There could be multiple reasons for this.Firstly, it could be related to the timing of the shifts.
If your shifts have different durations, then their individual OEE values will not all contribute equally to the total OEE.
For instance, if in one day you have two shifts of 10 hours and one of 4, then an anomalous availability, performance, or quality metric in the 4-hour shift will have less of an effect than if it were to occur during the 10-hour shift.
Alternatively, your shifts might not cover all 24 hours of the day, in which case the overall OEE will be a result of a greater set of input data (namely all of the 24 hours for each day) than your combined shift results.
Another possibility is that your shifts are overlapping, in this case the periods of overlap are represented multiple times in the average shift result, whereas each single point in time is represented exactly once in the total OEE result.Secondly, even if your shifts cover the full 24 hours of the day, all have the same duration, and do not overlap, it is possible (perhaps even likely) that the average shift result does not equal the OEE result.
This happens because the OEE and its components (i.e. availability, performance, and quality) of one period (e.g. a year) are often not equal to the average of those metrics over any periods within (e.g. the months).
That is to say, you could calculate that the 'AVERAGE MONTHLY OEE' of the last year was 85%, but that is a different metric than the 'OVERALL OEE' of the last year.
To explain that in more detail, we have to look at the manner in which the OEE is calculated.
To keep things simple, we will focus only on the performance component, assuming availability and quality are both 100% consistently.
The performance is calculated from the measured actual output and the theoretical output, with performance = ( ( actual_output / theoretical_output ) * 100 ).Now let's say you have three 8-hour shifts within the day that do not overlap.
The theoretical output during the first two shifts equals 2000, but the last shift operates against a theoretical output of only 20.
The actual output during both first two shifts meets the theoretical output at 2000, but the last shift only manages to meet half of it, leaving the last shift with an actual output of only 10.Now when we add up the numbers, we'll find that the final shift has a drastically lower performance rating and drags down the average shift average substantially,
but because on the whole, because they weren't contributing much to production anyway (with the theoretical output being 100 times lower), the overall performance for the day isn't affected much at all:performance shift 1 = ( ( 2000 / 2000 ) * 100 ) = 100%
performance shift 2 = ( ( 2000 / 2000 ) * 100 ) = 100%
performance shift 3 = ( ( 10 / 20 ) * 100 ) = 50%average performance shifts = (100 + 100 + 50 ) / 3 = 83.3%overall performance = ( ( 2000 + 2000 + 10 ) / ( 2000 + 2000 + 20 ) ) * 100 = 99.8%