High-Speed High-Output Diesel Engines
High-Speed High-Output Diesel Engines - 35 Years of Development of Railroad and Marine Applications
By Markus von Kienlin and G. W. Maybach
History
TO THE EXPERT, the name of Maybach Motorenbau immediately recalls a special class of engines characterized by the following outstanding features: high speed, light weight, small size, high fatigue strength (due to special design and production measures,) and low operating cost.
The unique development of the Zeppelin airship after the historic accident at Echterdingen in 1908, was decisively governed by the fact that Wilhelm Maybach, Gottlieb Daimler's collaborator, made available to Count Zeppelin an engine designed by his son, Karl Maybach. This was a 6-cyl 150 hp 1200 rpm engine with a fuel consumption of 254 g/hph and a weight-to-power ratio of no more than 2,99 kg/hp, in contrast to the engine installed in the first Zeppelin airship which had a weight of 26 kg/hp and a consumption of 508 g/hph. For the production of the new engine, Count Zeppelin and Wilhelm Maybach founded, on March 23, 1909, the Luftfahrzeug-Motorenbau G.m.b.H. which was affilleated to Luftschiffbau Zeppelin. The Maybach Motorenbau, which was entrusted to Karl Maybach from the outset, produced the engines for further Zeppelin airships and for aeroplanes. Among those engines, as early as 1917, highly rated high-compression high-altitude engines were built. Fig. 1 shows such a high-altitude aircraft engine. Fig. 2 is a sectional view of the cylinder liner and wrist pin area.
After the enforced interruption of almost the entire production at the end of World War I, new and related lines of production suitable for the skilled staff had to be found. The choice included Otto and diesel engines for a number of purposes. Research continued on Zeppelin airship engines including the well-known 550 hp VL-type engine (Fig. 3) for the airship "Graf Zeppelin," which accomplished the famous world tours under the command of the late Dr. Eckener, who until his death in 1954 was chairman of the Maybach Motorenbau. The first powerful high-speed diesel engine was developed in the Maybach workshops between 1919 and 1923. Construction of this G 4 engine (Fig. 4) was a remarkable venture, bearing in mind that at that time experience with high-speed diesel engines was actually only with submarine engines having a maximum speed of not more than 500 rpm, whereas the new 150 hp Maybach diesel engine had an operating speed of 1300 rpm.
Solid fuel injection was then still at its beginning, so that the G 4 engine had air injection. Fig. 4 shows the relative sizes of the diesel engine and the engine-driven air compressor needed for air injection.
Opinion at that time - which many still hold today - was that the high-speed engines despite their obvious advantages with respect to weight and reduced space requirements would have a shorter service life, have less favorable consumption figures, and be more prone to trouble. This opinion chiefly resulted from the fact that many firms tried to increase the output of their existent low-speed engines merely by increasing the engine speed and neglecting any improvements in design. Naturally, this led to setbacks which in turn gave rise to such widespread opinion. The Maybach Motorenbau, which even then had many years of experience in the field of high-speed Otto engines, adopted new methods by designing a diesel engine especially for high speed and by aiming simultaneously at a corresponding increase of operating reliability and service life, two requirements obviously necessary for airship engines.
Since the first results with high-speed diesel engines were very encouraging, the development of these engines was continued and their output increased. The output of the first 6-cyl engine was soon raised to 210 hp at 1400 rpm. followed in 1930 by a 12-cyl V-engine providing 410 hp at 1400 rpm which, at a later date, was brought to 600 and 650 hp by exhaust gas supercharging (Fig. 5). In 1931 two of these 410 hp GO 5 engines were installed in the "Fliegender Hamburger," the first high-speed railcar-train of the German State Railway (Fig, 6). Its schedule speed of 77.6 mph, no less than the regular daily top speed of 100 mph, on the route between Hamburg and Berlin was a sensation at the time among railroad experts.
These Maybach engines of the GO-type - primarily designed as power units for railcars - set the fashion for diesel train operation of this power class in Europe until the year 1940. Hundreds of railcars and diesel trains were equipped with these engines in Germany, France, Belgium, Holland, Sweden, Norway, and Spain. Moreover, the engines met with favorable reception in naval construction, and were increasingly used for high-speed passenger boats, customs cruisers, yachts, coastguard crafts, and other craft.
The GO-type were designed so that the cylinder blocks of 6.30 in. bore and 7.88 in. stroke, made in one casting with the cylinder head, were assembled in an aluminum crankcase. The engines had direct injection, and four valves per cylinder controlled by overhead camshafts. The pistons were made of aluminum. Convenient mounting of running parts posed a special problem at the time for these powerful engines. White metal bearings could not be used, and lead-bronze bearings were still in too early stages of development. Therefore, it was decided to provide both the crankshaft and big ends with roller bearings, the composition of material and thermal treatment for which were the subject of lengthy and painstaking development by the company.
In operation, these engines attained mileages between general overhauls of as much as 100,000 miles (about 2500 operating hours) and more, which prior to World War I was regarded as quite satisfactory.
After termination of World War Il the situation was entirely different. Under the influence of the immense advance of the diesel locomotive in the United States and elsewhere, railway companies became increasingly interested in converting their heavy traction service to diesel operation.
If European manufacturers wanted to take part in this development, they had to counter the American diesel locomotive with its heavy and slow-running engine by a type of vehicle which would be available for a more universal application, offer more favorable weight conditions, and which in respect of output and service life would at least equal the heavy engines.
For Maybach Motorenbau, as the representative of the European trend, the demand for long life of the engine and its components became predominant. New lightweight small-size engines had to be designed which would be capable of offering mileages between overhauls at least equal to those then attained in rail traction by the finest types of slow-speed engines. Beyond that, these engines were supposed not only to compete successfully in the field of diesel traction, but also to conquer new fields of application.
