Information on the Arnhem Sampler (BTMA)

(1)

Information

on the Arnhem

Sampler

(BTMA)

M. de Vries

Internal Report No. 3-79

I

Delft Universit

y

of Technology

Department

of Civil Engineering

(2)

I

·

1 ,

,.

·

.'

I

~

I

."..---

..

'

..

.

"

INFORMATION ON THE ARNHEM SAMPLER (BTIIA)

M. de Vries

Internal Report No.

3-79

Delft University of 'I'echno

Logy.

-.

Department of Civil Engineering

Fluid Mechanics Group

•

(3)

I

(

\

I .

.

1

I

t

I

:

I

:

1 I

I

1. Introduction

2. Description of the instrument

3. Brief h

i

story of the BTMA

page

3 A

nne

x:

Manufacturers leaflet on BTMA

Referoenaes

1. Versuchsanstalt für Wasserbau

der ETH Zürich

2. Delft Hydraulics Laboratory

3. Hamamor

i

, A.A.

4. Delft Hydraolics Laboratory

.'

-5. Delft Hydraulics Laboratory

6. Vries, M. de

"Calibration of Arnhem Sampler" (in

german), Zürich, july 1937.

"Calibration BTMA" (in dutch), Report

M601, Delft, Sept. 1958.

"A theoretical i

n

vestigation on the

fluctuation of bed-load transport" Delft

Hydraulics Lab ,, Report R4, 1962.

"Development of bedload samplers" (in

dutch) Report M601-II, Delft, May 1966.

"Calibration of bedload samplers" (in

dutch), Report M601-III, Delft, nov. 1969.

"On measuring discharge and sediment

transport in rivers" Delft Hydr. Lab.

Publ. 106, April 1973.

(4)

I

.-1-

.

I

1. Introduction

I

The Subcormnittee 6 of the Technical

Cormnittee 113

of

the International

Standardization

Organization

(ISO) is dealing with measurements

of

sedi-ment

transport

in open channels.

As a back-ground

paper for TCI13

this

report

is written

on the BTMA, internationally

named

the Arnhem

Sampler.

The BTMA has been developped

jointly

by Rijkswaterstaat,

Directorate

Upper

Rivers

in Arnhem

and the Delft Hydraulics

Laboratory •

I

.

1 I

The BT}!A (Bedload Transport

Meter Arnhem)

has a relatively

long history;

the proto~ype

was constructed

in the thirties of this century.

The BTMA

is still in use although

insight

in the mechanism

of bedload

transport

has

been

increased.

Hence also the insight

in the measuring

problem

and in

the use of the BTMA has undergone

changes.

I

This report

tory of the BTHA is only concisely

is kept short for a number

written

of reasons. Most of the early his-

down. Moreover

a substantial

report would very easily have the essential

information

burried

under a

great many details.

Finally

the writer

(who played

an active role in the

application

of the BTMA during

the last 25 years or so) would not have

the

time to write

the report.

I

Hence

this report contains

subjective

selection

of information

on the BT}!A

and its use.

I

2. Description

of the instrument

I

The BT}!A is a flow-through

instrument.

The basket

1S placed at the bed by

means

of a frame

.

(see Annex).

The basket

is shaped

in such a way that the

head-loss

over the small-mash

wire-netting

is compensated

due to the

under-pressure

behind

the

-

instrument.

Hence

they

hydrauZic coefficient

(n

H) should

tend to unity with

I

n

H - velocity

_ velocity

in the mouthpiece

1n the flow

I

More

important

is the

transport

.

'

coefficient

.

n

s

wi th

I

transport measured

by the BTMA

ns

=

actual

transport

(5)

--

-

-~--"

-I

1

1 I

I, I, j

1 I

· I

I

,

1

1,1

I

l

:1

I

1

1 I

I

-2-A principle

decision has been made with respect.

to the mouthpiece

• A

fixed

mouthpiece

will give the basket

always

the same shape

(nH is

constant).

However,

a fixed mouthpiece

will cause that the instrument

can be very badly placed on the (ondulated)

riverbed.

• A

fZ

e

xibZe

mouthpiece

will

in principle

lead to a variabie

nH. However,

the attachment

to the wavy riverbed

is much better.

During

the whole history

the BT}~ has been used with a flexible mouthpiece.

In other words

the transport

coefficient

has always

prevailed

over the

,

hydraulic

coefficient.

The principle

construction

of the BTMA has never been changed.

However,

:

the suspension

in the frame has gradually

be changed.

The drawing

in the

'

;'

bookof Graf

(1971)

is obsolete.

The Annex

shows also the whole

instrument

as

it has been in use for more than one decade. The idea behind it is that

af ter lowering

firstly

the frame 1S placed

at the riverbed.

