repo

rts

in a

pplie

d m

easu

rem

ent

Edit

ion

1/20

02

A new transducer for directly measuring the forces and moments which a skier transfers to the skis

The BELFA load application vehicle -measurement and loading techniques for the structural evaluation of bridges and sewers

Electrical strain gages, the essential features of which

were first developed in the nineteen-thirties, are today

counted among the most important sensors for the mea-

surement of mechanical quantities.

The property of electrical conductors to react to me-

chanical strain by changing their electrical resistance

forms the physical basis of such devices. However, their

success has only been made possible by spending many

years continuously developing strain gages themselves,

along with their associated measuring amplifiers and

software. HBM has made an important and significant

contribution to this process.

At the present time, those carrying out experiments in

stress analysis have at their disposal a rich variety of

strain gages in versions covering a multitude of applica-

tions. Transducers for mechanical quantities such as

force, mass and pressure are based on this technology

and operate to levels of precision that were inconceiv-

able just a few years ago. Powerful measuring amplifiers,

to an increasing extent controlled by complex software

systems, round out the range of available equipment.

However advanced the current state of development, it

does not automatically make errors of measurement and

analysis a thing of the past. As ever, the users of these

systems are the deciding factor and as such it is they who

are responsible for obtaining successful measurements.

It is their responsibility to select and use strain gages,

transducers, measuring amplifiers and software that suit

the problem under investigation, and to calculate and

evaluate the relevant mechanical quantities based on

the primary measurement results obtained. The latter

part of this process is often more difficult than the ac-

tual measurement, and can be guided by the manufac-

turer to only a limited extent.

Dear Readers,

”Reports in Applied Measurement” from HBM, in which

users report on interesting measurement and analysis

projects, help to publicize know-how and encourage the

exchange of experiences in this field of technology. In

line with the traditions of this journal, contributions are

drawn from widely differing areas and touch not only on

many topics of a practical nature, but also theoretical

aspects of metrology and engineering technology.

Above all it is our wish to publish interesting scientific

contributions which will help users to find a solution to

their own particular problems.

I hope you will thoroughly enjoy reading this issue of

”Reports in Applied Measurement”.

Sincerely,

Martin Stockmann

Dr. -Ing. habil. Martin Stockmann

Technical University of Chemnitz

Institute for Mechanics

Professorship in Experimental

Mechanics

3ram 1/2002

reports in applied measurement

IntroductionThe aim of the work is the design and manu-

facture of a transducer with six degrees of

freedom for measuring the forces and

moments transferred from a skier’s feet to

the skis (see Fig. 1).

Two possible configurations were analyzed

and tested in order to arrive at an optimal

solution. Both cases make use of sensors that

are fitted with strain gages arranged

between two aluminum plates. These sprung

elements are incorporated into the ski bind-

ing at the point where vibration absorbing

plates are normally found.

The first solution is based on the design spec-

ified by P. Jung in ”Reports in Applied

Measurement” [1].

The second variant represents an independent

development which is notable for having

fewer components and greater sensitivity in

measuring the six load components.

These sensors were used in the laboratory and

on the piste to test for the release of ski bind-

ings in accordance with national and interna-

tional standard specifications for insuring the

safety of ski bindings (ISO 9462) [2].

Initial observationsWith the aim of selecting the most suitable

configuration for the problem in hand, the

solutions known from published scientific

works were analyzed first. Special attention

was paid to monolithic sensors with three and

six degrees of freedom. These sensors are

noted for their compactness and for being

very sensitive to forces and moments.

One of the most commonly used designs is the

Maltese cross sensor [8]. Figures 2 and 3 show

Maltese cross configurations with three or six

degrees of freedom. In figure 2 the load acts

upon an inner flange (1) linked by four prism-

shaped struts (3) to an outer, circular flange

(2). A strain gage (4) is installed on the two

opposite sides of the x-y level of each strut. In

each case these strain gages form part of a

Wheatstone half bridge, which means there

are four half bridges in total.

A. Freddi, G. Olmi, D. Croccolo

DIEMInstitute for Machine BuildingUniversity of Bologna

Fig. 2: Maltese cross sensor with three degrees of freedom

Fig. 3: Maltese cross sensor with six degrees of freedom

x

y

z

M Z

M y

M x

F x

x

y

z

F Z

F y

1

2

3

4 z

y

x1

2

3

4

Fig. 1: The forces and moments which a skier transfersto the skis

reports in applied measurementramA new transducer for directly measuring the forces and moments which a skier transfers to the skis

The three axial forces (x, y, z) acting on the

inner flange can be determined by measuring

the tensile, compressive and bending defor-

mations of the prism-shaped struts. This con-

figuration makes it possible to compensate for

temperature, although four measurement

channels have to be used to analyze only

three axial forces. The transfer matrix there-

fore has a 4x3 structure and can be inverted

using the Moore algorithm.

