Please note:
The following article on piston rings was copied from digital pictures (that I found on the internet) into an html format. Most of the drawings were re-drawn to make them more legible. Some were not legible enough to re-draw and have been included in their original format. I have tried to copy everything exactly as shown on the pictures. The photos that are in the following article were cut from the original digital pictures and included here.
It has been a great source of information for me and I hope that others can get some use from it also.
This article has been completly translated from photo to html as of Jan. 18, 2008.

AN ESSAY ON PISTON RINGS

T. D. Walshaw


When buiding the SIRIUS engine recently I was asked to make a second for a friend living abroad. At the time supplies were difficult and his castings came without rings. He obtained a set of four from another source, and I noticed that they were somewhat stiffer - 0.049" radial instead of 0.039" as on the Stuart rings. When fitted, the engine was almost impossible to turn by hand, but eased off a littleafter 30 minutes of motoring followed by an hour under air, at 1200 rpm or so. However, when the pistons were drawn it was clear that the wall pressure was much too high, for, though not measurable, there was clear evidence of both ring and bore wear. This led me to check the wall pressure both by calculation and physically and the results were so surprising that I checked every piston ring I could find in the shop.

Ring tests

It can be shown that if a perfect ring lies in a truly circular cylinder exerting a uniform wall pressure, then the bending moment at any given point is given by:



       (1) ....M = 0.5p x w x D2Sin/2  


           Where; M = Bending Moment, ins.lbf.
                  w = Axial width of ring, ins.
                  p = Wall pressure, lbf/sq.in.
                  D = Ring Dia. ins.
                  ø = Angle between the gap and
                      the stress plane.
                  t = radial thickness, ins.


      If t is constant then the maximum stress occurs 
      when ø=180°, opposite to the gap. (see Expt 5)

By a fortunate quirk of mathematics, if such a ring is closed to the same installation gap by means of a band wound round (Fig. 1a) the band tension imposes exactly the same bending moment, the force being defined by the relationship:
Fig 1, formulae and drawings

This offers a means of checking the mean wall pressure and is common practice for large piston rings. However, it does present problems and for smaller sizes the arrangement in Fig. 1b is used, where the ring is closed by a diametral force, until the ring is closed to the bore diameter, D. This force (Fdd) is given by:

    (3) ....p=Fdd
          __________

           1.135w.D

This test is quite easy to carry out on small rings, as shown in fig 2. The depth gauge has been set to the bore diameter, held in the chuck of the machine, and brought down until the tip just touches the anvil on the scales. The latter record the diametral force.

Figure 2 The wall pressure can be estimated by calculation (for rectangular sections) if the properties of the material are known, but "book values" are not reliable, as cast iron does not obey Hooke's Law, and Young's Modulus (EE) varies with the stress. The relationship is:


(4) ....p =      g x E       
            7.06D (D/t - 1)3


g = difference between the free and closed gaps, ins.
E = Young's Modulus, ibf/sq.in.
Other notation as above.

Values of E can vary between 12 x 106 to 24 x 106 lbf/sq.in. and reference to the supplier is advisable. A figure of 18 x 106 is typical for 20-22 ton UTS, and between 15 and 17 x 106 for the 17 ton UTS centrifugally cast iron normally available for home-made rings.

To complete this exploration of the arithmetic, the stress in the ring can be calculated in the usual way from the bending moment, writing M = f.z, where z = wt2/6, using previous notation. However, this can be combined with the bending moment expression and simplified to:


(5)..... f = 3p(D/t)2   lbf/sq.in.


This gives the maximum stress at the heel of the ring - opposite to the gap - and offers a basis for ring design based on the maximum stress. As a matter of interest, a check on the maximum working stresses of a number of commercial rings of model size ranged from 40,000 to 49,000 lbf/sq.in. Almost certainly an alloy iron, but if standard "grey" cast iron, probably BS1452/1977 grade 350. The cast iron stick currently available has an ultimate tensile stress of 38,000 lbf/sq.in. - possible BS grade 260.

Test Results

The table below gives the figures. All are commercially made. Rings A to E are from Stuart Turner casting sets supplied between 1975 and 1990. Ring B is that supplied with my SIRIUS set and F the thicker one referred to above. G is a "paired" ring, one inside the other with the gaps opposed, source unknown.

