W.S. Graff Baker, Esq., A.C.G.I., B.SC., M.INST.T., M.INST.L.E.

Chief Mechanical Engineer (Railways) London Passenger Transport Board

and President of The Locomotive & Carriage Institution

I want to talk to evening about a subject which nobody can teach; about something that is in all of us which can be developed; about something which is essential in engineering. I refer to the question of design. I do not want to talk about stresses in materials, volts and amperes, or pounds per square inch, hut rather about the form and shape of things and their relation to their purposes. I will assume that in the course of your experience or studies you have learnt the whole of the theory of structures and machines and that you know how to make the various parts, but this knowledge alone will not produce good design.  To take an analogy, the artist must know good deal about anatomy before he can properly draw pictures of people or animals, but knowledge of anatomy is not likely by itself to give him a sense of composition or to result in drawings of beauty. Take, too, the question of architecture. It is quite easy for an architect to construct a building which will not fall down, provided he knows something about the principles of mechanics and the strength of materials, but it is rather exceptional to find a building which pleases and satisfies the eye, and relatively exceptional, too, to find a building which is functionally satisfactory in its arrangement.

What can be learned by study or practical experience is the anatomy of engineering and will not relate to the appearance or even, in fact, to the practicability of a design. I must insist that engineering design is an art just as much as is the work of a painter or an architect.

I have already said that this question of form in design lies within us all, hut it is a latent gift which requires development. It requires more than that, it requires a certain wish to develop it, and, frankly, it makes engineering design very much more difficult. It is easy enough to produce a machine that will work. It is not easy to design a machine that also looks well. One aspect of the question is that of simplicity. It can be said without any possibility of contradiction that if a machine is complicated in proportion to its function it is wrong; that is to say, that nothing but the utmost simplicity for a given function is right. It is no use expecting an automatic printing- press to be simple because its function is complicated and it follows that the assembly is complicated also. That is not to say that the functions of the separate parts of the machine should not be done each in its simplest way. On the other band, of course, it is difficult to separate the individual I functions of a complete machine and perhaps a little dangerous, inasmuch as by correlating them and finding mutual simplification rather than individual simplification the total result will be better.

Then in the question of simplicity in design there is the influence of the T –square - in other words, of the right angle. It is convenient to draw with a ‘T-square and a set-square, or with a draughting machine based on the right angle but what should we worship the right angle to the extent that we do? The shortest distance between two points is a straight line, not two lines at right angles. If you will look at engineering designs generally you will find that they are branded with the sign of the T-square. It does not matter in the machine shop today whether two things are drawn to he at right angles or not. Modern machine tools and marking-out appliances are perfectly capable of dealing with any angle, and, in act, one angle is not particularly to be preferred to another. I have from time to time threatened half-seriously to take away the T-square from my own draughtsmen. The net result of persistent effort in this matter has been an improvement. I might illustrate this in a very simple way. It was necessary to fit a hand—brake on some new cars. The handbrake wheel in the motor cab was determine in position by certain structural considerations in the partition on which it was mounted. The position of the pull rod on the underframe of the car, which ultimately operated the brakes, was also determined by structural considerations. The centres of these two points were hot above one another but displaced some six inches horizontally. The first design that was put up was one in which a vertical shaft, driven by a bevel gear from the hand wheel, was projected down and connected across the displacement distance by another little shaft and two more pairs of bevel gears. The obvious solution once one got away from the fetish of the right angle was to arrange the shaft from the hand wheel not vertically, but so inclined that it joined directly by bevel gearing from the hand wheel to the pull rod. The net result was the avoidance of one pair of bevel gears, the avoidance of two bearings and one shaft, and an increase in mechanical efficiency. This is a simple application of the principle of simplicity in design, and, I think, as clear an example as any of the advantages of getting away from the right angle.

Then there is the question of not being hidebound on principle. Because a certain part can best be made in a certain way is not by any means an indication that that method is suitable throughout the design. Taking, for example, roller hearings for axle boxes, these are common practice today and for various reasons. They are not adopted necessarily to reduce friction. After all, there is no finer hearing than an oil-film bearing, and at high speeds it is generally accepted that an oil bearing is better than a roller bearing. The question of hot axle-boxes has been satisfactorily dealt with by most railway companies, either on the basis of improved lubrication, or, in the case (if heavy services, by also providing really adequate surfaces. The cost of a roller bearing axle-box is higher than that of a plain bearing axle-box. The advantage, as it happens, comes from the fact that there is no wear on the axle journal and that the life of the axle will consequently he improved, as the axle will not have to be scrapped simply for wear before the end of its natural fatigue life. Generally speaking, in heavy service the end thrust on a plain journal axle-box is taken at the end of the axle. Experience shows that with this type of axle-box there is no wear at the end thrust bearing either of the axle or the hearing itself, so that there seems to be no point whatever in taking the end thrust load on an anti-friction bearing. The result of this is to simplify very considerably the design of the roller hearing axle-box by arranging that the vertical load only is taken on rollers; in other words, what is in my view the best axle-box is not a pure roller bearing axle-box but a combination of roller and plain bearings, each principle being used in accordance with the actual needs of the case.