Considering that for the aforementioned GO-type engines the roller bearings in the big end of the connecting rods were the limiting factor for the running performance, it was obvious that further development should aim at plain bearings for the big ends and a modernization of all running parts. The best solution in this direction appeared to be the discwebbed crankshaft, which inherently has greatly improved vibration characteristics, and which also provides ideal conditions for the entire bearing problem.
This was confirmed by the experience gained by Maybach Motorenbau during the last war with more than 50,000 Otto engines ranging from 300 to 700 hp at 3000 rpm, equipped with disc-webbed crankshafts and installed in heavy-duty military vehicles (tanks and heavy trucks). As a mater of fact, the adaptation of the disc-webbed crankshaft to the aforementioned pre-war GO-type, resulted in the change-over to the so-called GTO-"tunnel" type crankcase, representing a unique jump ahead in development.
Proof of this was the service results achieved by the German Federal Railway with its 600 bhp Maybach GTO type tunnel engines. It was revealed by spot checks that the bearings of these engines operating at 1400 rpm were still like new after distances of 300,000-400,000 miles. All other wearing parts showed such slight wear that a general overhaul was not anticipated before 600,000-700,000 miles, corresponding to about 12,000 to 15,000 operating hr. Early in 1955, these 600 bhp engines had exceeded an aggregate mileage of more than 6 million miles without any major engine overhauls (1).*
Today there are more than 1000 of these 12-cyl GTO engines rated up to 800 hp in service. Of these 840 are installed in shunting locomotives of the European standard type V.60, running for 20,000-25,000 hr before a piston check is scheduled.
Thus an optimum solution was found for the problems presented by the running gear of this engine type. For more powerful engines, however, with continuous outputs of about 100 hp per cylinder and with similar or rather better service behavior, one more step was necessary. That step was the introduction of the tunnel construction with roller main bearings surrounding the crank webs. The performance of this was proved first by thorough trials with powerful diesel engines of a speed range between 2400 and 2600 rpm. Eventually this led to the present Maybach MD tunnel engine, a design which has aroused great interest in the technical world. With this design, the problem of building a high-speed diesel engine with a running performance equal to, of better than that of a good slow-speed engine, can be regarded as solved.
Before giving a more detailed description of the MD design, it may be advantageous to discuss some general problems pertaining to high engine speed, and to report on the research techniques which contributed essentially to the development of the modern Maybach diesel engines.
General Questions Relating to High Engine Speed
It is, first of all, necessary to emphasize a fact often insufficiently realized by users: engine speed alone does not provide an adequate criterion for the genuine high-speed character of an engine. Even the mean piston speed does not provide such a criterion, since the mean effective pressure must also be taken into account, as well as whether the piston speed is used with high or low mean effective pressure at equal engine speed.
Some performance characteristics, important as far as engine loading is concerned, are now considered. Besides the absolute engine power N and the specific power per cylinder N/z, it is quite common to use the ratio of power per unit of swept volume N/V, characterizing the degree of utilization of the swept volume. The ratio of power per unit of piston area, N/F, also has a bearing in this connection: it signifies the total engine power related to the total area of the piston crowns. The piston crown is one of the engine parts subjected to the highest thermal load. To dissipate the heat accumulated here by means of conduction, radiation, or coolants is a matter of vital importance which may well be a limiting factor in increasing the specific engine power.
Strictly speaking, the ratios of power per unit of displaced volume, and power per unit of piston area can only provide a valid scale of comparison if the cylinders compared have the same size and shape. The influence of the stroke-to-bore ratio on these parameters has been investigated by Jaklitsch (2). Under certain limiting assumptions, he found that this influence is of the order of Vs/d, (s = stroke d =bore) a result derived statistically for diesel engines designed for aircraft and road vehicles, and it may be assumed, that this influence is equally applicable to larger types of diesel engines.
The fact that the ratio of power per unit of swept volume is influenced by the stroke-to-bore ratio can be explained by the fact that as this ratio changes there is also a change in the ratio of the heat-affected surface to the corresponding swept volume. With the longer stroke engine, where the piston diameter is smaller, the distance of heat flow from the center of the piston crown to the cylinder wall is smaller. Therefore, the temperatures at the center of the piston crown become higher if the piston diameter is increased. The results of similar investigations on the influence of cylinder bore and stroke-to-bore ratio on the piston temperature are plotted in Fig. 7. This graph shows that, even if mean effective pressure, piston speed, and stroke-to-bore ratio remain constant, an increase in the cylinder bore causes higher temperatures at the center of the piston crown. It will also be noted that the increase is steeper with short-stroke engines than with long-stroke engines. In order to obtain more realistic parameters for comparisons, the ratios of power per unit of swept volume and power per unit of piston area must be related to the stroke-to-bore ratio √s/d as follows:
"specific power per unit of swept volume" (N/V)
"specific power per unit of piston area" (N/F)
Apart from having such parameters as the specific power-to-volume ratio, and the specific power-to-piston area ratio, which are more indicative of the thermal loading of the pistons, it is also of importance to obtain criteria characterizing the mechanical stresses occurring in the rotating crankshaft assembly. This applies in particular to the big-end bearings, the loading of which might be characterized by the mass forces of the rotating and oscillating parts of the engine as well as by the maximum gas pressure. Assuming that, with the application of corresponding design principles, the weights of the rotating and oscillating engine parts vary with the third power of the piston diameter, a parameter a can be formed relating the mass forces to the unit of piston area, for example:
Jaklitsch calls this a value "running index" (2), and ascribes to it a major importance. The term "mass-force factor," however, appears to be more suitable as it relates more directly to the significance of the parameter. For it is accepted that, with a corresponding utilization of the cylinder diameter, the projected area of the big-end bearings increases with the square of the piston diameter, the massforce factor may well be accepted as a valid criterion for the bearing load due to the mass forces.
If the two engines compared are geometrically similar, that is, if the stroke-to-bore ratio remains the same, the above assumption is strictly true, as the masses in motion do vary with the third power of the piston diameter. In order to gain an idea in regard to the variation of the masses acting the big-end bearing When the stroke - to - bore ratio is modified, this ratio was altered in a V-engine and the variation of the masses acting on the bearing was calculated.