+Bore D
ins.
D/t D/w D/g t, ins w, ins free gap, g ins.Mean wall Press.
lbf/sq.in.
calc
Mean wall Press.
lbf/sq.in.
Measured
A 0.75 22 14 7.5 0.033 0.062 0.098 32 28
B 1.00 25.6 16 6.7 0.039 0.062 0.149 25 30
C 1.25 23 20 8.0 0.055 0.062 0.158 30.7 27
D 1.75 32 9.3 7.6 0.063 0.187 0.231 18 16.7
E 2.0 32 16 7.3 0.074 0.124 0.273 17 15
F 1.0 20.4 16 7.1 0.049* 0.062 0.140 40 41
G1 1.0 30 21 7.0 .030 0.047 0.142 - 19.3
G2 1.0 49 21 7.0 .032 0.047 0.142 - 19.3

* Note - Ring F measured after running, in worn state.

A ,B and E are for cylinders rated at 80 to 100 lbf/sq.in, piston speeds around 350 ft/minute. C and D are the intermediate and low pressure cylinders of a triple expansion engine. F is possibly intended for a petrol engine, but is listed simply as a "1" dia x 1/16" piston ring" by the suppliers.

Before commenting it should be realised that engines A - E were all designed from 60 to 80 years ago. At that time rings were made by cutting a gap in an oversize ring and closing it into the cylinder. Such would need a rather higher Mean wall pressure to compensate for the inequality round the ring, and to speed up running in. It is possible that the original dimensions were retained when the modern, heat formed, ring became available. Nevertheless the contrast between the wall pressure of rings A & B, and ring E, is difficult to explain.
The unsuitability of F is obvious. However, my own concern was not so much with the variation between these examples as the contrast with what I knew of full-sized practice.

Full Size Practice

My most direct personal experiance has been with stationary diesel and gas engines, the latter both high and low compression, but I have had indirect concern with large industrial steam plant. The table below gives some examples. Rings A, B, and C were internally peened to form the profile; D was heat formed. E was fitted with Lockwood & Carlisle internally sprung springs, whilst F was, I believe, the common "Ramsbottom" type. G is from a 1985 small "transit" type vehicle.

The data on ring H are obtained by calculations from the ring dimensions of a 5cc 4-stroke air-cooled aero engine designed by Prof. Chaddock for an attack on the world duration record for radio-controlled aircraft in 1967. (It then stood at nearly 9 hours) A - D & G, totally enclosed forced lubricant.

typeMax Press
p.s.i.
C/RMax bmep#
p.s.i.
Piston Speed
ft/min
Bore/str. in.Mean Wall
p.s.i.
No. of
Press. Rings
Notes
A Diesel75012.580105016 / 21 113**
B Diesel10501511015004.13 / 6 123**
C Gas4706.567100021 / 24 6-73**
D Gas100014.711015004.13 / 6 113**
E Steam120--85110045 / 54 2.51
F Steam Loco180-- ?140019 / 26 52
G Petrol400(est)7.5110199090mm / 69mm 251**
H Petrol?11?8090518mm / 20mm 172

** plus 1 stepped oil metering ring and 1 slotted oil scraper.
# "bmep" is the mean cylinder pressure related to the BHP rather than the IHP.

The engines A to D were unsupercharged and certainly not typical of modern practice. Nevertheless their cylinder conditions exceed those of almost all "model" I.C. engines and certainly those of steam plant. Engine "E", for example, worked something like 22 hrs/day, 5 and a half days/week, and was, at the time these rings were fitted, developing about 40% more than the original design power.

Engine "H" was bench tested at about 65 BHP/litre at 9,600 rpm. With an iron cylinder and alloy piston the rings proved to be quite adequate, and no oil control ring was fitted. Lubrication was flung from the crank.

Reference to "the books" confirmed that the ring pressures above were reasonably typical. Ricardo quotes a figure of 5-6 lbf/sq.in for spark ignition engines of 5.5/1 in the 1930's and Hepworth writing in about 1945 indicates about 9 lbf/sq.in for pre-war cars, but observed that high performance engines (piston speeds in excess of 2800 ft.min) might need up to 20. "Metering" oil control rings might need 10-15, but slotted oil scrapers might reach 30 or more, depending on the engine conditions.

