fRITZ fNGfNEER'~~G LABORATORY'LEHIGH UNIVERSITY

BETHLEHEM, PENNSYLVANIA

REPORT ON DUNAGAN'S BUOYANCY APPARATUS

By: J. M. Holme

lSI. f.

,;::

REPORT ON DUNAGAN'S BUOYANCY APPARATUS,

I - GENERAL PURPOSE OF APPARATUS,AND SHORT DTSCUSSIONOF OTHER tffiTHODS

The Dunagan Buoyancy Apparatus is designed for

the purpose of determining the constituents of fresh con­

crete.-'';rhe;: industry has long felt the need ,of an efficient

appara tus for maldng thi s determination as it would un- ~

, dOUbt~y reduce' the time element in our present cylinder

strength tests. Several engineers have advanced methods

but without exception they lack practicability.

R. L. Bertin of the Vfuite Construction Company

described a method for segregating fresh concrete in the

April issue of the A;C.I. Journal, as have W. I. Freel,

G. J. Griesenauer, H. C. Ross and C. A. 'Huges. While I

have not attempted to apply any of these methods, I am of

the opinion they all fall short of their purpose. Bertin's

method is the one most ~orthy of.~ote, but should be placed(!.v,;(:. ......... .h.> Y;-~~ .... ~ , .

2nd to Dunagan's. ~fundamental suggestion is to eliminate

the use of previously established apparent specific gravities

thruout a series of tests. He states that different samples

of aggregate will give great variations in specific gravities

determinations~ This point he stresses but passes over the

subjects of silt content and the cement retained on the

/'

/Ij .V!;lOO'sieve by sayingtha t the final results should be cor-

rected'for these items. From my experience t4is past

semester, I would say that he has stressed too highly the

point which involves least error and slighted .the sources

of greatest error.

II - DESCRIPTION OF DUNAGAN APPARATUS A1ID DISCUSSION OF ITS USE

2

Conslanf WQ~r/~,..e,.((Jrert!ow .s;pouf ,.~ _

. 1-' ~;;""";;;;:;:';:--I

lit- -J-

10

0\

~ 1t'/~=1))1~(~11~ ~=-~ =~\'\\ ~""'''''=~''I''''~\'i\,\'/~'''1''""""",,,.'\ ~= k ~::::7-' ....-:::,. -_."\ =,- --- .,Cross section of apparat1.1.s - Balanced v:rithout'load asshovm

.when Q is filled to overflow. containeft"C" is. immersed· and

\ pan B is in position.''''v:;J

:#;Weighing of samples.'-J. -',. , :.

(a) In air~

Sample is'weighed in pan B to an even kilogram<"(

by placing' slotted,. Kgwg~. o~r~eft-hand side of balance...?:;'. .\

Cb') 0eighing SamPl~) '~l1llnersed.,

Container C is removed. and 'partly filled with water ..

3

The sample is poured into C and stirred to remove entrained

air. C is then filled with water, the sample allowed to

settle for I minute and then replaced on beam and Q filled

to overflowing. The weight iwnersed is then obtained by1-

placing wgs in pap B and sliding the rider "R II• The beam

is graduated from zero to ten grams and the weight is

accurate to the nearest one half gram.

III - 'NECESSARY PRELIMINARY TESTS

Before beginning any series of tests certain pre-

liminary tests must be made. A - Apparent Specific Gravi ties.

The fir st of these is for the.apparent specific gravi ties

of the aggregates and of cement. Representative samples of

the aggregates should be secured, soaked in water for 24

hours'and allowed to surface dry. Using either surfaceCi--

dried coprse aggregate, surface dried fine aggregate, or

cement the sample is weighed in air and its weight recorded

= 1000 gms.

= 1000 gms.

2000 gms.

x·and its (Vfg; recorded as

---'

(}-(C'o]irse aggrega te =((Fine aggregate((Cement(

The sample is then weiShed iwnersed

·w~..J._-=Then.app. Sp. Gr. =(WI - w~

A - Consistency of Results

In our work we made a series of determinations for

the specific gravities of the aggregates and of the cement.

