Dynamic Response of a Double Effect Evaporator

H . ANDRE’ and R. A . RITTER’

Department of Chemical Engineering, Univerrity of Alberta, Edmonton, Alta.

The open loop dynamic response of a laboratory scale On a d k i t d’une maniire a d b a t e la r6ponne d y r -
double effect evaporator has been adequately described by mique, sow forme de boucle ouverte, d’un hvaporatenr de
a simple mathematical model based on the unsteady state faibles dimensions et 1 double effet. En utilisant un modhle
material and energy balances associated with the various mathhmatique simple et bas6 sur un r6gime irrhlier et
elements of the system. Excellent agreement has been des Cquilibres d’hnergie qui correspondent aux diven 616-
achieved between the experimentally determined response ments du systime, on a obtenu une exeellente concordance
of the evaporator to large disturbances in steam rate and entre la riponse exphrimentale de l’hvaporateur ides pertu-
feed rate and the predicted response based on the digital bations prononebs dans les vitesses de la vapeur et de
computer solution of the non linear model. I’alimentation et la r6ponse praite, laquelle htait ban& s u r
la solution d’un ealculateur num6rique de modile non.
linhaire.
A lthough the methods of design and steady state operation of
single and multiple effect evaporators are well documented
in the literature(*-*), very little attention has been given to the
transient behavior of these systems. Those publications which Theoretical Analyrir
are available indicate that the studies have been restricted to
actual single effect evaporator^(^^^) or hypothetical multiple In order to simplify the development of a mathematical
model, the evaporator was viewed as a combination of inter-
effect evaprators(7). The apparent lack of interest probably
acting elements which could be adequately described in terms
arises from the fact that evaporators of industrial size respond
of appropriate lumped parameters. Further simplification was
rather slowly to load disturbances and can therefore be ade-
then necessary to reduce the number of differential equations
quately controlled through manual or simple automatic mani-
in the resulting complete model to manageable proportions.
pulation of the input variables. Nevertheless, both the present
This was accomplished by substituting steady state algebraic
investigation and that conducted by Andersen et uP5) have
equations for those first order differential equations which
demonstrated that many interesting features of evaporator
displayed relatively small rime constants. For this purpose, the
operation can be explained in terms of the dynamic properties
of the system.
time constant was defined as the ratio of the coefficients of the
This paper outlines the theoretical and experimental analysis first order and zero order terms in the linearized differential
of the transient response of a laboratory scale double effect equations. I t is worth nientioning that considerable caution
evaporator designed to concentrate 180 Ib/hr of the 5 wt % iiiust be exercised in the application of this technique since
sugar solution to a final product containing approximately interaction between the elements may be such that the individual
1 5 wt % sugar. T h e experimental apparatus, illustrated sche-
time constants do not adequately represent the dynamic behavior
matically on Figure I, consisted essentially of a standard short of the associated elements. Thus, final justification for the
vertical tube first effect and a long tube forced circulation second method must be derived from experimental analysis and may
effect. Both feed solution and steam were introduced into the in fact only apply to the particular system under consideration.
first effect while the concentrated product was withdrawn from T h e following development of the mathematical model is
the second effect recycle stream. The second effect was further presented with reference to the block diagram in Figure 2
equipped with a cyclone separator and a water cooled total which depicts the interaction between the essential dynamic
condenser which was maintained at the desired operating elements of the system. The connecting lines between the
pressure by means of a steam ejector and pressure controller. elements represent the transfer of heat, liquid solution or vapor.
Non-condensible gases which entered with the feed solution
and accumulated in the second effect steam chest were continu-
ously purged from the system. T h e purging rate was controlled
in response to a differential vapor pressure transmitter located
c
A. Steam Calandria of First Effect: Assumin that the steam
in the calandria is saturated and that condensate oldup and heat
losses are negligible, the unsteady state material and enthalpy
at the top of the steam chest. balances may be written
The camposition of the product from each effect was ac-
accurately monitored by separate inline refractometers. In v,,dPS,
- = .F - s.,. . . . . . .
dt
addition, a continuous record of all pertinent flow rates and
inventories was maintained. It was, therefore, possible to
perform a complete steady state and transient analysis of the
system under a wide variety of operating conditions.
.
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...

