Availability studies for the undulator-based ILC source

Goal of availability studies: simulate integrated luminosity of linear collider designs if the components and their mean-time-between-failures (MTBF) and mean-time-to-recovery (MTTR) are given. The results should help to choose between different design options. However, the results depend significantly on the made assumptions, as it will be shown in the following examples.

Availability studies concerning the positron source with an auxiliary source:

Results under the assumption: `recovery time is proportional to time without beam'

Himel, Snowmass The impact of the positron source on the machine availability was first studied under the assumption that the recovery time after a shutdown is proportional to the time without beam.

The results showed, that with the inclusion of an auxiliary source, achieving between 1%--20% of intensity, practically all possible loss in integrated luminosity, caused by the undulator source, will be compensated. The used option ILC2 implies a conventional source (with 2 targets) and assumes that the mean time between failures (MTBF) can be improved for selected components by about a factor 2-20.

More details about further made assumptions can be found in this talk, given at Snowmass 2005 in the sessions of the 'Operations and Avalability' working group.

In the Positron Source White Paper 2005, it has been agreed that an undulator-based source with a keep-alive source of 10% nominal intensity suffices to meet all needed ILC requirements.

Tests of the simulation code with current collider data

`Warm-up' -- test of the frequency of failures:
First tests of the availability code via application on true collider data, for instance on HERA, have been done: as `warm-up' and under the assumption that the numbers for MTBFs were given, the simulated frequency of failures was compared with the actual number of failures. This comparison showed consistency; see the presentations `summary' and `more details', given at Snowmass 2005.
Schaetzel, Groemitz Test of the proportionality between recovery time and time without beam:
The made assumption that the recovery time after a down-time is proportional to the time without beam, has also been simulated with actual HERA data and could not be verified: no correlation could be observed between the time for luminosity-tuning and the time since the luminosity shutdown.

This assumption has also been tested for the PEP-II machine. But also in this case no definite correlation between recovery time and duration of the shutdown could be observed. More details about the made comparisons are given in this talk.

Results under the assumption: `recovery time is fixed tune-time (e.g. average value from e- arm)'

Schaetzel, DPG Motivated by the rather equally distributed luminosity tuning time at HERA, further simulations have been made under the assumption that the recovery time has a fixed tuning time: an average value for the tune-time from the electron beam.

The new results for the integrated luminosity differ from the results that were obtained under the proportionality assumption:

The difference in integrated luminosity between the undulator and the conventional source is practically equal to the time difference needed for further machine development (MD) in the undulator. Every ILC section needs about 1% (damping rings 2%) of total simulation time for MD improvement studies; more details.

Schaetzel, Ergebnisse, Groemitz
Analyzing the results in more detail shows that the machine uptimes are practically the same when using an undulator-based source or a conventional source.

These results also indicate that an auxiliary source might be redundant concerning the total uptime of the machine. The assumed time needed for MD causes the only difference between both types of sources. More details are given in this talk.

The studies also show how strong the impact of the basic simulation assumptions is on the results. Therefore further comparisons of availability simulations with true data from running machines are needed: for instance, an analysis of the experience with MD times at HERA.


Gudi Moortgat-Pick
Last modified: 24-August-2006