Principles of Modern Steam

Please click on the red arrows to open the text under each subject heading.


The celebrated Argentinean steam locomotive engineer Livio Dante Porta devised a classification for steam locomotive development as summarized below:

  • First Generation Steam (FGS) – practically all past locomotives;
  • Second Generation Steam (SGS) – new designs incorporating the best proven modern steam locomotive technology;
  • Third Generation Steam (TGS) – totally new formats requiring considerable research and development to achieve.

A more comprehensive description is offered by Dave Warale on page 24 of his book “The Red Devil and Other Tales from the Age of Steam“:

During his time at INTI Porta was not constantly involved in the day-to-day affairs of locomotive operation, and this gave him both the time and distance to develop new ideas for improved design.  Thus many of the advances in component design which were the practical outcome of Porta’s engineering research stem from those years, which were highly productive.  These advances could be grouped into three categories: those which could be applied with confidence without further research, those which still had some unknowns and which therefore required further development before application but which could be used immediately if the risks associated with the unknowns were accepted, and those which definitely required further research and development work before application. Examples of these three categories were piston valve design, the cyclonic furnace, and condensing below atmospheric pressure respectively. Porta also defined three stages in the development of the steam locomotive. First Generation Steam (FGS) included all existing designs. These could be improved without major structural alterations by the application of appropriate items of the new technology, but whatever was done to them their performance would probably still have remained on the FGS level because of their inherent structural limitations. Second Generation Steam (SGS) was defined as that which could be built new using those items of the then-available technology which required little or no further research: an example is the proposed metre gauge 2-10-0 of Fig.2. Third Generation Steam (TGS) was that which would have required a large research and development effort to materialize, such as (in the author’s opinion) the ACE 3000 design described in Chapter 6.3. Burning low-grade coal, modified FGS, SGS, and condensing TGS locomotives were expected by Porta to give year-round drawbar thermal efficiencies of about 10%, 15% and 25% respectively.

Note – Porta’s estimated drawbar thermal efficiency of 25% for TGS is significantly higher than the figure of 16.3% that Wardale suggests on page 501 of his his book “The Red Devil and Other Tales from the Age of Steam“.  Wardale makes it clear that his figure applies to non-condensing locomotives and that “higher efficiency could only be obtained by expanding the steam to sub-atmospheric pressure and low temperature by means of condensing to counter the negative effect on the cycle efficiency of the restricted inlet steam temperature as done in stationary steam plant”.



The α (alpha) Coefficient of a locomotive is defined as

In other words, it gives a measure of the average forces applied to the locomotive’s machinery as compared to the maximum forces that it is designed to withstand.  It could therefore be said to be a measure of “mechanical efficiency”, being the amount of use that is made of the metal that goes into the machinery, or more vitally, it is a measure of “capital efficiency” being the amount of use that is made of the capital cost of the machine.

A machine with a low α-coefficient is one that is designed to withstand forces that are much larger than the average forces applied to it and which therefore makes low usage of its size, weight and/or capital investment.  For instance, diesel engines have to withstand very high compression ratios and therefore tend to have a very low α-coefficient.  Porta quotes a figure of 0.11 for a two-stroke marine diesel engine, noting that diesel manufacturers have learned to artificially boost their α-coefficients through the use of compressors or turbo-chargers.

Single (simple) expansion steam engines might have an α-coefficient of 0.2 when operating most efficiently at 20% cut-off. However Porta compares these figures with that of an “advanced” compound expansion steam engine operating at 50% cut-off whose α-coefficient may exceed 0.5.

Thus compound expansion engines, by virtue of their ability to operate efficiently at long cut-offs make much better use of their their cost and the mass of the steel that goes into making them. Or to put another way, they deliver more power per tonne of engine weight, which is another way of saying that they have a higher power-to-weight ratio.


Several pages of this website include text and diagrams copied from Porta’s “compounding” paper, including the pages covering condensation/wall effects, steam leakage, clearance volume, incomplete expansion and triangular losses.  More specific references to his theories on compound expansion can be found on the Compound Expansion page.

Sincere thanks to Adam Harris of Camden Miniature Steam, publishers of “Advanced Steam Locomotive Development – Three Technical Papers” for allowing the sections of the book to be published on this website.

Efficiency (general)

Page Under Development

This page is still “under development”. Please contact Chris Newman at if you would like to help by contributing text to this or any other page.




Page Under Development

This page is still “under development”. Please contact the webmaster at if you would like to help by contributing text to this or any other page.’


The process of reducing constraints to steam flow is usually referred to as “Internal Streamlining”.

Internal streamlining produces two benefits:

  1. It raises the Front-End Limit resulting in increased steam flow through the cylinders which in turn increases the upper limit of power output for the engine.
  2. It has the effect of reducing flow resistance and thus energy losses at any given steam flow, thereby increasing efficiency, reducing steam and fuel consumption etc.

Streamlining may involve any or all of the following:

  • Increasing steam pipe diameter to reduce steam velocity and consequent pressure drops;
  • Removing or reduce sharp bends to lower pressure drops;
  • Increase steamchest volume to reduce steamchest pressure drop at high steam flow rates. Ideally the steamchest volume should equal the volume of the cylinder that it connects to;
  • Enlarge port openings as much as possible without excessive increase in clearance volume;
  • Rounding the edges of bridge bars and port edges and tapering the edges of valve lands (see diagrams below), can drawmatically reduce resistance to steam flow into and out of cylinders and minimize “wire-drawing” near the point of valve closure;
  • Increase valve travel increases the velocity at which the valve travels over the ports at any given cut-off, thereby reducing the period during which the valve throttles the steam flow as it closes over (or opens over) the port, thereby “sharpening” the valve events.
  • Enlarging the exhaust passages including blastpipe area frees up the flow of expanded (i.e. large volume) exhaust steam out of the cylinder, thereby reducing back-pressure and the quantity of steam retained in the cylinder at exhaust port closure, and thereby allowing more steam to flow through the cylinder.

Streamlining reduces pressure drops wherever it is applied, and in consequence it reduces energy losses. Streamlining therefore has the effect of increasing engine (and locomotive) efficiency, thus reducing fuel consumption for a given energy output, and increasing drawbar work produced by a given energy input.

It should be noted that “streamlining” the steam passages of a steam locomotive involves a lot more than “polishing the ports” of a car engine!

Porta illustrated his ideas for streamlining of valve land and port edges in a proposal that he put forward for improving the design of the A1 Tornado as below. He also calculated the reduction in pressure drop that could be achieved by such simple modifications:

In the case of the upper diagram, Porta calculated that the pressure drop through the improved design would be just 17% of the original, which would have the same effect as increasing the valve diameter by 2.4 times – i.e. 24” instead of 10”.

For the larger valve/port opening in the second diagram, Porta calculated the pressure drop through the improved design would be 31% as compared to the original, which would have the equivalent effect of enlarging the valve to 21.5″ diameter in place of 10″.



Edge Lands and Drifting Steam

It may be noted in the above diagrams that Porta illustrates a very thin edge land on the admission end of the valve head, and a rather thicker one on the exhaust end. The reason for this is that the land has to be thick enough to be able to resist the inertial and frictional forces that derive from the motion of the valve and which may be transmitted to the land by the ring that it retains.

At the admission end, the steam pressure inside the steam chest should be high enough to prevent the ring from contacting the land and imparting its own inertial and friction-derived forces onto the land. Hence the land should carry very little load. However steam pressure is very much lower at the exhaust end of the piston head with the result that the edge land has to be thick enough to withstand the ring loads that are applied to it.

It may be deduced therefore that valves with narrow admission edge lands rely on the presence of sufficient steam-chest pressure to keep the edge ring from coming into contact with the land. Most importantly, where such valves are used, it is essential that sufficient steam pressure is maintained inside the steamchest when the locomotive is drifting. This minimum required pressure is not likely to be high: for instance, in the case of the 5AT when drifing at 200 km/h, the steamchest pressure required to keep the admission rings from contacting the edge lands is estimated to be only 274 kPa. Nevertheless, failure to maintain steam pressure in the steamcest during drifting, may result in damage to, or even failure of, the land at the admission ends of the valve heads.




Lack of steam tightness is usually imagined to be associated with steam leaking from pipe joints and piston rod glands resulting in the familiar leaks that can be seen as a white plume of steam eminating from wherever the leak is occurring.  Steam leakage of this sort is indeed wasteful and deletarious to a locomotive’s performance, however much greater leakage can (and often does) occur that is not only invisible but unknown to a locomotive’s owner or operator.  Invisible leakage of this sort takes place when steam leaks past valve and piston rings and escapes (unnoticed) up the chimney.

The topic is discussed in some detail in two of the papers published in L.D. Porta’s “compounding” paper published in Camden’s book “Advanced Steam Locomotive Development – Three Technical Papers” including his short paper titled “Some Steam Leakage Tests on Locomotive NORA of the Ferrocarril Austral Fuegino”.  Avoidance of such leakage is covered in much greater detail in his unpublished papers on Tribology and the Design of Piston Valves.