It was found that when the stroke-to-bore ratio was increased by as much as 60% the increase in weight was no more than 2-3%. It follows that, within reasonable limits, the mistake made by assuming that the variation of masses is proportional to d3 is insignificant.
F r o m this may be concluded that the engine with the most efficiently utilized crankshaft is the one which has the highest mass-force factor a, and, at the same time, can prove to be reliable in service and to have the service life expected for its application (5).
The diagram in Fig. 8 is intended to show the significance of the mass-force factor. For instance, if an engine has a speed of 1300 rpm and a cylinder bore of 7.28 in., with a stroke of 7.78 in. the mass-force factor becomes 672 ft/min?. If the same engine is accelerated to 1800 rpm, the mass-force factor increases to 1289 ft/min?. With a slow-speed engine of, say, 250 rpm, 17.72 in. bore and 25.58 in. stroke, the mass-force factor is no greater than 199 ft?/min?. Since both slow-speed and high-speed engines under consideration have actually been found to be very reliable in practice, it follows that the crankshaft assembly of the high-speed engine is much more efficiently utilized than that of the slow-speed engine.
In order to explain the significance of the mass-force factor still more in detail the characteristic parameters have been determined for some of the most important present-time locomotive engines of about 1000 bhp and more. The data are listed in Table 1. Four-cycle engines have been selected only so as to make a simple comparison of the parameters possible.
Two engines with equal characteristic parameters can be assumed to show equal service performance of their piston and crankshaft assemblies. This, however, holds only for engines of similar design. Therefore the parameters should never be compared on their own, but only under simultaneous consideration of the designs of the engines involved.
It appears from Table 1 that the specific power-to-piston area ratio of the locomotive engines - irrespective of their power and speed - is within the limits 210 to 345 bhp/ft?, excluding the engines represented by items 8 and 9, which will be discussed later. For the turbocharged stationary engines, rated at continuous speeds from 250 to 600 rpm, the respective power - to - piston area ratios range approximately between 93 and 186 bhp/ft?. From this comparison it becomes quite evident that the effort to ensure small bulk of the locomotive engines to make them suitable for their application, imposes high demands on engine design as well as on material when operational reliability and life of the wearing parts of both engine groups is to be the same. Even more revealing are here the high mass-force factors for the locomotive engines which range between 603 and 915 f t /min', whereas the corresponding figures for stationary engines lie only between 215 and 431 ft ⅔ /min?.
These simple considerations show that locomotive engines which have proved satisfactory in practical operation must be types of a highly meritorious design, for they have withstood the imposition of quite severe demands. In order to permit a comparison between different engines on the basis of the criteria developed above, a common basis of "equal loading" has to be defined. "Equal loading" of two engines specifically means that the mechanical and thermal loading of piston and crankshaft assemblies of the two engines is the same. This is characterized by:
The conception of geometric similarity is understood to cover not only the cylinders, (stroke and bore), but also all those parts of the engine which are subjected to any stresses. In all these comparisons, the mechanical efficiencies of the engines are assumed to be constant. From the well-known engine power equation:
Assuming that for two geometrically similar engines (s/d = constant) the mean piston speed Cm the mean effective pressure Po, the swept volume Ve and the ratio N/ are constant, the correlation between power and number of cylinders becomes:
Fig. 9 shows this functional relationship indicative of the increase in output obtained with an engine of the same total swept volume by increasing the number of cylinders. For instance, if a plant hitherto driven by one slow-speed 12-cyl engine is driven, instead, by eight high-speed 12-cyl engines which have altogether the same total swept volume as the one slow-speed engine, the output is doubled although mechanical and thermal stresses in the piston and crankshaft assembly are the same. If it were intended to achieve this doubling of output by enlarging the swept volume of the single slow-speed engine, the swept volume of that engine would have to be increased 2.8 times.
Finally, regarding the significance of the parameter of specific power output per unit of piston area as a factor characterizing engine loading, it may be used to divide engines into different load or performance categories. For instance certain ranges of this specific power-to-piston area factor can be assigned to the category of engines of high specific output, to the category of vehicle engines, the category of marine engines, and so on. Under these assumptions can be derived for N/F/Vs/d = constant and (s/d) = const.:
for a given load category and geometrically similar engines, the specific power-to-swept volume ratio is inversely proportional to the bore.
Fig. 10 shows the relationship between the specific power-to-swept volume factor and the bore for different values of the specific power-to-piston area factor. For example, with geometrically similar engines of equal thermal and mechanical stressing of their piston resp, crankshaft assemblies, the doubling of the bore would halve the power per unit swept volume. In the same figure data are plotted for four high-speed engines belonging to four different performance categories, and one slow-speed marine propulsion engine.
The data of these engines are listed in Table 2.
Table 2 shows that a normal high-speed engine (for example, Engine No. 1) has the same power output per unit of piston area as the slow-speed marine engine No. 5, where the power of the latter must be regarded as being rather high compared with other similar engines. With high-speed engines, however, by means of suitable measures such as piston cooling the specific power output per unit of piston area can be increased for marine engines to 304 bhp/ft?, as shown for No. 3 engine. Yet this still does not exhaust the possibilities of increasing the specific power output. As No. 4 engine shows, by an increase in speed, mean effective pressure, and by other measures, the specific power-to-piston area factor of the same engine can be raised to beyond 600 bhp/ft?, a value which cannot be approached even remotely by slow-speed engines. This comparison clearly demonstrates the possibilities for high-speed engines to increase their specific power output; it also demonstrates the wide power range which can be obtained with one and the same engine.
The higher the specific power-to-piston area factor is of an engine which has proved to be reliable in actual service, the higher is the standard of technical development which the particular engine can be said to have reached. Hereby the measures are quite decisive by means of which higher specific power output per unit piston area has been attained, without exerting a detrimental influence on the reliability of the engine.