‘Present address: Department of Chemical Engineering, University of

Subsequent substitution of Fquation ( 1 ) into Fquation (2)
Calgary, Calgary, Aha. yields

The Canadian Journal of Chemical Engineering’ Vol. 46, Augurt, 1968

259
 PR 0 U CT

FIRST
EFFECT
P$ ~ ~ ~ T ~ ONDEN
~ ~ EE c CONDENSATE

II- STEAM

kcoND?EcTJ

FIRST EFFECT
ST

NMlt FE'ED

Figure 2-Evaporator-block diagram.

Figure 1-Evaporator-schematic diagram.

Finally, substitution of Fquations (5), (6) and (10) into

IQuation (1 1) yields

Tsr = -
.Si(H,i + 32) + ",l,4,T'l
__ _ _
Over thc tcnipcrature range of interest the enthalpics of 1.0s; + U,,d, . . . . . . . . ..(12)
steani and water at the saturation p i n t can be described with
an accuracy of better than 0.5% by the relationships B. Calandria Tubes in First Effect: Provided that longitudi-
nal temperature gradients in the metal tubes of the calandria are
H = 1066.0 + 0.4T. . . . . . . . . . . . . . . ..(4) negligible, a simple energy balance of the form
h = -32.0 + 1.OT. . . . . . . . . . . . . . . ..(5) d 1;'
where 7' is in degrees Fahrenheit. Wq C - = us,A~(Tsl
dt
- Tcl) - UciA1(Tc1- T1)*.(13)
Now assuming that a spatially averaged temperature, TCi,
can bc associated with the calandria walls, the rate of heat
transfer from the steam in the calandria is given by r
may be written to describe the res nse of the tube temperature.
Assuming now that the heat trans er coefficients are independent
of temperature, Rquation (1 3) is linear and the time constant
QsI = U,,Ai(T,, - Tc,)... . . . . . . . . . . . . . .(6) is defined by
Furthermore, since the steam is assumed to be saturated, the
steam density, p,,, is a unique function of steam temperature.
I f a linear rclationship of the form
ps, = +
aT,, 0.. . . . . . . . . . . . . . . . ..(7)
is assuined, substitution of Qiiations (4), (5), (6) and (7)
into I'quation (3) yields First Effect Calandria
Calundria Walls
a = 0.00075 Ibs/("F) (cu ft)
dT;,
__
Si(H,i
= -
+ 32.0 + l.OTsl) - U,,AI(T,,- T r l....
)
.(8)
A , = 9.3 ft' Cpc, = 0.12 B t r ~ / ( l h ) ( ~ F )
dl Vsl (0.4p,, + ~(1098.0- 0.6 T s I ) Si = 1.0 lbsimin LI,, = 8.4 Rtu/(OF)(sq f t )
T,, = 220'F ( iiii n )
U,, = 25.3 Btu!(min)("F) W,, = 97 Ihs
A Taylor serics expansion of the right hand side of Equation (sq f t )
(8) and the subsequent rejection of the resulcing non-linear V,, = 0.4 cu ft Va or Spare of the First
terms provides the follnwing expressions for the time constant C,, = 0.043Ibs/(cu ft) E d and the Steam Chest
defined earlier: of the Second
7 = - (-)a dT,, -I
Solution in the First Effert
BI = 1.69 Ibs/min
H,, = 1139.7 Btu/lb
Q I = 721.7 Btu/niin
aT,, dt CI = 0.0434wt fractioii Sc2 = 0.74 Ibs/min
C, = 0.0302 wt fractioii V1 = 1.1 CLI f t
or
7-=
V,, (0.4 p,l + a(1098.0 - 0.6 T,,)
Si +U~~AI . ;1 . . . . .(9)
F = 2.43 Ibs/min
Hol = 1139.7 Btu/lb
hi = 152.3 Btu/lb
LI = 14.0 Btu/min
pvl = 0.0216 Ih/cu ft

Second Effect Heating Tubes

A2 = 4.6 sq ft
0, = 0.74 Ib/min L l 2 = 4.55 Btu/('F)(sq ft)
If, in fact, this expansion is performed about the actual steady Ql = 944.7 Btu/min
statc values presented in Table 1, the corresponding value of the W I = 27.8 Ibs (mi n )
time constant is found to be 1.27 X 10-3 minutes. In comparison,
the time constants associated with the response of the sugar Solution .in the Second Efect
concentration in each effect to changes in the fced conditions B I = 0.08 Ibs/min
are approximately 10 minutes. T h e relative response rate of C2 = 0.0751 wt fractioii
the steam temperature in the calandria is, therefore, much faster Ho2 = 1126.1 Rtu/lb
hz = 118.4 Btu/lb
and Equations ( I ) and (2) may be replaced by their steady La = 55.9 Btu/min
state counterparts, 0 2 = 0.71 Ibs/min
W2 = 21.1 Ibs
s i = s,,.................... .( 10) X = - 66.1 Btu/(lb)(wt
and fraction)
S,H,i - S,,h,,, = Qsl. . . . . . . . . . . . . .(11)

