In supply chain management, the term “distribution-free” procedures typically refers to special forecasting and, above all, safety stock procedures that enable reliable statements to be made even if the prerequisite of a normal distribution of demand for articles does not exist.
All classical methods for determining forecasts or safety stocks assume that the frequency distribution of the demand quantities for an article follows a Gaussian normal distribution. In practice, this is not the case for 80% to 90% of the articles. The more sporadically an item is demanded, the greater the probability that there is no normal distribution. In this case, distribution-free methods are an important lever for obtaining reliable forecast values and safety stocks.
You can pragmatically recognise whether the demand for an item deviates from the normal distribution by comparing the mean value of a demand time series with its median. If the median and the mean fall apart, there is definitely no normally distributed demand.
For sporadic items and for spare parts, you should definitely use distribution-free procedures, at least for safety stock calculations.