In the high-speed Maybach MD engines, a great step forward was made with the introduction of pressure-oil piston cooling . Fig. I l gives several temperatures measured at different power outputs in the piston crowns and piston ring lands of a pressure-oil cooled and a noncooled piston. The comparison clearly shows that although the power output per unit of piston area was considerably increased, the temperatures in the compression ring area, due to the effectiveness of the piston cooling, did not exceed those measured in the conventional standard design. These temperatures determine if seizing of the compression rings will occur. The only temperatures which increased slightly were those in the center of the piston crown. This is due to the considerably higher thermal loading and to a slightly smaller conductivity of the steel crown. Since, however, the crown of the cooled piston is of heat-resistant material and not of aluminum, this temperature rise has no detrimental effect on service life.
So far, in the comparative evaluation of different engines, the common base, namely the engine loading, was understood to refer exclusively to the piston and crankshaft assemblies. However, cylinder heads and liners are both subjected to high stresses. This is due to the explosion pressure and also to the heating of combustion chamber walls during the combustion process.
To simulate the maximum stresses in the cylinder head and liner due to ignition pressure Pz, the tangential stress °mech acting on the external diameter of a thick-walled cylinder may be taken. This is calculated by:
mech = Pz d2 / 2 (d + b) b
where w h e re .
d= Internal diameter
b = Wall thickness
Correspondingly, maximum tensile stress due to thermal load Otherm of the cylinder head and liner may be taken to be proportional to the tangential stress on the external diameter of a cylinder that is internally heated and externally cooled.
° therm = 1 - U 8(d + б ) I n (1 + 26)
where:
E = Modulus of elasticity
v= Poisson's number
B = Coefficient of expansion
AT = Difference between inner and outer wall temperature
It becomes apparent from Eg. 4 and 5 that where the same material is used, stressing of the walls in cylinder heads and liners depends on the ignition pressure Pz, on the difference in wall temperature ST, on the internal diameter d, and on the wall thickness &.
As regards ignition pressure Pz it may be assumed that this remains practically unchanged where engines of different size and uniform mean effective pressure Pe are concerned. However, when the latter decreases, the ignitionpressure will also decline.
Given the same material, the difference in wall temperatures AT is directly proportional to wall thickness and specific heat flow. From the known engine equations the following proportions will result:
N ~ Pe Cm d2 N~ Q
o ~ Pe Cm d2 power output per cylinder
total heat flow through the walls
forming the combustion space
specific heat flow through the walls
forming the combustion space
Thus the following proportion may be taken for the wall
temperature:
AT ~ Pe Cm S (6)
Assuming uniform piston speed Cm and uniform mean effective pressure Pe,
AT ~ 6 (7)
In Fig. 12 a computation diagram has been plotted from Egs. 4, 5, and 7. This diagram, though based on conditions pertaining to the Maybach MD diesel engine, permits assessing for other engines the relationship between wall thickness of cylinder heads or liners, and mechanical and thermal stressing of walls at constant mean effective pressure. In this connection it should be noted that for the Maybach MD engine, of 7.38 in. bore and an average wall thickness of the cylinder head of 0.34 in., the mechanical and thermal stress figures have been entered as 100%.
Fig. 12 clearly shows that in a slow-running engine with
cylinders twice the diameter of the MD engine, with the
same mean effective pressure, and the same cylinder head
and liner wall thickness, the mechanical stresses in cylinder
head and liner walls are twice as high as in the case of the
quick-running engine. Thermal stressing of the walls only
differs very slightly. If, in this comparison, mechanical
stressing is kept constant in the slow-running engine, then
doubling the wall thickness will be required, in which case
thermal load will increase 2.4 times compared to the high-
speed engine.
For constant mechanical and thermal wall stressing and
equal piston speeds, a correlation between cylinder bore and
mean effective pressure was calculated from the above pro-
portions and plotted in Fig. 13. This shows clearly the ex-
tent to which the high-speed engine excels the medium- or
slow-speed engine in regard to the mean effective pressure
that can be realized. As an example, a diesel engine with
twice the cylinder diameter of the Maybach MD engine, and
having the same mean piston speed and the same mechan-
ical and thermal loadings of cylinder head and liner walls,
can be operated only at two-thirds the mean effective pres-
sure of the corresponding MD engine. This fact should be
taken into account because the possible thermal and me-
chanical loading of cylinder head and liner materials is prac-
tically the decisive factor today in limiting the specific power
output of a diesel engine.
For a long time, there was a widespread idea that high-
speed engines, because of their higher rpm and their gen-
erally higher piston speed have a greater liner wear than low-
speed engines. This idea has been discarded even by the
supporters of the slow-running engine.
It is argued today that the large bore of a slow-speed en-
gine also permits substantial overall wear before the cylin-
221
der liner must be replaced. Departing from the assumption
that the absolute wear per 1000 operating hr is about equal
in either case, the conclusion is that the cylinder liners of
a high-speed engine should be replaced at considerably shorter
i n t e r v a l s . thus resulting, as a rule, in higher costs for
spare parts, because any wear phenomena at this spot would
have an approximately constant relationship to the other
wearing parts of an engine (6). This conclusion, however,
stems from an assumption which in no way is represented by
actual existing conditions.
When studying technical literature, time and again one
finds the indication of 0.004 in. per 1000 per operating hr
as an average figure in regard to liner wear of large slow-
running marine engines. Depending on the nature of the fuel
used, this figure varies slightly upwards or downwards.
the other hand, there are operational results available today
from high-speed engines in railroad service which permit
certain conclusions. Fig. 14 shows the wearing quality of
some typical high-speed engines selected from a larger num-
ber of engines of the same type which for more than 5 yr
have been operated by the German Federal Railway (7).
engines concerned are 12-cyl Maybach engines, type GTO,
developing 600 bhp at 1400 rpm with a bore of 6.30 in. and
a stroke of 7.88 in., this being equal to a mean effective
pressure of 113.5 psi and a piston speed of 32.8 fps. These
engines still have uncooled pistons and therefore may read-
ily be compared with high-speed engines of another type.