To determine all six load components it is

necessary to use this particular Maltese cross

configuration (Fig. 3). Here the ends of the

struts (1) are connected to very thin mem-

branes (2) that are notable for their high radi-

al elasticity. Strain gages (4) are installed on

all four sides of the struts, requiring 16 strain

gages in all.

Here too, the strain gages installed on oppo-

site sides of the struts are interconnected in

half bridges. Thus there are eight half bridges,

though only six load components occur. For

calibration purposes the resulting transfer

matrix is 8x6 and this too can be inverted

with the aid of the Moore algorithm.

Another important configuration is the

Junyich sensor introduced in [4]. Its shape is

similar to the Maltese cross configuration,

but different wiring is used for the strain

gages in order to insure mechanical and elec-

trical decoupling and obtain a quadratic

form for the transfer/calibration matrix. The

diagram in figure 4 uses black points to show

that 16, 32 or 48 strain gages can be used.

The individual elements bij of transfer matrix

B=C-1 are set out in the following equations.

Here ε expresses the measured strain in

µm/m, where (εSG-number)Fi represents the

strain induced by the i-th load component

with F1=Fx, F2=Fy, F3=Fz, F4=Mx, F5=My,

F6=Mz (i=1-6).

(6)=b6i(ε11)Fi − (ε12)Fi + (ε15)Fi − (ε16)Fi

4

(5)=b5i(ε1)Fi + (ε7)Fi

2

(4)=b4i(ε4)Fi + (ε6)Fi

2

(3)=b3i(ε2)Fi + (ε3)Fi − (ε5)Fi − (ε8)Fi

4

(2)=b2i(ε13)Fi + (ε14)Fi

2

(1)=b1i(ε3)Fi + (ε10)Fi

2

4 ram 1/2002

Fig. 4: Junyich sensor, characterized by good decoupling and a quadratic calibration matrix

13b-

13-1

3a(1

5b-1

5-15

a)

14b-

14-1

4a(1

6b-1

6-16

a)

1b-1

-1a

(5b-

5-5a

) 4b-4-4a(8b-8-8a)

9b-9-9a(11b-11-11a)

2b-2-2a(6b-6-6a) 10b-10-10a

(12b-12-12a)

3b-3

-3a

(7b-

7-7a

)

SEZ. R -R

1 11

1

R R

SEZ. K -K

Fig. 5: Configuration of a Junyich sensor

y

xz

5ram 1/2002

reports in applied measurement

Regarding decoupling it should be noted that

when there is a combined torsion moment

and bending moment, the third bridge out-

put signal is zero (the SGs wired into the

opposite branches of the bridge are subject-

ed to the same strain). The same applies when

a vertical force is acting. Here the output sig-

nal from the fourth and fifth Wheatstone

bridge is theoretically zero, since the SGs on

opposing struts are subjected to opposite

strains. In order to increase measurement

precision, the number of strain gages can be

increased to 32. Then the bij-equations - with

the exception of

instead of

- have the same form.

Figure 5 shows how the cross element of a

Junyich sensor is designed with the inner

load flange and outer flange. At this point it

should be mentioned that roller bearings are

built in to prevent torsion of the struts under

the effect of torsion moments and bending

moments or tensile and compressive forces in

the event of longitudinal or transverse load-

ing. The sensors just described can be used

provided all forces and moments act upon

the inner load flange. However, in this partic-

ular application the load cell cannot be fitted

in a central position on the underside

because the vibration absorbing plates are

already located there. But since the underside

is mainly stretched in the longitudinal direc-

tion the above precondition is not fulfilled,

which would give rise to unacceptable mea-

surement errors and inaccuracies.

Using these sensors also leads to another

problem, which is a reduction in the stability

of the connection between the sensor and

the upper plate to which the ski binding is

fitted. International standards demand tests

which demonstrate that all adverse effects

on skiing and every risk of injury to the tester

are prevented.

Initial design and configuration fora transducerThe first design for a load cell, published in

[1], was fitted with seven separate sensors,

each sensor being loaded and relieved of its

load in a single direction. All sensors operate

as beam spring elements and have a load

imposed upon them by a force acting on two

diametrically opposed points (Fig. 6 und7).

The load is applied through a fork attached

to opposite ends of the struts, producing

decoupling in the other direction. The struc-

ture is isostatic. Four sensors measure the

vertical loads only (two moments and one

force). Two sensors measure the transverse

horizontal forces only (another moment or

force can be computed) and the seventh sen-

sor measures the horizontal loading in the

longitudinal direction only. Seven half

bridges were installed so as to obtain good

resolution and temperature compensation.

