deletions | additions       diff --git a/sectionPROPOSED_PRIM.tex b/sectionPROPOSED_PRIM.tex  index 95f9a18..e2b2ecd 100644  --- a/sectionPROPOSED_PRIM.tex  +++ b/sectionPROPOSED_PRIM.tex 
...
\subsection{Primary Failure Modes in Practice}  {\em ALBA} has no ``No-beam'' events are  defined ``no-beam'' mode, rather to start when the  beam trips current drops below a given limit. For  several facilities this is currently not case. ALBA, BESSY-II  and the time in which all   shutters are closed during user time are recorded as ``downtime''; LSLS-UVX consider  a typical example includes malfunction closure  of the injector which requires closing photon shutters by an interlock equivalent to a loss  of the shutters.  When running in decay mode at 120$\,$mA, electron beam.  For other facilities  a ``low-beam-current'' ``no-beam''  event is defined stops not  whencurrent in  the machine beam  is lower than 72$\,$mA, then back to   the nominal current $I_{\hbox{nom}}$ but when the interlock flags that prevent  the beamline shutters from opening are cleared.   Others add an amount of time to allow for the warm-up of the optical components at the beamlines.   These practices are oriented to use  a re-injection beam availability understood as photon beam availability.   Facilities are counting the time the photon beam  is required. readily to be used at the beamlines.  In top-up mode, with 100$\,$mA in this sense  the machine closure of the photon shutters for whatever reason has an  impact on the beam availability.   We propose to distinguish between a photon beam availability at the beamlines  and injections every 20 minutes, an  electron beam available at the storage ring.   A ``no-beam'' event would only be defined by  a low loss of the electron  beam current starts when and a  ``photon-shutter-closed'' event would be recorded only if  the current photon shutters are forced   closed while there  is below 95$\,$mA.  This not already a ``no-beam'' event.   The application of this new metrics will disentangle events of different nature.  The authors decided to allow arbitrary limits for ``no-beam'' events, to ease the application for different kind of facilities.  It does not make a large difference in practice: situations are rare at the evaluated facilities where the  current requires drops from  the closure nominal beam current to less than 50\% but not to zero.   ``low-beam-current'' events will vary significantly with the definitions  of the shutters $I_{\hbox{tol}}$ current limit.   At Spring-8,  for re-injection up to 100$\,$mA. example, a beam decay of about 0.1\% starts a ``low-beam-current'' event,   while at BESSY-II it starts only at 9\% beam decay.   The reason of this difference comes from the required current stability for the experiments  that can differ significantly between light sources.  The number and duration of "Low-beam-current" events do tell how well a facility   meets their promised beam current stability.  {\em BESSY II} does not distinguish between no beam and closed beamshutters:   in either case users do not receive photons on sample.   Thus, a ``no-beam'' event (``dark'' time) starts either when the beam current is below  $I_{\hbox{min}}$, or when the topup interlock enforces a closure of the beam shutters.  Whatever happens first will trigger the start of the ``no-beam'' event; it will stop  when the nominal beam current $I_{\hbox{nom}}$ is restored.  Short interruptions of the injector system are neglected, as long as the  beam current is not decaying to less than $I_{\hbox{tol}}=91\%$ of $I_{\hbox{nom}}$.   Below that threshold a ``beamdrop'' events is recorded.  Numbers for the limits are (a) in multi-bunch mode:  $ I_{\hbox{nom}}= 299\,$mA, $ I_{\hbox{min}}= 200\,$mA and $ I_{\hbox{tol}}= 272\,$mA and (b) in single-bunch mode:  $ I_{\hbox{nom}}= 13.5\,$mA, $ I_{\hbox{min}}= 8\,$mA and $ I_{\hbox{tol}}= 12\,$mA.  If BESSY II operates in decaying beam mode (low-$\alpha$, degraded single-bunch) no low-beam-current is defined.  Independent from the re-fill schedule insufficient beam current is accounted as ``no-beam'' by $I_{\hbox{min}}$.   In low-$\alpha$ mode $I_{\hbox{min}}$=5$\,$mA, in degraded single bunch mode $I_{\hbox{min}}$= 3$\,$mA  {\em Elettra} uses the term ``fatal failure'' for ``no-beam'' events.   It starts when the beam current is zero and stops when the software control over the insertion-device gaps is given back to the users.  The nominal beam current is kept constant during top-up with a delta of 1$\,$mA.   To maintain a current of 310 $\pm 1\,$mA at 2$\,$GeV, top-up is required every 6 minutes.   