With an average wear of 0.0063 in. related to 7500 hr, cyl-
inder wear in these engines is approximately 0.0008 in. for
every 1000 operating hr. These liner wear figures were all
measured where wear is greatest, namely, at the top dead
center of the topmost compression ring. As in the case of
cylinder bore, the permissible overall wear is in the neigh-
borhood of 0.016 in., reboring in connection with a major engine overhaul will not be required before some 20,000 op-
erating hr.
In this respect the objection may be raised that contrary
to the propulsion of sea-going ships, engines operated in rail
traction are not continously running at full load. This ob-
jection cannot be denied. However, it is a well-known fact
that continuous load changes, as they are the rule in rail
traction, subject the engine to increased thermal stressing,
and dust content of the induced air is another drawback for
railway engines. Wear figures obtained for more recently
installed MD-engines will be discussed later.
Apart from the wear resistance of an engine, the load on
the crank assembly is another factor of importance regarding engine service life, and here something of the influence
of increased engine speed may be shown. Characteristics
representing constant load on the main bearings of the crank-
shaft of an MD engine, the load being caused by ignition pres-
sure and mass forces, are plotted in Fig. 15 as a function of
torque and rotational speed. This graph shows that the load
on the main bearings at constant engine torque increases with
decreasing engine speed. Therefore, operating the engine
along a constant torque characteristic causes considerable
additional stressing of the main bearings. In order to obtain
power outputs as high as possible, it is from this point of
view more advisable to increase the speed rather than operate
at low speed and high torque. Operating the engine along the
propeller characteristic appears therefore to yield best serv-
ice results (8).
Based on the resulting bearing load the service life of the
roller main bearings was calculated. The lines of constant
service life have been plotted in Fig. 16 as a function of
power output and engine speed. It should be noted that this
calculation only applies to crankshafts with roller bearings.
From this graph can be seen that regarding service life con-
ditions are most favorable when with decreasing power out-
put engine speed is not reduced. It will also be noted that
if an engine is operated along the propeller curve, at a load
of about 83% of continuous power output, bearing life is
doubled, whereas with constant torque operation,e v e n a t
the lowest engine speeds there will be no increase in service
life. This again shows that increasing engine speed rather
than torque gives best results with respect to service life of
the crankshaft roller bearings.
The influence of high-speed on weight can readily be
seen in Table 1. When considering the column of weights
in this table the engines may be classified in three groups.
The first group with a weight-to-power ratio of about 20
to 22 1b/hp comprises the engines indicated under Nos. 1 and
These are 4-stroke engines with speeds up to 850 rpm,
built by European makers. Their specific power-to-piston area ratios and their mass-force factors are comparatively
low. The engines are quite heavy, thus having an essen-
tial bearing on locomotive weight. In the case of electric
transmissions the low speed will furthermore result in high
generator weights, unless a more expensive type incorporat-
ing two crankshafts and a step-up gear transmission for the
generator drive is adopted as for engine No. 2.
The second group refers to engines with a uniform weight
of some 16-18 1b/hp at 1000 rpm. Typical representatives
of this group are the two engine types nos. 3 and 4; these
are American 4-stroke locomotive engines (Alco and Cooper
Bessemer) of identical cylinder dimensions which are widely
used. Note that the weight reduction of these engines is
linked with an increase of the specific power-to-piston area
factor to about 344 bhp/ft?. For engines which have proved
to be satisfactory in actual service such values bear testi-
mony of a quite advanced stage in development. For com-
pleteness sake it should be mentioned, that into this cate-
gory belongs a widely used 2-stroke engine (General Motors)
16-cyl, weighing 15.8 tons, developing 1950 bhp at 835 rpm
corresponding to a weight-to-power ratio of 16.3 1b/bhp.
A definite step forward in regard to weight was the in-
troduction of engines of group III - engine nos. 5, 6 and 7 -
having a weight-to-power ratio of 6.5 to 1 1b/bhp.
These engines rotate at 1500 rpm and must therefore be classified
as high-speed engines. They originated in Europe and some
of them have gone through many years of development and
constitute an advanced form of powerful railcar engine.
In contrast to the United States, diesel rail traction in
Europe until World War Il was exclusively by the diesel rail
car. Here reference is made to the numerous high-speed
rail-cars and streamlined trains equipped with high-speed
Maybach diesel engines. The former German State Railways and other railroad companies operated such trains even
before the last war in multiple-units up to five cars powered by up to 2000 hp. This experience, supplemented by knowl-
edge obtained during the war years in all fields of engine
construction enabled the industry, by introducing engines
of this group, to set entirely new standards for the construc-
tion of powerful locomotives with a favorable power-to-
weight ratio.
Though it has not yet been generally recognized today,
these engines have no higher specific loading in regard to
their power ratings than typical low-speed engines, since
their specific power-to-piston area ratio, ranging from 278
to 334 bhp/ft?, is even lower than the corresponding values
of the American 4stroke engines mentioned before.
The mass-force factor of these engines, ranging from 790
to 915 ft/min?, is slightly higher; but this has not given
rise to any difficulties in practice, since the use of better
materials and improved machining techniques increased the
load bearing capability of the parts - especially of the con-
necting rod. Economically the use of better material was
possible as the bulk of the parts decreased.
Table 1 gives evidence of the substantial advantages of-
fered by high-speed engines in regard to weight. This re-
duced weight of the high-speed engine is an absolute ne-
cessity for the construction of powerful diesel locomotives,
and becomes even more important if for economic reasons
locomotives with not more than four axles are considered.