Porta describes the worst offenders being locomotives with American-type Duplex rings, and quotes Chapelon under whom measurements were carried out by SNCF that showed steam leakage losses of up to 12% of the maximum boiler evaporation in the case of a very well-maintained 141R.  He even mentions a figure of 20% steam loss from these locomotives, presumably in less well-maintained examples, and compares it with figures of 1.4 to 1.7% loss on the Rio Turbio Santa Fes when fitted with his design of valves and pistons, measured when the rings were life-expired.

The paper about FCAF’s Garrant “NORA” reveals even worse losses that by Porta’s estimation amounted to some 50% of steam generated by the boiler, adding that “when one considers that wall effects (on this unsuperheated engine) increase the indicated steam consumption by AT LEAST 100%, the actual steam/fuel consumption is roughly FOUR TIMES GREATER THAN WHAT IT COULD BE.  In turn this means that the boiler, and (water/fuel) tanks, etc, are four times larger than they should be – same for the annual fuel bill.”

[Following these and other findings, FCAF comprehensively rebuilt NORA, and under the guidance of Shaun McMahon, they incorporporated many “Porta enhancements”, as described by Shaun on a separate page of this website.  At the same time, the locomotive was renamed “Ing L.D. Porta”.]

In his “compounding” paper Porta comments about steam leakage as follows (with edits):

“To the best of the Author’s knowledge, only in France, after World War II, has leakage [past piston and valve rings] been measured. As a matter of course it has been much studied in Internal Combustion engine technology through which it has been well established that leakage occurs through the area left between the cylinder, the piston and the ring joint, not along the circumference, and this is confirmed by Chapelon.  In the case of the Author’s design, the leakage is very small. However, because of it being constant, its importance increases in the case of slow moving machines, such as those used in shunting

As a first approximation, leakage can be considered to be a constant volume of steam by-passing directly from the steamchest to the exhaust. It is proportional to the steamchest pressure and inversely proportional to the absolute steam temperature. But if conditions are such that the parts determining it are below the saturation temperature, its importance increases considerably because what leaks is condensate, whose density is more than a thousand-fold that of the steam. The same is obtained during the warming up period, which lasts roughly 20 minutes at full power, or may last indefinitely at low power.

Poppet valves are heavy offenders where leakage is concerned. The various claims against this statement have never been sustained by measurements or serious reasoning. Chapelon measured heavy leaks as reported in his book. The reason is that except in the Caprotti gear, the valves seat on the cylinder block, itself subject to widely differing (and varying) temperatures, and hence distortions. This aspect is so important that Stumpf developed elastic seats and the corresponding theory for them.

Because of it being constant, leakage may be significant for low-speed engines showing low piston speeds and low volumetric power. This is the case with shunting engines and many ships. The author’s technology since early times incorporated the concept of long strokes (American/GWR practice) and high rotation speeds – 504 r.p.m in AAR standards.”

In a later section of the paper, Porta goes on to summarize how “leakage is reduced through the application of the [Porta’s] advanced cylinder tribology, whose basic points are as follows:

  1. narrow, diesel type piston rings,
  2. as many rings as possible per valve head or piston (6 rings per valve head on Wardale’s ‘Red Devil‘),
  3. the valve resting on the liner so as to have a theoretical zero leakage,
  4. best “diesel” quality materials,
  5. elimination of abrasive material entering via blast pipe and moisture (use of antifoam: diesteraliamide),
  6. some piston rings made out of bronze (to aid lubrication),
  7. piston rods,
  8. oil injection “between” the valve rings (NEVER mixed with the steam),
  9. paraffin based oils,
  10. light weight piston valves,
  11. packing rings for the rods,
  12. liner cooling,
  13. adoption of wheels of the smallest possible diameter as allowed by the AAR standard (504 r.p.m.),

In summary, Porta makes the point that steam leakage necessarily increases steam consumption, and this this directly reduces a locomotive power output as illustrated by his oft-repeated equation:

Porta’s “compounding” paper addresses other closely associated factors that detract from a locomotive’s cylinder efficiency, including clearance volumewall effects (or condensation), incomplete expansion, and triangular losses as described elsewhere on this website.  More specific references to his theories on compound expansion can be found on the α Coefficient and Compound Expansion pages.

Sincere thanks to Adam Harris of Camden Miniature Steam, publishers of “Advanced Steam Locomotive Development – Three Technical Papers” for allowing the sections of the book to be published on this website.

For further information on tribology and piston and valve ring design can be found on the following pages:



The principles of “Compound expansion” are briefly outlined in the Simple/Compound page of this website which also describes the advantages of compounding as follows:

  • Use of compound expansion allows longer cut-offs to be used, thereby delivering more uniform wheel-rim tractive effort;
  • The ability of compound engines to operate efficiently at longer cut-offs increases their α-coefficient and thus their power-to-weight ratio.
  • The reduction in vibration (or knocking) achieved from the use of longer cut-offs removes the incentive to operate a locomotive with a throttled (partially opened) regulator, thereby allowing full boiler pressure in the steamchest;
  • More uniform torque delivered by compound locomotives reduces the propensity for initiation of wheel-slip at moments of (transient) peak torque.  This renders compound locomotives better suited to heavy haulage;
  • Reduced temperature differentials between steam entering and leaving a cylinder, minimizes heat losses and reduces or eliminates condensation, especially where the low-pressure steam is re-superheated.

Porta expands on these ideas in his paper titled “Fundamentals of the Porta Compound System for Steam Locomotives” as published by Camden Miniature Steam who have kindly given permission for extracts of text and illustrations from the book to be reproduced on this website.

Porta introduces the subject by explaining several important but seldom understood fundamentals that apply to equally to simple expansion.  These fundamentals are outlined on this website in the pages titled:

Porta points out that with compound expansion, steam leakage across valves and pistons will be lower than with simple expansion because pressure differentials are lower.  Also wall effects are less pronounced because cut-offs are longer (giving more uniform cylinder temperatures) and in the case of the low pressure (LP) cylinder(s), pressures, and therefore saturation tempatures) are lower.  However pressure losses are greater because of the the transference of steam from the high to low pressure cylinders, hence the importance of internal streamlining is greater.

The Ideal Compound Expansion EnginePorta offers the following insights into the nature of an ideal compound engine:

  • Equalily of Work in Cylinders: The aim should be to produce an equal amount of work (power) from the high pressure (HP) and low pressure (LP) cylinders.  He illustrates this with two simplified indicator diagrams as below, one describing two or four cylinder machines and the other 3 cylinder machines:

In the case of the two and four cylinder machines where the steam from each HP cylinder exhausts into a LP cylinder, the two parts of the diagram representing the HP and LP cylinder expansion are of equal area, giving a receiver pressure of around 25% of the HP steam inlet pressure.

In the case of the three cylinder machine where the steam from a single HP cylinder exhausts into two LP cylinders, the lower part of the diagram representing the LP cylinder expansion is of twice the area of that of the HP cylinder, giving a receiver pressure of nearer 50% of the HP steam inlet pressure.

  • Resuperheating: Resuperheating of the HP exhaust steam increases the work done in the LP cylinder while at the same time reducing the risk of condensationNote: the receiver pressure must therefore be lower where resuperheating is used, in order to maintain equality of work between the cylinders. The diagram below illustrates the effect of resuperheating.

  • Internal Streamlining:  Internal Streamlining is of great importance in the passages, receiver and resuperheater between the HP and LP cylinders in order to minimize the pressure drop through them and thus maximize the steam pressure entering the LP cylinder(s). He observes also that “what counts is the total pressure drop between the HP cylinder and the LP one, the latter including the triangular loss caused by the LP valve.  Thermodynamics tells us that this type of constant enthalpy transformation is more deleterious the nearer it is from the low temperature sink.  This calls for extra large valves and steam passage areas with a kind of “exaggerated” internal streamlining.”
  • Maximum Superheat Temperature: It is important to maximize the superheat temperature of steam entering the HP cylinder(s), consistent with the tribology employed. Porta recommends HP steam inlet temperature of 450oC, for which purpose, he recommends that his “and NO OTHER” cylinder tribology be adopted.
  • Clearance Volume: Interestingly, Porta recommends a clearance volume of around 16% for the HP cylinder so as to avoid over-expansion at low cut-offs.  This is very much higher than the 6% CV that he recommends for simple expansion locomotives.  Note: Porta mentions that in his own three-cylinder scheme, the clearance volume amounts to 35% because of the higher receiver pressure.  However, for the LP cylinder(s), Porta recommends that Clearance Volume be as low as possible (as for simple expansion cylinders) in order to minimize incomplete expansion losses.
  • High Adiabatic Efficiency in HP Cylinder: Porta notes that being a back-pressure engine, the incomplete expansion in the HP is quite low even when working at long cut-off. The resulting enthalpy increase is, in any case, partially recovered in the LP cylinder.  He notes also that the triangular losses can also be kept small through the use of internal streamlining thus giving very high adiabatic efficiency in the cylinder.
  • Maximum α-Coefficient: Porta strongly recommends against the practice of live steam injection into the LP cylinder, since this will either require that the LP piston and motion be designed for HP steam (thus lowering its α-coefficient).  Instead, Porta recommends the use of an automatic starting valve similar to that used on his 1948 experimental 4-8-0 compound as illustrated below.  The purpose of the valve is to deliver live steam to the low pressure cylinders at reduced pressure when starting, thereby allowing all cylinders to contribute to the initial starting effort at their designated working pressure.

