It is astounding that no matter how many times the importance of a time-weather-fault chart is indicated, and how such a chart often shows up a reason a fault occurs, people would rather waste an immense amount of time battling through the problems.
The idea of such a graph came from seeing a weather station 'predict' the local conditions for the following 24 hours. The data from a weather sonde (the repair of these brought loads of pocket money!) would be transferred to a special graph. The pattern of the lines would indicate a "Go", No-Go", or "Stand-By" day. Stand-by and Go days resulted in a radio broadcast to local farmers who would arm their silver-iodide rockets to "seed" possible hail bearing clouds that could potentially damage crops.
If such a chart can work to determine the conditions for a day's "fault condition", surely it would work for electrical faults too. It does! Don't wait for a fault to be elusive, the moment things appear out of the ordinary then start the chart.
The chart will consist of three lines being temperature, general weather, and consumption all plotted against time. The general weather line will go from 0 for absolutely putrid (but not necessarily cold) through to 10 for a beautifully clear (but not necessarily hot) day. The last plot made on the graph is when the fault occurs, and if there is a 'severity' attribute to the fault (e.g. happens more frequently) then this is plotted.
Should a fault occur at a set time, this will be indicated by evenly spaced markers on the graph. Should it be happening under a certain set of conditions, this too will be clearly seen.
Armed with this information one can plan when to next go to site to 'trap' the fault occurrence, together with the appropriate equipment for trapping the type of fault.
For one chap who did heed this advice, it proved to be a huge time saver. The chart clearly indicated the fault to occur at or around 23°C. It was then traced to a control circuit on an air-conditioning unit where a 2° window existed for switching between heating and cooling. If neither heating or cooling was needed, then, under the specific load condition, the fault occurred.
Another case had the investigator notice that the fault always occurred approximately two days after rain. With electricity, one would expect any rain induced faults to occur pretty soon after it rains, but the chart clearly showed the delay. In this case it was a capillary action of water in a wall holding a faulty electrical joint.
The samples are endless. Learn from others!