Simulation

The simulation of valve gears is not new: many drawing offices had the use of large adjustable frames on which to evaluate new valve gear designs and aid the design in detail. Simulation by computer relieves the tedium of complex mathematics and provides the means of speedy evaluation and adjustment; nevertheless the adherence to principles and the responsibility for skilled attention remain with the user.

Simulators have been devised largely from enquiry into the kinematics and not from a viewpoint of knowledgeable design skills. Some early examples used rather basic formulae and most retain internal interpretations of principle that deny their use for critical analysis. For design purposes this factor also has drawbacks and perhaps the greatest misuse of a simulator is to expect it to have some magical means of initiating the design in unskilled hands. One of the symptoms of this is the failure of any two simulators to agree exactly in detail.

Perhaps the longest established and most prominent simulator programs are those by Charlie Dockstader, Dr.Allan Wallace and Prof. Bill Hall. None of these gentlemen, generously donating their programs to the public domain, claim to be valve gear experts. Dockstader has enthusiastically formed a formidable collection of over 30 gears in an excellent display, a healthy number of which may be used during individual design. Any problems are assiduously and promptly attended. Wallace has developed presentations of Walschaerts’, Stephenson’s and Baker gears in addition to his earlier DOS programs. Ready features enable automatic or manual distinction and the effects of a worn gear may be assessed. Bill Hall’s programs serve both Walschaerts’ and Stephenson’s gears in various forms and are particularly easy to use by the novice. All these programs are direct and convenient replacements for the drawing office large models and as such must be considered as confirmatory, and not design programs. They produce listings of the piston/valve relationships throughout a cycle at any position of the reverser.

Most have built-in constraints, often for good reason, that may curtail pure analysis of existing designs and all place the skills of design firmly with the user. However, the process of evaluation is speedy and the ability to make adjustments with immediate appraisal brilliant. Crucially, there is considerable scope for the users to begin to understand relative sensitivities of adjustments and their effects on kinematic performance, unattainable before the computer age. Steam locomotive engineers regrettably never had such advantages and it is to their credit that so many first class designs were laid down on the drawing board without such aid. Some appear to have designed from pure principle, accepting as inevitable the errant ways of impure geometry, whilst the more astute and skilled knew all the ways of overcoming the little difficulties. The modern designer has little excuse.

The design spreadsheets on the DOWNLOADS page automatically generate most of the requirements of both Wallace and Hall simulators. By this means the basic design work can be quickly accomplished and the simulator fed with sensible parameters ready for fine tuning. Both diagrams and figures then provide immediate verification of any detailed alterations and confirm expected performance, once the user becomes conversant with the program. The Hall simulator automatically sets the valve for equal leads, but with some choice in the case of Stephenson’s, even if the gear design does not support equal leads. The others allow adjustment of the valve on its spindle.

In general the model engineer, because the steam engine is inherently so tolerant, currently underrates the value of the simulator and its usefulness in securing the best possible steam distribution and power output. Perhaps the long history of pursuing the hobby in spite of the lack of valve gear knowledge (see the MINIATURES page) suffuses a natural wish for improvement in this respect. The means are now available and by regular use of a simulator the user can learn a great deal.

The main use would appear to be the checking of a proposed design: it cannot itself design without containing too many constraints that limit the scope of analysis and curtail the ability and skills of the design engineer to work a little outside bald principles where prudent.

Perhaps the greatest asset is the ability to fine tune from the design’s nominal nature to a comprehensive and detailed conclusion. Designers of valve gears have not left a written legacy portraying their skills and in any case may not have been called upon often in a whole career to produce a totally new arrangement. The simulator allows of small adjustments, with due knowledge, and immediate results.

It does, not unexpectedly, require considerable conversancy to understand what alterations may be sensible and their effects on the whole. Regular users will discover the finer points and be able to pinpoint the sensitivities of the variables. This is particularly useful in cases of gear compaction or some physical restraint affecting ideal proportioning. The analysis of existing designs reveals the skill level of the designer and may show why some part of the design appears at first glance to be asymmetric to the untutored eye. Certainly it can reveal how near to perfection the steam distribution can be. For locomotive applications the best is both achievable and obligatory.

There is a viewpoint that the precision of the computer simulator is incompatible with a 19th century invention, but analysis soon shows that excellence was achieved in many old cases and that the use of DRO machinery by the model engineer can be productive of the precision able to render the results given by the simulator.

Manipulation:

The first task, which may appear to be a stupid statement, is to enter the designed figures. Newcomers may not find this simple firstly because one has to understand the parameters, named differently in different simulators. Secondly, some entries require negative values and in addition the user may have to devise a workaround where the layout has peculiarities ‘foreign’ to the given one. The best advice is to aim for a diagram that looks right! Those who have had a go will understand this and realise that a wild diagram is easily produced even with experience.

Some do not realise that in the many cases of inclined cylinders the simulator treats the centreline horizontally, and it is good practice to construct (not ‘draw’) the gear in CAD alongside the process. This verifies dimensional accuracy for the inputs, some of which may have to be gleaned from the CAD construction process – it depends on how the particular simulator wants its inputs.

Don’t proceed until the diagram looks right and valve setting, at least as near as possible, has been accomplished. Now take stock of the results given. Many of the items fed in are nominal dimensions from the design work and will require fine adjustment and valve resetting before any more serious detailed adjusting. At all times the simulator is reacting purely to user input – it has no magic wand to wave over poor design.

Predictably, manipulation within the simulator largely consists of trial and error until the skills are built up. The user will find the equalisation of leads fairly straightforward, but the equalisation of other events much more of a challenge.

Moving the disparate parameters relative to the piston travel can be achieved in a number of different ways or combination of ways, but most of these measures result in producing coincidence only at one specific cut off. Altering the curve of front port activity to fit that of the rear port, or vice versa, is considerably more complex since it involves so many variables with dependency on the initial design elements. Nor are the results of a single alteration always what one might logically expect because all parameters are affected in the process.

As an example, related to actual practice, suppose the test (Walschaerts’) for zero valve movement over the range when the crank is on dead centre reveals a slight movement. In practice the length of the eccentric rod would be altered by the required amount. However, simulation may show that this has rectified the problem at one dead centre but not the other, without giving a clue as to the reason, except that it must be another part of the mechanism (any one of many) which is at fault. It may also be found that the test is verified for the forward range but moves in the reverse range. That there is no definitive answer here shows that the author, too, is still learning after considerable use of the simulator! Rome was not built in a day.

This example is interesting not just because it was regarded as a principle. L.D.Porta discovered that contrary to historic presumption the requirement for lead with long lap employment was in most cases greater at slow speed slogs to prevent pounding than at higher speeds on shorter cut offs. He therefore produced variable lead in Walschaerts’ gear by reducing the return crank length. In effect this single alteration advances the expansion link component in full forward gear and lessens towards mid gear, probably courting negative lead in full back gear. In practice of course, the altered timing will affect all the valve events slightly, since all are shifted in relation to the piston and in different degrees at different states of the reverser.

Because the majority of fine tuning moves will affect both the leads and their equality it pays to check and reset the valve at each alteration, otherwise a true picture is not maintained. Scale plays a part in the decision making, as 0.001″ is of less import as the scale increases.

In spite of this screed now filling three pages of A4 the whole design and simulation process can be completed in a very short time, with more accuracy and guarantee than any other method.