Calibration was carried out in the laboratory,

where each sensor was subjected to a series

of defined loads.

This process made it possible to determine

the mathematical relationship between the

measurement signals and the forces con-

cerned. The precision of the regression lines

to the measured values was found to be

between 99.9 % and 100 %.

(8)(εk)Fi

(7)(εka)Fi + (εkb)Fi

Fig. 7: ... and in cross-section

F

F

Fig. 6: Single-axis sensor in 3D view ...

F

F

L

The diagram in figure 8 shows some typical

laboratory test results. They represent the

development of moments during a test for

binding release due to rotation of boot and

leg. The diagram illustrates the simultaneous

occurrence of torsion moments and bending

moments together with their peak values.

Results of the skiing experimentsSki trials were carried out in the Dolomites

(Trento, Italy). Typical measurement results

from a test run are shown in figures in figure 9.

The new transducerThe new transducer was intended to be note-

worthy not only for its extreme compactness

but also for its safety and sensitivity. It was

important to have sufficient length in the

torque arms in order to increase measurement

precision, to avoid adverse effects on the skiing

and to protect the tester from the risk of injury.

By way of preparation the previously deter-

mined experimental results were first analyzed

and processed, and the actual peaks for the

longitudinal, transverse and vertical forces act-

ing on the sensors were computed.

The results shown in the diagrams were

processed using a mathematical model to sep-

arate out the loads acting on the four sensors.

In these diagrams the four color-coded lines

represent the evolution of the vertical loads on

the sensors. To make the most of the advan-

tages of a modular configuration, each sensor

was manufactured using the same electroero-

sion machine with identical settings.The new

design placed the main emphasis on

mechanical decoupling.

6 ram 1/2002

Fig. 8: Measurement results during a test for binding release due to rotation of bootand leg

Fig. 9: Test results for the giant slalomabove: forces; below: moments

0 0,5 1 1,5 2 2,5 3 3,5 4Time [s]

200

150

100

50

0

-50

-100

-150

-200

Mom

ents

[Nm

]

1000

800

600

400

200

0

-200

-40060 65 70 75 80 85 90 95 100 105 110

Time [s]

Forc

es [N

]

120

100

80

60

40

20

0

-20

-40

-60

Mom

ents

[Nm

]

60 65 70 75 80 85 90 95 100 105 110Time [s]

Bending moment My

Bending moment Mx •10Torsion moment Mz •10

Vertical force Fz

Transverse force Fy • 5Longitudinal force Fx • 5

Bending moment My

Bending moment Mx • 5Torsion moment Mz • 5

7ram 1/2002

Fig. 12: View of the new sensors showing the configuration and wiring of the SGs

Fig. 11:Belleville springs placed under the headsof the screws that make the connectionbetween the sensor struts and the loadapplication device to prevent moments tobe transferred when the load is applied

Fig. 10: Three-dimensional view of the new sensor with three degrees of freedom

7

4 56

3

1 2

8

7

3(L)4(T)

1(L) 2(T)

6

5

V V V

2 3

1 4

Channel 1

6

5

Channel 2

8

7

Channel 3

reports in applied measurement

Fig. 10 shows a domed surface and a

tapped hole at the point where the loads

act. The strain gages are installed in three

areas to measure deformation. One measures

chiefly the vertical deformation , one the

longitudinal and another the transverse

. The four small tapped holes allow the

sensor to be screwed to the under plate,

while the fifth central hole is intended to

save weight.

Fig. 11 shows that Belleville springs are

placed under the plate and under the heads

of the screws that make the connection

between the sensor struts and the load appli-

cation device. This prevents a moment being

transferred to the transducer when the load

is applied. This degree of freedom is also

achieved by means of a gap between the

screw shank and the hole in the upper plate,

and by means of an elongated drill-hole at

the position of the Belleville springs. Wiring

as four half bridges and four full bridges

insures high resolution and provides temper-

ature compensation. The configuration of the

strain gages is illustrated in Fig. 12.

Strain gages 1 and 3 are sensitive in the

longitudinal direction (longitudinal arrange-

ment L) while SGs 2 and 4 are sensitive in

the transverse direction (transverse arrange-

ment T).

These SGs are wired to a full bridge. In con-

trast strain gages 5 and 6, which mainly sense

transverse forces, and strain gages 7 and 8

for vertical forces, are wired to half bridges.

