Conference held by André Chapelon in the university of Paris - La Sorbonne in 1933.
This conference follows the firsts enormous successes gained by Chapelon in the beginning of his brilliant career; In 1926 with the creation of the KylChap, 1929 with his first series of PACIFICS and 1932 with the launching of the first series of 480s.
It's the more exhaustive and concise text I've found about the history and development of the compound mode in general. I hope this translation is readable.
Note the (I believe) very interesting preamble regarding Chapelon's views about his scientific and technological approach. It's a very short introduction to an mostly unknown aspect of Chapelon's work, I mean Chapelon as a philosopher and historian of sciences and technology.
Reading, as time permits, as much steam related material as I can find on the w.w.w. and perceiving through it some lacks of tangible reference elements, I believe that it's time give some place to the history of steam machines and modern thermodynamics as well as to remember some apparently forgotten famous names and works. Can be the starting point of long and fascinating encyclopedic researches... In any case, that's what I highly suggest!
Chapelon had a quite peculiar style, I made my best to try to respect it. Might improve with time... and your help, may be! Many thanks to those helping me to improve my Anglo-American!!
Words or sentences in italics as in the French text. As usual, all the drawings and photos can be enlarged by clicking them.
By Mr. André CHAPELON,
graduate engineer from the "Arts et Manufacture" ("Arts and Factories")
school,
Engineer at the Hardware Service of the Compagnie d'Orléans.
Even if the compound system has a brilliant history, it has not escaped from the vicissitudes of fate. After the fights of the beginnings, which brought it to the greatest successes came the times when discussions starts again and where confidence seems to be restricted.
Science itself can't avoid the inexorable law of fortune. The best theories are only temporary, because their truth is often only relative to an interval of conditions. Theories became false when the limits of that interval expand.
In fact, to use a stylistic device borrowed to geometry, we only know a few tangents to the curve of realities.
When great progresses are made, it's sometimes possible to replace two or three of these tangents by an envelope; It's the vanishing point of a few theories and their replacement by a new one, more general, that one will perhaps abandon when something better will be found.
If the achievement of Science, its ideal being of course the knowledge and the complete understanding of nature's laws, seems not to cease to escape from us, we can get over our absence of complete understanding because by simply making laws whose significance is limited but useful and consequently convenient, Science, which is human, serves our needs of expansion, of development and of progress in accordance with evolution principle by making technical realizations feasible, humble but tangible goal of Science.
Technology, more than Science, is subject to fluctuations; Its complexity makes it more arbitrary and, sometimes submitted to a blind empiricism it takes on unexpected shapes and can follow an evolution opposite to the Logic: It's in this way that it happens to technology to be under influence of fashion.
Engineer's part is to lit technology to the light of scientific principles. He must look to conciliate the teaching of the theory and results of experiments, enforcing Descartes' method (n.o.t.: Descartes is a major French mathematician, physicist and philosopher of the 17th century, one of his most famous work being his "Speech about Method" or "Le discours de la méthode" written in 1637) and pursue his goal with confidence, because success always crown so well oriented efforts.
History of compound locomotive shows many examples of it.
What is the compound machine? To know it, it is good to refer to its origins.
Contemporary of James Watt, it is the creation of Jonathan Hornblower, builder of machines in Newcomen (G.B.), who patented on july the 13th., 1781 (#1298), the use of two cylinders to operate expansion. He uses, he said, steam after its action in the first "container" (n.o.t.: retranslated, don't know the word used in the patent) to act in the second one by its expansion.
The idea was quite innovative and would have been, in those times, incomprehensible if it hadn't gave a way out to a then new issue: The expansion.
Lacking of a proper mechanism to cut the inlet in the single cylinder machine, one solves the problem by first filling completely a small cylinder, then by sending exhaust steam of that small cylinder in a large cylinder, its capacity being about twice the one of the small cylinder: In this way, the amount of expansion was of two.
