226 Beam-based Correction and Optimization for Accelerators
Beam-based optimization of LMA was first demonstrated on the ESRF
storage ring [81]. In the experiment, the objective function was defined as
f (x) = −τ (x)
Iσ z0 σ y,0
I 0 σ z (I)σ y
,
(8.5)
where σ z (I) is the bunch length calculated with an impedance model, σ z0 =
σ z (I 0 ), σ y is the average vertical beam size for the measurements at 13 locations distributed around the ring. Care was also taken to wait for at least 30
minutes after each injection of a fresh beam injection to allow the beam spin
polarization distribution to settle in an equilibrium state as it could affect
the Touschek lifetime by up to 15%. With the beam density factor separated
out from the objective function and spin polarization effect stabilized, the
optimization targets only the LMA.
Two sets of sextupole knobs were used in the experiments. In one setup, 12
sextupole correctors were used for a fill pattern with many low-charge bunches.
The lifetime was increased from 47.5 hr to 75.5 hr by RCDS with about 5
iterations in 40 min. In the other setup, 10 sextupole knobs that consisted
of 5 main sextupole families and additional sextupoles with individual power
supplies were used on a fill pattern with 16 high-charge bunches. The lifetime
was increased from 11 hr to 17 hr. The optimized lattice had been used in
operation.
Beam-based optimization of LMA was first demonstrated on the ESRF
storage ring [81]. In the experiment, the objective function was defined as
f (x) = −τ (x)
Iσ z0 σ y,0
I 0 σ z (I)σ y
,
(8.5)
where σ z (I) is the bunch length calculated with an impedance model, σ z0 =
σ z (I 0 ), σ y is the average vertical beam size for the measurements at 13 locations distributed around the ring. Care was also taken to wait for at least 30
minutes after each injection of a fresh beam injection to allow the beam spin
polarization distribution to settle in an equilibrium state as it could affect
the Touschek lifetime by up to 15%. With the beam density factor separated
out from the objective function and spin polarization effect stabilized, the
optimization targets only the LMA.
Two sets of sextupole knobs were used in the experiments. In one setup, 12
sextupole correctors were used for a fill pattern with many low-charge bunches.
The lifetime was increased from 47.5 hr to 75.5 hr by RCDS with about 5
iterations in 40 min. In the other setup, 10 sextupole knobs that consisted
of 5 main sextupole families and additional sextupoles with individual power
supplies were used on a fill pattern with 16 high-charge bunches. The lifetime
was increased from 11 hr to 17 hr. The optimized lattice had been used in
operation.