In order to reduce the number of channels,

the double-grid strain gages for the longitu-

dinal forces (T rosettes) are installed on the

two forward sensors only. The strain gages

for transverse forces are installed on the

same side as the two front and two rear sen-

sors, and are wired to two full bridges. The

total number of channels then comes to

eight (four for vertical forces, two for trans-

verse forces and two for longitudinal forces).76

5

3

2

1

Rear

section

Front

section

Initial calibration of the newtransducerDuring initial calibration each sensor was

loaded with predefined forces and the mea-

surement signals related to the three bridges

(or two for heel sensors) were analyzed. The

measurement data was used in the appropri-

ate transfer matrix, where good linearity and

reproducibility were observed.

By way of example, Fig. 13 shows the regres-

sion curves for vertical forces as determined

by the calibration test.The calibration results

were used to define and normalize the trans-

fer matrices. Normalization was achieved by

multiplying each column with the measured

peak value for the force concerned.

Computation of the coupling coefficients CC

(by dividing each element of the normalized

transfer matrix cij by the sum of the elements

in this line) shows good sensor decoupling.

The transfer matrices were inverted for com-

putation of the calibration matrices. For the

purpose of determining the definitive 6x8

matrices that define the relationship

between the six degrees of freedom and the

eight load signals, the matrices were linked

as shown in Fig. 14a and b.

Global calibrationFor global calibration of the load cell, the four

sensors were interconnected with the aid of an

upper plate as shown in Fig. 16 and then

loaded with three forces and three moments.

In this special case eight Wheatstone bridges

were used, resulting in an 8x6 transfer matrix

B and a 6x8 calibration matrix C.

=c ij

Σ j c ij

(9)CC

0 50 100 150 200 250Vertical force [N]

2000

1500

1000

500

0

Stra

in [µ

m/m

]

TheroySensor 1Sensor 2Sensor 3Sensor 4

The number of rows in a transfer matrix is

equal to the number of measurement data

channels, whereas the number of columns

matches the degrees of freedom (load com-

ponents). In the case of the calibration matrix

the situation is reversed, since the mathemati-

cal relationship C=B+ can be represented as

C=B-1 when C and B are quadratic matrices.

Otherwise C is treated as pseudoinverted

matrix B and can be computed using the Moore

algorithm.

Calibration is carried out in the following

sequence: The upper plate is loaded with a

longitudinal force (e.g. 10 N) and the eight

measurement data channels are analyzed. The

force is then increased in steps until the peak

intensity value is reached. This loading

process must be repeated at least twice. The

mean values then have to be computed in

order to obtain a high degree of precision.

The procedure enables eight functions to be

FxFyFzMxMyMz

∆ε1∆ε2∆ε3∆ε4∆ε5∆ε6∆ε7∆ε8

=

3.148896088 0.020754301-2.551861142 0.04465757-0.572254861 0.004654152

3.222587 0.031136-2.54098-0.04447-0.56982 0.006982

-0.08143 0.0487 0.064283 4.51E-05 0.014415 0.010921

-0.000552593-0.000165641 0.172801372-0.003024024 0.038750708-3.7145E-05

-0.006260992-0.000907351 0.169897454 0.002973205 0.038099504-0.000203474

0 0.049743958 6.83397E-05-9.27603E-06-1.53252E-05-0.011155083

0-0.00061 0.165678-0.0029-0.03715 0.000138

0-0.00113 0.161563 0.002827-0.03623 0.000254

8 ram 1/2002

Fig. 13: Regression curves for vertical forces

Fig. 14b: Association of measured quantities

Fig. 14a: Computation of the load components from the measured strain

1

2

3

4

5

6

7

8

No. of the Wheatstone bridge

∆ε1

∆ε2

∆ε3

∆ε4

∆ε5

∆ε6

∆ε7

∆ε8

Output signal

9ram 1/2002

recorded. It also enables the computation of

eight regression curves which describe the

trend in the measurement data (measurement

data compensation) and which are character-

ized by the fact that they show an increase.

These coefficients are entered in the first col-

umn of the transfer matrix. They refer to the

first load component and represent the signal

for a 1 N loading.

The above procedure is then repeated using a

transverse force. The values determined in this

way are entered in the second column of the

transfer matrix. Finally this procedure is

repeated for all the other load components.

The last three columns of the transfer matrix

then contain the results arising from the out-

put signals from the Wheatstone bridges in the

case of the three loads with a moment of 1

[N·m].

In order to compute a calibration matrix the

transfer matrix then has to be inverted. Deter-

mining the calibration matrix is the most

important objective of this work, since compu-

tation of the six load components acting on

the newly developed load cell is directly

dependent upon it.