260 The Canadian Journal of Chemical Engineering, Vol. 46, Augurt, 1968

r =
qb'
................. (14)
UIIAi + UG,AI
Again utilizing the steady state values provided in Table 1,
it can be shown that the time constant pertaining to the calandria
tubes is 0.04 minutes and, therefore, negligible. As a result,
the temperature of the tubes is only a function of the steam and
su ar solution temperatures and the rate of heat transfer to the
f
so ution may be expressed in terms of an overall heat transfer
coefficient:
Q1 = U1dl(Tq - TI)... . . . . . . . . . . . . . ( I S )
C. Solution in First Effect: The dynamic response of the
first effect and, indeed, the entire evaporator system is highly
dependent upon the solution circulation rate through the cal-
andria. Substantial concentration gradients will be established
across the calandria if circulation rates are low in comparison
with the feed and/or evaporation rate. While this problem
may become significant in large evaporators involving viscous
solutions, estimated circulation rates of 140 lb/min@*D)and
experimentally measured solution concentrations above and
below the calandria indicated that the first effect in the present
system behaved essentially as a perfectly mixed cell. Thus,
the material and enthalpy balances relating to the solution in
the first effect may be written,
d WI
- = F - B, - ............
0 1
dt
d-W-ICI
dt
- FC, - B i C i . . ............
That is, the enthalpy of the vapor in the combined vapor space
is equal to the enthal y of the vapor just leaving the boiling
solution in the first ef! ect. Therefore, since it can be assumed
that the boiling point rise of the sugar solutions encountered here
is negligible, the temperature of the vapor in the steam chest
of the second effect is equal to the temperature of the solution
in the first effect.
The tubes which provide the heat transfer surfaces in the
second effect are like those in the first effect, of low thermal
capacitance. Consequently, the tube temperature again responds
rapidly to load disturbances. By way of a similar argument to
that used previously, it can be shown that the instantaneous rate
of heat transfer to the solution in the second effect is given by
Qn = - Tz)... . . . . . . . . . . . . . ( 2 5 )
UZ-~Z(T,
E. Solution in Second Effect: It is reasonable to a s s m e that
all of the sugar solution in the second effect experiences perfect
mixing since the liquid circulation rate of 200 Ibs/min is ex-
tremely large in comparison with both the throughput rate of
3 Ibs/min and the total holdup of 30 Ibs. As a result of this
simplification the material and energy balances may be written
. .(26)
dWsCz
-= BlCl - B&. .....
dt
d- W ah
=
dt
QZ + Blhl - Bzhn - OzH02 - LZ.. . . . .(28)
Through appropriate substitution Equations (27) and (28)
can be reduced to

D. Vapor Space of First Effect and Steam Cheat of Second

Efieet: In the event that a high resistance to vapor flow
dl = (BI(C1 - CZ) + ole,).. . . . . . . .. ( 2 9 )
w 2dC1
exists between the two effects, the dynamic response of the and
first effect vapor space and the second effect steam chest must
be considered independently. However, under the present Wz dhz
- = Qz
dt
+ Bi(hi - h t ) - Oz(H02 - hi) - L1... ( 3 0 )
circumstances the measured pressures in the two chambers were
essentially equal suggesting that the resistance was, in fact, If the vapor in the second effect is saturated and if the boiling
very low. The entire vapor space was, therefore, considered as point rise in the solution is negligible, the temperature of the
a single element for purposes of dynamic analysis. The appro- vapor and solution are equal. Furthermore, since the temperature
priate material and enthalpy balances relating to the vapor and pressure of the saturated vapor are related, the enthalpy
space may be written of the solution in the second effect may be completely defined in
terms of the pressure and concentration. In general, the re-
lationship between these variables may be stated as
dhz ahzdC2
- = -- aha dPz
dPrlK1
VI - OIHOI
dt
.
.Sczh,, .Qa. .. , . .( 2 0 ) dt aCz dt
--
4-dPz dt
. . . . . . . . . . . . (. 3 1 )