Automatic starting valve used by Porta in his experimental compound 4-8-0 of 1948

The live steam 1 (from the HP steamchest) flow to the receiver 2 until the pressure coming via communication 3, plus the pressure in 2, equals the pressure acting on the underside of the valve 4. The various sections are defined by the pressure ratio P1/P2. The piston 5 has a labyrinth seal.
  • Optimum HP and LP Cylinder Sizes: A secondary argument against the use of live steam injection into the LP cylinder is that it necessitates a reduction in the size of the LP cylinder.  Porta recommends that the LP cylinder be as large as possible in order to minimize incomplete expansion losses, and notes that had Chapelon not allowed 10 bar live steam injection into the LP cylinders of his 140Ps, he would have doubled the LP piston swept volume while using the same motion, hence giving better expansion at high powers, reducing specific steam consumption and thus delivering greater power. Porta recommends that the cylinder diameters and stroke be sized to meet low-speed adhesion requirements.  He further recommends that the HP motion be designed for the pressure differential between boiler and receiver, and that the LP motion be designed for the pressure differential between receiver and exhaust/back-pressure.
  • Determining HP and LP Cylinder Sizes: Porta offers some simple formulae for determining the sizes of HP and LP cylinders based on avalaible adhesion.  He begins with the “standard” TE formula for simple expansion locomotives – viz:

T.E.=\frac { 0.85.p.{ d }^{ 2 }.s }{ D }

where p = boiler pressure, d = piston dia, s = stroke and D = wheel dia.

This equation can be rewritten by introducing π/4 as follows:

T.E.=\frac { 0.85.\pi .p.{ d }^{ 2 }.s }{ 4 } \times \frac { 4.s }{ \pi .D }

and introducing adhestive weight and friction coefficient:

\mu .W=\frac { 0.85.\pi .p.{ d }^{ 2 }.s }{ 4 } \times \frac { 4.s }{ \pi .D }

However π.p.d2/4 = piston thrust P, hence

\mu .W=0.85.P.s\times \frac { 4.s }{ \pi .D }


P=\frac { \mu .W.\pi .D }{ 0.85.4.s }


    • P = the force in the piston rod in Newtons;
    • μ = the nominal adhesion coefficient (as selected by the locomotive’s designer);
    • W is the adhesive weight in Newtons;
    • D the driving wheel diameter in metres; and
    • s the piston stroke in metres.

Porta then goes on to determine the cylinder/piston diameters for the high and low pressure cylinders in a compound locomotive (dH and dL respectively) as follows:

If ΔpH is the pressure drop between the boiler and the receiver (between the HP and LP cylinders) and if ΔpL is the pressure drop between the receiver and the exhaust (cylinder back pressure), then:

\frac { { d }_{ H }^{ 2 }.\pi .\Delta { p }_{ H } }{ 4 } =\sqrt { \frac { \mu .W.\pi .D }{ 0.85.4.s.2 } }

from which [for a 2-cylinder compound]

{ d }_{ H }=\sqrt { \frac { \mu .W.D }{ 0.85.s.2.\Delta { p }_{ H } } }


{ d }_{ L }=\sqrt { \frac { \mu .W.D }{ 0.85.s.2.\Delta { p }_{ L } } }

For a 3-cylinder compound the two low pressure cylinders’ diameter is deterimined from:

{ d }_{ L }=\sqrt { \frac { \mu .W.D }{ 0.85.s.4.\Delta { p }_{ L } } }


  • Three Cylinder Compounds: Porta makes the following recommendations for 3 cylinder compounds:
    • Higher receiver pressure than for 2 and 4 cylinder compounds (as discussed above);
    • HP cylinder offset to the left-hand side to give a more straightforward front-end arrangement;
    • All cylinders of roughly equal volume but with the LP cylinders being a little larger than the HP cylinder (perhaps to account for pressure drop in the receiver);
    • All three cylinders driving a single axle, with inclined centre cylinder.  Porta claims that he has shown it to be possible to arrange a crankshalf with with sufficient strength to support 6000 h.p. or more (4400kW), the stresses being equal to those allowed by AAR for straight axles.

    Porta notes that the design of his 4000 h.p. 2-10-0 for the metre gauge Belgrano Railway in Argentina, the crank angles turned out to be 120 degrees, but that this may not always be the optimum setting.  Note: whilst it would have delivered an uneven exhaust beat, Porta considered that it would not have affected the performance of the boiler because of the use of GPCS.

  • Valve Gear and Cut-off: Porta does not define any firm recommendations for the arrangement of cut-off for the HP and LP cylinders, saying only that cut-offs should be adjusted first by calculation and then verified in operation to equalize the work done in the HP and LP cylinders.  He suggests that if the HP and LP cut-offs are tabulated, their relationship will be approximately linear allowing the adoption of a fixed setting of the valve gear to maintain that relationship.

…… text still being written as at 28th Jun 2011 ….


Several pages of this website include text and diagrams copied from Porta’s “compounding” paper, including the pages covering condensation/wall effects, steam leakage, clearance volume, incomplete expansion and triangular losses.  A more specific reference to his theories on compound expansion can be found on the α Coefficient page.

Sincere thanks to Adam Harris of Camden Miniature Steam, publishers of “Advanced Steam Locomotive Development – Three Technical Papers” for allowing the sections of the book to be published on this website.

For a historical perspective on compound locomotives in the UK, an interesting article on the subject published in the July 1992 issue of Steam Classic magazine can be downloaded here.


Combustion & Exhaust

Page Under Development

This page is still “under development”. Please contact Chris Newman at if you would like to help by contributing text to this or any other page.’




The application of GPCS or “Gas Producer Combustion System” to locomotive fireboxes is explained in great detail in pages 78 to 92 of Dave Wardale’s book “The Red Devil and Other Tales from the Age of Steam“.  Basically it is a 19th century technology involving the blowing of oxygen over heated coal to make “Producer Gas” or a mixture of carbon monoxide and nitrogen through the exothermic chemical reaction:

2C + O2 + N2 → 2CO+N2.

Producer gas was sometimes used for domestic heating, powering motor vehicles and power generation.  It is reportedly still widely used in industry because it can be made with cheap fuel (coal, wood or biomass) – see

L.D. Porta was the first engineer to apply the principles of producer gas to a locomotive firebox, first demonstrating its advantages on the locomotives of the Rio Turbio railway in Argentina in the early 1960s.

The advantage that the GPCS offers is that it allows the quantity of “primary” air passing through the grate to be significantly reduced perhaps by as much as 70%, thereby reducing the upward air velocity through the firebed that causes lifting of unburned coal particles off the fire surface and discharged through the chimney.  The reduction in unburned fuel loss that can be achieved – and consequent increase in thermal efficiency – is significant especially at high steaming rates that approach the grate limit.

Greater quantities of “secondary” air have to be fed into the firebox above the firebed to burn off the carbon monoxide (CO) that is produced.  A significant proportion of this can be fed in through the firedoor (which must be left continuously open), but additional quantities are usually required to be fed in through air inlet tubes penetrating the sides and crown of the firebox. The position of each of these secondary air inlet tubes should be selected to:

  1. Deliver secondary air as uniformly as possible over the over-fire combustion area – i.e. with some of the air penetrating to the centre of the firebed;
  2. To create mix turbulently with the producer gas (not with other airstreams entering the firebox);
  3. As far as possible to separate any rising coal particles from the gas stream and to throw them onto the firebed;
  4. Direct the airstream high enough to prevent it skimming coal particles off the firebed or to interfere with the firing process.

(See page 90 of Wardale’s book for more specific details).

Another important requirement for effective gas production is that the firebed be deep.  Ideally its depth should be at least 15 coal particle diameters in order to ensure that the maximum amount of carbon monoxide is produced from the reaction between the carbon in the coal and the available oxygen in the primary air.

However, an increased firebed depth combined with the exothermic (heat producing) reaction results in higher firebed temperatures and therefore an increase in clinker formation from ash fusion.  This can be countered by the introduction of “clinker control steam” into the ashpan which has the effect of reducing the firebed temperature as it passes through the coal particles because it undergoes and endothermic (heat absorbing) reactions in the form of:

C + H2O → H2 + CO

C + 2H2O → 2H2 + CO2

Both reaction result in hydrogen gas which gets burned off with the secondary air, and the first reaction produces additional producer gas (CO) which also gets burned off above the firebed.