The effect of welding on the construction of rolling stuck is very marked. It results in a lighter and more rigid construction, but it has the not inconsiderable disadvantage that it is much slower in construction than the old method of riveting. It is true that the riveted construction means that all the pieces have to have a large number of holes drilled to take the rivets, but on relatively small parts of the underframe this is done in a shop specially arranged for the purpose, and not much floor space is occupied until the frame begins to be assembled, which by riveting is done comparatively quickly. With a welded underframe the whole of the floor space occupied by the frame is occupied the whole of the time the welding is in progress, and the welding of a large frame is a matter which has to be dealt with piece by piece to avoid distortion of the whole. As a result, when a considerable number of cars require to be produced, it is found best not to weld the whole frame, but to rivet together parts of a reasonable size, which have been previously constructed by welding. This, again, shows that it is necessary sometimes two adopt two mutually supporting principles rather than to ask either principle to carry the whole weight.

The aspects of the subject of design are innumerable and the methods of dealing with the problem are almost equally varied.  There is the trick, if you like to call it so, of turning a design inside out.  If a design is unsatisfactory, this is one of the most certain ways of attack and the result is commonly right.  As a case in point, I could cite something of my own experience, that of the design of a tripcock for a train  .A tripcock is a valve, which is opened by an attachment fixed at the signal, so that if the train over-runs the signal at danger the brakes are applied.  As the engagement between the arm on the train and the arm on the track may vary, it is desirable that this valve should be self-opening once it has started to open.  The slide on the screen shows how this valve was originally designed.  You will see that it has right in the middle of it a stuffing box, so situated that its dimensions can be barely adequate for its work, and so that any necessary adjustment can he achieved only by knocking round the gland nut with a chisel.   The other side of the slide shows the same valve turned inside out.  You will see that the need for a gland has vanished, the valve works equally well, is simpler, is easier to look after and altogether more satisfactory.

The next slide shows you an example of inverting design in cars.  With the introduction of sliding doors in saloon type cars it was found necessary to provide draught screens each side of the doorways to protect the passengers in the seats from the draughts. from the open doors.  In view of the large doorway opening that was cut, particularly in tube cars, it was found desirable that the draught screen should form part of the basic structure of the car in order to reinforce the structure.  To this end, as you will see from the slide, it was built up as a steel framed panel with windows.  To the front of the frame was secured a grab-pole, and up the side of it was run an air pipe leading to the passenger emergency valve.  The result is clumsy looking and not particularly cheap, but merely perhaps the obvious way of doing the job.  Consideration of this design revealed the fact that all that was needed in the way of structural strength of the car was a vertical stanchion taking the compression stresses in the body due to the shear.  It was necessary, too, for there to he a screen for the passengers and some way of dealing with the air-pipe.  The result was the design shown on the slide.  The grab-pole has become a permanent structural part of the car, made in the form of a tubular strut, which is made adjustable by screwing the end so that it can be made to take up its due proportion of load during the construction of the car.  The fact that it was a tubular strut made possible its use as a conduit for the air, and the draught problem was taken care of by the use of a plain piece of glass supported between horizontal members and the car side.  This newer type of draught screen fulfils all the requirements that I have laid down, and not least of all the requirement that it should look well.  The adoption of this design in place of the other has the effect of increasing the apparent size of a car which is admittedly small and which cannot afford any restrictions in its appearance.

To take another case in car design, it has until recently been the practice to fit a footboard along the sides of those cars on the London Transport services which do not run in the tubes.  The reason for this footboard is to provide a means of preventing accidents due to passengers getting their feet down between the car side and the platform, reasons of clearance rendering it necessary that the car should not be of sufficient width all the way up to prevent this difficulty.   It is only practicable to have a projecting piece at the low level.  Such a footboard tends to collect dirt and makes the car more difficult to clean.  Recent cars for this service have been constructed with the sides swept out in a curve at the lower region to the same point as the outer edge of the old footboard.  This gives a better appearance to the car and provides for the necessary functional requirements of operation, while leaving a continuous surface which not only does not get so dirty but is easier to clean.