Arthur Woolf started again Hornblower's idea in 1803 and had the chance to succeed, thanks to the slightly higher pressures used in his times and to the increase to expansion that he raised to 6 to 8 volumes. It must be said however that, if Woolf's machines had such a great success, it's mainly due to the lowered difference of pressure between the two sides of the piston, which eased hugely the disastrous effect of leaks in times when machines building and maintenance was rudimentary.
Hornblower's and Woolf's machines used two cylinders arranged side by side; Both pistons drove the same crank and moved in parallel.
In 1824, Joseph Eve patented a "compound" machine in which steam, after its usage in a first high pressure engine, was sent in a second, low pressure engine where it expanded.
In 1834 Ernest Wolff took a patent (#6000), describing the compound machine as we know it today, using an intermediate receiver. This machine even featured the inlet of live steam to the intermediate receiver to help starting, technique used on every actual compound locomotives.
Woolf's machines and the compound machines of Wolff almost gave off in
the Marine and took, around 1860, following the most resounding results
(consumption reduced of more than a half), the immense development than
we know; They were only abandoned to leave place to triple and quadruple
expansion machines, ultimate development of this expansion mode.
|
Figure one schematically explains the working
of Woolf's machines and of the Wolff's compound machine. We see noticeably
what is the influence of fractionating expansion on condensations inside
the cylinders, condensations depending on the difference of temperature
between inlet and exhaust. One can see that, from this point of view the
compound with receiver is far superior to the Woolf's machine which has
few differences with the single cylinder machine but benefits as the compound
of the reduction of harmful areas.
|
 |
The most curious thing about evolution of the compound machine is that it was almost accomplished in total ignorance of the wall action phenomenon.
The genius of Watt had the premonition that something particular happened in the cylinders of a steam machine, the water consumption he recorded being four times the one corresponding to the volume moved by the piston; But he allocated this surplus to condensations due to cooling that happened against cylinder wall during expansion, idea that, besides, leaded him to invention of the envelope (1763).
Rankine himself attributed the presence of water in cylinders to the condensation of dry steam during adiabatic expansion, according to the newly born theory.
It's Hirn, in 1855, who demonstrated first, in a positive manner, usefulness of the envelope to lower consumption of machines; But at first, he carefully avoided to give the scientific explanation of it, because none had been found yet.
The English engineer Clark seems to be the first to have fully understood the wall action phenomenon and its influence on efficiency of machines. He has given the explanation of it in his masterly "Railway Machinery" of 1855, were he relates numerous experiments made in 1852 on british locomotives.
"It's today proven that, in many cases, to an increase of expansion is corresponding an increase in consumption and a decrease of developed power", he said; It's that when expansion increase, the losses due to condensation also increase and lower the efficiency.
Clark recommend as a remedy the increase of speed of the piston, method whose value has been since demonstrated.
He points out, supporting this theory, that british drivers had noticed that 20% was the most economical setting for valve gear on their engines. Tests made in France by Desdouits thirty years later gave exactly the same results.
In the new edition of his book published in 1877, Clark clarifies his thinking: "In practice, the only hindrance reventing to adopt favorably long expansion in a unique cylinder, consists in condensations occurring in the beginning of stroke, when steam is introduced against colder piston and cylinder areas".
Hirn was the first thermodynamicist (1876) to look after real functioning of machines in collaboration with Leloutre; He recorded losses of 30 to 70% due to condensation in the machines he studied.
The Theory of the real machine was given the shape we know today in 1878 by Dwelshauvers-Dery who thus achieved the masterly work started by Hirn.
Thus, while the steam engine was spreading at an incredible rapidity, the theory of this machine progressed feeling one's way.
From 1824 to 1850, incomplete researches related to an inexact yet theory are made. Carnot and Clapeyron illustrate this heroic era: Carnot by making physic and thermodynamic theory of fire machines, Clapeyron by giving a graphical representation of cycles.
From 1850 to 1860, the exact thermodynamic theory is established by Rankine and Clausius. It's the theory of the ideal machine, to which Thompson, Combes and Zeuner were also devoted.