Then the spring

at which

the mouthp

Lece.i

i.sfixed presses

the basket

at the bed

.

When raising

the instrument

firstly

the basket

is lifted and then the frame

.

In.

_

this

way

it is prevented

that the basket

scoops

sediment

from the bed.

The efficiency

of the instrument

depends

on the amount of sediment

caught.

Too large fillings

should therefore

be avoided.

On the other hand a small

catch implies a short measuring

time. Hence

the relative

accuracy

of

determining

the measuring

time decreases.

As the BTMA is developped

for the use on the Rhine branches

the transport

conditions

in this river have guided

the design.

The standard measuring

time for an individual

measurement

being a comprom1se

between

the two

extremes

indicated,

is kept at

two minutes.

Usually

10 individual

observations

were

taken in one vertical.

This number

is mainly

based on the measuring

scheme:

the wish

to measure

the transport

in a cross section during one working

day.

(6)

-I

I

,)

I

:'

1 I

I

,

I

'

I

-3-3. B

'

r[ef history

of the BTMA

" '

.'.

:.:'

Develloped

in 1~36 the Arnhem

Sampler was calibrated

in 1937 in Zürich

[IJ.

It is known

that amongst others H.A. Einstein was involved

.

It appears

that

the instrument

was calibrated

without

the frame. This gave later reasons

to criticism.

Therefore

when the high-discharge

flume of the

,

Delft Hydraulics

Laboratory

(Laboratory de Voorst)

became available

it was decided by

Rijks-waterstaat

to recalibrate

the BTMA. The second calibration

tests were

carried

out in 1957 and 1958

[2].

In these calibration

tests catches with

the BTI1A were compared with

the average flume transports

as if the flume

was representing

á

river. The measurements

show much scatter and the report

concludes

that the value

n

=!

has to be taken as the best approximation;

s

a random error of a factor 2 for an individual measurement

can be assumed.

Later the scat ter in the individual

observations

of the BTMA recognized

as due to the natura

1 behaviour

of bedload

transport

(via ripples

and dunes)

were

studied. For a schematized

triangular

dune the probability

distribu-tion of the instantaneous

transport

(s) is given by

"

Pl{S/~}

=

0.5 (s/s) with 0

< sis < 2

₍₁₎

Hamamori

[31 developped

a theoretical

probability

distribution

p

2

{S/S}

for the case in which

secondary

ripples

'

propagate over primary

(triangular

dunes).

He found

0.25 (s/s){1

+

In (s~~)}

(2)

Later comparison

with

laboratory-

and field data showed good results

es-pecially

with P2{S/~}

for the field data with

the Arnhem

sampler

(Fig. 1)

.

99,9 % I I I

:I

LABORATORY • DATA

-

' MEASUREMENTS PANNERDEN

_l

CHANNEL BY BTMA

,

r

I

/.~

~.

~ (sIs)

.J

~,~

I

~~)'p,

(s/~ )

\ ,,/ '

I

I "

Thîs has led to a completely

new

approach

of calibrating

bedload

samplers

[4}.

This can be explained

briefly

as follows

.

An

individual

catch with

the BTMA 1S practically

a sample of the

instant:!aneoustrans-port

(s) as the measuring

time (2

minutes)

1S small compared

to the

periods

of the bedforms

.

Calibration

means

that the individual

catch

(c)

60 La

20. 10

,I

5 01 02 0.4 0.6 10. 2,0 1.0 60 - RELATIVE TRANSPORT

Fig.

I Probability

distributions

of

(7)

I

-4-I

'

I

--

-

~

..

I

of the sampler has to be compared with the instantaneous

transport

Cs)

~

-

lP

resent. Comparisons

of average catches

CE) and average transports

C~)

are only valid for the relation between c and s if the relation ~

=

fl(E)

is

Zinear.

As this is not obvious the calibration

relation

s

=

f2(c) had

to be sought.

,

.

...

As it is not possib

le

to measure

s directly

a procedure was developp

e

d

to compare the probability

distributions

p{c} and p{s}. The relátion

s

=

f2(c) is the result of this comparison.

I

_

The probability

surements on a sandbed in a flume in which a statistical-stationary

distribution

p{c} can be measured

by taking

r

ando

m

mea-

bed-laod transport

is present.

In the third series o

f

calibration

tests

[5]

carried out in

1967

and

1968

this new approach was applied. In these t

e

sts

the probaóility

distribution

p{s} was measured by mean

s

of a small

sand-trap suspended

in the large sandtrap of the 3 m

e

ter wide high-dis char

ge

flume.

The small sandtrap had a width of 8~ cm (equal to the width of the BTMA

mouth piece) and it was empti

e

d automatically

every

2 minutes

(the

standard measuring

time of the BTMA). The values of s were determined

by weighing

the small sandtrap continuously.