Based on the above, the main features of a GPCS firebox are:

  • Perforated grate with pin-hole perforations giving a free-air area of between 5% and 10%;
  • Lowered grate to maximise the depth of the firebed;
  • Secondary air holes through the sides, top, front and/or rear of the firebox to allow “secondary” air to enter above the fire;
  • Pipework to deliver small quantities of exhaust steam into the ashpan.

Wardale converted the fireboxes of both his SAR developments (Class 19D No 2644 and Class 26 No 3450) to GPCS operation.  He achieved considerable (though variable) success with both conversions, as recounted in detail in his book, noting that the cost of converting a firebox to GPCS may not be warranted on a locomotive that is not going to be steamed anywhere near its grate limit.


The concept of the Kordina named after its inventor Zsigmond Kordina (see below), is described on page 153 of Wardale’s book “The Red Devil and Other Tales from the Age of Steam”, its function being to:

“expand the release steam to low pressure and high velocity at the point where the exhaust flows from the two cylinders join, preventing backflow of release steam into the opposite cylinder which occurred whenever the flow are of the exhaust passage junction was much larger than the blast nozzle tip. This was normally the case with First Generation Steam locomotives and showed as a jump on the back pressure line of the indicator diagrams at approximately mid-stroke position, the most unfavourable point.”

Wardale adds the interesting observation that:

“Ideally, some of the great amount of energy in the release steam could be used to create an ejector pump action on the exhaust from the opposite cylinder, lowering the back pressure in this cylinder, even to below atmospheric pressure.”

In the case of Wardale’s Red Devil, he describes its exhaust manifold being formed such that

“the exhaust passages from each side first branched into those for the front and rear chimneys respectively, and at the very top, the dividing walls separating the exhaust steam flows from the right and left cylinders ended. At this point the combined flow section was equal to the total exit section at the blast nozzles and this was the essence of the Kordina.”

Plate 16 from Wardale’s book is copied below, showing the arrangement of the Red Devil’s Kordinas:

In the case of the 5AT, in FDC 12 line 301 Wardale describes the Kordina as “… essentially the reduction of the total flow area of the exhaust steam passages from right and left cylinders to that of the total blast nozzle tip area at the point where these passages combine below the blast pipe”, and he illustrates this in Fig 12.5 (copied below), showing Kordina to form the upper end of the exhaust manifold in such a way that the (circular) passage from one cylinder fits inside the passage from the other cylinder, forming an annulus around the first, each passage having exactly the same cross-sectional area, each being the same as the total blast nozzle tip area.

Richard Coleby has incorporated this detail into his 3D “Solidworks” drawing of the 5AT cylinder block as illustrated below:


Zsigmond Kordina

In his book, Wardale states that Zsigmond Kordina was a Russian-born engineer, however he must have spent some of his life working in Hungary since his name is mentioned in a paper titled “The Role of Engineers and Milestones of Industrial Development in Hungary” as follows:

MVAG started locomotive manufacturing in 1873. The 1000th locomotive displayed at the Millennium Exhibition in 1896 is a proof of boost. With the leadership of Zsigmond Kordina, an internationally recognised locomotive designer team was working in MVAG at that time. The locomotives manufactured here could run at the speed of 100 km/h in 1900. Recognition of locomotive manufacturing was well demonstrated at the Paris World Exhibition in 1900: MVAGs twin-cylinder express locomotive was awarded a Grand Prix.

In his book “Compound Locomotives” [pp 37-38; Atlantic Publishers, UK, 1994; ISBN 090689961 3], John van Riemsdyk mentions Kordina as being the original designer of a numerous and long-lived fleet of tandem-compound 4-4-0 locomotives built for the Hungarian State Railways between 1890 and 1904.




Page Under Development

This page is still “under development”. Please contact the if you would like to help by contributing text to this or any other page.’

As stated on the Exhausts page under the “Terms and Definitions” section of this website, Wardale defines the exhaust system as “thermodynamically the heart of a locomotive which must therefore be as good as possible within practical limitations”.

The development of locomotive exhausts took a leap forward in 1926 when André Chapelon developed his Kylchap exhaust which incorporated a Kyala spreader (second stage nozzle) and third stage cowl between the blastpipe (first stage nozzle) and chimney. The name “Kylchap” recognizes the contributions of Chapelon and the Finnish engineer Kyösti Kylälä who developed the Kyala spreader.  The Kylchap exhaust is perhaps best known for its association with the LNER A4 Pacific “Mallard” which broke the world speed record for steam traction in 1938 with the assistance of a double Kylchap exhaust.

Ing. L.D. Porta subsequently improved on the Kylchap exhaust by developing the “Kylpor” exhaust which he successfully demonstrated on the 2-10-2 Santa Fe locomotives of the Rio Turbio Railway – see Hugh Odom’s Ultimate Steam Page for details.

Porta subsequently developed the Lempor exhaust which he claimed to be superior to the Kylpor.  The theories behind both systems are complex and beyond the scope of this website to explain.  However Porta wrote a paper in 1974 which provides the mathematical foundation of the Lempor, titled “THEORY OF THE LEMPOR EJECTOR AS APPLIED TO PRODUCE DRAUGHT IN STEAM LOCOMOTIVES“.

This paper can also be found on Martyn Bane’s website from where the above copy was sourced.

The paper is also available through Hugh Odom’s Ultimate Steam Page which also offers a link to a spreadsheet for designing a Lempor exhaust system (see

Hugh Odom’s page also includes a list of corrections to Porta’s Lempor Theory paper as provided by Dr. Jos Koopmans in January 2006.  It should be noted however, that David Wardale disputed some of Koopmans’ claims, expressing his views at the Modern Steam conference held at York in December 2006.  At Wardale’s request, his documented views have been posted onto this website where they may be found in the FAQ section, together with associated correspondence between himself and Koopmans.

Note: Jos Koopmans is the author of a book titled “The Fire Burns Much Better … ” (see also



Mechanical Enhancements

It may seem counterintuitive that locomotives with smaller driving wheels should achieve higher efficiency than those with larger driving wheels, but it is a fact that high engine rotation rates offer a number of technalogical benefits.   It may also seem counterintuitive that long strokes (such as used on GWR 2-cylinder locos) also offer advantages over short stroke engines.

High Rotation Rates:  In many of his technical papers, Livio Dante Porta stressed the importance of long strokes and high rotation rates, and in the latter case recommend the adoption of a maximum rotation rate defined in AAR standards of 504 rpm (= 8.4 Hz, or revs per second), when determining the size of a locomotive’s driving wheels.

The advantages of high rotation rates can be summarized under three headings:

  • Leakage:  Where not properly controlled, steam leakage is one of the principle causes of efficiency loss in steam locomotives.  Leakage past piston and valve rings is hard to detect or to measure, but it tends to be a constant amount regardless of engine speed.  Hence its importance increases in slow moving machines such as those used in shunting­, and its importance decreases as engine speed increases.
  • Wall Effects:  At high engine speeds, steam has less time to lose heat through contact with the cylinder walls, hence a reduced efficiency loss. In his Compounding paper, Porta claims that the temperature drop (where there is no condensation) is roughly inversely proportional to the rotational velocity to the power of -0.3 (i.e. ω-0.3).
  • Lighter Hardware:  Small wheels imply lighter hardware, hence higher power-to-weight ratio

The 5AT’s 1880 mm (6′ 2″) driving wheels were ideal in terms of meeting the AAR requirement when operating at its maximum continuous operating speed of 180 km/h (113 mph).

Note: It is much more difficult to design high wheel rotation rates into a heavy freight locomotive given the fact that even if fitted wth very small driving wheels of 1.22 (4′ 0″) diameter, the locomotive would need to operate at 115 km/h (72 mph) to achieve a rotation rate of 504 rpm.  Big end clearance restrictions make it impossible to design wheels that are small enough to achieve this rotation rate at “normal” freight train speed.

Long Piston Stroke: Advantages of long stroke, small diameter cylinders include lower clearance volume and lower heat transfer losses (i.e. lower wall effects) – see Questions about the 5AT Technical Specification.  In his Compounding paper, Porta implies that long stroke, small diameter cylinders help to reduce steam leakage, however he does not specifically state this, and (given the nature of leakage which in Porta’s designs is limited to the ring gaps, it is hard to see that cylinder diameter would have any significant effect on leakage.




The Oxford Dictionary defines tribology as the branch of science and technology concerned with surfaces in relative motion, as in bearings. It is therefore inseparably associated with the subject of lubrication.

Ing. L.D. Porta was the first engineer to apply the science of tribology to the design of steam locomotives, seeing it as an essential function in the application of the high temperature steam that is needed to deliver improved cylinder performance.

Porta studied the topic extensively, researching papers and records relating to the application of tribology in the development of the modern Internal Combustion (IC) engine. Porta recorded his findings and his theories in two papers, the first written in 1975 and the second, an updated version of the first, in 1995.