For the cement we obtained a consistent result of 3.1. For

the aggregates results were averaged in both cases and varia-~ ..

tions from the mean.noticed~~ The largest variation from the

mean was in no case greater than 2 parts in 260. This is

certainly a striking contrast to figures Bertin cities vlhere-

in he notes variations from the mean as large as 9 parts in

260. I am inclined to believe with Dunagan that sampies

taken from the same stock pile will give a resulting apparent

specific· gravity that will not cause any appreciable error

in the results. At any time results are in doubt tb~ ag-

gregates can be checked for the specific gravity by weigh-

ing immersed, surface drying and weighing in air.

B - IVioisture Determination

The determination for free moisture or absorption

is based on the previously determined ~pparent specific

-gravity.

is positive. If the result is a negative vilue it indicate~

the amount of moisture the aggregate ,is capable of absorbing.

These tests should be made before actual working tests and

should be checked from time to time during the work. cor,:,­

rections can readily be made and resulting errors are

negligible.

C - Silt Determination

The remaining preliminary tests are for the purpose

of determining the percentage of silt in the aggregates and. 'I

the percentage of cement retained on the(l)lOO sieve. Silt·!.,,~I c'\.

is considered to be any of the aggregate passing the'@lOO

sieve. I shall outline the method Dunagan suggests for

these determinations.

A representative sample of aggregate is secured and

is weighed immersed.. Recorded Wl . The 'sample is then trans-. n -"

ferred to the ne sted sieve s {(1J4 on &rioo) and wasl:-ed until

free of silt. It is then weighed immersed again - Recorded

The percentage of silt = Wl - W2

Wl

C - Determination of Cement Correction

In his manual Dunagan. makes mention of the cement

retained on #100 sieve but fails to state a method for

determining its percentage. It may be obtained in two ways.

1 - weigh a' sample immersed - record Wl . Transfer to #100

and wash. Weigh immersed the cement that does not pass the

sieve and record W2 •

percent retained W2

2 - Use a mix of fi ne aggregate (wi th si 1 t previously re-

moved by washing) and cement in the proportions they occur

in the working mix.

Weigh the washed sand immersedjrecord Wl .. rt

.Add cement and recordtotolweight im.11lersedf\W2 •

Transfer tO~lOO sie~eand wash until all cement has passed

that will pass and then weigh immersed'/ record W3

./ (foJ retained = W3 - Wi J~t ~ W2 - Wl

These last determinations;(% silt passing the #100

and %cement retained on the #100) are the sources of error

that make determinations of the constituents of concre~e

obtained by this method of douQtful value .. I shall discuss

them after giving an example of an actual test run ..

IV - PROCEDURE FOR ACTUAL TEST

The preliminary tests described above are assumed

to have been run. A sample of concrete is secured.

aggregate)f~

\/1100 (fine..1

w~· immersed = W2 ./

It is washed on the nested 'Sieves(#~ and ~OO until free of\.j, f1..

cement and silt. The aggregate retained on (#7'4 (coJirsef/.J J.

is weighed im.11l~rsedtrecord w3~f~~e aggregate on

aggrega te) is added and total V>lJ W4 obtai ned •.

/f:sp gr j

W3 = Wj of C.A. immersed - W3 Sp gr- QJ, = wg of

C A in air. /

~jW4 ~ :- VI]3 = wfJ of FA immersed (W4-W3 sp gr- FA= Vl9

of FA in ai r. vi'

(fP gr )W2 - Vi = wg of cement immersed (W2- VV4 ) sp gr-l cel7?enl;:='4

QOEont =wg of cement in air.

v - DISCUSSION OF RESULTS AND PROBABLE ERRORS

The wg of the original sample in air minus the sum

of these last 3 wgs in air = the wg of the water. If no cor-

rections be made any silt that ·was present in the aggregates-'

in recorded as cement. Secondly any cement retained on the

#lOO is recorded as aggregate. These two factors do not

cancel each other and the result is the w/c is incorrect.

Corrections can be made for both of the se factors but the

corrections are of uncertain value.

In one series .of tests which we made to determine

the percentage of silt in fine aggregate the mean equaled

0.8% silt. The greatest variation from the mean was 1.0%.

