A-Labs: How our students helped Downer improve efficiency
A project to help Downer improve its operations gave Coco Wu a taste of life as a working data professional during her studies.
Downer Group is one of Australia and New Zealand's biggest integrated services providers, covering a wide range of operations including road management and maintenance.
The company is also proficient in using analytics in its daily operations, and its Road Services business partners with Melbourne Business School to host Full-time Master of Business Analytics students on Analytics Labs, or A-Labs, to gain practical experience by helping to solve business problems.
The challenge for the 2021 A-Lab team – consisting of myself, Imalka Rangala and Venkatesh Vijayaraghavan – was to work with Downer's Road Services team to optimise the way pavement materials are produced and delivered to job sites each day.
I benefited greatly from the A-Lab experience, as it helped me get a taste of working as a data professional. As well as having the opportunity to apply classroom knowledge in solving real-world problems, I also got experience in stakeholder management and communicating my ideas clearly to people without a strong technical background.
Here's what was involved in the project.
A key operation of Downer's Road Services team is to schedule trucks throughout the day to pick up the required asphalt mix manufactured at their asphalt plants and deliver them to specific job sites. The company operates with human allocators who develop schedules through extensive discussions with multiple parties involved in the process, and the outcomes usually rely heavily on their experience.
The 2021 A-Lab team worked closely with the pavements production team and developed a set of decision support tools aimed at helping plant allocators determine an optimal job schedule on any given day that minimises overall cost, while still meeting operational and customer needs.
Raw data such as a job list of customer demands, material, production plant and truck specifications, were transformed and inputted into the model. The optimisation was done in two stages: job allocation to asphalt plants, and production and truck scheduling.
Stage 1 determined the materials that met the job requirement and allocated jobs in a way that minimised the production and transportation cost based on the distances between plants and job sites.
Stage 2 dived into each asphalt plant and optimised truck scheduling to meet demand rates, while minimising the transportation costs by optimising the number of vehicles required. With the optimal truck schedule, the optimal production schedule can then be backtracked based on the production rate for different materials.
Ten different constraints related to the business operation were identified and taken into consideration, such as the plant production limitations, transport capacities, and specific requirements from customers.
Multiple mathematical formulae were created to represent the decisions at different stages and to accurately capture the objective functions and the constraints. The (mixed integer) linear programming models were then coded and solved using the Julia programming language, which outputted the result of the optimal job arrangement and schedules, ready to be used for decision-making.
Realising the potential impact this project could have for Downer was the most rewarding part of the A-Lab. The optimisation models we built can reduce production and transport costs for operations across the State.
This is done by optimally allocating jobs to the most relevant plants and creating schedules that optimise the number of trucks used for delivery each day, while still meeting the required demand.
The models are also a helpful tool for the human allocator at the source plant. With the optimal schedule automatically generated once the job list is inputted, they can focus on the implementation and execution of the schedules and deal with any potential changes that happen during the day.
Having the two-stage model can also provide Downer with both the macro and micro outlook on the business operation for strategic decision making. For example, determining potential future source plant locations by evaluating the outsourcing pattern.
A more efficient production and delivery schedule also means less wastage when producing asphalt, less truck usage throughout the day, and less waiting time for truck drivers. Not only could this increase profit, it also helps Downer become more carbon-friendly.
What the experience was like
My A-Lab experience was very positive as Downer has a supportive and friendly culture. I learned and grew enormously during the six weeks with the help and guidance from the Road Services team, especially two of the Full-time Master of Business Analytics alumni who work there.
During the A-Lab, we could decide on the directions to take and the method to use throughout the project, and we were also given the contact details of all stakeholders involved and were encouraged to contact them directly for clarifications.
We constantly interacted with our stakeholders to clarify issues, discuss findings, and keep them informed of our progress. Through this, I learned to always present our methods and results in a non-technical manner and incorporate storytelling to make the technical information more compelling.
I felt supported by the regular check-ins arranged by our supervisors to exchange ideas and overcome roadblocks. As a result, we were able to experiment with different approaches and find a solution to problems regardless of their difficulty. I really appreciated the autonomy and supportive culture at Downer, as I felt respected and trusted.
We also had an immersive experience visiting one of Downer's asphalt plants in Somerton. We were able to see the process of asphalt production, from inputting the raw bitumen and aggregates into the manufacturing plant, to loading the trucks with asphalt ready for delivery.
We had the opportunity to consult with a plant allocator and learn about their day-to-day roles and responsibilities. The interactive experience allowed us to better understand the business operations and helped us greatly when incorporating different real-world factors and constraints into the model.
We were also given the chance to meet and network with senior managers from different areas, who were all very approachable, willing to help, and enjoyed sharing their experiences and insights. Through this, we had a clearer view of the industry and learned about the different functions of the road services business.
The A-Lab allowed me to see the value of the fundamental analytics skills I learned in the Full-time Master of Business Analytics program, and how to apply them, as well as helped me in developing strong communication and stakeholder management skills.
Hopefully the benefits from this project will help the organisation as well, and may further encourage Downer to accelerate its analytics capacity and discover deeper value through what analytics can unlock.
For more analytics information and research, visit our Centre for Business Analytics page.