Porta introduces the second of these papers by pointing out that steam locomotives traditionally achieved only around 1000 hours’ life for piston rings whereas large marine diesel engines regularly achieve 20,000 hours despite working at pressure five times higher than steam, and at temperatures of 2000oC (as compared to ~400oC). His paper attempts to address the differences between the two technologies and puts forward recommendations for improvements to steam locomotive piston and valves, many of which have been applied and field-proven by David Wardale in his iconic Class 26 rebuild, No 3450 “The Red Devil”.

Porta’s paper focuses mainly on locomotive piston valve rings since these are subjected to the most demanding service, both in terms of high temperature and of low velocity in that most of their movement is too slow to generate the hydrodynamic conditions on which good lubrication depends. The paper also makes recommendations for piston rings for which conditions are less demanding, temperatures being significantly lower and rubbing speeds higher.

His tribology papers include some very instructive diagrams. A particularly useful one shows the variation in temperature over the surfaces of valve and cylinder liners as measured on his experimental locomotive No 1802. The diagram is reproduced below, illustrating the very high temperatures experienced by the liner, valve heads and valve rings on the admission side of the port:

Porta offers several principles for improving both the life and seal of valve and piston rings, all of which have proven successful on his own locomotive modifications and on others’ (notably Wardale’s in South Africa).

1: Minimizing the temperature of the oil film:

  • Do not mix atomized oil with live steam. Inject oil in liquid form directly onto rubbing surfaces.
  • Arrange the piston valve heads so that some of the exhaust rings rub over the admission edge of the liner port to facilitate transportation of heat from the hot (admission) side to the cold (exhaust) side of the liner;
  • Use light-weight (small section) rings of diesel engine quality, to maximize the conduction and transport of heat as described above;
  • In the case of valve heads, ensure that oil is injected “between the rings” (the coolest zone) from where it can spread over the liner surface;
  • Use relatively heavy bridge bars to limit temperature rise from contact with live steam and to better transport heat away from the hottest valve rings. [Note: A wide bridge bar is required at the bottom of each liner to support the valve head and ring joints.]
  • Cool valve liners (locally) using saturated steam –e.g. by passing it through grooves formed in the outer faced of the liner – see Porta’s A1 valve diagram on the Valve Design page and the following images taken from Wardale’s book and the 5AT FDCs.
  • Keep live steam in the steamchest separated from the outer surfaces of the main cylinders.

2: Maximizing the seal between rings and liner:

  • Fit as many rings as possible to each piston and valve head;
  • Use light-weight (small section) rings of diesel engine quality to better conform to the shape of the liner.
  • Minimise the gap between the (valve or piston) crown and liner and minimise the ring gaps.
  • Do not use tail rods on valve pistons. Fit ring retainers to keep ring gaps aligned along the bottom of the valve head – i.e. at the point of contact with the liner – to prevent steam leakage through the gaps.
  • Use articulated valve spindles so that both valve heads rest evenly on the liner, contact being between the bottom of the valve head and the liner to ensure zero steam leakages through the ring gaps which are aligned along this point of contact.
  • In the case of pistons, tail rods should be fitted to prevent direct contact between piston and liner. Where no tail rod is fitted, piston crowns should have bronze facings to minimize wear. Rings should be left free to rotate on the piston crown.

3: Maximizing lubrication:

  • Provide an sufficient allocation of oil to ensure good lubrication: German State Railways’ allocation of 1.4 g/km for both valves and cylinders should be taken as the absolute minimum; Porta’s C16 1802 with an allocation of 11 g/km is a safer quantity.
  • Small reductions in surface temperature deliver large gains in lubricant life and performance;
  • Use a mixture of bronze and cast-iron rings, bronze rings providing a polishing action on the liner surface;
  • Weld bronze facings to valve crowns that rub directly on the bottom of their liners; also to piston crowns where tail rods are not fitted.
  • Use high-viscosity paraffinic oils, bearing in mind that there is a trade-off in that high molecular weight oils decompose more rapidly that lighter oils.
  • Use additives such as neatsfoot to aid emulsification of condensate, and colloidal graphite (in pumped lubrication systems) to enhance high temperature lubrication and to fill scratches in liner surfaces.
  • Drive oil pump(s) from top of the combination lever to make the rate of oil delivery proportional to cut-off.
  • Provide means to (manually) increase oil delivery whenever priming occurs.

4: Minimize Abrasion

  • Avoid use of oils that carbonize rapidly at high temperatures.
  • Avoid situations were oils might carbonise from being in contact with very hot surfaces for long periods.
  • Prevent the ingress of abrasive material either from steam contaminants (through the use of high-performance antifoams, or from smokebox ashes being drawn down though the blast nozzle (through the technique of drifting in mid-gear with the throttle “cracked open“.

5: Miscellaneous

  • Avoid gum formation through oxidation as a result of oil contact with air. For instance, avoid the use of snifting valves or other means of allowing air to enter the steamchest or cylinders.




Various other pages of this website are devoted to specific aspects of piston valve design including:

This page serves to summarize the overall principles covering valve design as defined by Porta including linked references to the above pages. It also includes the following sections:

Note: All discussions about valves on this website relate to piston valves which were favoured over other types by both Porta and Wardale. A separate series of pages are devoted to an extensive discussion by Wardale on the relative merits of Piston Valves vs. Poppet Valves.

Porta’s Principles relating to Valve Design

Porta’s principles relating to valve design focus on two disparate criteria:

  1. Limiting energy (pressure) losses as far as possible;
  2. Minimising friction and wear.

In Wardale’s design of the 5AT, a third criterion was added – viz: keeping the masses of the reciprocating parts (valves, rings, valve rod etc) as small as possible so as to minimise interial forces.

Taking each of the first two criteria in turn:

Limiting energy (pressure) losses – This can be achieved by:

  • Using the maximum possible diameter of valve to give the longest possible port area. In the case of the 5AT, Wardale achieved this requirement through the use of twin Φ150mm valves driven through a rocking shaft which provided the same port area as a single Φ350mm valve. The twin valves served to reduce substantially both the reciprocating masses and the clearance volume which would have been excessive had a Φ350mm valve been mounted above a Φ450mm cylinder.
  • Reducing the clearance volume as far as possible (without excessive restriction of the steam passages).
  • Streamlining the valve heads and ports to minimise wire-drawing pressure drops and triangular losses – see the Internal streamlining page;
  • Optimising valve gear geometry, valve travel and valve events – including lap and lead – to best suit the intended duties of the locomotive.
  • Minimise steam leakage, especially past valve and piston rings by using multiple narrow rings with small ring gaps, and aligning ring gaps along bottom of valve heads.

Minimising Friction and Wear – This can be achieved by:

  • Applying lubricants directly to the rubbing surfaces rather than mixing it with steam being fed into the steamchest. In the case of piston valves, it is important that lubricant be applied “between the rings” – see Lubrication page;
  • Limiting the temperatures of rubbing surfaces by
    • use of heavy bridge bars to more rapidly conduct heat from the admission side of the liner to the exhaust side;
    • use of multiple narrow rings to more efficiently transport heat from the hot (admission) side of the liner to the (cooler) exhaust side;
    • use of a diffuser mounted outside the valve head on the exhaust end to direct cooling exhaust steam over the hotter surfaces at the admission end of the valve head;
    • where very high temperature superheated steam is used, provided saturated steam cooling of the rubbing surface on the admission side of the valve liner.
    • use of bronze rings and crown facings to polish liner surfaces and reduce wear.

Porta incorporated most or all of these principles into the diagram that he sketched for inclusion in his proposal for the A1 Tornado, as below. The sketch is probably unique in that it shows how a valve may be enlarged without altering the outside dimension (or appearance) of the valve casing.

Fig 34: Fabricated Cylinder End
as proposed by Porta for A1 Tornado

The legend from Porta’s paper explaining each itemised component, is as follows:

“The outline adheres to the original one [for the A1]. The liner (1) is pressed by the cover (2) secured by SAE 4340 studs (3). Dimension ‘a’ is kept to a minimum, thus for a given ‘b’, the valve diameter can be maximized and the clearance volume made as small as possible.

(4) is a diffuser also performing a cooling action for the (valve) crown (5). (6) is a sealed insulation; (7) saturated steam liner cooling grooves;

(8) and (9) are hollow, welded-on BRIMs filling the clearance spaces. The space (10) is filled with glass wool mattress. (11) is a dummy protrusion for appearance.

The valve liner is 30% larger than the original; valve lap is 55mm (against 38), and the flow coefficient is 0.96 instead of ~0.60, all making ~2.5 times greater flow area at short cut-offs.

Hooks (12) are a trick to keep unchanged the valve chest cover diameter. The width of the steam passage (13) has been increased from 50.8 to ~60mm, thus resulting in a ~300cm2 area. Were it not for respecting the outermost dimension of the cylinder, it would have been desirable to increase it to 400cm2 so as to have 1:4.6 ratio with piston area: this affects the performance at very high speeds (see Section 20.2).

Clearance volume is estimated as 8% of cylinder volume, a rather small figure due to brims (8) and (9).