At last, in 1876 and 1878, a complete theory of the steam machine is established by Dwelshauvers-Dery; It's the theory of the real machine to which works of Mr. Nadal and their application to locomotives added more brightness.
As one may well think, these new ideas were not well agreed everywhere with favor: It's because the famous explanation of condensations by water driving with steam coming from boilers was strongly anchored and that one doesn't abandon such a "lazy" explanation which was taught yet in 1900 and of which I wouldn't dare to affirm that not any supporter still exists.
It is for those reasons that one couldn't pay too much tribute to the daring and prolific work started as soon as 1873 by the distinguished engineer that was Anatole Mallet, in particular for the application of compound mode to locomotives.
Mallet had understood, one of the very firsts, this dark phenomenon for so many others of wall action and he has seen in the compound mode the most efficient mean then to put it right.
In his remarkable memorandum of 1877, read at the Société
des Ingénieurs Civils de France (n.o.t.: Civil engineers society
of France), he threw light on the value of compound mode and on its value
in the improvement of locomotives, remembering the teaching of Thomas,
his master at l'Ecole Centrale, and supporting himself on the works
of Leloutre and Hirn in particular which occurred brightly almost at the
same time but without succeeding to convince the greatest number.
|
Whatever, Mallet got the better on the by then
more than resistant opinion of the railroads, and he achieved in 1876 to
apply the compound mode on two small 0 4 2 engines of the "Chemin de fer
de Bayonne à Biarritz", which were built in Le Creusot (Industrial
- mainly mechanical building and related metalworking, thanks to iron and
coal mines of this area - town of the Massif Central, tr. note.). It's
one of these engines that show you the figure 2.
|
Figure 2 |
Figure 3 |
Figure 3 shows the so simple arrangement of this two cylinders compound,
which so much looked like an ordinary machine that many were mistaken and
congratulated Mallet, in front of the one of these machines on display
at the "Exposition de Paris" (Paris industrial Show) in 1878, to have gave
up his revolutionary ideas and to have at last made an engine like any
other.
|
This machine naturally raised the most violent criticisms. Above all, her dissymetry was condemned and worth her the qualifying of "Wobbly". Nevertheless, a great future was reserved to the two cylinders compound and the "Wobblies" might work during more than twenty years all the Prussian expresses; It was also, as Anatole Mallet himself has said very successfully, the wedge that has allowed the dual expansion to enter the very hostile then circles of the railroads.
Although France had been the cradle of dual expansion applied to the locomotives,
it is not at all in her place that the compound locomotive has covered
all the phases of her development; It's in Germany almost, thanks to the
persevering efforts of von Borries, that the compound mode spread the most
easily. In 1890, ten years after the first application of the system in
this country, not less than 430 two cylinders compound were counted.
Figure 4 |
France get started right away in the 4 cylinder compound mode; It's in
this way that the first compound intended to a major company was set up
in 1886 (10 years after the first Mallet's machine) on the network of the
Nord; She had been built upon the initiative of Mr. de Glehn, engineer
of the Société Alsacienne de Constructions Mécaniques,
in the factories of this company. It' s the locomotive 701 that the figure
4 shows you.
|
|
The figure 5 indicate to you what was the arrangement of her cylinders;
Her two driving axles were not coupled but driven each other by a group
of cylinders.
|
Figure 5 |
 |
The figure 6 shows you the photography of this locomotive which constitute
together with the Nord 701 an historical type.
|
|
Figure 6
|
|
The success of these prototypes was such that from 1890, it was no more built non compound fast locomotives in France (one Etat type and one P.O. type excepted), so well that in 1902, 1 128 compound could be counted on our seven major networks of which 23 were Woolf types, 2 were three cylinders and 16 two cylinders, the 1087 left being of a four cylinders of an almost standard type known under the name du Bousquet de Glehn, or french type, whose Nord 2121 locomotive built under the directives of the chief engineer du Bousquet and of Mr. A. de Glehn, set up in 1892, constituted already the accomplished type (1) (Click the figure to reach the note)
The locomotive which is may be the most characteristic of this time was the "Atlantic Nord" of 1900 that shows you the figure 7, which allowed to reduce to 3.15 h. the duration of the journey Paris-Calais, duration shortened since of five minutes only.