Data on steam engine practice is sparse as to wall pressures, and complicated by the fact that all relates to the ordinary "Ramsbottom" ring. However, Spooner, 1908 remarks that "...3 or 4 lb is all that is needed..." and describes the "Allen" internal spring ring which achieves this figure. However, his rules for Ramsbottom rings (t=D/30 to D/40, gaps D/10 to D/7.5) imply wall pressures ranging from 4 to 8 lbf/sq.in. D.A. Low, writing in 1927, uses a relationship in terms of D P/100 which works out at T=D/32 to D/34, giving about 7 lbf/sq.in at g=D/10. However, in his book on "Heat Engines" Low observes that with a well-fitting ring the wall pressure "...need not exceed 2 lbf/sq.in". These wall pressures may, at first sight, seem to be very low, but it must be remembered that the spring pressure of the ring will be augmented by the gas or steam pressure which builds up behind the rings as gas leaks through the axial clearance. (See "M.E." 16AUG74, p. 824, Mr. Smart).

Model Piston Ring Practice

References to piston rings, other than in the body of articles, are very sparse in "The Model Engineer" - less than a dozen over the last 60 years. it has not been possible to search every I.C. or steam engine construction series, but in all those looked at the only specification has been the axial width of the ring. Ian Bradley, 22OCT1942, gives a detailed account of the manufacture of tapered Ramsbottom rings, but the range of D/t ratios is very wide, as is that for ring gaps. Mr. E.B. Rix, ME 3JUL56, shows an interesting piston valve design and attempts to analyse the stresses, but his initial assumption (that the neutral axis of the unstressed ring should be equal to the piston diameter) is dubious. However, he makes a useful contribution on the merits of the tapered ring. M.W. Smart, 16AUG74, makes very useful observations on ring operation, but the bulk of his article concerns "composite" rings, the rubbing surfaces being of PTFE. Nowhere is there any mention of wall pressure at all over this period.

Prof. Chaddock, (21APR67) gives "rules" for the radial thickness and gap for I.C. engines which would, with E=17 x 106, provide wall pressures in the range 13 to 28 lbf/sq.in. He also describes a "heat forming" method suitable for amateurs, and emphasises the need to machine the O.D. after forming. The rules, that t=D/25 to D/30, and that gap=4t, have been the I.C. Standard for model rings ever since. However, the stress does need checking at t=B/25 as it does work out higher than the normally available 17-ton iron will stand. Mr J.M. Tulloch, in "Engineering in Miniature", Oct 1979, uses this ratio also. However, his procedure is flawed, in that he does not machine the O.D. and, as I shall mention later, his heat treatment is incorrect.

The only reference to wall pressures which I have been able to find is in a series of articles in "Strictly I.C." from FEB/MAR 1989 to JUN/JUL 1989 by Mr G.S. Trimble. This concernes the rings for his very fine V-8 racing engine, 1" bore and stroke - but, unfortunately, neither the speed, mean or maximum pressures, nor the power output are stated.

He describes a long series of experiments and offers a very elegant "stress quadrilateral" which can simplify choice of gap, radial thickness, etc, as well as a "foolproof" device for splitting rings before heat treatment. The wall pressure finally selected is 45 lbf/sq.in. This surprised me, especially in view of the experiance with a SIRIUS ring with a similar pressure. Unfortunately Mr. Trimble did not reveal the mathematical basis he used, though I imagime that it was the classical "curved beam" theory, but I did notice some flaws in his assumptions. First, he rejects any wall pressures less than 30 lbf/sq.in, in using a fallacious argument to justify this. He rightly observes that the manifold depression (i.e. the suction pressure) on a throttle controlled engine may be high. He quotes 20" Hg,(though in fact can be as high as 28") and assumes that his wall pressure must be higher than this to avoid "sucking the rings off the cylinder wall". This is an error; the piston ring does not know what is causing the pressure difference accross it, and large gas engines may run for an hour or more with wall pressures as low as 5 lbf/sq.in. under this condition. He allows a factor of safety of three to reach his figure of 30 lb.