The equations for computing the strain from

the loading and for computing the loading

from the strain are as follows:

Multiplying by the BTmatrix gives

If the experimental or mathematical problem

is given the right conditions, the resulting B

is an 8x6 matrix of rank 6. Then BT B is a 6x6

matrix with full rank which can be easily

inverted as follows:

On comparing (11) and (9) it is noticeable that

the 6x8 matrix (BT B)-1 BT is identical to C

and represents the calibration matrix B+

which expresses the relationship between the

strain signals and the loads concerned.

(BT B)-1 BT ε=F (13)

BT ε=BT B F (12)

F=C ε (11)

ε=B F (10)

Fig. 16:above: Individual sensors;below: The complete trans-ducer; ready for global calibration

Fig. 15, right:Second generation sensorwith HBM strain gages, e.g. LY11 3/350

Longitudinal force on the front left sensor

Longitudinal force on the front right sensor

Transverse force at the toe

Vertical force on the front left sensor

Vertical force on the front right sensor

Transverse force at the heel

Vertical force on the rear left sensor

Vertical force on the rear right sensor

Measured quantity

reports in applied measurement

SummaryThis paper presented two transducers with six

degrees of freedom, suitable for measuring

the forces and moments that a skier transfers

to the skis.The first transducer was fully test-

ed in the laboratory, during calibration and

during skiing trials under real load condi-

tions.The second new and independently

developed transducer was designed to

increase compactness, safety, precision and

sensitivity in relation to the six loads.

At the same time satisfactory decoupling

effects were achieved in three ways:

- mechanically by means of the HBM

strain gages which are optimally tuned

to the geometry and stresses of the

sprung elements.

- electrically due to wiring the strain

gages in a special way to compensate for

unwanted signals.

- mathematically by computing the cali-

bration matrix.

The individual sensors in the new load cell

were fully calibrated. A global calibration

then has to be carried out in the laboratory,

during which forces and moments are simul-

taneously applied to the whole configuration

along the x, y and z axes (Fig. 16). Skiing tri-

als can then be carried out under real load

conditions.

References

[1] Peter Jung: Applying metrology to skis, Reports in

Applied Measurement, 11 (1975) Vol. 2, pp. 34-37,

Vol. 3, pp. 59-62

[2] Alpine ski-bindings, Safety requirements and test

methods, International Standard, ISO 9462

[3] Lu-Ping Chao, Ching-Yan Yin: The six-component

force sensor for measuring the loading of the feet in

locomotion, Materials and Design, 20 (1999) 237 –

244

[4] Lu-Ping Chao, Kuen-Tzong Chen: Shape optimal

design and force sensitivity evaluation of six-axis

force sensors, Sensors and Actuators A Physical, 63

(1997) 105 – 112

[5] Gab-Soon Kim, Dea-Im Kang, Se-Hun Ree: Design

and fabrication of a six-component force/moment

sensor, Sensors and Actuators A Physical, 77 (1999)

209 – 220

[6] Tom Boyd, M.L.Hull, D.Wootten: An improved accu-

racy six-load component pedal Dynamometer for

cycling, Journal of Biomechanics, 29 (1996) 1105 –

1110

[7] D. Gorinevsky, A. Formalsky, A.Schneider: Force

Control of Robotics Systems CRC Press, New York

10 ram 1/2002

11ram 1/2002

IntroductionTo demolish and rebuild road bridges that

have been found by calculation to be unable

to support heavy loads consumes resources

and disrupts traffic. With the aim of econo-

mizing on materials, and in order to rational-

ly perform loading experiments using the ex-

perimental load-bearing method known as

EXTRA on bridge structures and sewers, a

group of researchers at the Universities of

Bremen (team leaders), TU Dresden, HTWK

Leipzig and the Bauhaus University of Weimar

worked with the Eggers group and other com-

panies to develop a special vehicle that could

be used in flexible ways to apply loads to

structures. Known as BELFA, its name is an

acronym from the German word for this type

of vehicle (”Belastungsfahrzeug”).

BackgroundThe national and local road networks in

Germany include more than 80,000 bridges of

various kinds and some 300,000 km of sewers

of widely differing ages, some of which are in

doubtful condition. By way of comparison,

there are around 760,000 bridges throughout

Europe. About 90 % of the bridges are solid

structures, and around 70 % of these have sin-

gle spans of less than 18 meters [1]. A broad

estimate of the total cost of currently out-

standing repairs and maintenance to bridges

and sewers runs into double-figure billions of

euros.

The investment required to replace these

structures with new ones can only be brought

down to a sustainable level by successfully ex-

tending their useful life to a significant de-

gree despite further increases in traffic vol-

umes. This can only be achieved with objec-

tive, experimental proof of adequate load-

bearing capacity, especially in the case of

load-bearing structures that are already dam-

aged or have been singled out for some oth-

er reason, followed up as necessary with long-

term observation and monitoring.