I

1 As a result of these third calibration

tests the conclusion had aga~n to

be drawn that

n

=!.

The conditions

present during these tests were

s

2 ~

s

<

.

6 m2/day

for three sand mixtures with

D

=

6.4 nnn

(D

go

=

14.7 .

nnn);

D

=

2.5 _

nnn

(D

go

=

5.8 mm) and

D

=

3.1 mm (D

go

6.6 mm) respectively.

Th

e

tests did not

'

indicate a nonlinear~relationship

for s

=

f2(c).

1 I

I

with an.

Finally

Ldeal:

some information

sampler has to be mentioh.~d. To obtain his engineering

obtained for the accuracy

.

of individual catch

d

es

eg

ree

at Delft University

of Technology H. Nijdam

(amongst other things) si

m

ulated

individual

catches

in using the Hamamori distribution

(Eq.

2).

In this w

a

y

the variance

of sis (being the relative error of s) as a function of th

e

number n of individual

catches. The results are based on k individual

g

r

ou

ps

of n observations.

Figure

2

· shows the results.

1 I

I

(8)

I

'~ \'.'

I

'

I

,

I

...

~

...

-

.

" 0", i» ,_. .I \: 1:' ;>,' '-",W <.',',' \_\_{__._.}_{~_... ....t_}_._--._•. _{_} 0,10 !

,

I

...

,

_

.

I'---.~,

-!"-r-,

_~.

• K=50

..

• K=1OO 0.05 0,03 1 5 1) 20 50 IX) ~ NI.M8ER OF OOSERVATICtoIS

Fig. 2 Relative error 1n bedload

measurements

-,

-5-

_'.'

.

..~

.

This study indicates

that the

usual choice of n

=

10 observa-tions in each vertical

leads to a

c

onsiderable

relative error for an

ideal

instrumènt due to the

physi-cal ph

e

nomenon.

This result is based on the assu

m

p-tion of random sampling

.

It is

necessary

to setup the bedloa

d

measurements

in such a way [6]

that this condition

1S fulfill

e

d

.

The measuring

procedure

1S

(9)

I

v

.

ESSEN

I

DELFT

y

I

-~

8~

>.~

C'!

i5

.

_~

_'J

_C

~

OUl

fO

f2

'$

ID.qo

.

~~'

...Ct)

o

Ct) ,....;H

t>~

Cl) Ol

E-4G1

~

I

t--_IN_S;._T_R_..:;U_M_E_N_TE_N_F_A_B_R_I_E_K_V_A_N_E__;S;._S;._E_N___:;_B_.V_._~,EL....:.F02T...:.:lt1;.;;.;~.23

--I

~:':':~~

îir

05:._ND3_)

BED-LOAD TRANSPORT METER "ARNHEM"(BTMA)

l:j

DE LF T

f

ig

.

l

Purpose: To measure bed-load of coarse sand and fine gravel in rivers and other watercourses.

Principle:

A

frame mounted sampler is pressed on the river-bottom by a leaf spring. Behind the mouth of the sampler, opening 0.085 x 0.05 m2,is a basket of fine wire meshing. The form of the basket causes

a low pressure behind the instrument so that water and transported bed material enter the mouth with the same velocity as if in undisturbed flow. The bed material particles which are too

coarse to pass the meshing are caught.

1. Introduction

Bedload-transport is difficult to measure. Since the development of the Bedload Transport Meter Arnhem (BTMA) in 1936 and the Zürich calibration (1937) large numbers of measurements have been carried out including calibrations by the Delft Hydraulics Laboratory. Since 1936 considerably more insight is obtained into the mechanism of bedload-transport and consequently into the way of measuring it.

2. Descri tion of the instrument

e lnstrument conslsts 0 a sampling body

suspended from a frame by six brass chains and a stainless steel leaf spring.

The brass sampling body has a streamlined shape

and has been dimensioned in such a way, that the

amount of water entering the mouth is equal to

the quantity th at would pass a corresponding

cross-section when the instrument would not be

present. The mouth is a rectangular metal pipe with a length of 0.04 mand internal dimensions of 0.050

m

high and 0.085 m wide. This is con-ne~ted with the actual sampling body by gradually diverging flexible tube with a length of 0.12 m, attached to the body by another rectangular me-tal pipe with a length of 0.021 mand internal dimensions of 0.071 m high and 0.102 m wide. The cross-section of the sampling body rapidly

increasing to a height of 0.093 mand a width of 0.150 m decreases the mean velocity,which enables the sampled material to settle.

The remaining 0.30 m of the body consists of fine monel gauze, converging up to the end, which is formed by a brass tail piece with emplying plug. A zone of low pressure surrounds the monel gauze part of the body and stimulates the velocity of the effluent water.