Second he had difficulties with excess oil consumption. I deal with this later, but whilst a pressure of 45 lbf/sq.in. may be neccessary for the scraper ring it is certainly not needed for a pressure ring. "Horses for courses" - let each ring be proportioned for its own duty. Thirdly, like Mr. tulloch, he uses too high a stress relieving temperature, sufficiently high to alter the value of Young's Modulus, so that it may well be that the actual wall pressure was lower than he calculated. The more so as he seems to be using an alloy iron, with UTS in the region of 41 Tons (Imperial, not US)/sq.in.

It could be that there is some reference to wall pressure in Model Engineer, buried in a constructional artical, but it is clearly impossible to read every piece over the 110 volumes available! The matter must be approached by deduction, in the hopes that some member with adequate facilities can carry out experimental work to confirm or reject the conclusions.

Criteria for model engine rings

Considering steam engines first, it is fairly clear that the "duty" of a model ring is less onerous than that of the prototype. Even if fully loaded the stationary engine model is working at a lower piston speed and, in most cases, at a lower admission pressure. The duty period will be less - hours/month rather than hours/day! The small locomotives work at admission and mean effective pressures well below full size, and the mean piston speeds are much lower - around 200 ft/min against 1200-1600, even though the rotational speeds may be higher. Duty runs seldom last more than 30 minutes without stop - IMLEC is probably the longest non-stop run made. Annual mileage is relatively small.

As to I.C. engines, all will tend to lower maximum pressures than prototype, simply because there are proportionately higher heat losses. Piston speeds, again, will be lower in all scale or semi scale engines. so far as the classical model horizontal engine is concerned, whether gas or petrol, this applies completely, with bmep down to 55 lbf/sq.in. and piston speeds at 450 ft/min or so. Free-lance "hard work" engines still run at lower piston speeds, but both maximum and mean pressures are of the same order as those of the "softer" current types of automobile engines. Piston speeds are down to less than 800 ft/minute against 2600 for a "family" car - the maximum speeds are lower, too, about 4500 rpm compared with 5600. When we come to the "high performance " engines, at first sight the conditions do appear to be arduous. The best figure I have found for a model is about 60 BHP/litre, with a bmep of 120 lbf/sq.in. at 6500 rpm - piston speed about 1600 ft/min - and air cooled at that. However, these units do have a short duty period as a rule - 20 minute per trip, perhaps. The only "long duration" high performance model I have been able to find is that by Prof. Chaddock, referred to earlier, and it will be noted that the ring pressure he used is lower than any of those that I tested from model steam engines! These conditions are nowhere near those for a high performance prototype engine. (A Porsche 968 develops 80 BHP/litre, bmep at 168 lbf/sq.in. and piston speed up to 3800 ft/minute, whilst the 911T offers 97/litre, bmep 220 lbf/sq.in. at 3350 ft/min!)

It is clear that the high performance engines may need special attentioin and experiment when dealing with piston rings, but for the rest there seems to be only one reason for these small engines to need higher wall pressures than in full size practice. Most full size I.C. engines are run 10 - 15 hours to bed down the rings properly - or a 600 mile "running in" period on a car; during this time the pistons of a 1.5 litre engine will run no less than 250 miles up and down the cylinders. The same applies more or less with a 10" X 12" diesel engine. Further, there are certain deficiencies in the manufacturing method used with amateur made rings (which I deal with in a moment) which make tham less than perfect. it may be necessary, therefore, to stiffen up the rings a little to speed up running in - if only to get the loco fit for IMLEC in time! Stationary models - steam or I.C. - can, of course, be motored without difficulty, but not locomotives.

One more consideration should be looked at, that of the so-called ring flutter. Ascar engines developed in performance it was found that at certain speeds excessive blow-by past the rings occurred. The received explanation is that at these speeds the inertia force on the rings towards the top of the stroke exceeded that due to gas pressure. The ring "floated" in its groove, and failed to seal. Raising the ring pressure from 9 to 15 lbf/sq.in. effected a cure, though some ring manufacturers favored designing rings which exerted a higher wall pressure only in the region of the gap. In very high performance engines even higher pressures were used, but later a dual ring was found more effective with less friction.

This phenomenon need not concern the model maker. The inertia force depends on the mass of the ring, the square of the rotational speed, and the engine stroke. Flutter in a model will, therefore, not occur until its speed reaches scale X prototype flutter speed. for a model car engine at 1/4 scale this speed will be of the order of 16000 rpm.