This variation would cause an error in the w/c ratio of

0.2 gals,. per sack. For 2000# concrete this error (if it

were the only error) might not be serious. In another series

of tests on different sand we found a mean of 4.83% silt

with the greatest variations from the mean = 2.1%. This

variation would cause an error in the w/c ratios of 0.4 galsl

sack. An error of this magnitude would probably render the

7

concret@ 'que,'s'ti·onabJe~Errors 0 f thi s nature fallon the wrong

side of safety, since they give a W/C ratio of a lower value

than the actual W/C. Consequently a strength determination

based on this method would indicate too high strength concr'ete.

The question of cement retained on the #100 sieve

is dependent on several factors. 'More cement is retained

when washed mixed with sand than when washed alone. The time

washed with sand in proportions of the mix the

These second washings were done as nearly as

done in actual test.

and thoroughness of washing ,also have

In tests ,:we made (When washedr ~0.8~. WhenV.I, Q .(l. --~ %re t '= 1.6%.

,,/-- ._~

possible to the way they would be

a decide.-. d ~ffec~~ ,oP", ,~'i~__.J.J

alone) the ~\retained =-' : ,-,1 1\ '

.' ..... 1

How-

, ever" Mr. Nettles determined that by continued washing he

could grea tly reduce the percentage retained. lr~ashing a

sample without the addition of sand he obtained 0.9% ret., '

Washing a second sample together,with sand and continuing

washing long after'the lait trace of cement passing could. ,',..

be detec ted he obtained a figure 0 f 0.9,3% retained whi'ch is

in close aggreement with his first figure. I am of the firm

opinion and our tests bear witness that with oIdnary washing

the percentage of cement retained on the #100, s~e isa

function of the sand (both quantity and kind).

'VI - EXAMPLES Al\lJ) DISCUSSION OF RESEARCH AND FIELD TEST RESULTS

, In handling the apparatus during the coursel of a

test (that is weighing and vlashing) considerable care must

aggregate will cause an increase

An error in weighing againin

this may be/either direction.

//

l~S of

';}: ,

~J,.6'--

'V'

I\

).I

be exerCised.; Any, !' I,

in the .cementJ determinat ion.- { ~--)

I\;ffects tbe ~llc }but. ..;-~-+"...",

On the accompanying sheets, 1,2,3 are the results

of five tests we made on one batch of concrete. It may be

seen in each case the water cement ratio is somewhat less

than the actual, water cementra tio used in preparing the mix.

The largest vai~iation amounts to 0.31 gals. per sack, while

th~ average variation is 0.17 g~ls/sack.

On sheet 4, I have ·li sted .several of the results

taken from actual test done for the Clinton Conveyor by Mr.

Dunagan. These samples \vere'in each case, taken from a mixer

and in each run the samples received approximately the same, . ' ~

amount of mixing. The results show great i~consistenc~s.

. r ,In two case s the\ W/¢) is approximately 1 gal. low. In an-

. '---.._/ ..

other case it is 1.72 ga1./~ack too high. 'A third value is

0.28 ga1s/s~ck low and two 'additional values 0.08'ga1s. per'

sack high.

,From the comparision of these few figures )'Nhieh are

symbolic of both sets 'of te sts in general) I f e'e1 that our

results are by far the more' consistent and show that they

are the result of extreme laboratory care. The second set

of figures (sheet 4) are such as might be expected from

field tests made with the apparatus. It is decid,ed1y im-

Corrections t:; I .rO ~ - U1 • .r-! . °,rO Weight Submerged Q).p H P lQ ~ 'J1 • • r.:: r-1 .ptU) Z• 0'

Q) +' ..c:rd Q) rl ° .p • I:) r.:: +' rlrl 0 ° H·r-!..- •0 +' tU) I:) bfIl) .r-! • I +'H .r-! ..c: oe:- ro ° ..c: o A .r-! :> OHX CHZ ..c: H <I) 'r-!b.O rl H . '¢.r-! .p tU) .r-! X [;J .r-! b.O :>U1 +' • AO.r-! Q)S b.O <I)' .... Per Silt Cement H <l)H A b.D H bJJm !.is .r-! .p lQ .......... .p .r-! 'r-! H 0: ° .~ 0::<I) <I) .r-! S H~ ..-l .b.D .r-! 0:: <I) ai • 0 H <I) ord ° p HCH ..

