) (c ) 

where γ is the thermal expansion coefficient of the invar rod tape, and σ∆t is the standard of air-
temperature change during the measurements.
For the case of a small height difference h , for instance h < 50 m , the third variance components
h 2σ32 is insignificant.

EXPERIMENTAL STUDIES
As said in the introduction, the study has been carried out on three geopolygons, „NOVO SELO“,
„TVRĐAVA NIKŠIĆ“ and „OSNOVICE PODGORICA“. Each polygon has its specific properties
which will be described now.

1. GEOPOLYGON „NOVO SELO“ was projected specially for this study so that the control
network for dam „NOVOSELKA“ was extended to reach a maximum distance of 2.1 km. The
measurements were performed with level Wild NA 3003, No 94173. The main characteristics of
this polygon are the following:
- maximum height differences in network max h = 125 m ,
- fixation of bench marks (points) pillars,
- supports for rods pegs (~ 70%); foot plates (~ 30%),
- maximal distance in network 2.1 km,
- average length of levelling section 0.55 km,
- ground soft (~ 70%); hard(~ 30%),
- average height difference in levelling section h = 55 m
- average length of line of sight d = 11.4 m ,
- average number of readings in series nč = 5
- average number of stations per 1 km of section n = 44
 Epoch 96: 11°C  25°C; mean := t 17.3° C 
- temperatures during measurements  
 Epoch 97 : 21°C  32°C; mean := t 25.0° C 
The measurements were performed with a given upper limit of empirical standard deviation for the
arithmetical mean from nč readings of 0.02 mm. Thus the number of readings is unknown and the
average number of readings was assessed to be nč = 5 .
The variance components are estimated from the differences dh= h' − h" by using the model

σ2dh =4σ2h =σ12 + Lσ22 ,


where L is the levelling section length. Based on these variance components the estimate of the
height - difference standard is obtained as a mean from the forward - backward measurements
σh =0.30 mm , for L = 1 km with f = 60 d.f.
In order to estimate the levelling accuracy from the adjustment of the measurement results model
(7) is used, but with a modification that the third variance component is not taken into account, i. e.
the following model is used

σ2h =σ12 + Lσ22 (7')

(where Lσ22 is used instead of nσ22 because n from (7) can be replaced with L / d , d = 11.4 m ).
So, based on the adjustment of the measurements from both epochs for the height differences an a
posteriori accuracy estimate is obtained (Vučkov 2001):
σ h, L=1 km = 0.32 mm , for L = 1 km with f = 14 d.f.

Using model of variance components (7) and (8) with d = 11.4 m , n = 44 , h = 55 m , nč = 5 ,


σ STR = 0.2 mm , and replacing σ∆t by ∆ t = 20° C − (17.3° C+ 25.0° C) / 2 = 1.15° C , where 20° C
is the temperature of calibrating rod graduation, for the height differences we obtain the a priori
accuracy estimate
σ h, L= 1 km = 0.33 mm , with f = ∞ d.f.

The test confirms the equality of the two variances, a posteriori and a priori.
Note. At a station the height difference is determined only once, hence we cannot derive the
accuracy estimate from the measurements at the station.
2. GEOPOLYGON „TVRĐAVA NIKŠIĆ“ is almost completely on the rock mass. In the
measurements digital level LEICA DNA 03 No 332735 is used. The main characteristics of this
geopolygon are the following:
- maximal height difference in network max h = 30 m ,
- fixation of bench marks (points) pegs in rocks,
- maximal distance in network 0.41 km,
- average length of levelling section 0.092 km,
- ground rocks (~ 80%); hard (~ 20%),
- supports for rods rocks (~ 80%); foot plates (~ 20%),
- average height difference in levelling section h = 5.7 m
- number of readings in series nč = 4
- average length of line of sight d = 8.3m ,
- average number of stations per 1 km of section n = 60
- temperatures during measurements 15° C  39° C; mean : t= 25° C

Note. At air temperature of 37° C the level due to overheating could not measure.
From the differences of height differences at a station d ∆h = ∆h′0 − ∆h′′0 , using the variance
component model
σ2d ∆h =4σ∆2 h =σ21 + d 2σ22 ,
the estimates of the variance components are obtained σ21 =0.0003023mm 2 and
σ22 =0.00001466 mm 2 /m 2 , with f = 150 d.f., based on which the a posteriori estimate of
standard height difference at a station is obtained as a mean from two measurements:
σ∆h =
0.018 mm , f = 150 d.f., - for d = 8.3m , and

σ∆h =
0.039 mm , f = 150 d.f., - for d = 20 m .
According to (2) and (3) we obtain the a priori estimate of standard height difference at a station:
σ∆h =
0.054 mm - for d = 20 m .
The a priori standard value σ STIL is valid a soft ground, hence σ∆h a priori is higher than σ∆h a
posteriori.
Based on the differences ''forward - backward'' the estimate of the standard deviation for the height
difference h from (6) is obtained as the mean from the measurements ''forward - backward'':
σ h, L= 1 km = 0.17 mm , with f = 12 d.f. ,

and from the adjusrment, using the measurements in one way, for the same h from (6) the a
posteriori estimate of standard deviation is obtained:
σ h, L= 1 km = 0.21 mm , with f = 17 d.f.

