Search

Ferromagnetic shields for solenoidal superconducting magnetic energy storages (SMES) are optimized with respect to the stray field or the mass of the shield. Two optimization strategies of zero order are compared: the deterministic downhill simplex method and the evolution strategy, a stochastic method. Due to the comparison a combination of both methods is tested. To guarantee short computation times the underlying nonlinear field analysis is based on an integral equation method.

Small high power SMES in toroidal and solenoidal geometries are compared. The toroidal 1.4 MJ SMES currently under construction at Technische Universitat Munchen serves as example. The hypothetical equivalent solenoid is surrounded by a ferromagnetic shield, so that the stray field of the toroidal SMES is not exceeded. The shape of the shield is optimized with respect to the mass by using a stochastic strategy in combination with a deterministic method. Finally the superconductor losses and the eddy current losses in the shield are calculated for a sinusoidal load cycle. >

Zur Berechnung rotationssymmetrischer Probleme der Magnetostatik wird die Integralgleichungsmethode verwendet, die nur eine Diskretisierung des magnetisierbaren Materials erfordert. Um kurze Rechenzeiten zu gewahrleisten, werden die Magnetisierung in jedem Element als konstant angenommen und die Fernfeldwirkung von Spulen sowie der magnetisierten Elemente geeignet angenahert. Die erreichbaren Genauigkeiten werden diskutiert und die erforderlichen Rechenzeiten aufgezeigt.

The authors describe a small energy storage plant for experimental tests. The centre of this energy storage plant is a small, fast acting superconducting magnetic energy storage (SMES) of 1.4 MJ. The SMES is composed of six superconducting (SC) coils arranged in a torus. The SC coils are cooled indirectly by supercritical He. To avoid eddy currents as far as possible all important parts are made of glass-fiber reinforced plastic. Calculations show the losses of the small, fast acting SMES during periodical load cycles depending on different charge and discharge times. >

In recent years considerable emphasis has been placed on the application of computer control to substrate feeding in baker's yeast production. Many investigators have studied the dynamics of different control loops and also first steps have already been taken to optimize the control algorithms. As far as the experimental verification of different strategies is concerned, the published work is, however, restricted to laboratory scale only. From current practice in baker's yeast industry which is most often based on open loop control with fixed schedule for carbon substrate addition, it is well known that mixing of substrate is playing an important role. The effects of the dynamics of mixing process in mechanicaly stirred tanks upon the performance of baker's yeast fermentation with fixed schedule have been investigated earlier by using a circulation model (Bajpai and Reuss, 1982). In the present contribution this framework is used in conjunction with a closed computer control of carbon substrate feeding based on exhaust air analysis. From experimental results for computer controlled fermentation in 10 and 2 000 l scale a decrease of productivity with increasing circulation times of liquid is observed. These results can be confirmed with the model calculations.

In recent years considerable emphasis has been placed on the application of computer control to substrate feeding in baker's yeast production. Many investigators have studied the dynamics of different control loops and also first steps have already been taken to optimize the control algorithms. As far as the experimental verification of different strategies is concerned, the published work is, however, restricted to laboratory scale only. From current practice in baker's yeast industry which is most often based on open loop control with fixed schedule for carbon substrate addition, it is well known that mixing of substrate is playing an important role. The effects of the dynamics of mixing process in mechanicaly stirred tanks upon the performance of baker's yeast fermentation with fixed schedule have been investigated earlier by using a circulation model (Bajpai and Reuss, 1982). In the present contribution this framework is used in conjunction with a closed computer control of carbon substrate feeding based on exhaust air analysis. From experimental results for computer controlled fermentation in 10 and 20001 scale a decrease of productivity with increasing circulation times of liquid is observed. These results can be confirmed with the model calculations.