 |
|
|
The figures 8 and 9 show in a striking way what was the influence of the
introduction of the compound mode in France on the speed increase of trains,
on the Nord network first, cradle of dual expansion, then on the Orléans
network which joined to this system in 1900 only.
 |
 |
|
|
|
 |
 |
|
|
|
 |
|
|
The introduction of superheating on the locomotives was the starting point of a memorable controversy between the supporters of the single expansion and those of the compound mode, controversy to which will forever remain bound the names of two distinguished German engineers, Garbe and von Borries; The first, fierce supporter of the single expansion machine, superheated, using a relatively low boiler pressure (12 hpz. or 171 psi.), the later bound to the compound mode that he achieved moreover in the well known four cylinders arrangement, with only two valve gears but with four combination levers, arrangement applied to more than 1 500 machines of the P.L.M..
It's impossible for me to relate fully the history of the development of superheating and of its fight against the compound mode; Let's only say that the superheating went out of the fight indisputably winning, but that its victory was only complete when it was allied, or if you prefer, added to the compound mode.
The tests made on this topic by the P.L.M. between machines identical in every detail, excepted with regards to the engine which was in both case four cylinders either single expansion, either compound, and using in the first case superheated steam (S.E. engine, n.o.t.) and in the second case (compound engine) either saturated steam, either superheated steam, seems to have settled the question almost definitely.
Two Pacific locomotives, identical excepted for the mode of use of the
steam, were set up in 1909: One single expansion with four equal cylinders
using superheated steam at a pressure of 12 hpz (171 psi.) (fig. 13), and
the other, four cylinders compound using saturated steam at a pressure
of 16 hpz.(228 psi.) (fig. 14).
 |
 |
|
|
|
1 - Superheated compound machine;
2 - Saturated steam compound machine;
3 - Single expansion, 4 cylinders superheated compound;
The superheated compound, as well working freight trains as working 'rapids' passenger trains saved 10% in comparison with the single expansion, equally superheated 4 cylinders machines.
As a conclusion the P.L.M. transformed all its single expansion, four equal cylinders machines into compound, keeping their boiler pressure of 12 or 14 hpz ( or psi); The machines transformed in this way use 10% less fuel than before their transformation.
Just before the war (the first world war, n.o.t.), the Prussian State railways made analogue conclusions comparing their ten-wheels locomotives type S.10, one being single expansion with four equal cylinders, the others superheated compounds; They said that they enhanced the superheated machine with the compounding; In France, it was the compounding which was enhanced by the superheating; This common outcome caused for reflection, of course.
The fig. 15 show you the compound locomotive which was built in 1914 in conclusion of its trails by the Prussian State.
The things have nevertheless changed much since as, during and after the war, the german railways have exclusively built three then two cylinders single expansion locomotives; They tested however in 1924 some 4 cylinders compound Pacific which have been found slightly superior only at high powers and were not generalized. The networks of Bade, Bavaria and Wurtenburg remained favorable to the compound mode and the Bavarian Pacific, derived from the prototype designed and built by Krauss-Maffeï has been admitted very recently among the standardized types of the Reichsban. This latter has just been put under test a few Decapod and Pacific compound locomotives whose boiler pressure is of 341 psi (24 kg. per sq. cm.). May be is it the clue to a revival of the compound mode in this country.
The war had also and generally this result to urge to the simplicity rather
than to the perfection, increasing, even in France, the favor of the two
cylinders single expansion superheated machines.
|
Thus, a large number of these locomotives was built even in the case of
the high powers by the companies of the Midi (Pacific), de l'Etat (Mikado)
and of the P.O. (Pacific & Mikado). The figure 16 & 17 show you
a locomotive of the Midi whose boiler pressure is 185 psi (13 kg per sq.
cm.) and a locomotive of the P.O. "american type" built in 1921 by the
ALCo. featuring a boiler pressure of 171 psi (12 kg. per sq. cm).
|
Fig. 16
|
Before to show however to which value of perfection have been carried the superheated compound machine, it is necessary to emphasize in a accurate way the respective benefits of the compound machines and of the single expansion machines.