Here also, calculating the variance components a priori (7) and (8), with d = 8.3 m , n = 60 ,
h = 5.7 m , nč = 4 , σ STR =
0.1 mm (bench marks in rock mass), σ STIL =0.2 mm (rod on rock or
on peg) and replacing σ∆t by ∆ t = 20° C − 25.0° C = 5.0° C , for the height differences we obtain
the a priori accuracy estimate
σ h, L= 1 km = 0.16 mm , with f = ∞ d.f.

3. LEVELLING BASE LINES „SADINE“ and „ZLATICA“ in Podgorica, each 1 km long, are
set on a soft flat, and horizontal terrain. For measuring digital level LEICA DNA 03 No 332735 is
used. On levelling base line „SADINE“ as rod supports pegs were used and on „ZLATICA“ foot
plates were used, whereas the measurements were carried out with the following line of sight
lengths:
d = 12.5 m , d = 25 m , d = 50 m and d = 100 m .
The temperatures during the measurements were: for “SADINE” from 12° C to 30° C with
=t 25.2° C , i.e. for “ZLATICA” from 2° C to 19° C with= t 14.1° C . On levelling base line
„SADINE“ the analysis of measurement accuracy is obtained from the station differences “forward
- backward”, i.e. on base line „ZLATICA“ from the station differences from two measurement
series. Based on these results the estimates of the standard errors in the height difference h from (6)
are calculated per 1 km of levelling-section length. In such a way the a posteriori accuracy is
obtained:
- for „SADINE“ (with pegs): σ h, L= 1 km = 0.24 mm , with f = 69 d.f.

- for „ZLATICA“ (with foot plates): σ h, L= 1 km = 0.48 mm , with f = 141 d.f. ,

whereas according to (7) and (8) the a priori accuracy is obtained: σ h, L= 1 km = 0.33 mm ( f = ∞ ).

Apparently, the measurement with pegs was significantly more accurate than that with foot plates
which, of course, is also in favour of the equality test for the two variances for the confidence
probability of 0.99.

CONCLUSION

If in the realisation of experiment the conditions given in the description of the method of precise
civil engineering structures levelling are respected, then on the basis of the analysis presented above
and the experimental results we can conclude the following:
• Accuracy of height difference, σ h, L= 1 km , 0.2 mm − 0.3 mm can be achieved, but

• It is necessary to use pegs as rod supports!

ACKNOWLEDGEMENTS
The complete logistical support for carrying out the thorough measurements upon GEOPOLYGON
„NOVO SELO“ in 1996 and 1997 was given by geodetic engineers Mr. Dimitar Džonov, Director
of RGU Macedonia and Mr. Vančo Postolovski, Assistant Director of RGU Macedonia.

REFERENCES
[1] FCE (Faculty of Civil Engineering - Department of Geodesy), Podgorica (2006): Elaborat
merenja na GEOPOLIGONU „TVRDJAVA NIKSIC 06“.
[2] FCE (Faculty of Civil Engineering - Department of Geodesy), Podgorica (2006): Elaborat
merenja na BASE LINE „SADINE 06“ .
[3] FCE (Faculty of Civil Engineering - Department of Geodesy), Podgorica (2006): Elaborat
merenja na BASE LINE „ZLATICA 06“ .
[4] GUGK (Glavnoe upravlenie geodezii i kartografii pri Sovete ministrov SSSR ) (1974):
Instrukcija po nivelirovaniju I, II, III i IV klassov. Nedra, Moskva.
[5] Pelzer H and Niemeier W (Eds) (1984): Precise Levelling, Contributions to the Workshop
on Precise Levelling. Dümmler Verlag, Bonn.
[6] Perović G (1989): Adjustment of Calculus, Book 1th Theory of Measurement Errors. 2nd
revised and enlarged ed., Nau~na knjiga, Belgrade (in Serbo - Croat).
[7] Perović G (2005): Least Squares (Monograph). Author, Belgrade.
[8] Torge W (2001): Geodesy, Third completely revised and extended edition. W. de Greuter –
Berlin – New York.
[9] RGU Makedonije (1996): Elaborat merenja na GEOPOLIGONU „NOVO SELO 96“ .
Skopje.
[10] RGU Makedonije (1997): Elaborat merenja na GEOPOLIGONU „NOVO SELO 97“ .
Skopje.
Professor Ph. D. Sci. Gligorije Perović, Geod. Eng.,
Faculty of Civil Engineering, Department of Geodesy, Belgrade,
Bulevar kralja Aleksandra 73/I, 11000 Belgrade
SERBIA
E-mail: perg@grf.bg.ac.yu; Tel/Fax ++381 11 337 0293

Assoc. Professor Ph. D. Sci. Stojančo Vučkov, Geod. Eng.,


Faculty of Civil Engineering, Department of Geodesy, Skopje,
Partizanski odredi bb, 91000 Skopje
FYR MACEDONIA
E-mail: vuckov@gf.ukim.edu.mk; Tel: ++389 23 116066/150, Fax: ++389 23 117367

Assistant Trainee Darko Anđić, Geod. Eng.,


University of Montenegro, Department of Geodesy, Podgorica,
Cetinjski put bb, 81000 Podgorica
MONTENEGRO
E-mail: angel@cg.yu; Tel: ++381 81 242 630, Fax: ++381 81 268 617

Potrebbero piacerti anche