At 2.4$\,$GeV, the beam current of 160$\pm 1\,$mA requires top-up every 13 minutes.   Thresholds of 307.5$\,$mA at 2$\,$GeV and 158$\,$mA at 2.4$\,$GeV are used to initiate a ``low-beam-current'' event.   After the ``low-beam-current'' issue has been solved, the nominal beam current mode   is recovered above 240$\,$mA at 2$\,$GeV or 110$\,$mA at 2.4$\,$GeV.   For lower currents, top-up can be activated to maintain constant beam current.  To reach the nominal beam current a new injection must be agreed with the users.   {\em LNLS-UVX} combines ``beam trip'' and ``delayed beam delivery'' into ``no-beam'' events.  Such events occur when the beam current falls below the ``no beam'' threshold (60$\,$mA for the UVX multibunch operation)   or when the beamline shutters are closed by a machine interlock flag during user time.   The scheduled injection period is 30 minutes; beyond that a ``no beam'' period starts and lasts until the interlock flags  that prevent the beamline shutters from opening are cleared.  Early delivery is not considered for reliability purposes but is considered for   yearly beam availability, which is calculated based on the recorded events in the datalog.   It may happen that by some reason the current is partially lost and although it is still above the ``no-beam'' threshold it is low compared to the expected for that time.  When this kind of event occurs the users are consulted about re-injection. If a new injection takes place the problem is then accounted for as a ``no-beam'' event. Otherwise just a "Fault" event is recorded.   {\em PETRA III} starts ``no-beam'' events (or ``downtime'') when the beam current   is below 75$\,$mA (75\% of the nominal beam current of 100$\,$mA).   It stops 1 hour or the length of the downtime (whichever is shorter)   after the nominal beam current of 100$\,$mA is reached again.  This time is added to allow for the warm-up of the optical components in the beamlines.   The nominal beam current is kept constant by top-up:   The beam is accumulated to 101$\,$mA.   As soon as it decays to 100$\,$mA, the injection is started automatically.   Injection stops as soon as 101$\,$mA are reached again.   Short interruptions of the injector system in the order of a few minutes are neglected,   as long as the beam current is not decaying to less than 75$\,$mA.   Below that threshold this is called a ``no-beam'' event.   ``Low-beam-current'' is not a criteria in the statistics of PETRA III.  {\em SPring-8} starts ``no-beam'' events (labeled ``downtime'') when the beam is aborted   and stops ``downtime'' when the main beam shutters (MBS) for photon beam lines are unlocked.   This is because the users cannot open the MBS during the beam injection for full current storage due to radiation safety.   The nominal beam current of the SPring-8 is also kept constant within 0.03\% by top-up.   For top-up the injection current is kept at 0.03$\,$mA, and the shot number per one injection is one.   ``Low-beam-current'' events are recorded when the beam current decays below 0.1$\,$mA of the nominal 99.5$\,$mA.   To keep one shot per one injection, the interval of the injection is not fixed but variable.   SPring-8 defines $I_{\hbox{min}} = 0\,$mA, since low current is not experienced during user operation except for a beam abort.   {\em SLS} starts ``no-beam'' events (``downtime'') when the beam current is below 20$\,$mA and   stops when the control over insertion device gaps is given back to the users.  The yearly beam availability is not calculated based on ``no-beam'' events,  rather it is calculated from the beam current, as archived by the control system.   The reason is that these automated availability calculation were used long before   the automatic event logging system was in place.  The beam current is kept constant at 402$\,$mA during top-up,  as soon as it decays to 400$\,$mA the injection starts automatically.   Injection stops at 402$\,$mA.   Short interruptions of the injector system on the order of minutes are neglected,   as long as the beam current is not decaying to less than 399$\,$mA.   Below that threshold we record a ``low-beam-current'' event, called ``beamdrop''.   The total duration of injector outages for the yearly statistics is calculated from ``beamdrop'' events.  \subsection{Primary failure mode limit comparison}  Table~\ref{tab:pf-limits} shows the current limits for the primary  failure modes for the seven facilities.  The column $I_{\hbox{nom}}$ shows the maximum current in the given mode. 
...
meaningful comparison of the reliability of the injection process.