THE TURBINE DRIVEN CENTRIFUGAL PUMP PROBLEM

The problem of directly connecting the steam turbine to drive the centrifugal pump arises from the great disparity between the spouting velocity of steam and the peripheral velocity required of the pump impeller. Steam discharged from ordinary boiler pressure to a high vacuum attains a velocity of about 4,000 feet per second. The buckets of an impulse turbine should, for the best efficiency, run at a little less than half the steam velocity which in a single-stage turbine, gives rotative speeds of 10,000, 20,000 and even 30,000 r. p. m., depending upon the capacity. The rotative speed may be reduced by various methods of multi-staging, but there is a limit to the speed reduction that may be obtained with high efficiency. For each turbine output there is some one rotative speed which gives the highest efficiency. For a turbine developing 1,000 shaft horse-power, this speed lies between 4,000 and 5,000 r. p. m. To reduce the speed below this with a given number of wheels will cause the duty of the turbine to fall off because of high velocity in the steam as it leavees the buckets, while to reduce the speed by adding more stages will increase losses by steam friction unless the diameters of the wheeels be reduced, which reintroduces leaving losses. The use of a large number of stages also increases the first cost of the turbines. Suppose a 1,000-horse-power turbine operating at 4,500 r. p. m. is to be directly connected to a centrifugal pump delivering water against a head of 200 feet. The peripheral velocity required of the impeller of this pump will be in the neighborhood of 115 feet per second, and as the capacity of the pump will be about 30,000 gallons per minute, the speed will be approximately 650 r. p. m., that is, about one-seventh of the turbine speed. By using two pumps, each of half the capacity, this speed can be increased in the ratio of the square root of 2 to 1 or 1.4. 1.4 times 650 is 900 r. p. m. Or by using three impellers in parallel, it may be increased in the ratio of the square root of 3 to 1, or to 1,100.

As this is still considerably short of the best turbine speed, the only way remaining, if the turbine is to drive the pump directly, is to compromise between the two speeds, that is, reduce the turbine speed and increase the pump speed. The sneed of the turbine may be reduced by adding more stages, and by running below the most favorable speed for the given number of stages. The speed of the pump may likewise be increased by increasing the entrance velocity and using a smaller impeller. Each has a detrimental effect upon the respective efficiency, as already explained for the turbine. As for the pump, the increased velocity of entrance is secured at the expense of the suction head and this velocity head must be reconverted into pressure head at the delivery. Inasmuch as this reconversion is not perfect, an additional loss of work is imposed upon the pump over what w'ould be necessary were a normal entrance velocity used. Obviously, this method of increasing the pump speed is the more unfavorable the less the total head to be overcome. For moderate and large capacities this method of compromise involves the use of several impellers operating in parallel and running at high speed with high water velocities, which is apt to introduce mechanical and hydraulic unbalancing and vibration, which consequent pitting and corrosion, necessitating frequent renewals. The solution of the speed problem which has been most successful is the use of a mechanical speed-reducing gear between the turbine and the pump. This permits of running each member of the unit at its most favorable speed, having regard solely to its own characteristics. If the turbine be operated at the speed corresponding to its maximum efficiency, it is possible to vary the speed considerably, while affecting the steam economy very little. This feature is useful where a pump is to be operated at different rates of delivery, as will be apparent from the following: In the figure herewith are shown characteristic curves of a centrifugal pump running at three different speeds. Taking for instance the curve marked 1,700 r. p. m., it will be seen that at no delivery the pump generates 130 foot head, but as the delivery is increased, still maintaining the same speed, the head generated becomes less, until at the point of maximum efficiency, it is about 95 feet. As the speed is changed, the delivery at which the highest efficiency is obtained varies directly with the speed, while the head varies as the square of the speed. At 1,500 r. p. m., the maximum efficiency is obtained at a delivery' of a little less than 1,500 gallons per minute and a head of about 75 feet, while at 1,200 r. p. m., the maximum efficiency point is at 1,200 gallons per minute with a head of 48 feet. If the resistance against which the pump delivers is made up wholly of frictional head, this relation is very fortunate, since the frictional head increases as the square of the rate of flow. In other words, as the speed is increased, the volume delivered will increase as the speed and the head generated as the square of the speed, so that the head will be just sufficient to overcome the frictional resistance. If the load consists largely of static head, however, as when pumping into a reservoir, there will be comparatively little variation in the head to be overcome at different rates of delivery. We may suppose, for instance, that the pump represented by the curve of the figure herew-ith will be required to deliver 1,900 gallons per minute against 80 foot head, which it will do when running at 1,700 r. p. m. with an efficiency of 71 1/2 per cent. If now only 1,550 gallons per minute are required against the same head, it may be obtained by dropping the speed to approximately 1,550 r. p. m. with an efficiency of 75 per cent. By dropping the speed to 1,450 r. p. m. the delivery may be reduced to 1,100 gallons, still retaining a pump efficiency of over 70 per cent. In other words, the pump will deliver against constant head at approximately the same efficiency over a considerable range of deliveries and speeds and if the turbine is running at the best speed for efficiency somewhere in the middle of this range, its efficiency will not fall off much at the higher and lower speeds. However, if the turbine were originally running too slow for maximum efficiency, the decrease in speed would effect it seriously, the efficiency falling off almost directly as the speed.

If you are a current subscriber, login here to access this content.

If you would like to become a subscriber, please visit us here.