Hence the number of generating units should generally be reduced to the minimum that will satisfy the following condition, viz.: that the total number should be such that one unit can go out of service without seriously overloading the others, dividing its load among them. Good modern generators usually are able to carry 25 or 30 per cent over their normal rated loads for some hours without serious heating, so that ordinarily four or five generators suffice. If the load upon the plant is normally moderate, two or three generators may serve admirably until such time as the normal full load nears the rated capacity of the machines.
In the case of a transmission or distributing network fed from several stations the total output rather than the individual output of the separate plants should be considered in deter mining the size of units to be used. for as re gards reserve capacity it makes little differ ence whether it is in one location or another. Thus in some of the large hydro-electric net works 8 or 10 plants co-operate and the size of the units in each is determined not by the possession of a theoretical reserve capacity in a particular station, but by considerations of convenience, though some of the stations may have only one or two generators where such an arrangement suits the hydraulic conditions, and any station is called into or out of action by the lc,;:d dispatcher just as if all the gen erators were concentrated under a single roof.
In steam-driven stations past practice has tended toward great aggregation of capacity, but more recently, since widespread distributing networks have become fairly common, the ad vantage from the standpoint of distribution of several generating plants has gradually been recognized and in planning these it is again the total capacity on the system which counts. The vital point is to avoid on the one band putting too large a proportion of the possible output into a single unit, and on the other splitting the output into a number of units too small for the most economical operation. Stations with more than five or six generators are now gen erally relics of an early stage of design, except in certain cases of hydro-electric plants where the hydraulic conditions are such as to intro duce the dilemma of choosing between rela tively small units of normal design, or large units of special and sometimes disadvantageous design. Such cases arise in working at low heads where the available power from a single wheel or pair of wheels is comparatively lim ited. Cases may also arise in which generators for different purposes, as for general transmis sion, railway and electro-chemical service, may have to be installed in the same 'plant. In this
case the number of machines for each use would be determined as if the other machines were located somewhere else. In plants in cluding raising transformers the capacity of these is determined on the principles already laid down for the generators. Recent practice tends toward the installation of a single three phase transformer or bank of three single phase transformers for each generator without further subdivision. As between the two, prac tice is drifting toward the latter on account of the more convenient size of the separate units in large output and from the fact that in case of need two of them can be operated on open delta connection, thus giving an additional ele ment of reserve capacity. Broadly, large sta tions should differ from small ones only in the capacity of individual units, so that if a 2.000 kilowatt plant is equipped with four 500 kilo watt units a 20,000 kilowatt plant would be equipped with four 5,000 kilowatt units and so on.
In either case the same allowance for re serve capacity would naturally be made, but evidently the larger plant would have the advantage not only of greater economy in the larger units but of reduced attendance cost per unit of output, and would also cost less per unit of output.
Large power plants then evidently can pro duce electrical energy at a lower cost than small ones. As the size is increased, however, the relative gain is diminished so that a size will finally be reached at which further increase in capacity ceases to be beneficial. Indeed since a very large station usually implies a large area served, the average ti e distribuon losses will i be reduced by serving it from several intercon nected power houses, so that the maximum economical capacity of a single power house is less than would be determined by considering only the effect of size on fuel and labor.
For example, if a certain considerable dis trict requires 200,000 kilowatts, it is altogether probable that two 100,000 kilowatt plants will serve it more economically than a single gigan tic power house and unless the district is of moderate area, four 50,000 kilowatt stations would probably do even better, provided they could all be well located. In other words a power plant of 50,000 kilowatts or thereabouts comes so near to giving maximum economy of operation that the difference is easily offset by lessened losses in distribution. The only cases in which size may be economically increased without limit are those in which the plants must for one reason or another lie far with out the area served, as in great hydro-electric plants like those at Niagara, or in plants using coal at the mines for transmission of power on a colossal scale.