Wall Action
 |
The entropy diagram (fig. 18) gives us the answer by showing that
a perfect machine working according to the Rankine's cycle exhausts in
strongly wet steam, even for pressures as high as 242 psi (17 kg per sq.
cm.) and already very high superheating temperatures of 752° F. (400°
centigrade).
|
Besides, if we try to analyze even briefly the real phenomenon of the exhaust, we do rather disturbing observations concerning as much the absolute value of the temperatures measured on the exhaust as the temperature of the steam remaining in the cylinder at the beginning of the compression.
Let's consider the figure 19; One can see that the temperature is continuously variable during the anticipated exhaust. The quantity of steam remaining in the cylinder at each instant has expanded adiabatically, either by pushing the piston up to the end of the stroke or by throwing the steam out of the cylinder. The temperature of this steam thus correspond point to point to the temperature given, at equal pressure, by the adiabatic. As for the steam exiting the cylinder, it takes successively a temperature slightly lower than the one of the steam remaining in the cylinder because of the expansion without production of work that it undergo when entering the atmosphere. The phenomenon of walls action raise more or less these temperatures; - But one see that the readings of the thermometer located at the exhaust are difficult to interpret, and that the temperature of the steam remaining in the cylinder is relatively much lower than the temperature measured in this way (i.e. of the exhaust steam). There is even great chances, as can reveal the entropy diagram, for the steam remaining in the cylinder at the end of the exhaust phase to be saturated steam at a titre even quite low (0.95) when the pressure of the inlet steam and the degree of expansion are high enough.
We know that the condensations in the cylinders, or simply the losses due to the effect of the walls, depend of the difference of the temperatures between inlet and exhaust and that, from this point of view, the compound mode which shares between the H.P. cylinders and the L.P. cylinders the total temperature drop and offers in its H.P. cylinder harmful areas, to which are equally proportional the actions of the walls, about twice smaller than the ones of the unique cylinder of the equivalent single expansion machine, presents considerable advantages.
Be:
T1 the temperature of the steam when it enters the H.P. cylinder;
T'1 the temperature of this steam on the exhaust of the same cylinder;
T''1 the inlet temperature at the L.P. cylinder;
T'''1 the temperature of the steam at the exhaust of this cylinder.
We can assume that T'1 = T''1.
It result from the works of Mr. Nadal that the amount of heat q given up to the walls during the inlet and taken back on exhaust is, in particular in the case of the locomotives, more or less proportional to the area s of the harmful spaces and to the difference of the temperatures T1 - T' between the inlet and the exhaust.
We can thus write: q = Ks (T1 - T'1) in the H.P. cylinder of the compound, q' = K's' (T''1 - T'''1) in its L.P. cylinder and q = K1S1 (T1 - T'''1) in the cylinder of the equivalent single expansion machine.
The loss of energy, so due to the fact that a certain amount of heat q given up to the walls of the H.P. cylinder at the temperature T1 and taken back at the temperature T1 can only be used in the L.P. cylinder between the temperatures T'1 = T''1 and T'''1 , is equal to the difference between the works which would have been made by a Carnot's machine setting into action this amount of heat and working first between the temperatures T1 and T'1, next between the temperature T''1 = T'1 and the temperature T'''1.
As the work produced by this Carnot's machine would be respectively equal to:
(q (T1-T'1)) / T1 and (q (T''1-T'''1)) / T
the loss of work searched is:
or:
Likewise, the loss of work in the L.P. cylinder is:
If we assume, in first approximation that k=k and that s1 = s/2:
Such is the approached expression of the loss by wall action in a compound locomotive. In the equivalent single expansion, working between the same temperature interval, it's easy to see that this loss takes the following value:
the harmful area being more or less the same for the only cylinder of the single expansion machine and the L.P. cylinder of the equivalent compound machine.
