178 NEW DEVELOPMENTS IN HIGH VACUUM APPARATUS. 



Assumed that a single-stage air ejector is designed to produce a vacuum of 

 2-inch Hg. absolute. When starting this ejector, the absolute pressure at the air 

 intake is the same as that at the air discharge, namely, atmospheric pressure. The 

 steam will leave the nozzle with considerably less velocity than that which it would 

 obtain under normal operation. The efficiency of the nozzle will be decreased, and 

 the entrainment surface (outer surface of the steam jet) will be decreased. The 

 air to be entrained is in a less rarefied state, hence the loss due to the impact dur- 

 ing entrainment is also greater. Consequently the velocity of the mixture is smaller 

 and its volume is larger. As this mixture has to pass through the throat of the 

 diffuser, the sectional area of the diffuser would have to be enlarged for starting 

 conditions. If this is not done, starting will become impossible. Assumed that the 

 steam jet after leaving the nozzle does not touch the diffuser walls, but leaves an an- 

 nular opening or channel between the outer surface of the steam jet and the inner 

 walls of the diffuser through which the air is entrained (as shown in Fig. 4, Plate 

 70), a vacuum will be produced which will gradually increase with the decrease of 

 the losses. The velocity of the steam will increase with the vacuum. This in turn 

 will retard the enlargement of the diameter of the steam jet. The increasing dif- 

 ference of pressure in the diffuser will recede farther towards the diffuser throat. 

 And if the designed vacuum is obtained before the steam jet touches the diffuser 

 wall, the ejector will start spontaneously. 



If, however, the jet touches the diffuser walls before the designed vacuum is 

 obtained, the entrained air cannot pass. It will re-circulate, as illustrated by the 

 arrows in Fig. 5, Plate 70, and the ejector will never start. This will happen if a 

 single-stage ejector is designed for 2-inch Hg., because its diffuser throat area will 

 be entirely too small for starting conditions. 



If increased in sectional area to overcome those, the section will be too large to 

 be filled out with the co-mingled steam and air at 2-inch absolute, and atmospheric 

 air will rush back into the entrainment space and the desired vacuum will not be 

 obtained. 



To overcome this difficulty such a single-stage high-compression ratio ejector 

 would have to be provided with either a flexible diffuser or special arrangements for 

 starting. 



In all ejector designs, whether for low or high compression ratio, the above 

 facts have to be given careful consideration. The diffuser throat is generally en- 

 larged over the required size for design conditions. This enlargement represents a 

 compromise between diffuser losses and the self-starting capacity. The latter is 

 more important than high efficiency. In practical installations vacuum conditions 

 vary, and lack of this self-starting characteristic will cause the ejector to stop 

 working, which may have serious consequences. Experiments have shown that the 

 maximum ratio of compression for a single-stage ejector working in connection 

 with condensers should not exceed i to 7. In other words, when it is desired to 

 have more than 255^-inch vacuum with 30-inch barometer, it is advisable to arrange 

 two ejectors working in series and divide the ratio of compression. All high- 