Bed material which is tOD coarse for the mesh to pass remains in the sampling body. The 51-mesh gauze has wires of 200~m and openings of about 300_,.Ám.

Therefore it is evident that the sampler is not

suited for very fine material , which would soon clog the meshes.

The frame is constructed of steel, consisting of two symmetrical halves to which a rudder may be attached. The purpose of the rudder is to keep the mouth perpendicular to the flow direction.

The rudder is fitted with a curved plate to rest on the bed.

The legs are also fitted with supporting plates.

The apparatus is lowered to the bed by means of a steel cable attached to the lever arm. The lever can move up and down although the upward movement is limited. When the frame is hoisted up or lowered, the lever is in the upwards position. When the frame rests on the bottom the lever will move towards its downward position allowing the mouth to sink with a re-duced speed.

A leaf spring connects the halves of the frame, halfway supporting the mouth of the sampler.

From the foregoing description it is evident that the mouth of the sampler does not touch the bed before the three supports of the frame Characteristics: Dimensions:

Max. length of frame and rudder

Max. width of frame Max. height of frame Length of basket Max. width of basket Length of trough Width of trough

Height of trough on frame

Weight: frame + rudder + basket trough + frame 1.85 m 0.81

m

0.42 m 0.53 m 0.15 m 0.75 m 0.53 m 0.78 m 32 kg 22 kg

Accuracy: With frequently repeated measurements the order of magnitude of material transport can be established. However, it should be realized that the figure obtained depends greatly on the position of the sampler in the bed-form (ripples, dunes, plain bed, etc.). This implies a careful scheming of the measurements, while the results are best evaluated by comparing with results of observations under different flow conditions. Advantages: Robustly built for handling in the fie1d.

Disadvantages: Instrument must be handled by a davlt or derrick due to its weight and dimensions. Bed-load sampler BTMA consists of:

Instrument frame in the following parts: one tailfin, one frame with lever

Stainless steel leaf spring with phosphor-bronze spring set

Sampling body with six nickel plated brass chains, shackles and rings

1 Trough on steel frame (in 8 parts)

16 Stainless steel bolts and nuts for the trough frame

Measuring glass of polypropylene 500 mL Measuring glass of polypropylene 250 mL Set of spare parts (see overleaf)

Price Dfl.

(10)

rest on the bed and the cable is fully released. Moreover a flexible double leaf-spring connects

the sampler mouth with the big leaf-spring. All these provisions make it possible for the position of the mouth to be adapted to the

contours of the bed.

3. Mechanism of bedload-transport 3.1 General

Usually ripples or dunes are present at the riverbed. Therefore bedload-transport fluc-tuates locally and temporarily. In the trough of a dune the local transport is almost zero, whereas at the crest the transport is 2, 4 or more times the average transport. As the bedform is propagated relatively slowly the "period" of a bedwave is large. Much larger than the two-minutes measuring time which is standard for an 1ndividual BTMA-observation.

3.2 Random Samtlin~

When the B MA 1S lowered at the riverbed, then its position with respect to the bed-form can not be controlled nor measured accurately.

Depending on the depth and flowvelocity the BTMA.reaches the bed somewhere in a reach.

Two cases do now exist.

(i) If this reach L is large with respect to the length of the bedwave

(

/l

)

,

so

if L)À then random sampling is reached also when the survey vessel has a fixed position (fig.2a)

(ii) In all other cases random sampling is not obtained from an anchored vessel but only if the vessel takes a different position at each individual observation. (fig.2b)

Note: From this follows that a successful -- measuring-scheme with the BTMA

requires a longitudinal sounding in order to get informed about the bedform present.

3.3 Calibration and accuracy

Accord1ng to ca11brat10n the BTMA has an efficiency of 50%.

This means that the transport in the measured zone of the stream (i.e. to a distance of 0.05 m. from the bed is two times higher than follows straight forward from the cathes.

The instrumental error in the individual observation can be estimated at + 30%. However, the error due to the restricted number of observations (i.e. originating from the phenomenon) can be larger. This samyling error amounts to (for random

samp11ng

for 10 observations sampling error + 30%

for 20 observations sampling error

+

20%

for 100 observations sampling error ~ 9%

4. Measuring erocedure

The clean 1nstrument is lowered from a vessel (sometimes from a bridge). The instrument has to be lowered slow enough to be sure

that the frame gets its right position

compared with the current. The emptying of the instrument can be carried out in the trough. Squirt the inside of the sampler

by means of a deck wash. The catch is measured volumetrically.

f

i

g

.2

a

Spare parts:

1 Phosphorbronze spring set

1 Rectangular rubber sleeve for sampling body 5 m Nylon cord to fix the sleeve to the body