Liner(s) (7) are of the through type from end to end (in three parts) thus avoiding “torturing” valve rings at the conical entrances. Radiuses (14) through (17) are important to have not a “vena contracta”. (18) is the position of the platform above the outer cylinder.

This basic design was first applied by the writer on prototype engine No 3477, FCGR, on Rio Turbio engines, on prototype No 4674 FCGB, and by Wardale in prototypes 2644 (19D) and 3450 (Class 26), SAR.”

Note: The trapezoidal ports are not the type that Chapelon recommended in which the ‘vertical’ faces are sloped in order to graduate the ‘release’ of exhaust steam so as to reduce firebed disturbance (see page 97 of Wardale’s book). In this case, it is the bridge bars that are inclined in order to spread the area of contact with the valve rings thereby reducing their wear rate.

Multiple Narrow Piston and Valve Rings

Porta recommended the use of multiple narrow valve and piston rings, the word “multiple” meaning “as many as possible within the length of the valve of piston crown”. In his Fundamental Design Calculations for the 5AT, Wardale specified multiple narrow rings for the purpose of “minimising the sealing duty of any individual ring, reducing ring-liner friction and wear, and enabling steam tightness to be maintained for extended periods of service without inspections or maintenance”.

Porta recommeded the use of multiple narrow rings in several of his papers, describing their advantages in the following generalized terms:

  1. Narrow rings are more flexible and therefore provide better contact with the liner when thermal and other stresses cause distortions to the ring shape.
  2. Narrow rings can more readily pick up heat from the hot rubbing surface of the liner (on the admission side of the port) and transport it to the cooler exhaust side of the port, thereby assisting in the process of cooling the liner (and improving lubrication).
  3. Use of narrow rings allows more rings to be fitted on any given length of crown. If sufficient rings can be fitted to ensure steam-tightness, there may be space to add some bronze rings to help polish the rubbing surface of the liner and thus improve lubrication.
  4. Lower mass and therefore lower reciprocating inertial forces – low enough that steam pressure can prevent the outer ring on the admission side from touching a thin (streamlined) edge land.
  5. Ring-mounting stresses and distortions in narrow rings are less than in wide ones.
  6. The steam pressure differential across the ring set is distributed between the rings, thus the use of as many rings as possible reduces the steam pressure on each ring and thus leakage past each.

The illustration below (Plate No 31 from Wardale’s book) shows one of the valve heads from the Red Devil on which 12 rings were mounted. The left side is the admission edge with a very narrow streamlined land as per Porta’s recommendations. The exhaust edge land is much more robust because it has to carry inertial loads from the outer ring.

Wardale’s description of the image is as follows:

“Close-up view of a valve head showing the thin yet adequately robust admission edge land which allowed ring-control of steam admission and cut-off whilst giving a good steam flow coefficient at all port openings Also visible are the set screws for holding the ring-cuts at the botom of the valve when the assembly was mounted in the locomotive.”

Valve Liner Cooling using Saturated Steam

Where very high superheat temperatures are to be used, Porta recommended the use of saturated steam to cool the rubbing surfaces of valve liners. He illustrated the concept in the diagram (above) that he prepared for the A1 Project.

Wardale adopted the idea for his Red Devil and also for the 5AT. Below are two illustrations from Wardale’s book showing the grooves machined into the outer faces of one of the valve liners for the Red Devil. Below them is the sketch that Wardale prepared to illustrate the liner cooling system for the 5AT.

Plate 28. 26 Class No 3450: cooled through-type pearlitic cast iron valve liner

Plate 29. Detail of the valve liner cooling passages.

Diagram of Valve Liner cooling system
from 5AT Fundamental Design Calculations

Post Script: In his Fundamental Design Calculations for the 5AT, Wardale devoted 180 lines of calculation to the design of the cooled valve liners. He concluded the calculations with the following note:

“The principle of cooling by saturated steam is applied in these calculations to the valve liners, which are the hottest of the various engine rubbing surfaces and therefore in the greatest need of cooling. However the same principle can be applied to any other rubbing surfaces, such as the cylinder liners and piston rod / tail rod packings, should cooling of these items be shown by service experience to be necessary. Cooling using boiler feedwater under pressure, by piping some of the feed from the feedwater pump to cooling passages for the surfaces concerned then back to join the main feedwater flow upstream of the feedwater heater, is also a possibility (similar to water cooling of i.c. engines). This may be particularly suitable for piston rod & tail rod packings.”




This page covers briefly a number of topics related to the enhancement of valve events to improve a locomotive’s performance. These come under the following headings:

A separate page covering valves and valve gear can be found in the Technical Terms section of this website.

Valve Events

Valve Events have been defined elsewhere as the four defining points in the cylinder power cycle – viz: (1) Admission; (2) Cut-off; (3) Release and (4) Exhaust Closure. These four points, in turn, define the intervening periods of Admission, Expansion, Release and Compression as shown on the idealised Indicator Diagram (below).

The same page also defines the terms Lap and Lead and describes the purpose of each.

The design of valves and the timing and nature of their associated “events” have a strong influence on a locomotive’s performance through their association with triangular losses and incomplete expansion losses in the cylinders. Thus any modifications to valve and/or their events should be aimed at improving the steam flow through the valve, most especially by maximizing the area available for steam flow is vital, as is streamlining of the path that the steam is to take.

Specific examples of improvements that can be adopted (some with relative ease) include:

Associated changes that will help to improve valve performance include:

  • Increasing the size of the steamchest so as to minimise the steamchest pressure loss during admission.
  • Improving the exhaust system to minimise back-pressure (remember that when back pressure is lower, the steam’s specific volume is higher, so a larger volume of steam has to pass through the valve)
  • Optimising clearance volume.

Most of the above are to some extent interlinked – for instance increasing the steam lap usually necessitates increasing valve travel; and streamlining valve heads and rings usually involves making the events ring-controlled.

Any alterations to valve events should take into account the nature of the work that the locomotive is intended to undertake. For instance

  • Where a locomotive is to be used for high speed operation, then increasing both lead and steam lap is likely to be beneficial. (Note: Increasing lap is likely to require changes to the valve gear geometry to increase valve travel.)
  • Where a locomotive is to be used for shunting or heavy haulage, then a shorter lead may be appropriate.
  • Where a locomotive is required to deliver high power outputs at high wheel rotation rates, then variable lead may be advantageous.
  • Where a locomotive is to be operated mainly in one direction, then the valve gear geometry might be arranged to optimise the indicator diagram when running in that direction.
  • Where a locomotive (e.g. a tank engine) is to be operated in both directions, then the valve gear geometry should be optimised so that it produces similar indicator diagrams in both directions.
  • If the exhaust system is producing so much draught that it diisturbs the firebed, it may be advantageous to increase the exhaust lap to reduce the cylinder pressure at release. (Note: increasing exhaust lap will also advance the point of Closure and thus increase steam compression.)


Optimising Lead

Porta’s views on Lead

The subject of Lead is discussed in depth by L.D. Porta in his paper titled “Notes on the Optimum Value of Lead in Steam Locomotives” in which he summarizes the optimum value of lead as “that which gives the least inlet pressure drop”. In generally this occurs when the compression line of the Indicator Diagram reaching the steamchest pressure when the piston reaches “dead centre”. He illustrates this with the diagram below:

Fixed Lead

Walschaerts valve gear is commonly described as having a fixed lead. In fact it is fixed only in linear terms, in that the linear position of the valve remains fixed regardless of cut-off when the piston is at dead centre. However when looked at in relation to the crank rotation angle, it will be discovered that as cut-off increases, the lead decreases in angular terms as may be seen in the diagram below (adapted from Fig 46 of The Red Devil). In other words, at any given rotation speed, the time available for steam to enter the cylinder in advance of the piston reaching its dead-centre position, reduces as cut-off increases. This is desirable for starting purposes, but may not be optimal for hauling heavy loads at high speed.

Variable Lead

In the same paper, Porta mentions the idea of introducing variable lead to Walschaerts valve gear by shorting the eccentric crank. Wardale refers to this in FDC 5 line 34, saying: “Variable lead can be applied to Walchaerts valve gear by (i) shortening the eccentric crank length or (ii) slotting the combination lever top and suspending the front of the radius rod from a link held in one arm of a crank, and by raising or lowering it varying the effective length of the combination lever dimension between the radius rod and valve spindle joints.”

Introducing variable lead using Porta’s shortened eccentric crank has the effect of increasing “angular lead” at long cut-offs thereby reducing a locomotive’s starting capability but increasing its ability to haul heavy loads at high speed – i.e. produce high power at high speed.

According to Wardale, the slotted combination lever and suspended radius rod concept was adopted by the Denver and Rio Grande Western Railroad on their L-105 4-6-6-4 Mallet locomotives. He describes its purpose as follows: “By having the crank cam-operated and linking the cam rotation to cut-off, any reasonable desired variation of lead with cut-off (such as zero lead in full gear, then high lead in the long to mid cut-off range, decreasing for shorter cut-offs) could be arranged. This system is potentially advantageous, but it would be more so on a freight locomotive required to make difficult starts and habitually working in long cut-off”.