Development of Measuring Techniques for High-Speed Engines
Formerly, in development and testing of internal com-
bustion engines one had to rely fully on results obtained by
test bed endurance trials and on ensuing experience gained
i n p r a c t i c a l o p e r a t i o n . With this kind of testing it could
often take a long time before any weak spots in the engines
c a m e t o l i g h t . However, especially in the last 10 yr, elec-
trical measuring techniques have been so improved that it
is now possible to actually measure the mechanical load to
which an internal combustion engine is subjected. In the
case of high-speed engines the requirements with respect to
measuring methods are considerablym o r e s u n d e n o w i n ?
to the higher frequencies, so that recent development of
measuring techniques is of particular benefit to the high-
speed engines.
In order to anticipate the reliability of individual ma-
chine members, it is of prime importance to know the fa-
tigue strength of the material. The results from fatigue
strength tests can be presented in the form of a diagram in
which the number of load alternations prior to the fatigue
fracture is entered as a function of the ultimate strength.
was recognized by Wohler as long as a hundred years ago that
there is an interrelationship between the fatigue strength of a
steel and the load cycle, and that a material which has with-
stood a certain load level for a definite number of alternat-
ing stress cycles will not break. As can be seen from Fig.
17 this load level does not alter even when the number of
load cycles is increased from 10 million to 200 million (Fig.
18). For both high-speed and low-speed engines the number of 10 million load alternations is attained relatively quickly,
so that for the design the fatigue strength and not the tensile
strength has to be taken into account. It is quite natural that
when investigating the fatigue strength of a material there
will be some dispersion owing to irregularities in the surface
structure, and it so happens that in endurance tests not a
single fatigue characteristic is obtained but rather a range,
of which the lower stray field limitation supplies the prac-
tical values for calculating the operating reliability. In such
investigation regard must be paid to whether there are pure
tension-compression or bending stresses, or if there is a non-
uniform stress distribution. In the case of applying a bend-
ing load on small diameter rods or rods with notches, higher
alternating stress amplitudes might be acceptable, because
residual stresses as a result of machining the surface layer
may become particularly perceptible here.
In the Maybach stress laboratory various grades of steel
have been examined in respect of their endurance strengths.
In Fig. 19 the bending fatigue strength of various steels has
been plotted as a function of the tensile strength. The in-
dividual values clearly indicate that the service life of an-
nealed (hardened and tempered) steels increases with the
tensile strength. Therefore, the alloying constituents are
only of importance inasmuch as they ensure perfect and
thorough hardening. As a comparison with these measure-
ments, a characteristic curve "b" by Wellinger and Gimmel
(9) has been added to Fig. 19, showing general agreement
with the findings of Maybach Motorenbau as represented by
curve "a."
Apart from the influence of tensile strength on fatigue
strength, exhaustive studies were also made regarding the
influence exercised by surface finish and material purity.
Some results of such an investigation are shown in Fig. 20.
The graph shows that machining may have a considerable influence on the fatigue strength of the material, and that
under certain circumstances the advantage of high strength
as offered by high tensile steels may be offset entirely
by their poor machinability. As indicated in the graph, the
decrease in fatigue strength in the case of a forged surface
may be quite considerable. Similarly unfavorable condi-
tions are to be expected where impurities and heat-treat-
ment defects are present. However, by the use of modern
test methods it is possible today to ascertain a desired de-
gree of surface finish and purity of a material so that, when
properly inspected, the minimum fatigue strength of the lat-
ter can be realized. Apart from the surface finish, the in-
fluence exercised by fretting corrosion on fatigue strength
is also of considerable importance.
In this connection the bending fatigue strength of a high
tensile steel clamped between two jaws of unhardened car-
bon steel was determined by the Maybach stress laboratory.
It is shown in Fig. 21 that due to the action of fretting cor-
rosion phenomena the bending fatigue strength of this rod
was reduced to as low as 28,400 psi. Another test speci-
men of the same material, soft-nitrided, attained a bend-
ing fatigue strength of 71,000 psi under the same circum-
s t a n c e s . Consequently, the elimination of fretting corrosion
by nitriding and the compressive stresses thereby induced in the surface, increased the bending fatigue strength of this
material by 150%.
Once the fatigue strength of a material is known, the de-
termination of the actual loads on individual structural mem-
bers is necessary, and particularly the pattern of the stresses
induced by the loads. An excellent way to ascertain the
direction of major deformation and/or principal stresses, even
in the case of structural members with complicated contours,
is the brittle lacquer method which, as long ago as 1924,
was evolved by Maybach Motorenbau in connection with the
development of engines for the airship "ZR III," the "Los
Angeles," which on Oct. 13, 1924, started for its epic flight
across the Atlantic Ocean.
In this method, the component considered is coated with
a special lacquer, in the surface of which cracks - the so-
called elongation lines - will show up at stresses below the
elastic limit of the structural members. These elongation
lines, originating always vertically to the tensile load, first
appear in the surface area which is subjected to highest
stresses. As an example, Fig. 2 shows the image of elongation
lines on a flat bar under a bending load. In the area under
tensile load (bottom of Fig. 22) the elongation lines are di-
rected perpendicular to the tensile strains, and in the area
of compressive load (top of Fig. 22) they are directed in line
with the compressive strain. When the stress pattern has been
determined, a suitable strain-measuring instrument may be
used to measure the actual magnitude of the stresses.
Practical application may be shown by the images of
elongation lines in a fork connecting rod in Fig. 23. In the
rod on the left hand side of the figure the lacquer was ap-
plied prior to applying the tensile load, and on the right
hand side prior to relieving the load. In this way, the first
and second directions of main stresses are indicated.
subsequent strain measuring with strain gages pasted perpen-
dicularly to the elongation lines, the stress distribution in
the connecting rod can be determined. Local stress peaks
thus found can be eliminated or reduced to acceptable lim-
its by suitable modifications in design.