Oil Control

This is needed only with totally enclosed engines (steam or I.C.) with splash or pressure bearing lubrication. Oil from the main and crankpin bearings is thrown onto the cylinder walls - in fact, almost all the oil supplied by the pump will land there. This flood can overwhelm the pressure rings, no matter what pressure is used, and some control must be effected. This is done by supplementary rings which scrape off surplus oil.

First, however, a few basic points. (a) The pressure rings must be lubricated. There are many recorded cases where excessive oil consumption has been caused by too fierce a scraper ring.. The pressure rings "scuff" through lack of oil, and the resulting grooving of the cylinder disables the oil control ring. (b) Unlike a commercial undertaking, the cost of lubricating oil is not a consideration. So long as the plugs do not oil up, a little extra oil does not matter a great deal. A commercial stationary (or marine) engine may hope to reach an oil consumption of perhaps 0.75% of the fuel used, and car manufacturers only garantee an oil consumption of about 2% - 3% of the overall fuel figure. A two-stroke may demand 4% or even more. Many model engine builders are far too concerned about this - far better a little oil smoke than a ring failure! (c) It is quite useless to estimate oil consumption by motoring the engine with the cylinder heads removed. Under these conditions the pressure rings act as a very efficient oil pump - in the upwards direction! Finally, (d) it is a common mistake to assume that a smoky exhaust during or after idling is a symptom of deficient oil control. Under these conditions the charge temperatures are very low; little fuel is being burned and, moreover, the mixture is heavily contaminated with residual exhaust gases. So, instead of burning cleanly, the oil which does reach the combustion chamber (quite normally) "smokes". If the exhaust is reasonably clean under power, light load smoke should not cause alarm so long as the plugs are not disabled.

The most common types of oil control ring are shown in Fig. 3. At (A) is the bevelled ring, which rides over the oil film on the upstroke, but scrapes when running downwards, the oil passing through the piston through the holes shown. It has the advantage that the width of the contact face can be made very narrow when new, but as it wears and becomes more effective the wall pressure is reduced. The disadvantage (for model makers) is the stepped scraper (C). The ring is the same section as the pressure ring, but the width of the step is adjusted to give a higher wall pressure. Again, oil escape holes are provided. This is a very effective ring indeed, and has the advantage that experiment is very easy. This type, with a step from 1/2 to 2/3 of the axial width, is used as an oil metering ring above the gudgeon pins and below the 3 pressure rings on large diesels. During development the width of the step is determined to give just enough lubrication for the pressure rings and no more. However, it is quite adequate as a main control ring when speeds are low and oil-flinging moderate.

(C) is the slotted oil scraper. it is, in effect, two narrow rings with an oil escape path between them. Its effectiveness is more due to the duplicate scraping action rather than high pressure. A similar type, easier to make, has oil holes rather than slots through the ring. This type of scraper usually has the same radial thickness as a pressure ring, but is wider to accommodate the slots and associated grove.

So far as models are concerned I would expect the stepped scraper to perform adequately in all but the most extreme cases. Only where the step width has to be less than 1/3 of the axial ring width would I consider the slotted (or drilled) type. However, there is an important proviso. NO scraper ring, however fierce, will get rid of the oil if the escape holes are inadequate. Designs I have seen in Model Engineer (and elsewhere) with no more than half a dozen 1mm holes (or, in one case, No. 70 - 0.71mm!) are asking the impossible. The diameter of the holes through the piston should be not less than 2/3 of the axial ring width. Holes through the ring are more constrained, of course - in fact, this is the reason why manufacturers changed to slots, despite the added difficulty of manufacture. The number of holes should be "as many as you dare, within reason". I would expect to see 12 in a one-inch piston, (in each row with a slotted ring) and more in larger pistons.

As to wall pressure, 30 lbf/sq.in. should be more than adequate and in the interests of reduced friction and wear I would start at 20 during the development stage. Rather than put up wall pressure, advantage can be taken of the fact that the stepped ring also acts as a pressure ring. The line-up can then be one plain pressure ring, one stepped ring, and one slotted or drilled scraper below that.

Finally, to repeat earlier comments. Too high a scraping specification simply adds friction to no purpose and can be dangerous. So long as the performance is not spoiled by plugs oiling up it is wiser to tolerate what may appear to be a high oil consumption.

continue with part 2

HOME