CAMS: A Central Asset Management System with a Data-Driven Flare


Community infrastructure assets represent a vast investment built up over many generations. In Australia alone, infrastructure assets are valued at approximately over $3 trillion. Maintaining these assets requires an understanding of the deterioration process, which is complex, involving many factors. Unfortunately, historically we haven't invested in capturing accurate condition data on these investments.

The typical practice of maintenance of infrastructure assets has been quite reactive. When something fails or someone complains, remediation has happened. Organizations managing infrastructure assets often will do yearly inspections and make maintenance decisions for that particular year, but after that, they would discard the data. The result is that authorities managing community infrastructure often only have a rough idea of what the future holds and budget over runs due to unplanned repairs are common.

At the Royal Melbourne Institute of Technology (RMIT), where I work as Deputy Dean for Research and Innovation in the School of Engineering, we sought to solve this problem by developing a Central Asset Management System (CAMS). This began as a three-year research project funded by the Australian Research Council in partnership with six local councils and the Municipal Association of Victoria. The CAMS team of civil and infrastructure engineers at RMIT developed degradation models for community building assets that usually have a life expectancy of 50 to 100 years.

With CAMS, we were able to predict the future of these building assets using a probabilistic approach with around 80% accuracy, which is impressive for infrastructure. And that accuracy is constantly improving: We can dynamically recalibrate the predictive models as new data sets come in, so it’s a live predictive model.

At the end of our initial three-year project, we received another grant to develop a cloud-hosted platform that would incorporate these deterioration models. Then we went into the commercial world when the City of Melbourne started using the system. From that process, we received increasing amounts of data, which improved the accuracy of our algorithms, but it also introduced us to the issue of how we collect and present data to the end-user partners of our work.

Moving Beyond Research, with the End User in Mind

As we worked with the City of Melbourne, we found our council partners had difficulty translating our modeling outcomes into implementable actions. I realized this as I saw them requesting an increasing number of scenarios that my team would then develop algorithms for, which was not the best use of our resources.

Predictive models are nothing without a platform to make sense of it all.

Additionally, we found the data sets collected by city consultants were not clean. The discrepancies from one inspection to another hindered the accuracy of our models. We wanted to develop a mobile app to give inspectors the asset list for each building so the data set would be consistent across inspections.

What we needed was a reporting dashboard for inspectors, as well as an analytics platform that would automatically generate reports to empower our stakeholders, meanwhile freeing my engineers to focus on the modeling. Qlik became that reporting layer for CAMS. 

Seeking a Flexible Solution with Drill-Down Data

One reason we went with Qlik is we had worked closely with Melbourne-based Cast Solutions and had built a relationship with them. If Cast hadn’t introduced us to Qlik, we might have followed a different path. But Cast Solutions showed us exactly what Qlik can do in filtering data for a diverse array of reports and outputs. In addition, Qlik offered an exceptional user experience, as well as smart and simple functionalities that enabled insightful analysis. That’s why Qlik ultimately won out in our critical business analysis comparing it against other platforms like Tableau and Power BI.

Say, for example, the council wants to see the condition of all the air conditioning systems specifically in the city-run childcare centers over a period of five years. They can drill down to that level with Qlik. Sometimes, council also has questions about the quality of inspections on city assets. Was an inspection truly thorough, or was it actually quite cursory? With Qlik, councilors can drill down into the inspection data to see how much time the inspector spent at the facility for each inspection.

Empowering Stakeholders to Make Informed Choices

We now see there is an educational layer we have to add to our predictive modeling in order to help stakeholders translate outcomes into implementation. Let’s say CAMS tells you your air conditioning system will require replacement in 10 years with probability around 80%. But what does 80% probability mean when it comes to implementation? 

It depends, based on the thresholds you set and the risk you are willing to take. Do you want to wait until the air conditioning system has 80% probability of failure, or do you want to take action when that probability teeters at around 50%? It likely depends on the resources available to you.

Based on the thresholds they set, a city council can then budget around a spike in expenditure for system maintenance in, say, four years. They can do that because they have inventoried all their current assets in inspections through CAMS, our modeling predicts the probability of failure, and we can present understandable reports through Qlik.

Structural assets like bridges present even more complex challenges for understanding the full implications of an asset’s condition, and this is where Qlik’s reporting flexibility is vital. We’ve just released CAMS for bridges, working with the road authority here in the state of Victoria.

It’s important to note that applying predictive analysis isn’t purely about reactively planning for maintenance expenses. Maintenance actions can include steps like this to defer maintenance for a few years. The important thing is to understand the costs.

When we talk about costs, these are not solely financial. Our analysis includes how the condition of an asset impacts the community. If the road authority has to close a bridge, what are the total social and environmental costs of detouring vehicles, freight, public transit, and emergency first responders? Making decisions about such important pieces of infrastructure requires slicing and dicing the data in ways that identify the patterns and highlight the important areas to attend to. That complete intelligent business analysis is done by Qlik.

Complex decision making requires a business intelligence tool that lets you slice and dice data to understand everything.

Most people, including the people on council we task with making decisions about these public assets, are not civil engineers. By adding easy-to-understand reporting onto our modeling suite, Qlik helps non-engineers make informed policy choices about complex engineering conditions based on our predictive analysis. 

Development Partnership: Getting CAMS to the Most People

By managing the recording, intelligence, and data, Qlik enables us to focus more on the scientific and technical aspects of CAMS’ predictive capabilities without worrying about how we present that data to the customer.

One of the things that attracted us to Qlik from the beginning was that they were willing to enter into a development partnership with us. Cast Solutions worked with us to develop our business strategy with Qlik as an integral collaborative partner along this journey. This OEM partnership has enabled us to adopt an alternate cost-efficient licensing model, meaning we can deploy CAMS to a larger number of partners.

I should also mention that CAMS has two products: CAMS Mobile is the inspection app and then the CAMS predictive modeling suite is our main scientific output. Some inspectors want only the inspection app—they don’t need the full predictive modeling—so for them Qlik becomes very handy because they simply want to present the data set in a variety of ways. Providing Qlik proved more effective than attaching the CAMS predictive modeling suite, which comes at a higher cost. It’s just another example where our OEM partnership has helped put CAMS in the hands of more people.

Investing in the Future, Improving Community Satisfaction

We are expanding fast, currently looking at opportunities in Japan, China, and the European market. We’ve now got Asian Development Bank funding to implement the system in seven local councils in Sri Lanka.

Keep in mind that RMIT has around 330 final-year undergraduate students in civil engineering, and these final-year students require work placements to complete their program. With CAMS, we are creating opportunities for these students to work at the local council level engaging in inspections.

Right now, about half my team receives its funding from commercial income, the other half from research income. I have the feeling that by the end of next year, the entire team will support itself from commercial income. At that point, we’ll be able to create an entity outside the university. There will still be a research component at the university, but it will be separate from the commercial entity.

But, ultimately, this all comes back to the councils we serve. When you reduce the downtime of assets because you have a system to predict and plan for maintenance, you optimize your maintenance expenditure. Councils tell us their inspection time reduced by half for hard assets and 90% for immobile assets after we introduced RFIDs.

Rock-solid data = better decision making and happier end users.

But the indicator I’m most proud of is that when the City of Kingston, just outside Melbourne, conducted surveys, they found a 30% increase in community satisfaction with their facilities after implementing CAMS to manage their property portfolio. That’s the best measure, I believe, of how we help our partners maintain their assets and keep buildings working as they should. Qlik helps our partners understand the hard data from CAMS, making for better decisions, and, ultimately, happier end users.