or more conveniently as
provided that the vapor is saturated.
I t is apparent that Equations (19) and (20) are entirely dhn
-- =
dt
-dt + Y dPz
x dCz - . . . . . . . . . . . . . ..( 3 2 )
dt
analogous to Equations (1) and (2) and, since the corresponding
parameters are of the same order of magnitude, it follows that
the time constants pertaining to the vapor space are again Since the pressure in the condenser associated with the second
negligible. The appropriate steady state equations are in this dP2
instance effect was, in fact, under control, the quantity, -, may be set
dt
01 = s"z.................... '(21)
and equal to zero, whereupon Equation (32) reduces to
OlHol - Sczhlz = Qz... . . . . . . . . . . . . . . ( 2 2 ) dhz
Furthermore, if it is assumed that condensate holdup is negligible,
-=
dt
x-dCz
dt
...................
then.
Scz(Hvl- h l n ) = Qz... . . . . . . . . . . . . .(23) The simple relationship between Equations (29) and (30)
defined by Equation (33) may be used to derive the following
-
where (Hvl &) is the latent heat of vaporization of warer. expression for the rate of vaporization in the second effect:
Thus from Equations (21), (22) and (23) the result is obtained
that
Hal = Ho1.. . . . . . . . . . . . . . . . . . . . ( 2 4 )

The Canadian Journal of Chemical Engineering, Vol. 46, Augurt, 1968

261
 24

2.2
9 - - IST.-- EFFECT
. PRODUCT
- 0

2 .o

\
I g 1.8
J
z
0 - PREDICTED
I -PREDICTED
0 EXPERIMENTAL
1.6 0 EXPERIMENTAL
a
z
0 3
~
0 7 1 2 1.4
LL

5 1 1.2

49

i------
I .o
IST. EFFECT

0.8
0 20 40 6Q 80 160 120 I40 160 I60
4 0 20 40 60 80 100 120 140 160 180
TIME -MINUTES
TIME -MINUTES Figure L R e s p o n s e of the first and second effect liquid
Figure 3-Response of the first and second effect composi- product rates to a step up in steam rate.
tions to a step up in steam rate.

F. Inventory Control in Each Effect: In order to maintain a
constant inventory of sugar solution in the evaporator, each
effcct was provided with liquid level control. While for this
purpose the flow to or from the effect may be alternatively
chosen as the manipulated variable, the latter arrangement was
adopted for the present study. The feed rate of dilute sugar
solution to the system was thereby retained as an uncontrolled
load disturbance.
The action of the conventional proportional plus reset control
dcvices used in this connection is adequately described by the
exprcssion
na = - K , [e + Jedt] + mo... . . . . . . . ( 3 5 )
These equations, together with appropriate empirical relation-
ships between controller output and flow rate from each effect,
then complete the mathematical model of the system.
state performance and dimensions of the system or they were
predicted on the basis of well established correlations. O n the
other hand, certain essential quantities, including the overall
heat transfer coefficients and the heat losses, were not readily
measured and were, therefore, determined indirectly. Values
of the heat transfer cocficients were derived from enthalpy
balances around each effect while discrepancies in these balances
were attributed to heat losses. I t is interesting to note that the
model was relatively insensitive to changes in either of these
parameters. Cmnsequently, estimates of heat transfer coef-
ficients and heat losses within 2S% of the actual values were
quite acceptable.
The required enthalpy data for sugar solutions within the
desired concentration and temperature range, from 0% to 15%
sugar and from lW°F to 2m0F respectively, were estimated
from an empirical equarion
h = T(1 - 0.454C) + 0.6C - 32.1 ... (36)

G. Transport Lag: Thc computer simulation of the evaporator
system was generalized to some extent by incorporating into
the progrim a procedure for estimatin the transport lag associ-
B
ated with the flow of sugar solution rom the first effect to the
second effect. For this purpose plu flow conditions were
f
assumed to exist in the line connecting t e two effects. I t should
be pointed out in this connection that experimentally determined
transport lags were in the order of two to three minutes(10) and
did not thercfore significantly influence the dynamic response of
the system.
Computer Simulation
In general, the data required for the solution of the foregoing
equations were either obtained directly from the initial steady
based on data reported by Higbie(11).
Simulation of the dynamic response of the open loop evapo-
rator system was carried o u t on an I.B.M. 7040 digital computer.
For this purpose, the application of a fourth order Runge-
Kutta-Gill integration routine with a step size of 0.2 minutes
proved to be quite efficient in that 45 seconds of computer time
corrcsponded to 180 minutes of actual evaporator operation.
The initial conditions required for the simulation were
provided by the computer solution of a simple steady state
model of the evaporator. Actual values of feed rate, feed com-
position, second effect pressure, heat transfer coefficients and
heat losses encountered in the laboratory were used in the
calculation. As a result, a common basis was provided for the
theoretical and experimental analysis.