Ring-Control of Valve Events

In most locomotives, the opening and closing of the ports to admit steam to the cylinder and to exhaust steam from the cylinder, occurs when the edge land (the rib that holds the valve ring in place) passes over the edge of the port. This is termed valve-control (or land-control) of the valve events.

One of Porta’s many innovations was to introduce the concept of ring-control in association with the streamlining of the corners, edges and surfaces that the steam has to negotiate its way around as it passes from the steamchest and into the cylinder, and on its way back out to the exhaust system. The concept is illustrated below, in this case comparing control of admission steam with the (traditional) edge land and with (Porta’s) ring control.

Two important observations may be made from these diagrams:

  1. Ring control provides a much more precise timing for the event. In the case of land control, steam will begin to find its way into the cylinder before the ring passes over the edge of the port, leading to wire-drawing and “triangular” loss. With ring control, admission to steam is almost instantaneous and the steam flow much more streamlined (hence much lower triangular losses).
  2. The edge land on the admission end of the valve head can be very thin because the steam pressure serves to keep the ring away from the land thereby reducing (or eliminating) intertial ring pressure on the land. Thicker lands are required at the exhaust ends of the valves (as shown on the similar diagrams on the Streamlining page).
  3. The right-hand diagram illustrates Porta’s recommendation that the bottom of the valve head rubs on the liner. See Piston Valve Design page for further explanation.

The above diagrams illustrate the principles of streamlining which are covered in more detail on a separate page.


The multiple element Piston Rod Packings fitted to The Red Devil are described on page 176 of his book as follows:

Fully floating five-element metallic piston rod packings replaced the single element Paxton-Mitchell type which did not guarantee steam tightness over long periods without attention. Some time previously Porta had sent me a drawing made in 1975 of a piston rod packing to his design for the Indian Railways WG class 2-8-2. ….. The principle was to put as many sealing elements in series as possible, the sealing function being shared by all elements, it being the same principle that dictated the maximum number of piston and valve rings. The lower load taken by each element reduced its wear rate correspondingly and enabled steam tightness to be maintained over much longer periods than with simple packings.

The proposed WG class packing was used as the basis for the new design [was] produced both the piston rod and valve spindle packing drawings. A split packing housing derived from the British ‘Britimp’ packing was used. This ele­gant arrangement minimized the number of steam-tight faces required and made assembly and removal of the packings extremely simple. The split housing halves were located by dowels and tightened together by Allen screws, and could be fitted in place of the existing packings without any alterations to the back cylinder heads. Each packing element consisted of two rings made up of four segments located together by a peg and groove arrangement which held the joints in each pair of segments at 90° to the joints in the adjacent pair, thereby eliminating any leakage path past the joints. The segments were made from cast P66 alloy (a lead – copper – sulphur alloy), the same material as used for the Paxton-Mitchell packings. …. Plate 35 (below) shows the packing rings and one split housing half in position on a piston rod during assembly.



A locomotive must be designed so that its tractive effort is not so high that it “loses its feet” on starting.  Equally importantly, its tractive effort should be limited so that it does not slip at speed. This latter becomes a much more important consideration for Second Generation Steam locomotives like the 5AT which maintain a high tractive effort at high speed (see Tractive Effort page and T.E./Adhesion diagram below).

In the past, locomotives were commonly designed to operate with a minimum frictional coefficient of 0.20 (or 20%) on starting.  Through careful specification of its sanding system, Wardale was able to increase this to 25% in the case of the 5AT.

In the Fundamental Design Calculations for the 5AT, Wardale took pains to evaluate the available  adhesion not only under average tractive forces, but over the full range of tractive forces that apply between wheels and rail over each revolution of the driving wheels.  As explained on the Tractive Effort page of this website, traction forces vary by significant amounts over each wheel revolution, most particularly on 2-cylinder machines, but also on multi-cylinder locomotives. The calculated variation in tractive force delivered by the 5AT over a 360o driving wheel rotation at various speeds, including at the maximum speed of 200 km/h, should be compared to available adhesion, as illustrated below (taken from FDC 1.4):

In consequence of this tractive force variation “self-controlled” (i.e. momentary) slipping is likely occur, especially in conditions of low adhesion, four times during each revolution resulting in uneven wheel-rim wear, the greatest wear coinciding with these peak loads.

Furthermore as such wear develops, it is likely to promote slipping through the momentary reduction in reaction force between rail and wheel at the points of greatest wear, thereby exacerbating the rate of wear.  In consequence, Porta always contended that adhesion would be improved AND wheel-rim wear reduced by regular machining of the rims to maintain their circularity, preferrably by regular in-situ reprofiling.  Porta also recommended maintaining the outside rim diameter of the driven wheels some 4mm larger than the coupled wheels in order to even out their wear rate.  For these purposes, Porta set up a wheel rim profiling machine in the workshop at Rio Gallegos at the eastern end of the Rio Turbio Railway for routine in-situ maintenance of the driving wheel tyres of the railway’s Santa Fe locomotives.

For optimising adhesion L.D. Porta “High Adhesion Wheel Profile” for use on steam locomotives, further information can be found on Martyn Bane’s website at  This wheel profile incorporates a drip-groove around its outer edge, designed to prevent oil drips from the wheel hubs from contaminating the rail surface.

Please refer to the Adhesion page in the Steam Loco Definitions section for general information on adhesion and tyre wear.

Franklin Self-Adjusting Spring-Loaded Wedges

Self-adjusting spring-loaded wedges, supplied by the Franklin corporation, were designed to maintain zero clearance between axleboxes and horn guides, thereby preventing the shock loads or “knocks” that were once commonly heard from run-down locomotives as the clearances between axle boxes and hornguides were taken up with reversal of piston thrust.

In Clause 5.23 of the 5AT Specifications, Dave Wardale wrote as follows:

“The driving and coupled wheel axleboxes shall be provided with wedges as follows: driving axlebox; parallel guides at the rear and spring-loaded wedges at the front: leading coupled axlebox; fixed wedges at the rear and spring-loaded wedges at the front: trailing coupled axlebox; spring-loaded wedges at both front and rear (to allow for mainframe expansion between the driving and trailing coupled axles due to radiation from the firebox). The position of the fixed wedges shall accord with the exact leading coupling rod lengths between bearing centres, and shall be lockable. The spring loading of the wedges shall be of the Franklin lever type, mounted transversely on frame cross-stays, as necessary to clear the underhung springs.”

The 5AT axlebox and hornguide arrangement is illustrated below from a CAD arrangement drawing by Richard Coleby.



Franklin-type Engine-Tender Buffer Mechanism

A Franklin-type radial buffer provides a rigid connection between a locomotive and tender, allowing no relative movement in the fore-and-aft direction but allowing relative radial movement for negotiating curves.

The diagram of a Franklin radial buffer (below) illustrates the basic principle where two wedges are pressed towards each other by coil springs, forcing the radial buffers (one on the engine and the other on the tender) to maintain contact with one-another through their respective buffing plates.

More detailed information, including a marked-up version of this diagram and photographs of Franklin radial buffers, can be found on Hugh Odom’s Ultimate Steam website at

David Wardale specified a Franklin-type radial buffer between the engine and tender of the 5AT in order to use the inertia of the 80 tonne tender to minimize the fore-and-aft “rocking” sensation often experienced with two-cylinder locomotives, this rocking motion being caused by unbalanced inertial forces from the reciprocating masses – viz: pistons, piston rods, crossheads and connecting rod ends.

The amount and position of balance weights on a locomotive’s driving wheels are often adjusted to partially offset the horizontal inertial forces resulting in radial imbalance that produces hammer blow which can cause damage to track.

Wardales specification for the 5AT included the requirement that “the engine-tender drawgear shall be of the ‘solid’ (unsprung) type, with a spring-loaded friction-damped intermediate radial buffer (e.g. of the Franklin type), this arrangement giving an engine-tender coupling that allows the mass of the tender to act effectively to reduce fore-and-aft accelerations due to the unbalanced reciprocating masses”.



Other Improvements

Porta’s Water Treatment is an extension of the TIA system that was successfully used in France and the UK in the later days of steam.  Porta’s water treatment is claimed to:

  • stop all water-side corrosion;
  • prevent scale formation, even when using the hardest of water;
  • prevent deposition of sludges and thus the need for frequent boiler washouts;
  • reduce the need for blow-downs;
  • ensure purity of steam leaving the boiler.

Porta Treatment achieves these benefits from a combination of:

  • very high levels of dissolved solids in the range of 10,000 to 20,000 ppm (or even higher);
  • very high alkalinity of around 3,000 ppm (as CaCO3);
  • addition of tannins;
  • use of special antifoams to prevent priming.

Continuous use of Porta’s water treatment can extend boiler lifespan almost indefinitely and dramatically reduce boiler maintenance costs.