rl ~ <l)p Cent -No. +No. Total 0 s .p A . a>r.:: :s: 0:: rl rl 0 ~ en A <It p-!o U1A H :s:: ~ 0 P rl W A :s:: .r-! I:) X m A p Q 0 <I)8 * en 100 100 ::J ::J CJi ---.. ~ I:) ·bJJ ° ~ ~.p H ~ by by' +'m * u: ~** ~ p ---..+' H .0 ::;. p-! p 0U) '* **

~.

wt. ~.V.s: s: II p-! ....... Z.- _. .-

j 1 "'/f93.1l :;.27&,,0 1/38"£t. ;I2 3039.() ..

3 ,z39.? 239.0 /?,Zh /7,28'~-_.

G.1C/() --4 */6/t?-5" /h, O'6'~5 'J<t76£.3 :'J<~3

~""'7 r"f',? 9?/.~ Ff'''X~g3 /5~'U) 3,~bt) 6/2'.0 -#./9 3,9~54-f.k~1---'--6 6~-1,2 :*;'2 f-h,~ -~~ f,2,Z ~-?~,4 ~'%6"..<S j~5tJ.() .z.-4.2tl f103.~ !z9,/4 fZ·5F5 ;2. ,!,?'l

7 .;l. f~,5' ·.~,6 -4,7-~,tg f ~~ -6,9 k?$~ ~~"'76 Aj~~() /.tJt/tl 11317.4 9,4,z ,/, t?t?O ~~3

8 ~,j'ft' tt,~ '3 6,7£~---

,Z 1 .I?'?/. 7 -r-3 /.I"3', ~ Ij;;..J,? /(2 .z~79.1

3 'z33/1 .:l.3'j/1 /7,7tJ J',2.'64 *J.5:2tJ,5 t£ ..3f'

* 8'~Z. 7 ~*t'.2 ~*~6~3-

5 f~.~ -+2,{, '?'8s.~ /4-1£? '3,694 5tO,1W.2,4! 3,CJtJ5 44.'1r 6.37. g' it'*/'Z +7.7 ,-4.E 1-3,4 G4/,2. **./- ~..2..f' :;~-1Z.~ 2,~b2- 40(J.r ::10,33 ;),5'iS" 2i',787 ,27/, :;l. l<~0 -2.(,;-7,7 -1-4-3 -~.O .;:Lr;~-. ~ 'i<~476 39/,"1 /,t/lJtJ /;lb, z. '1,55 J, df)O 9,2-3

, 8 I !0.59'! G.7.3 : I ~,7-1

****

Test DataKey: - 1. Total Sample

2. Cement + Sand +3. Water4. Stone and Sand

** Preliminary

'stone

Test Data5. Stone6. Sand7. Cement8. Water-Cement

OperatorSample) (1Taken ) (2

Corrections t) I .rd >.. -- Ul • 'M . 0rd Weight Submerged Q).p ~ .,olQ ~ fJ} . • s::: ..--1 .p~ Z• .,

Q) .. -P ..Qd Q) ,-j 0 -P • 0 s::: .p ,-\..--1 0 0 H'M •0 -P QD otu:Q.> 'M • I -PH 'M ..G ot- co 0 ..G o Pi 'M :> OH>< tt-lZ ..G ~ Q) 'M~ .-I H . ..G 'M -P ~ 'M >< OJ 'M QD :> Cf.l -P • Pi 0 'M illS QDQ)Per Silt Cement H Q)H Pi till H ~co co 'M

-P lQ __-P 'M 'M ~ 1J2 0 ~ 0::Q) Q) 'M S ~~ 'M • till 'M 0:: Q) co • . H Q) .rcj 0 ,0 ~tt-l .,

.-l !A Q)p bent -No. +No. Total 0 s .p Pi • <J>S::: ~ 0::.-1 .-l 0 ~ UJ . Pi <r\ P-lo Cf.lPi H ~ ;:j D ,0 ,-\ uJ PI ~'M 0 >< co Pi ,0 • 0 Q)8 * . if, 100 100 ~ ~ CfJ --.. >.. 0 · QD 0 >.. -:t:.p H >.. by by' .pco * u: ~** ;?: P -..-P H .0 ~. P-l P <:I~/) *~.

