WALSCHAERTS’ VALVE GEAR DISTORTIONS IN TIMING
Any locomotive valve gear has to consist of two components which are combined so that the valve can operate the two ports of a double-acting cylinder. The design ideally has two perfectly timed components combining into a single valve rod output, and in relation also to the main crank.
Absolute perfection is not available, nor, with a little bit of reflection, is it desirable. The main crank has rotary motion converted by the connecting rod from linear motion of the piston. In doing so, the changing angle of the connecting rod from front dead centre to the piston’s mid position (travelling in forward gear) accounts for less than 90o of crank turn and the piston’s next half travel, to rear dead centre, takes the crankpin more than 90o. A similar upset can occur at any swinging link in the gear, and any drive originating from the crank (eccentric or lever) to feed the valve also carries the timing error of the main crank.
So, the nature of the problem of valve gears is not a requirement to produce perfect timing but to emulate the imperfect timing of the main crank. Any rotary parts (eccentric), or part-rotary (swinging arms or an expansion link) not only have to match the erratic crank but also generate their own errors to compound the problems. This is the complex nature of valve gears so little understood or appreciated, the upshot of which has been too much turning a blind eye against the difficulties and deciding to consider the small hitches as inevitable.
If this presents a hopeless minefield, as it seems to have done inexcusably even in many drawing offices, one should remember that some designs, particularly of Stephenson’s gear, are incredibly immaculate and give events of less than 1% difference between front and rear ports anywhere in the full quadrant range.
The Walschaerts’ drop link, being a fixed link, provides a convenient level from which to drive the combination lever in direct synchronisation with the piston. The bottom pin, acted upon by the crosshead in each half-stroke, produces lap + lead movement of the valve if we so proportion the pin distances of the lever, and the upper pins are reversed for outside admission so that the valve travels the opposite way. As the combination lever’s lowest pin is forced to describe an arc it is necessary to provide a flexible link – the anchor or union link – in the driveline. This union link itself suffers an angularity in a cycle dependent on its length and also that of the drop link, thereby introducing a distortion, and, as it happens, a most useful designer’s means of pitting the quantity of timing distortion of the whole component against that derived from the return crank. It is not an overstatement to say that this is a most critical and sensitive means of adjusting the equality of steam distribution without destroying or compromising the principle of equal leads.
The combination lever and its elements are physically supported either by a valve crosshead or a hanger. Both methods involve a minute distortion at the radius rod joint which is fortunately of little moment.
The other component also suffers its own points of distortion. The return crank and eccentric rod couple suffer the conversion of rotary to linear motion. In addition, they pick up the main crank distortion and the excursions of the expansion link end of the eccentric rod are non-linear. If the expansion link tailpin lies on the gear’s horizontal centreline, like many American designs, the return crank is at 90o to the main crank but the long tail needs a greater return crank pitch circle to produce the required amount of swing unless the trunnion is asymmetrically displaced. If the tail is shortened, as in most cases, the eccentric rod drive is inclined and therefore the 90o is compromised accordingly.
There is no premium to pay for the inclination, except perhaps mechanically. Geometrically the bonus is a smaller pitch circle for a given travel and a decrease in the theoretical angular backset of the tailpin. The smaller pitch circle is of particular advantage in cases of long valve travels and inside admission. This is the primary reason why events with today’s long travel valves are often poorer than those from the old short travel slide valves.
The dieblock has to slip in the expansion link to cope with the circular swing of the link – differently according to the type of reverser arrangement. What is more it is different in each half of the swings. All is felt at the radius rod front end and delivered to the valve. An extended radius rod and slot arrangement, favoured by the LMS, LNER and GWR, generally works well and the designer should aim for a maximum lifting arm angle of about 28o. The mechanical ‘kick’ at each dieblock reversal may be more harmful beyond this limit.
The front pin of the radius rod inputs all these distortions to the valve via the combination lever and its representation of the other component. Where the rear of the radius rod is supported by a lifting link, or hanger, the die slip springing from the expansion link swings becomes modified by the swing of the hanger, be it ahead or astern of the expansion link. Designers should dimension these parts so as to counter other disturbances and this may involve a change of backset also. Clearly there is an opportunity to ameliorate difficulties or compound them according to the skills of the designer. The number of variables is such that a mathematical solution cannot easily address each individual case. The essence of the design is to controllably vary the depth in gear through a cycle to equal out the distortions.
The effects on the steam distribution need little imagination. Walschaerts’ gear, though fairly straightforward in concept, invokes much more skill in detailed design and the interaction of its elements than does Stephenson’s gear. It is small wonder that the gear was poorly understood in its early days and frequently underachieved much later with long travel valves.