262 The Canadian Journal of Chemical Engineering, Vol. 46, Augurt, 1968
 2wI b- - - IST. EFFECT PRODUCT

- PREDICTED

i'j
o EXPERIMENTAL

0
E'
8
v)
b 2 ND. EFFECT

- PR ED1CT ED
0 EXPERIMENTAL

4
1*21i
I .o

c
20
0
d180 40 60 80 100 120 140 160
TIME -MINUTES
0 20 40 60 80 100 120 140 160 180
TIME- MINUTES Figure 6--Response of the first and second effect liquid
product rates to a step up in feed rate.
Figure LResponse of the first and second effect cornposi.
tions to a step up in feed rate.
it to a wide variety of evaporator installations. T h c modular
structure of the model facilitates the introduction of additional
Experimentul Analysis and Results elements and permits the necessary modification of the equations
In order to test the validity of the open loop dynamic model, to account for changes in the relative time constants of the
the predicted response of the laboratory scale evaporator was existing elements .
compared with the observed response under a variety of oper- In subsequent publications it will be shown that the model
ating conditions. More specifically, the transient behavior of presented here provides a suitable basis for the design and
the product rate and the product cornposition from each effect analysis of evaporator control systems and provides some ex-
was determined following a step chan e in steam rate of approx- planation of the peculiar operating characteristics of industrial
B
imately 20% and a step change in eed rate of approximately evaporator systems.
10%. T h e results of these analyses are shown graphically in
Figures (3) to (6). Acknowledgment
In spire of the rather large load disturbances imposed upon
The authors wish to express their appreciation to the National Research
the process, it is apparent that the agreement between the Council for partial support of this project and to acknowledge the fact
predicted and measured response is quite satisfactory. The that the research portion of the project was carried out in the Department
of Chemical and Petroleum Engineering at the University of Alberta.
minor dicrepancies evident in the response of the product
concentrations probably result from the assumption of perfect Nomenclature
mixing of 'the solutions in each effect and the small transport
la associated with the composition measurement. The some- A = heat transfer area, ft*
R
w at wider deviations between predicted and measured product
B = liquid bottoms flow rate, Ibs/min
C = concentration, wt fract
flow rates are primarily due to inaccuracies in the interpretation C, = heat capacity, Btu/(lb)('F)
of noisy orifice differential pressure recordings. e = error signal, percent
F = feed rate, Ibs/min
H = vapor enthalpy, Btu/lb
Conclusions h = liquid enthalpy, Btu/lb
K, = controller gain
The dynamic response of a double effect evaporator can be L = heat loss, Btu/min
adequately described in terms of a simple mathematical model m = controller output, percent
comprising the material and cnthalpy balances associated with ma = initial controller output, percent
the process components. In particular, the model consists of 0 = evaporation rate, Ibs/min
P = ressure, psia
five first order differential equations, appropriate equations to
describe the control action and a number of algebraic equations
Q =I eat transfer rate, Btu/min
S, = condensation rate, Ibs/min
to define those elements which exhibit a relatively fast dynamic Si = steam rate, Ibs/min
response rate. Ti = controller integral time, min
T = temperature, "F
Although the model was developed for a specific system, it t = time, min
is reasonable to assume that minor modifications would adapt U = heat transfer coefficient, Btu/(ft%)("F)

