A methodology to determine a conditional probability density distribution surface from S–N data