The main ground rules for the experimental

analysis of present load-bearing performance

were acquired in recent years by cooperation

between the organizations mentioned above

in developing and testing the EXTRA method

on more than 300 bridges, major construc-

tions and sewers [2 ... 5]. The method con-

Marc Gutermann

Rolf Sommer

Klaus Steffens

Fig.1: Load testing with the aid of loading frames at Dassow, Germany (1993)

Fig. 2: BELFA on the road

reports in applied measurement

reports in applied measurementramBELFA load application vehicle -measurement and loading techniques forthe structural evaluation of bridges and sewers

cerned has since been given legal status in the

form of a guideline [6].

The experience gained from many of the ob-

jects investigated made it clear that the time

required for preparatory work and related

phases of the experiments, such as setting up

and dismantling the load application engi-

neering, represented some 70 % of the total

time needed for a typical object and was an

obstacle to wider application of the method.

In the case of bridges there was the associat-

ed factor of relatively lengthy disruption to

traffic (Fig. 1). The development, construction

and testing of the BELFA prototype is intend-

ed to meet the need to reduce investigation

costs and bring the disruption from road

works down to one or two days.

Operating principleThe vehicle is 2.75 meters wide and 3.50 me-

ters high. It consists of a towing vehicle, a

welded main chassis comprising two longitu-

dinal and 16 transverse girders, ballast water

tanks, and a deck crane with adjustable ex-

tension and height (Fig. 3 to 5).

Four hydraulically movable feet positioned

over the neutral abutments/pillars are used to

align the vehicle horizontally and vertically.

Five test rams with fittings on extending slides

generate the test loads. The rear-mounted

control cabin and associated diesel-electric

drive for the soundproofed hydraulic motor

complete the list of main elements. Before the

BELFA can be placed in the test position it has

to be pulled forward by the towing vehicle un-

til it is over the load-bearing structure that is

intended to be examined (Fig. 4 to 5). In this

situation the towing machine load can be re-

duced to the minimum 23 tonnes by tem-

porarily removing ballast so that it is even pos-

sible to cross load-bearing structures on class

12 bridges. The heavy five-axle trailer vehicle

at that point remains outside the test area. On

reaching the test position the vehicle is

- Ability to carry out loading investi-gations using the EXTRA method inone lane of bridge classes 12 to 60as defined in DIN 1072 (12.85); upto bridge class 30 this would involvenot using a back tie, that is notdrilling through the waterproof layerto install tie rods, and for bridgeclass 60 two back tie rods would beinstalled per abutment

- Ability to operate as a special vehiclewithout significantly restricting traf-fic (all-up weight 80 tonnes, axleloads ≤10.5 tonnes, turning circle D = 22 meters)

- Largely autonomous compliance withall requirements relating to the load-ing procedure and operating safetyin the course of preparing and exe-cuting the investigation

- Use of the self-locking loading prin-ciple when using hydraulic forcegeneration in connection with a rigidversion of the BELFA’s main chassis(max. f = 25 mm at a maximum testloading of 1.5 MN)

- Variable span adaptation where l =4.00 to 18.00 meters by means of in-finitely variable telescopic action

- Ballasting with 20 tonnes of masswith the aid of flexible water tanks

- Axle loads on towing vehicle to bekept down to ΣF = 230 kN whencrossing class 12 bridges

Following restricted tendering, the order for

construction including a target time for plan-

ning and construction (9 months) was award-

ed to Eggers Fahrzeugbau GmbH, vehicle-

builders, in 28816 Stuhr, Germany.

SpecificationsConstructing the load application vehicle involved a number of complex engineering

requirements, the most important of which were the following:

12 ram 1/2002

Fig. 3: BELFA at its premiere in Neu Kaliß, Germany

chocked and aligned vertically in the trans-

verse direction. The towing vehicle and trailer

are raised and trimmed with the aid of the bal-

last, increasing the available test dead weight

to 80 tonnes (or 100 tons using the water bal-

last). The weight is equally distributed over the

feet (Fig. 5 -6).

The test rams are then moved into their in-

tended positions and vertically aligned if nec-

essary. Three of the five rams represent the

three axles of a heavy goods vehicle (HGV) as

defined in DIN 1072, while the two outer rams

are used to apply a loading to the remaining

surfaces fore and aft of the HGV area.Each of

the last two rams mentioned can also be used

in a separate test for determining bearing

strength in the event of lateral forces in the

vicinity of the support structure. If the avail-

able dead weight of the BELFA is insufficient,

pre-tensionable back ties must be installed in

the abutment area using tie bolts fixed in the

abutment itself. The dead weight and pre-ten-

sioning of the back ties then form a total avai-

lable test load capacity of 1,500 kN, which is

adequate for g-fold loads up to bridge class

60.