** ES:i§: II P-l .....- wt. ~.V. z-- -1 I'" :.2. tJ3'6' !/.;..1a?3 7691,5" -2 127.28:.< -_.3 ,?~q.·8 :J-09',8' /6,9/ 17,2~ d4 */"(1'z.r5 r·~53.5 '*tJ,~ -/",2,& +'z,~ gS{;,/ 1'f'-1"j, ~J3 1/.JfK,~ a. g"(),j. 15#/,9~J.h~ 3.9tJS l4~i6.~>.--.--6 5'!r,J :*/.Z of ?,I -4./ 1-~{) .s-9,g, 3 /l<~•.~.a5' 9~,.! .e,r;/7 371/,2-12q,~ 2.5"f5" b~,?f

7 ',254.; /.6 -.2.6-?-1 -1-4./ -5:;; ,,21'9./ Pi'~..f7~ 3617.7 /.~tl{) 11/8,IP 9,5;; /,0lnJ Q,:J..38 0,57/ (;.-1$ -- .---- . -

.4 1 '-'l1l4J. 3 Ih()6",t /3sg,7 H_2 1<:96),93 .e3K,/ 1.::<3~, I Il,j?- 17.2'64 *156-t.,Z

1*91h',5' 1+:*&.1 . ~~v;3'3-

5 .r2,7 -I-A'/ .9/5.2 1#;:'4$ 5'.654 579.3 42,'4 3.9tl!l 41.~1r. &5/,7 1Ii*1. ;e -f7,g- -4.5' -I- 3. "3 ~.5:0 Pt:*A6ZS VtleU-1 :a,6t?:J,- 409·4 30,/S ;2,:;-8S'1.18'.7¥'7 Z'?J./ <~6 -,2./-),! -I-4,~ -~.O Z77,/ rt~--17'" 4(J9.~ /,tl~ /3//1 9,71 I, ()()~ tj. '2.7

! 8 I 1 it),5¥;J... b,5"5'

9tJ6/3, '.

* Test Data*** Key: - 1. Total Sample

2. Cement + Sand +3. Water4. Stone and Sand

** Preliminary Test Data5. stone

stone 6~ Sand7. Cement8. Water-Cement

OperatorSample) (1Taken ) (2

.........

Corrections 8 I 0

to ~ - U2 • •...i . 0to Weight Submerged (J)~

~ ~~.0 l!J ~ 'il . 0 5=: ....-1 ~till :z;

• a.> -P .Qd 0 -P . 0 l=: -P r-Irl 0 0.. -. H·...i .0 -P QD ot:uW .r-;l • I -PH .,-1 .s:: ot-- ~ 0 11 0 A ''-; :> OHX (j.-fZ ..s:::: H a.> .,-;t-J) r-I H • ..s:::: .,-1 .p QD ....i X rJl .,-4 :> U2 -P . PiO •...i Q)

S t<o a.> Per Silt Cement H a.>H A till H QDctl as .,-1.p l!J __ .p •...i •...i H t:r: o ~ 0:a.> a.> •...i E H ;:s:a.> .,-1 • till .,-1 ll:: Q) aj • 0 H Q) otO 0 .0 H(j.-f ..

.-l !A a.>.o Jent -No. +No. Total 0 s ~ . IV5=: ~ ll:: rl r-I 0 ;s 8] A <!j P10 IJ)A H is: ::s 0 .0 r-I rrl A ~'...i Q X as A .0 • 0 ())s * if, 100 100 ::s ::l UJ

---- ~Q . b.O 0 ~ <:t: +-=> H ~

,--. .p

ctl * U). ~** ~ .0____-P

H P 3': P1 .0 by by 1.0U) * ** ~ ~ II P1 --- wt. i\.,V. \Zl

17FCJ.~-l' '-' -

S 1 :1J/tJ(r,.~ /3/},(} /7'2 2.'V3.4 -3 ..;J.3:;".~ ~32. (p /7,66 /J,;z~ .-_.

