YAWL4ProductRecall

Background

Product recall and food recall procedures deal with how to withdraw defective products from the distribution networks (Trade level recall) and/or from the consumer (Consumer recall). Timely and effective response to potential recall scenarios are essential to ensure a successful product recall. In Australia, the Product Safety Policy Section of the ACCC is responsible for monitoring the conduct and the outcomes of voluntary consumer goods safety recalls in Australia. The food recall process is governed by Food Standards Australia New Zealand (FSANZ) and the protocol to follow during a recall is set out in the Food Industry Recall Protocol document. Organisations involved in the product and food manufacturing and distribution chains need to understand their responsibilities and to ensure that they have a recall plan that complies with the current protocols. We are interested in examining how a sophisticated business process automation solution such as the YAWL framework can support some of these requirements from product recall domain, namely compliance with protocols, timely reporting, and traceability.

Project description

We have developed a YAWL process model to depict the responsibilities of sponsors when recalling food products in accordance with the guidelines set out in the Food Industry Recall Protocol (5th Edition June 2004 pp.14). Figure 1 shows the activities expected to be carried out by a sponsor during food recall. Figure 2 shows a subprocess of this model which depicts filling out a Product Hazard / Risk Assessment Form that is used to decide whether a product recall should be carried out. These process models can assist in communicating the responsibilities to the sponsor in a clear and concise manner and to ensure that the sponsor's own recall plan complies with the prescribed model.

Figure 1: Control flow visualisation of Sponsor responsibilities process in YAWL

Figure 2: Control flow visualisation of Assess risk subprocess

As YAWL process models are executable, we have also captured information (data) requirements of these models in XML as well as the person responsible for each activity (resource). Figure 3 shows a dynamically generated Risk Assessment form displayed to the user to capture the required information. As the standard does not mention who in the sponsor's organisation should take on the various activities, the current model utilises the role-based resource allocation mechanism which indicates that an employee with the 'Recall sponsor' role can carry out the activities.

Figure 3: An example Risk Assessment form for Product recall

After carrying out a number of practice runs with this executable YAWL process model in the YAWL engine, execution logs of this model are then generated and these logs are analysed using the Processing Mining framework (ProM). This analysis enables the sponsor to provide evidence for compliance with the protocol as well as to gain insights into the timeframe required to complete the recall. Figure 4 shows the various event filters that can be applied to the logs to observe the behaviour of the interested activities in more detail. Figure 5 depicts a bar chart that shows the wait times and execution times of various activities within the product recall process.

Figure 4: Filtering log events in ProM

Figure 5: A bar chart which shows the time it takes to perform each activity by various staff members

Project Members

• Colin Fidge
• Arthur ter Hofstede
• Marlon Dumas
• Moe Thandar Wynn

This work is being carried out in conjunction with the research carried out as a part of the ARC Discovery Project on rapidly locating items in distribution networks.