Hydraulic loading unit and associ-ated monitoring equipment The hydraulic loading unit in the strictest

sense consists of five test rams, the hydraulic

motor complete with power generator and

the associated programmable controller. This

unit alone is responsible for the controlled

generation of the test load. All other essential

hydraulic functions, such as extending the

four feet, are specific to the motor vehicle and

are perceived by the various separate modules

as being independent of the loading unit.

Each test ram is linked to the main chassis of

the BELFA by an extensible positioning slide in

such a way that its axis is rigid in the trans-

verse direction and jointed in the longitudinal

direction, though lockable (Fig. 6). Not only do

the hydraulic rams each have a nominal force

of 500/250 kN (compression/traction) but

each ram is fitted with a 500 kN tensile/com-

pressive force transducer at the end of its pis-

ton rod.

All data streams, whether from the transduc-

ers installed on the load-bearing structures

under investigation or from the force and tilt

13ram 1/2002

Fig. 4, above:Extending the frame increases the usable length to 18 meters

Fig. 5, left:Views of the BELFA;above: in traffic; below: in test position

reports in applied measurement

measurements performed by the load appli-

cation vehicle, are brought together in the

rear test control point. The latter has a

dynaSax II PLC for analyzing the measurement

signals and a PC controlled UPM 100 multi-

point measuring device from HBM (Fig. 7).

When a test is in progress the PLC passes the

current force values from all test cylinders to

the multipoint measuring device. A down-

stream analysis program provides an online

Approach

Crossing

Test

l < 18m

Super-structure

W = 80 tL = 22,5m

W = 100 tL = 35,5m

SensorMeasuring

base

Fig. 6:Test ram with extensible positioning slide, force transducers and cross-beam

Integrated force measurement

Hydraulic ram2

1

1

2

14 ram 1/2002

Fig. 7, left:19" rack with PC, monitor and controller for load application and vehicle hydraulics, together with UPM100 multipoint mea-surement amplifier from HBM

display of the independently measured defor-

mations of the load-bearing structure and the

force/deformation functions. The test supervi-

sor and a metrologist in the control cabin are

provided with all the information they need re-

lating to the test in order to operate the BELFA.

For the sake of maintaining operating safety

the PLC monitors the behavior of the vehicle

throughout the test loading procedures. Not

only the forces acting upon the supports but

also the longitudinal and transverse tilt at both

ends of the main chassis are continuously

recorded and compared with threshold crite-

ria (Fig.9). If a warning is ignored and an abort

threshold is exceeded, the hydraulic motor

automatically cuts out. The test load is reduced

in the shortest possible time. In the process the

system monitors both absolute and relative

values.

The PLC automatically reduces threshold values

in accordance with the situation of the super-

structure, for instance in the event of pro-

nounced longitudinal tilt.

Experiences during initial imple-mentationSince BELFA first entered service in March 2001,

it has been used to examine the structures of a

total of eleven bridges in five German federal

states. Thanks to the experimental analysis car-

ried out on the load-bearing structures with

the aid of BELFA, it was possible to reclassify all

these bridges at least one class higher than

would have been possible using conventional

calculation methods. Thus these structures

meet the requirements out of today’s traffic

revenue and they were able to be cleared for

higher loads. The cost saving due to economiz-

ing on materials and continued use when com-

pared with new replacement structures was de-

clared by those in charge of construction to be

between € 250,000 and € 1 million. Disruption

on the bridges amounted to one or two days or

less, no more than half the amount of time

specified.

ProspectsOnce the research project is complete it is

intended to set up a scientifically-based

scheme to offer the services of the BELFA load

application vehicle. Current information can be

obtained from the web site www.belfa.de.

Acknowledgements The project presented above was kindly en-

couraged by the Federal German Ministry for

Training and Research, Bonn, and managed by

the Deutsche Zentrum für Luft- und

Raumfahrt e.V. (German Aerospace Center),

Bonn, under reference 01 RA 9901/0. Our spe-

cial thanks are owed to the participating

districs for providing the pilot objects and the

necessary infrastructure.