4 * r. "/5/~,

5 'l'-a~6,S- :"'0.3 1-~,6 +~,6 8'~q,/ "X&?33 ,14/9.2-' 3.~~ r5o./ ~/,77 ,3,t;tlS- 11,GI~---

6: ~--14. ~ :*;;z -I- 7.'7 -"'1,S" +.1,"'2- ~4J,:z. *~6~5" /IJ.J'/.7 2,6/$ 1M.S 30·7/ :2 J6"¥S'" :23'.7r _.

? -Z7?,5 .'/, t&> -.2,6-1,7 -I--4.S" -s:~ ,:27,2,7 *~-f76 4/J2,5' /, tltlCJ /.29, g' ~f6 /J~ .9dLf-._-

8 a.~7~ 6,~()'".=-= ~~--_._.-. b==- - . .. - .=

1... "I'

-2 ... "_..

..

3-,.-. , , -:,._'!

.: .'.

4 *,.- .. :".., ..

5 * 1<* **....~- ; ~ -- -' -., . /

C "* **.7 :'1' **

8 I.. , !

****

Test DataKey: - 1. Total Sample

2. Cement + Sand +3. Water4. Stone and Sand

** Preliminary

'Stone

Test Data5. Stone6. Sand7. Cement8. Water-Cement

Gperator3ample) (1Tak-en ) (2

/ ~) 7,0

,/brCenl- A.6.so/ure Yo/u.I77e II//x bv WeIQ.h;: "

,

Wafer Cemenl Sand Ii'ocK Cemenf .jf,,.,d '/?ock S'fR ~9a~KSltrndard /R..34 9.s"e ~S:t64 4~.S-O I ..z.~3 4./' ~,49 'hOt)

/fun 1 /5:"8 9./h ..e().9~ 54..10 / /.98 ,£()5' 7.03 ~,/z

/'7- 40 9.:19 24.00- 4!J./5 / 2.2/, 4-4~ S:~? ~.7Z

Run2/9.27 9-78 ,2".83 ..,-4,02 'I 2.36'" 3,79 6,/4 7,08

/8.24 /"""'1 ,e.4.4-e "'''./8 / /,97 3.•1 ,J;61J '.08

/i'un3'ZI.,z,Z F.fl3 .,eB.B9 47.()~ / .2.z-'f 4,54 6·78 r.7z/8.08 9.2~ .24.76 47-90 1 2, '9/ 44() ~.7/ '7.03

1.I:f

possible to use the care and take th~ time necessary in field

tests that can be used in laboratory research work.

Therefore operating the apparatus under the present

suggested method results such as those on sheet 4 are to be

expected.

VII - TECHNIQUE A1~ ADEQUATE SIEVES

I feel that some errors may be eliminated if a more

rigid code for operation is set forth. It is my stiggestion

that the sieves used be changed and a time element be intro-

duced into the washins. The accompanying figure sho'V1Ts theI

screens as I. think tb.ey might be arr'snged.

i'o",.p fils confa/ner waler'/-I9hl.

t}//ndriCa/conla/ner

~;%.s"pQce beTween sieYt:?S'/ anolconfa/ner .

f---!'------!

Both sieves fit tightly against sides of container

If an arrangement of sieve of this type could be

devised Rcleanu water could be used in washing; the sieves

could be shaken violently and no losses could occur ~~th

-

reasonable cars being exercised. If the pr eliminary tests

were made using these sieves the percentage of cement re-

tained on #100 could, I believe be somewhat controlled by

f .

using .the 2nd method I .have suggested for its determination

and by washing and shaking the sieves in as a consistent

manner for a certain established length of time.

VIII - CONCLUSION

I do not feel that the apparatus is without pos-

sibilities. It's greatest fault comes from the lack of ~

i' ~

check on the two most important iiems cement and water. ,AJ~~'1 f~¢J

As it stands therefore I do not think the apparatus serves /Ji1,. M""~"I ........\ ~.

its purpose to an appreciable extent. However since the \

sources of error are well known a study devoted to their

elimination should certainly bring about results that will

tend to increase the efficiency of the apparatus to a con-

siderable extent.

/.5