Estimating temperatures in refrigerated containers and trailers : how many randomly placed cargo temperature sensors are necessary?

How many randomly placed cargo temperature sensors are necessary to gain a sufficiently accurateestimate of the temperatures of the complete cargo during refrigerated transport in reefer containersand (single compartment) refrigerated trailers?To answer the above question it is necessary to first determine what exactly the actual temperatureprofile in a standard container is. To that end a physical model of the temperature dynamics in reefercontainers and refrigerated trailers was constructed. The model has not been validated explicitly,but the simulation outcomes compare reasonably well with what the authors, experts on temperaturemanagement during refrigerated transport, would expect. This physical model is then simulated fora reefer container shipment scenario representative of a banana transport. The scenario includesa pulldown phase because it assumes stuffing non-precooled bananas. Then a statistical modelof the temperature distribution in a reefer container is defined. Inputs of the statistical model areany number of cargo temperature measurements collected at unknown locations in the cargo space.The statistical model simply calculates the cumulative distribution function of the measurements,without making any assumption about the type of distribution function. The difference betweenthe statistical model output and the physical model output is analysed for a range of sensor numbersand different sensor-placement methods.The results indicate that there is little benefit in placing more than 30 randomly placed sensorsper container load for the purpose of gaining a sufficiently accurate estimate of the temperaturesof the complete cargo during refrigerated transport in reefer containers, when using these recordingsas inputs for a statistical model.It is not to be expected that the situation will be a lot different for refrigerated trailers, because thereis no radical difference between refrigerated trailers and containers in terms of equipment size orin terms of the typically occurring pattern of temperature distribution.Care should be taken, as the conclusions depend on many assumptions.Random sensor placement is an ineffective sensor placement method, as two smartly placed sensorscan be more informative than e.g. 40 randomly placed sensors. Smart sensor placement in a containeris: one in the bottom of the coldest pallet (position 2 or 3 from unit-end) and one in the centre ofthe warmest pallet (door-end).The main research question in this study is very generic, and therefore hard to answer adequately.For future studies it is recommended to be more specific in formulating the purpose and applicationdomain of temperature recording during transport.