References

[1] Steffens, K. (Ed.): Experimentelle Tragsicherheitsbe-

wertung von Brücken in situ zur Substanzerhaltung

und zur Verminderung der Umweltbelastung; Koope-

ratives Forschungsprojekt 01-RA 9601/6 EXTRA II,

Abschlußbericht, Hochschule Bremen, Eigenverlag 1999

- A defined longitudinal or transversetilt is exceeded

- There is too great a difference be-tween each of the supports in a pair(right and left)

- The total load acting upon all sup-ports is too low (lifting/slipping)

A test procedure is aborted inthe following circumstances:

- A defined minimum force is not pre-sent in one of the supports(lifting/slipping due to excessive lon-gitudinal tilt)

Fig. 8: Extensible positioning slides

messtechnische briefe

Published by:Hottinger Baldwin Messtechnik GmbHIm Tiefen See 45D-64293 DarmstadtTel. +49 6151 8030Fax +49 6151 803 9100www.hbm.com

EditorDr.-Ing. habil. Martin StockmannTechnical University of Chemnitz

Please address any questions about RAM articles to the publisher.

Please address any questions about HBM products to your local HBM sales office.

CopyrightCopying is only permitted with theadvance written approval of the editorialoffice or the publisher. Copying or repro-duction in the form of photocopies,microfilm or other means for commercialpurposes is not permitted.

All rights reservedNo liability can be accepted for the pro-cedures and circuits described and thenames used in respect of the infringe-ment of patents or trademarks of thirdparties.

reports in applied measurement1/2002Issued: april 2002

“reports in applied measurement” is issued two times a year

Printing: 100% chlorine-free bleached paper has been used to protect the environment

imprint

Fig. 9:One of the two front supports with integrated force measurement

[Experimental in situ evaluation of the load-bearing

capacity of bridges for the purpose of economizing on

materials and reducing the environmental load;

Cooperative research project 01-RA 9601/6 EXTRA II,

final report, published University of Bremen 1999]

[2] Bellmer, H., H. Brandtl, J. Quade, G. Kapphahn,

P. Schwesinger, K. Steffens: Experimentell gestützte

Untersuchung des Trag- und Verformungsverhaltens

der Brücke über die Stepenitz im Zuge der B 105 in

Dassow; Bautechnik 71 (1994) H. 11, S. 782-791

[Experimental investigation of the load-bearing and

deformation behavior of the bridges over the

Stepenitz along the course of the B 105 in Dassow;

Bautechnik 71 (1994) Vol. 11, pp. 782-791]

[3] Bucher, Ch.; R. Ehmann, H. Opitz, J. Quade,

P. Schwesinger, K. Steffens: EXTRA II - Pilotobjekt

Weserwehrbrücke Drakenburg. Experimentelle Trag-

sicherheitsbewertung von Massivbrücken. Bautechnik

74 (1997) 5, S. 301 - 319 [EXTRA II - Pilot object,

Weser bridge Drakenburg. Experimental evaluation of

the load-bearing capacity of solid bridge structures.

Bautechnik 74 (1997) 5, pp. 301 - 319]

[4] Steffens, K.; Ch. Bucher, H. Opitz, J. Quade, and P.

Schwesinger: Experimentelle Tragsicherheitsbewer-

tung von Brücken. Bautechnik 76 (1999) 1, S. 1 - 15

[Experimental evaluation of the load-bearing capaci-

ty of bridges. Bautechnik 76 (1999) 1, pp. 1 - 15]

[5] Steffens, K. (Ed.): Experimentelle Tragsicherheitsbe-

wertung von Bauwerken: Grundlagen und Anwen-

dungsbeispiele. Bauingenieurpraxis, Ernst&Sohn

Verlag, Berlin November 2001, ISBN 3-433-01748-4

[Experimental evaluation of the load-bearing capaci-

ty of bridge structures: Basic features and typical

applications. Bauingenieurpraxis, Ernst&Sohn Verlag,

Berlin November 2001, ISBN 3-433-01748-4]

[6] Richtlinie für Belastungsversuche an

Betonbauwerken. Deutscher Ausschuss für

Stahlbeton, Fassung September 2000 [Guideline for

load tests on concrete structures. Deutscher

Ausschuss für Stahlbeton (German Committee for

Reinforced Concrete), September 2000]

ram 1/2002 15

You are invited to contribute an article or articles to RAM, and will receive anauthor’s fee if your article is published.

E-mail, editor: Martin.Stockmann@Mb1.TU-Chemnitz.deE-mail, HBM: klaus.bathe@hbm.com

Reply fax to the editor: +49 371 531 1471

Name

Company / Dept.

Street

Town, Postal code

Telephone

Fax

I am interested in having an article published in RAM and

would ask you to kindly contact me

Hottinger Baldwin Messtechnik GmbH

Im Tiefen See 4564293 DarmstadtGermany

Tel. +49 6151 8030Fax +49 6151 803 9100

E-mail: info@hbm.comInternet: www.hbm.com