If we pose for example:
T1 = 400° + 273° ;
T'1 = 250° + 273° = T''1 ;
T'''1 = 130° + 273° ;
comes:
and:
Thus:
The loss by wall action would be so, in first approximation, more than two and a half times higher in the single expansion machine than in the equivalent compound.
The nature of the fluid and in particular its degree of humidity or of superheating take greatly place in the absolute value took by this loss, the coefficient k having very different values according to the condition of dryness of the cylinder walls and according to the condition of the steam.
There take place the considerable advantage of a strongly superheated steam and equally the advantage of a sufficient degree of superheating in the L.P. cylinders of the compound for the losses by wall action in these cylinders doesn't reduce in a sensible way the efficiency of the whole machine.
It shows also that, despite the very high superheating temperature that
one can use on a single expansion machine working at a pressure close to
20 kg. per sq. cm. the wall action can remain if the expansion is
pushed far enough, so that droplets of water cover the cylinder wall in
abundance by the end of exhaust.
Throttling
Dead Spaces
 |
It's sufficient, to demonstrate it, to apply the principle apparently
discovered by Clausius of the break-down of the real cycles, in principle
of arbitrary shape, into Carnot's elementary cycles, merging in the diagram
of Clapeyron with an adiabatic, each one corresponding to an heat input
dQ having an individual efficiency of:
|
T being the initial temperature and T' the final temperature of this elementary cycle.
We are thus able to analyse in all their details the real cycles and look for the conditions of the best use of the heat to convert it into work, taking into account the multiple constraints restricting the possibilities to carry out a practical cycle such as the temperature, the space available and the pressure limited by the conditions of use of the materials.
The break-down of the cycles into Carnot's elementary cycles has been may be the most fertile process made available to the engineer by the thermodynamic.
Let's see the figure 20. If we suppose the cycle of the machine limited during the compression by the adiabatic E F meeting the the equality condition of the areas F A B B' = D C C', we see that, if we replaced this adiabatic E F by another, E' F' shifted to the left of the graph, the consequence would be the entry in the machine of of an additional weight of steam of which the use would be lower since the efficiency of the Carnot's cycle, merged with the adiabatic E' F', is lower to the common efficiency of the elementary Carnot's cycles making up the cycle limited to the contour E F B C', cycle having by construction the same global efficiency than the one that we are studying.
Likewise, if one used a compression stronger than the one corresponding to the adiabatic E F, the cycle studied being limited for example by the curve of compression E'' F'', we see that due to this (stronger compression, tr. note), we would cut out from the cycle limited by the contour E F A B C D elementary Carnot's cycles of an efficiency higher than the average and therefore, the final efficiency of the cycle would be equaly lowered.
Knowing how to determine the best compression ratio to use in a machine, we can compare what happen from this point of view in a compound an in the equivalent single expansion machine.
We find that, for an equal work produced, a pressure of 20 kg. per sq. cm. (280 psi) and a superheating to 400° C. or 752 Fahrenheit the single expansion machine uses 6 liters or 12.68 U.S pints per H.P. per hour with the inlet cut at 10% while the equivalent compound uses 5.1 liters or 10.77 U.S. pints with inlet settings of 30% in its H.P. cylinder and 45% in its L.P. cylinder of a capacity equal to the one of the S.E. machine and double of the one of the H.P. cylinder; The saving is
One understand indeed that the dead spaces in the L.P. cylinders give rise to a loss that can't be recovered in others next cylinders, are filled with steam at the pressure of the intermediate receiver only, many times lower than the boiler pressure, this reducing considerably the importance of the corresponding losses since the capacity of the space to fill in is the same in the case of the compound that in the case of the single expansion, the L.P. cylinder of the compound being nothing else than the one of the equivalent single expansion machine.
(Pressure) Drop at the intermediate receiver
........ To be continued ...