additional work is unnecessary. The same method can
be applied to individual turbines pumping their own oil from a tank in
the bedplate; the return oil, as previously described, being temporarily
prevented from running back to the supply.

The causes of excessive oil consumption by bearings are many. There is
an economical mean velocity at which the oil must flow along the
revolving spindle; also an economical mean pressure, the latter
diminishing from the center of the bearing toward the ends. The aim of
the economist must therefore be in the direction of adjusting these
quantities correctly in relation to a minimum supply of oil per bearing;
and the principal factors capable of variation to attain certain
requirements are the several bearing clearances measured as annular
orifices, and the bearing diameters.

It is not always an easy matter to detect the presence of water in an
oil system, and this difficulty is increased in large circuits, as the
water, when the oil is not flowing, generally filters to the lowest
members and pipes of the system, where it cannot usually be seen. A
considerable quantity of water in any system, however, indicates its
presence by small globular deposits on bearings and spindles, and in the
worst cases the water can clearly be seen in a small sample tapped from
the oil mains. There is only one effective method of ridding the oil of
this water, and this is by allowing the whole mass of oil in the system
to remain quiescent for a few days, after which the water, which falls
to the lowest parts, can be drained off. A simple method of clearing out
the system is to pump all the oil the whole circuit contains through the
filters, and thence to a tank from which all water can be taken off. One
of the ordinary supply tanks used in the gravity system will serve this
purpose, should a temporary tank not be at hand. If necessary, the
headers and auxiliary pipes of the system can be cleaned out before
circulating the oil again, but as this is rather a large undertaking, it
need only be resorted to in serious cases.

[Illustration: FIG. 64]

It is seldom possible to discover the correct and permanent temperature
rise of the circulating oil in a turbine within the limited time usually
alloted for a test. After a continuous run of one hundred hours it is
possible that the temperature at the bearing outlets may be lower than
it was after the machine had run for, say, only twenty hours. As a
matter o