Due to the smallness of present-day strain gages it is easy
to measure the stresses even in lubricating oil holes having,
for example, a diameter of only 0.315 in. The effects of
such holes in crankshafts have been determined by the May-
bach stress laboratory (10). As may be seen from Fig. 24,
stress concentrations are best removed by widening the hole
in a slightly conical form. In spite of this, however, round-
ing off the rim of the hole will still be necessary to obtain a faultless surface. Care must be taken to ensure that the
location of the hole subjected to maximum stress is not on
the outside of the rim, but somewhat more to the inside.
Thanks to the high standard of modern measuring tech-
niques it is also possible to determine the stresses in the
crankshafts of engines in motion. In this respect, investi-
gations were made at Friedrichshafen on a 16-cyl MD en-
gine. Strain gages were applied to each web of the crank-
shaft, and the leads were brought to the outside through slip
rings. By way of an automatically operating multiple switch
gear these gages were connected with an oscilloscope, which
filmed the load diagrams of the individual strain gages in
rapid succession. At the same time, by means of a torsional
vibration gage developed by the Works, torsional vibration
amplitides were measured at the free end of the engine crank-
shaft and likewise filmed on an oscilloscope.
The load diagrams established at the crank webs are so-
called "cumulative curves" of the local crankshaft loading.
These curves were evaluated under two aspects: (1) the oc-
currence of maximum loading, (maximum crankshaft torque
in regard to magnitude and location), and (2) by harmonic
analysis, in order to obtain a comparison between the mea-
sured local load and the results of the torsional vibration computation supplemented by the measured vibration am-
plitude at the free end of the shaft.
In Fig. 25 two series of typical load diagrams are shown
for the eight throws of a 16-cyl V-engine crankshaft (crank-
pin 1 on the driving end). During this measuring procedure
the engine output was 2500 bhp at 1800 rpm. The diagrams
to the left of the schematic shaft were plotted for the shaft
not fitted with a torsional vibration damper, whereas those
directly to the right of the crankshaft were plotted for an
assembly fitted with a very effective type of viscous damper.
The test record sbove each series of load diagrams repre-
sents the corresponding torsional vibration amplitude mea-
sured at the free end of the crankshaft. When examining
the illustrations it must be borne in mind that owing to vary-
ing degrees of sensitivity of the strain gages, the scale for
the amplitude in the load diagrams is not the same for all
t r o w s . The diagrams at the extreme right represent the
summation of the crank torques as calculated from the in-
dicator diagrams of individual cylinders. The scale of the
amplitude of the calculated torque diagrams always coin-
cides with that of the corresponding measured-load diagrams,
whereas the scale of the abscissae is considerably compressed.
It appears from the comparison that the measured torques
in the throws, whether or not a damper is fitted, are actually
smaller than those computed. In addition, there is evidence
that the stresses when a damper is used are smaller still;
and the shape of the load diagrams for the shaft having an
efficient damper will thus be approximate to the shape of the
corresponding calculated diagrams.
In order to give an idea of the stress distribution along
the shaft, the maximum positive and negative stresses se-
lected from the corresponding load diagrams (Fig. 26) have
been plotted against the length of the shaft for operation
both with and without a torsional vibration damper. It will be noted that maximum torque occurs in the middle of the
shaft. For comparative purposes, the calculated torques
have been shown as well.
From this investigation one may appreciate that with an
installation having favorable torsional vibration character-
istics, the installation of a damper may bring little further
reduction in the stresses beyond those shown in Fig. 26: on
the other hand, lack of a damper in an installation with un-
favorable characteristics may bring higher stresses than those
indicated in the figure.
The amplitudes of the torsional vibrations at the free end,
and the load diagrams measured at the crank-throws, were
analyzed electronically in order to check the procedure by
w h i c h t h e s t r e s s e s d u e t o t o r s i o n a l v i b r a t i o n s c a n b e e s t i -
mated. Operation without a damper resulted in greatest
loads occurring close to the third natural frequency; for this
resonance range the measured and computed load ampli-
tudes of the harmonics of higher orders are plotted in Fig.
27 against the shaft length. Here harmonics of the 4th and
6.5th orders showed actual values up to 145% above those
calculated, whereas for all other orders the values deter-
mined were lower than those calculated. It must be borne
in mind that the correspondingly analyzed components of
the forces arising from gas pressure and inertia are not con-
tained in the calculated amplitudes of the harmonic oscil-
lations, but are included in the analyzed amplitudes actually
measured. As regards order of magnitude, it may be ap-
preciated from these measurements, an approximate deter-
mination of the stresses in the crankshaft can be obtained
by the vibration calculation for the case of resonance, sup-
plemented by the measurements of the vibration amplitude
at the free end of the shaft.
An electric indicator of Maybach's own design with a
seismic system is used to measure torsional vibrations. It
consists of a heavy symmetrical mass which revolves to gether with the engine crankshaft. The mass itself is con-
nected to the crankshaft by means of four spiral springs and
a damping device. Whenever there is any torsional vibration
of the crankshaft, the mass will maintain its uniform rotary
motion and the crankshaft will thus move against the mass.
This movement of the crankshaft causes a self-inductance
variation in coils forming a bridge fed with alternating cur-
rent. Supply of feeding voltage and pick-up of measuring
voltage takes place through a slip ring transmitter. The tor-
sional vibration indicator is operated over-critically. By
adequately tuning both mass and springs its natural frequency
could be brought down to 5 cps.
It is an advantage of this type of torsion indicator that
observation is possible during the test, so that increase in
vibration limits can be noted at once, and the test values
recorded remotely.
