Unscheduled Maintenance Events
- August 18, 2003 at 6:11 am #33081
Dear Sir,I monitor aircraft unscheduled maintenance events for an airline.The airplane systems are subdivided by labels identifying major systems. For eg. 29 for Hydraulics, 32 for Landing Gear, 22 for AutoFlight etc.The pilot reports malfunctions and maintenance crews fix them.I use what is considered industry practice to monitor these events. It goes like this.Say for Hydraulics there are:Jan 2002 = 3 events Feb 2002 = 0 events Mar 2002 = 5 events etc.These events are divided by flight hours for each month to get rates, R(Jan), R(Feb), R(Mar)….Over a period Jan to Dec, the rates are averaged, say ie. R(x)/12 = X.The sigma is calculated, say S.A UCL of X+2S is calculated for 2003.Assignable causes are investigated when say, R(Feb 2003) > UCL. UCL. UCL.Is this a valid form of SQC?Is there a superior SQC method? If YES, which one?This question stems from SQC literature that detail X bar, R bar, p, np, c, u charts that do not seem to detail this form of SQC.Your learned opinion and discussion will be greatly appreciated.Thank you.0August 18, 2003 at 12:24 pm #88950
DaveGParticipant@DaveG Include @DaveG in your post and this person will
be notified via email.
Although I am not a reliability expert, I see the following opportunities for you:
Setting an upper control limit is a reasonable thing to do, but yours has no relationship to any criticality standard and could leave you at risk. In addition, why would you limit your root cause investigations? I would want to know the source of all unscheduled maintenance, for safety reasons if nothing else.
The question is not whether you have a valid SQC method, but rather the value of what you are measuring.
First, you must establish what your objective is: Safety? Minimal downtime? Minimal cost? Then translate these into operational standards.
Second, unscheduled maintenance can be considered special causes (about which you have imperfect knowledge). Do you have an adequate definition for a malfunction, and is it rigorously adhered to? Is it enough that the pilot reports malfunctions? Are there other sources of data, such as observations by the maintenance workers? Are there systems to detect problems before they occur? Are there indicators of potential problems, such as fluid pressure, wear on rotating components, electrical load, etc.?0August 18, 2003 at 1:42 pm #88954
Looks like you’re using the ATA Chaps to follow up with malfunctions.
As aircraft systems are complex sets of subsystems it’s better for you to follow by subsystems , e.g for Nav ( ATA 34 ) ATA 34-11 , ATA 34-15….etc , just as examples ( ADF, VOR, ILS….etc )
Today’s central computers allow for a quick tracing of malfunctions and those are taking you straight to the subsystem or the sub-subsystem…..
After gathering all the data and finding the frequency of occurence for each then you can begin the stat journey with a better picture. This is my personal opinion.
0August 18, 2003 at 3:50 pm #88961
Please watch for “conveince” repairs in your “unscheduled Repair” database.
After many yyears in the aircraft business I’ve found that many airlines and repair stations perform scheduled repairs off schedule because it is convenient to do so, not because it is required.
For example if a brush life on an aircraft generator is scheduled for repalcement every 2000 hours and an aircraft is in for some other service at say 1900 hours, mechanics perform the brush replacement at the same time and log it as unscheduled repair.
This is very prevalent in the industry.0August 18, 2003 at 8:21 pm #88973
Thomas C. TribleMember@Thomas-C.-Trible Include @Thomas-C.-Trible in your post and this person will
be notified via email.
Selva:You ask:”Is there a superior SQC method? If YES, which one?””This question stems from SQC literature that detail X bar, R bar, p, np, c, u charts that do not seem to detail this form of SQC.”All the respondents provided good suggestions, but your question above has been addressed, in part, by Don Wheeler in his book Understanding SPC. Dr. Wheeler suggests plotting time between events, when the event is rare, or infrequent, e.g. accidents, fires, contamination, etc. You didn’t mention frequency, but plotting time instead of events is another way at looking at the data – for analysis, as an aid to understanding. I can’t think of any reason why you could not plot a control chart of the number of events themselves, or time between the events – if the events are infrequent. The control chart will tell you about the stability of the system of unscheduled maintenance. I recommend Individuals and moving Range charts, or XmR charts – since no logical subgroup may exist. I agree with the respondent who suggested that you exercise caution about “what is an unscheduled maintenance event?” I suggest that you creat an operational definition of “unscheduled maintenance” so that your data collection step is valid – that you do not mix apples with oranges.Good Luck.TC Trible0
The forum ‘General’ is closed to new topics and replies.