tem of governor regulation, however, it is
possible, considering again for a moment the case of three machines in
parallel, by decreasing the sensitiveness of one governor only, to
accommodate nearly all the total variation in load by means of the two
remaining machines, the unresponsiveness of the one governor to change
in speed maintaining the load on that machine fairly constant. By this
method, at any rate, the variation in load on any one machine can be
minimized down to, say, 3 per cent, either side of the normal full load.

There is another and more positive method by which a perfectly steady
load can be maintained upon one machine of several running in parallel.
This may be carried out as follows: Suppose, in a station having a total
capacity of 20,000 kilowatts, there are three machines, two of 6000
kilowatts each, and one of 8000 kilowatts, and it is desired to carry
out a steady full-load test upon one of the 6000 kilowatts units.
Assuming that the test is to be of six hours' duration, and that the
conditions of load fluctuations upon the station are well known, the
first step to take is to select a period for the test during which the
total load upon all machines is not likely to fall below, say, 8000
kilowatts. The tension upon the governor spring of the turbine to be
tested must then be adjusted so that the machine on each peak load is
taxed to its utmost normal capacity; and even when the station load
falls to its minimum, the load from the particular machine shall not be
released sufficiently to allow it to fall below 6000 kilowatts. Under
these conditions, then, it may be assumed that although the load on the
test machine will vary, it cannot fall below 6000 kilowatts. Therefore,
all that remains to be done to insure a perfectly steady load equal to
the normal full load of the machine, or 6000 kilowatts, is to fix the
main throttle or governing valve in such a position that the steam
passing through at constant pressure is just capable of sustaining full
speed under the load required. When this method is adopted, it is
desirable to fix a simple hight-adjusting and locking mechanism to the
governing-valve spindle. The load as read on the indicating wattmeter
can then be very accurately varied until correct, and farther varied, if
necessary, should any change occur in the general conditions which might
either directly or indirectly bring about a change of load.

Preparing the Turbine for Testing

All