Porta wrote a lengthy (and not easily understood) paper on the subject titled “Steam Locomotive Boiler Water Treatment” which has so far not been published.

Wardale has described his use of Porta Treatment in his book The Red Devil and Other Tales from the Age of Steam, and his views on the subject are summarized in the notes of a meeting with Chris Newman in Beijing in Oct 2003.

The subject of Porta’s Water Treatment was the feature of an article by Martyn Bane and Shaun McMahonin Steam Railway Magazine No 335.

More comprehensive infomation is available from Martyn Bane’s website at




The temperature of feedwater entering the boiler should be as high as possible in order to maximize steam production and to minimize thermal stresses in the boiler structure

Feedwater heating can be achieved in many ways. The “steam injectors” commonly used on First Generation locomotives provides a degree of preheating, however they only work at relatively low water temperatures. Where high temperature feedwater heating is used, the water must be pumped through the heater. In the case of the Red Devil, Wardale provided a single feed pump but retained the original injector as an independent standy, the injector-fed water bypassing the feedwater heater.

Feedwater temperature should be as high as possible to maximise energy savings. Ideally it should be 100oC or above which necessitates the use of a “closed” or pressurized system. A typical closed system takes the form of a tube-and-shell heat exchanger in which feedwater passes through the tubes while exhaust steam is fed into the outer shell and condenses on the tube surfaces. The Red Devil’s heat exchanger was of this sort, being in the form of a ø550mm cylindrical casing with a tube plate at each end (one fixed, one floating) with six banks of 23 x ø20mm copper tubes through which feedwater passed in a sequence designed to maximise heat transferby exposing (as far as possible) the hottest water to the hottest steam and vice versa.

Two similar but smaller heat exchangers were proposed for 5AT each exchanger containing 6 banks of 15 x ø20mm copper tubes. Wardale’s sketch below illustrates the counterflow principle:

The “economizer” (described on a separate page) is another type of feedwater heater which may be used on its own or in conjunction with an exhaust steam heater. Porta combined both in his more advanced designs.

Advantages of Feedwater Heating:

Improved performance and fuel economy:

As Wardale says on page 156 of his book, “Normally a feedwater heater was viewed as a device for saving fuel and water, savings of 10-15% in both being generally claimed. However its importance, like that of having the best possible exhaust system, seems to have been rarely appreciated [by FGS designers]. The significance of both becomes fully apparent only when locomotive performance in terms of the power generated from a given mass of hardware was pushed to its limit”.

A calculation is presented under the heading of Specific Enthalpy which demonstrates how a 10% saving in energy requirement can be expected by preheating feedwater to 100oC, however Wardale shows how feedwater heating can produce a very much greater saving in fuel consumption when a locomotive is operating at or close to its grate limit. He illustrates the point in the diagram below:

Fig 41 - Effect of Feedwater Heating on Boiler Output

Water saving through Recyling of Condensed Steam

Wardale also points out that exhaust steam feedwater heaters deliver significant water savings through the recycling of condensd steam. In the case of The Red Devil, he noted that “it was estimated that some 13.5% of the cylinder exhaust steam would pass through the heater, equivalent to approximately 12.5% of the total evaporation, this latter figure representing the water saving due solely to the feedwater heater system. This was in itself a significant figure contributing to a greater operating range between water stops.”

Cautionary notes:

  • The application of feedwater heating to an existing design will result in reduced superheat temperature as mentioned on page 161 of “Wardale’s book“. This is because less heat needs to be generated in the firebox to boil preheated water, and therefore there will be less heat available for superheating the steam that is generated”.
  • From Wardale’s FDCs for the 5AT: “It is essential that the piping from the cylinder exhaust passages to the heaters allows [the required amount of steam to be delivered to the heat exchanger]. As the exhaust steam flow to the heaters is in parallel with that to the blast nozzles (and the combustion air preheater) this becomes more difficult as exhaust design is improved and the blast nozzle size increases. With a free exhaust, exhaust steam escapes more easily from the blast nozzles and correspondingly less will go to the feedwater heaters if the piping to them is restrictive. At the detail design stage analysis must be made to ensure that the flow resistance of the pipes concerned relative to that of the exhaust passages to the blast nozzle tips will allow [the required proportion of exhausted steam to be reached]. An approximate comparison of the flow resistances involved suggests that the exhaust steam pipes to the heaters, each of length ≈ 4~5 m, must be absolutely not less than 100 mm bore diameter and should preferably be larger. The take-off for the feedwater heater steam should be just downstream of the steam chest exhaust chambers: a suitable arrangement is to have a take-off in each exhaust passage just inside the frame portion of the cylinder block, the two pipes from each cylinder joining at a Kordina-type joint.  These pipes can be seen (coloured cyan) in the cylinder/steamchest drawing below:


An “economizer” is a supplementary water heater consisting of the front part of the boiler which is partitioned off from the rear section by a thin steel plate baffle, closely-fitted inside and around (but not rigidly attached to) the boiler barrel, tubes and flues. Thus, the pressure on both sides of the baffle is the same – the baffle simply serving to retain the colder incoming water to give it time to gain temperature before it “overflows” into the evaporative (rear) section of the boiler.

Keeping the colder water at the front of the boiler serves two purposes:

  • it maximises the temperature differentials across the tubes – i.e. between combustion gases and boiler water – allowing the cooler water in the front of the boiler to extract heat from the gases that have lost much of their energy evaporating the water at the back of the boiler.
  • segregating “the cold” from the hot water, minimizes thermal stresses within the boiler shell.

Economizers were not commonly used in the past, being introduced (invented?) by Chapelon in his experimental 2-12-0 freight locomotive 160A-1 designed in 1936.

The 5AT design incorporated an economizer section at the front end of its boiler, described in the introduction to FDC 9 “Feedwater Heating” as follows:

“A Chapelon-type economizer at the front of the boiler barrel.[2] This is a partition at the front of the barrel formed by an intermediate tubeplate and in free communication (by an overflow hole(s)) with the rest of the boiler. This free communication means zero pressure difference between the two and there may therefore be no need for pressure-tight joints where the boiler tubes pass through the intermediate tubeplate – a good fit of the tubes in the holes to stop gross flow of water between the two sections may suffice, e.g., for parallel tubes, to be achieved by drilling all tubeplates to the same template and aligning them up with some dummy tubes in place when welding to the barrel. The boiler clack valves are situated in the economizer section, all incoming feedwater entering the boiler at the front where the combustion gasses are at their lowest temperature, and therefore acting to give the highest temperature difference, and consequently highest heat transfer, in this part of the boiler, improving its absorption efficiency.”

[Note: the Franko-Crosti boiler used on some of the BR 9F 2-10-0s was a singularly unsuccessful form of Economizer.]


There have been many and varied practices adopted by drivers and/or imposed by railway authorities, for “drifting” of locomotives – i.e. running at speed without power such as when descending a bank. On page 101 of his book “The Red Devil and Other Tales from the Age of Steam“, Wardale emphasise the complexity of the problem by listing no less than 13 criteria to be satisfied by any applied method of drifting – viz:

  1. Minimum (ideally zero) cylinder steam consumption;
  2. Minimum (ideally zero) cylinder power;
  3. No net negative cylinder work;
  4. Cylinder pressure to remain above smokebox pressure to avoid sucking smokebox gases into the cylinders;
  5. Adequate cushioning of the reciprocating masses to be provided by compression at the end of the stroke;
  6. Require no snifting or by-pass valves of any type, since the former result in cooling of the cylinders and spoilt lubrication through oxidation of the oil, and both were prone to leakage;
  7. Require no separate drifing valve;
  8. Give smooth riding of the locomotive without damage to any of its parts and without longitudinal vibrations that could be transmitted to the trailing load;
  9. Require no special attention on the part of the driver, i.e. be such that the technique could be expected to be correctly implemented in normal service;
  10. Where piston valves with thin admission edge lands were fitted, maintain adequate steamchest  pressure (approximately 400 kPa) at high speeds to prevent the admission edge valve rings from touching these lands;
  11. Must not cool the cylinders;
  12. Give no excessive high cylinder temperature, thus maintaining good lubrication conditions;
  13. Give minimum oil consumption in the case of locomotives fitted with variable stroke drive for the mechanical lubricator.

In order to determine the optimum drifting method for his Class 19D conversion, Wardale took sample indicator diagrams when running using three methods of drifting:

  1. valves in mid-gear; throttle shut;
  2. valves in full-gear; throttle cracked open.
  3. valves in mid-gear; throttle cracked open.

In case 1, he detected a momentary vacuum in the cylinder; in case 2, the steam flow became far too high; case 3 was found to be satisfactory, and was therefore recommended as the operating regime to be adopted for both his 19D and Class 26 locomotives.

In his reply to a question about the high piston speed of the 5AT, Wardale included a supplementary  observation that “Drifting has to be with a limited amount of drifting steam which carries away friction-generated heat which would otherwise build up during high-speed drifting”.

See also “Snifting, Drifting and By-Pass Valves“.