On the electrical side, too, simplification can he made on the basis of practical requirements.  For example, it has been customary to fit hand-operated main switches on electric rolling stock, and until recently the desirability of this has riot been questioned.  A main switch is electrically a very short distance from the main circuit breaker, line switch, or first contactor.  The effect of opening the switch when current is off is merely to isolate the short piece of cable running from the main switch contact to the circuit breaker, etc.   Hand-operated main switches are not designed to interrupt flowing current.   The amount of isolation secured by opening the switch is negligibly more than arises from tripping the circuit breakers or the line switch relays, which is invariably arranged to he done by remote control.  It will thus be seen that the main switches have no appreciably useful function.  Repair work is seldom done on a train with the current on, and when an electric train is in the car shed or repair shop it is not standing on current rails.  The less electrical apparatus one has the better from the points of view of first cost, maintenance expenses, and liability to breakdown.  It would therefore seem desirable to omit main isolating switches, and on recent rolling stock this has been done.

Careful examination of the functions of wiring generally has led to certain changes.  For a long time it was customary to run positive and negative wires in separate conduits on rolling stock having an insulated return.  Systems of this kind are subject to the circumstance that, in the event of one pole grounding, there is not a short circuit created between the poles, as the current from the ground has to get back to the other pole by the insulation resistance of that other pole.  This means to say that the current is limited by the insulation resistance of the whole of the other pole throughout the system to a figure, which, while not constituting a short circuit, will create a very destructive arc.  It therefore appeared desirable to change the system by which wires were run in separate conduits and arrange for wires of a given circuit on opposite polarity to run in the same conduit, so that a ground would develop as quickly as possible into a short circuit and blow the fuses, thus isolating the fault.

The examples that I have cited of modifications, improvements and simplification in design are all drawn from my own practice and experience.  If I might cite one simple thing outside my own experience, I would refer you to the design of driving mirrors for motorcars.  Generally speaking, these are a very crude job.  I have slides showing what I would regard as an example of good design in this connection, and what I would regard as bad design.

In the matter of design there is always the question of considering whether a thing should he made adjustable or not.  Generally speaking, I am averse to making anything adjustable if it can he avoided.  Adjustments are almost inevitably made by screws, and, in the nature of things, screws used for adjustment cannot be properly locked against working loose.  The ordinary principle of a lock nut or a setscrew is simple and cheap, but mostly unsatisfactory.  It is far better to design so that adjustments can be avoided altogether, or, if they have to he made, to arrange for them to be made by inserting shims or slim pieces of metal between the parts that need taking up for, say, wear.

The fastening of things together is another difficult matter.  Where things require to be permanently fastened together, it is well to consider whether the apparatus could not be made in one piece; next, whether the parts could be welded together; next, whether they could be riveted together; and, finally, whether it is essential for them to be bolted together or fastened with screws.   Set-screws are things to beware of, as they are difficult to lock.  If they are undone a few times the thread in the underlying portion may become worn and require re-tapping or the portion may even need scrapping and replacing.

I see I am getting down to details, and this is a thing I wanted to avoid. I want rather to keep to main principles.

A vast deal of work is done in making drawings of things that are unduly costly when it comes to manufacture or maintenance.  Many draughtsmen appear to be satisfied when they have made a drawing of something that will work.  This is underrating the object of making a drawing.  Perhaps I should have used the word ‘design’ rather than the word ‘ drawing’.  A designer should consider these points:

1.        Will it work?

2.        Is it as simple as possible?

3.        Could it easily be maintained in service?

4.        Can it be manufactured?

5.        Does it look well?

If these points have been taken care of in the design there is little doubt that everyone will be satisfied.  If any one of them is omitted, the design is in fact poor, perhaps useless.  A drawing is commonly a picture of one of a large quantity of things to be made.  Generally speaking, it would pay to spend very much longer on the design on the drawing board than appears to be spent usually, and not only to spend more time, but more thought and more care and more energy in insisting on getting the right answer.  Once a drawing is made, the articles are many and may last for years.  A drawing is perhaps made in ten hours; each of the articles it represents may take, say, fifty hours to make and last fifty years.  The draughtsman sees the thing once and never again, but someone has to look at, live with, and maintain the finished articles through the whole of their life.  It does not do, even from a material point of view, to hurry or skimp drawing office and design work, while from aesthetic point of view it is invariably a tragedy.

All the examples I have given you indicate quite clearly, I think, the kind of lines on which I feel that engineering design should be studied.  You cannot go wrong if you aim at the utmost simplicity, and if you design with a view to making an article that is not only finished but also looks finished, and a complete whole and not made up of little hits stuck on one another as if by a series of afterthoughts.  Do not he in a hurry to adopt your first thoughts - they are seldom right.  When you are creating something (and this is what designing always aims at), something that will perhaps last longer than you will, it is surely worthwhile to spend time and keep on trying to produce the best thing and not to be content until you are satisfied and feel that everyone else who sees it or uses it is satisfied, too.

October, 1938.

- End -

.