Validation of Australian food quality traceability technology (Smart-r-tag): Quality development of two perishable fruits; Strawberry and Avocado

Food quality is influenced by abiotic conditions such as: temperature, relative humidity, gasses (oxygen, carbon dioxide, etc.). These were monitored in experiments with strawberry and avocado by Smart-r-tag sensors manufactured by SensaData and provided to WUR. The data from the sensors were used as input for prediction of fruit quality and shelf life with a quality loss model. The objective of this research is to test if SensaDatas sensor tags are able to capture abiotic conditions as input for quality prediction for strawberry and avocado. The study did not include developing new quality models based on the acquired data. Models described in literature and developed by WUR are used for quality prediction. The sensors used in the study are the Smart-r-tag Ver1, capturing temperature and relative humidity information and the Smart-r-tag Ver2, recording temperature, relative humidity, oxygen - and carbon dioxide concentration. In two experiments, one with strawberry and one with avocado, the quality of the produce was evaluated and the abiotic storage conditions were monitored using the Smart-r-tags. During storage the strawberries showed different levels of decay depending on the storage temperature, especially the storage condition at 20 °C affected the fruit severely. Avocados stored at different temperatures showed different levels of firmness loss. During the periods in which temperature was high (22 °C and 18 °C) the decrease in firmness was the highest. The Smart-r-tags are able to measure and log the abiotic conditions (temperature, oxygen and carbon dioxide) in which the produce was stored. However, for relative humidity there are also some nonrealistic readings, readings above 100%. Furthermore concerning the monitoring of oxygen and carbon dioxide contents (inside modified atmosphere packaging), the carbon dioxide measurements are inaccurate when the actual carbon dioxide contents are higher than 10% and/or when relative humidity in the packaging headspace is saturated. For quality modelling purpose, the parameter temperature was used as input variable. This data was as input useful for quality prediction. The quality prediction did not exactly match the observed quality, as the quality models were not optimised for these specific produces and abiotic conditions. The models can be adapted or other models could be used to fit the data better. The following recommendations can be given based on the work that was performed: 1. Validate relative humidity sensors when measuring at high humidity. In supply chains with perishable product like fruit and vegetables humidity is commonly above 90% RH and often higher than 95 % RH. It is important that the sensors operate well in this RH range for them to be useful in practice. Certainly a humidity cannot be higher than 100% RH. 2. Validate if the carbon dioxide sensor is measuring the correct concentration when measuring under high humidity. We found a discrepancy between our reference and the output of the Smart-r-tag ver2 sensor. 3. Select and use a quality prediction model that fits the need and support the decision making of the intended customer for the tags. There are many models described in literature, but they serve a certain purpose. Generic models are relatively easy to use, but might be too general for the case on hand. This has to be evaluated in a follow-up project, with practical pilots. In a possible follow-up WUR is willing to assist SensaData with selecting and setting up the best quality prediction models in combination with the needs of the customer.