The Canadian journal of Chemical Engineering, Vol. 46, Augu I6#, 1968 263
 v - volume, cu ft Referericer
w = weight of liquid holdup or calandria, Ibs
x = partial derivative of liquid enthalpy with respect to con- Badger, W. L.: “Heat Transfer and Evaporation”, Chemical Catalog
centration, (dk?/dc2) Company, New York, 1926.
Y = partial derivative of liquid enthalpy with respect to Brown, G. G.:“Unit Operations”, John Wiley and Sons, New York,
pressure, (dh,/dP,! ----.
1036
Leibovic, K. N.: Chem. Eng. Progr 54 No 3 1958
P = density, Ibs/cu ft Standiford, F. C.: Chem. Engineer& 40, Nb. > l 1963.
r = time constant min Andersen, J. E., Glasson, L. W. ahd Lees, F.’ P.,: Transactions
S.I.T., March. 1961.
Johnson D.E.: I.S.A. Journal 7, No. 7 1960.
Subscripts Manczak K.: Archiwum Autdmatyki I ’Telemechaniki, 1, No. IX,
I = refers to first effect 1964.
Levy. S.: J. of H.T Trans. ASME May 1960.
a = refers to second effect Lottes. P. A. and d i n n , W.S.: NAc. Sd. and Eng., 1, No. I, 1956.
CI = refers to first effect calandria Andre, H.:“A Mathematical Model of a Two Stage Concentrating
c2 = refers to second effect heating tubes Evaporator”, Ph.D. Thesis, Univ. of Alberta. Edmonton, 1966.
refers to feed Hi bie, H. E.: Ind. & Eng. Chem., 47, No. 1. 1955.
f = Rafston, A. and Wilf. A. S.: “Mathematical Methods for Digital
01 = refers to vapor produced in the first effect Computers”, John Wiley and Sons, New York, 1960.
01 = refers to vapor produced in the second effect calandria
SI = refers to the steam in the first effect calandria Manuscript received September 12, 1967; accepted February 26, 1968.
s1 = refers to the steam in the second effect steam chest
si = refers to the input steam
UI = refers to the vapor in the first effect and the second effect
steam chest * * *

Dissolution of Carbon Steel Scrap in

Hydrochloric Acid

TYSON RIGG

Peace River Mining & Smelting Limited, Edmonton, Alta.

The dissolution of carbon steel in hydrochloric acid has
been investigated from the point of view of large scale
manufacture of ferrous chloride. It has been found that
the rate of dissolution at 95°C. can be represented by the
equation :
R = k ([HCI] + 0.58 [FeClJ)o,’6
where k varies with the physical and chemical nature of the
steel. The activation energy varies with the particular
experimental conditionn over the ranse 12.0 to 18.1 K.cal/
mole. The rate of reaction is not influenced by agitation.
The mechanism of the dissolution process i n dincussed
together with practical aspects and pilot plant results.
On a Btudib la dissolution de I’acier nu carbone dans
I’acide chlorhydrique, en rapport avec la fabrication du
chlorure ferreux Bur une grand khelle. On a trouvk que la
vitesse de dissolution i 95°C peut dtre represent& par
I’bquation suivantc:
R = k ([HCII + 0.58 [FeCU)’.*
oii K varie avcc la nature physique et chimique de I’acier.
LUnergie d’activation varie avec lea conditiom expbrimen-
tales particulitwes dans l’intcrvalle dc 12.0 ii 18.1 K. eal/
molicule. La vitesse de reaction n’est pas influencb par
I’agitation. On discute le mbcanisme du procbdb ainsi que
les aspects pratiquee et lee r&ultats d’une urine-pilote.

I n connection with the proposed utilization of carbon steel

scrap as a raw material for the roduction of high-grade iron
P
powder via the RCA ferrous choride reduction process(’) it
was necessary to obtain some kinetic data on the dissolution of
such material in hydrochloric acid. T h e action of hydro-
chloric acid on iron has been studied previously by many
investigators but in virtually all cases the experimental con-
ditions were vastly different from those required for the de-
liberate dissolution of large quantities of metal. The present
work was undertaken because of the dearth of information
relevant to the development of large-scale dissolvers suitable
for commercial use.
Experimental
Most of the iron samples consisted of ordinary 5/16-in.
plate washers purchased from an engineering supply house.
These washers appeared to have been made by cold-punching
from a thin hot-rolled sheet. The washers were degreased
before use by extraction with carbon tetrachloride in a Soxhlet
apparatus. Analysis of a large number of the washers showed
that they could be divided into a high carbon group and a low
carbon group. The high carbon group contained 0.524 f
0.014%c and the low carbon group contained 0.031 f
0.007%c.T h e inangancse contents were respectively 0.77%

The Canadian Journal of Chemical En#ineerin#, Vol. 46, Aumaa, 1968

264