Pressures in the cylinders of high-speed engines are pref-
erably measured today by indicators based on the piezo-
electric principle. Very exact measurements are possible
with such indicators, provided that the length of the hole
connecting cylinder and indicator is shorter than its diame-
ter. Should this not be the case the test results will be grossly
falsified due to induced compression waves in the con-
necting channel. Fig. 28 shows a typical test record of an
8-cyl Maybach MD 440 engine operating at 1000 bhp at
1500 rpm where five items are recorded simultaneously: noz-
zle lift in the unit injector, plunger stroke in the unit in-
jector, pressure distribution in the precombustion chamber,
pressure distribution in the main cylinder space, and crank
angle marking. Valve lift and stroke of the plunger were
measured directly in the unit injector by inductive means,
the two pressure diagrams with the aid of piezoelectric quartz
indicators installed in the cylinder head, and the crank-an-
gle marking was obtained by means of make-and-break con-
r a c i s . All five test results were transmitted to a reed-type
oscillograph where they were recorded photographically. The diagram of the nozzle valve lift in Fig. 28 clearly
reveals with the marked kink the actual commencement of
injection (8.5 deg btdc). Regarding pressure distribution in
the precombustion chamber, a rapid rise approximately 4
deg crank angle after opening of the nozzle valve can be
recognized. This 4 deg crank angle, corresponding to 0.5
millisec, represents the ignition delay of the fuel.
When comparing the pressure distribution in the precom-
bustion chamber with that in the cylinder it will be readily
observed that during the process of combustion there is com-
pression-wave action between precombustion chamber and
cylinder. This, however, is less perceptible in the cylinder
than in the precombustion chamber.
A further example of measuring instruments which have
been employed for the progressive development of Maybach
engines is the air-speed indicator of the firm's own design.
This device can be arranged immediately before the inlet
valve in the intake pipe and is used to measure the speed
and the direction of air flowing through the inlet valve. In
this way, it is possible to supplement the pressure measure-
ments in cylinder, exhaust and intake pipes carried out to
determine optimum scavenging conditions. In principle, the
air speed indicator consists of a small baffle plate which, by
a torsion rod, is connected with two small soft-iron plates;
these in turn are arranged in the air gap of inductive coils.
Under the action of the air flow impact on the baffle, the
small soft-iron plates are displaced, whereby self-induct-
ance variations are set up in the coils forming the bridge cir-
cuit of a carrier frequency amplifier. The self-inductance
variations (proportional to the variation in air speed) are
transformed by the amplifier into changes in voltage which
are recorded by an oscilloscope. The natural frequency of
the air speed indicator is above 100 cps. Damping of the
coil system is achieved by filling the latter with a viscous
damping compound.
General Description of MD-Engine Range
Having dealt with the history of Maybach Motorenbau and with the techniques available today for continuing research now given on and development of these engines, certain deails are now given on the progressive development of the MD-engines.
The experience gained with high-speed diesel engines since 1923, and particularly tests with powerful diesel engines in the speed range of 2400-2600 rpm, eventually led to the Maybach tunnel engine of the MD type running generally at a speed around 1500 rpm. Like earlier Maybach engines, the MD tunnel-type engines are single-acting 4-stroke engines. Ranging from 4- to 16-cyl, they are built either as in-line engines or as V models, naturally-aspirated or turbocharged. Outstanding design features of the MD engines are:
1. Forged one-piece disc-webbed crankshaft carried in a single piece tunnel housing.
2. Pressure-oil cooled aluminum pistons (forced circulation cooling system) with detachable steel crowns.
3. Individual cylinder heads with central pre-combustion chamber, and six small well-cooled valves, with automatic backlash adjustment.
4. Unit injectors for each cylinder.
5. Tunnel crankcase fabricated of steel castings, steel plates and profiles.
6. Large number of repetitive components for the whole engine range.
These points will now be discussed more in detail.
Disc-Webbed Crankshaft and Tunnel-Form Crankcase - In the MD-design, almost the entire length of the engine is used for main and big-end bearings; the crankwebs, which in a conventional engine must be regarded as a loss in this
respect, are here encircled by roller main bearings.
results in the following advantages:
Short engine length with minimum distances between cyl-
inders - this is of great advantage regarding elastic defor-
mation and vibration problems.
Considerable reduction of bending stresses to which the
crankshaft is subjected, due to the crankwebs forming the
Particularly good conditions for all crankshaft bearings
as these are generously dimensioned.
Wide connecting-tod bearings of width nearly equal to
diameter are used without danger of edge loading. There-
fore liquid film lubrication can be assured under all oper-
Due to the comparatively small dimensions of such a disc-
webbed crankshaft, questions of material, flow of fibers,
surface hardness and quality are easier to solve than they
are for larger-sized crankshafts, especially of slow-running
engines.
As a result of its rugged construction and particularly of
its shorter length a disc-webbed crankshaft is at least 40%
stiffer than a conventional crankshaft having a pin of the
s a m e diameter. In order to demonstrate this effect more
clearly the natural frequencies of a conventional crankshaft
and a disc-webbed shaft have been compared in Fig. 29.
The two shafts are illustrated in Fig. 30. The conventional
crankshaft of the Maybach GO engine has a stiffness between
two adjacent piston masses of c = 18.2 × 106 in-lb/rad. Thus the natural frequency of the engine alone, with flywheel
but without vibration damper, becomes n,= 5870 min"!.
Within the speed range between 500 and 1400 rpm the 6th
order which is very strong is located at 980 rpm. On the
other hand, the disc-webbed crankshaft of the rebuilt GTO
MD engine has a stiffness of c = 53.2 × 106 in.-lb/rad
and a natural frequency of the engine alone without vibra-
tion damper of ny = 10,550 min"!.
With this high first nat-
ural frequency the 6th order lies outside the working speed
range, and since higher orders do not need to be taken into
consideration, the whole range from 500 to 1400 rpm is free
from resonance.
The advantage of the disc-webbed crankshaft design (11)
is not as easy to see for torsional loads as for bending loads,
because transfer of force from the crankpin to the web can-
not be avoided even in a disc-webbed shaft. Nevertheless,
the disc-webbed crankshaft still has superior form factors Ok
related to torsional loads: ax = 2.32 for the disc-webbed
crankshaft as compared with 0x = 2.79 for the conventional
crankshaft (Fig. 33).
Form factor refers to the ratio between the stresses actually
measured in the fillet and those calculated for the pin.