ANALYSING EXISTING DESIGNS
The analysis of existing valve gear designs provides valuable information on several counts. It provides evidence of the skills employed by locomotive engineers, shows by direct comparison the efficiency and equality of the steam distribution available both from different gears and different dispositions of the same basic type of gear, and reveals deficiencies and their causes.
The primary source of information has to be an original motion drawing, though this may or may not be an accurate representation of what was actually produced in practice. Interpretation has to begin with the nominal parameters stated on the drawing – nominal lap, lead, full gear valve travel and cut off. Frequently, drawings are deficient in supplying all the dimensions and sometimes there are conflicts requiring resolution. The most common occurrence concerns apparent assumptions that the draughtsman’s pencil line joins two or more centres which CAD construction cannot support.
CAD reconstruction is the initial step in analysis, quickly finding any geometrical conflict and its exact amount. Resolution will depend on the case in question and sometimes a familiarity with the idiosyncrasies of the gear and reconciliation of the nominal parameters quoted on the drawing assist in that resolution. Neither CAD nor the simulator can ‘hide’ dimensional inaccuracy, but occasionally it is necessary to assume an obscured or illegible dimension in order that simulation input can be completed and this usually leads to verification or otherwise of the assumption.
Swindon drawings are normally in the piston mid stroke position but most favour dead centre. This is particularly necessary in forming Walschaerts’ valve gear, since the position of the front end of the radius rod is indeterminate at other crank angles until the whole linkage has been defined. Transfer to the simulator is straightforward but no results are reliable until valve setting has taken place. This operation soon determines the gear design’s ability or otherwise to support the principle of equal leads. Where this is impossible it will generally be found that valve setting to equal leads simply compromises all the other events, whereat it becomes incumbent on the analyser to reach the best conclusion, just as the ‘real’ valve setter would have to do in practice.
The sole difference is that the probable cause of the discrepancy, be it faulty design or deliberate compromise, may be ascertained by simulation. There is rarely recourse to the original designer in this respect but it is reasonable to suppose that both he and the valve setter, the latter probably working to set drawing office instructions, attempted to produce the best results in any given case.
In some ways the simulator mimics the large valve gear models used in drawing offices to verify a design, though the skills of the respective operators may differ. The stream of readily available accurate measurement by computer assists but does not replace experience. The simulator operator has to build up a knowledge of what to adjust and why. The process may also include suitable translation to apply propensities to a model quite unrealistic in the prototype.
It is a matter of good housekeeping to update the CAD drawing where simulation has shown amendments to be efficacious, either in a model or a prototype, but to retain the analysis as a historical record. Several analyses are appended here for interest and may be augmented from time to time.