Reducing the errors when measuring railway noise and vibration

When measuring the noise and vibration from trains it is necessary first to establish how much data needs to be collected in order to achieve the required level of accuracy. Existing measurement standards tend to concentrate on controlling the train speed, the condition of the track and the condition of the train, but do not indicate how much data is required. This paper sets out to establish how much data is necessary for a particular type of measurement situation. Using the measured distributions of noise and vibration data, the Monte Carlo method is used to determine the amount the sample mean may vary from the population mean as a function of the number of samples. This information can then be used to determine the number of trains that would need to be measured to achieve a particular level of accuracy. It has been found that if noise measurements are taken for trains comprising identical vehicles then, if the sample mean is to be within approximately I dB of the true value, then measurements would need to be made on at least 34 vehicles. When the trains contain different types of vehicle, the number of measurements increases. With ground-borne vibration it was found that at least 40 trains, each comprising nine coaches and a locomotive, were required to be able to measure the vibration dose value with a degree of accuracy that is equivalent to +/-1 dB.